From rob.clewley at gmail.com  Sun Jun  1 14:26:17 2008
From: rob.clewley at gmail.com (Rob Clewley)
Date: Sun, 1 Jun 2008 14:26:17 -0400
Subject: [SciPy-user] ANN: PyDSTool 0.86 released
Message-ID: <a749952d0806011126q537cf3d6laea9f36b9af55ffb@mail.gmail.com>

The latest update to the open-source python dynamical systems modeling
toolbox, PyDSTool 0.86, has been posted on Sourceforge.
Major highlights are:

 * Now compatible with Python 2.5 and Numpy 1.0.4 / Scipy 0.6.0
 * Decreased overhead for simulating hybrid models
 * Improved efficiency of VODE Generator in computing trajectories
 * Interval class now supports discrete valued intervals
 * Improved diagnostic reporting structure in Generator and Model classes
 * Inclusion of intuitive arithmetic operations for Point and Pointset classes
 * Various bug fixes and other API tidying

This is a minor update in preparation for a substantial upgrade at
version 0.90, which will move symbolic expression support over to
SymPy, support much more sophisticated data-driven model inference,
and greatly improve the implementation of C-based ODE integrators.

You can download the latest version from
http://www.sourceforge.net/projects/pydstool/
For installation and setting up, as well as some tutorial information,
see http://pydstool.sourceforge.net
The download contains full API documentation, BSD license information,
and further details of recent code changes.

As ever, all feedback is welcome as we try to find time to improve our
code base.


From phhs80 at gmail.com  Sun Jun  1 17:15:00 2008
From: phhs80 at gmail.com (Paul Smith)
Date: Sun, 1 Jun 2008 14:15:00 -0700 (PDT)
Subject: [SciPy-user] OpenOpt errors with ALGENCAN
Message-ID: <cb4c5879-48b8-4831-80d8-7d6b4d92f69c@m45g2000hsb.googlegroups.com>

Dear All,

I am trying to run the nlp_1.py example of OpenOpt on F9 with ALGENCAN
solver, but getting these errors:

========================
starting solver ALGENCAN (license: GPL)  with problem  unnamed
Traceback (most recent call last):
  File "nlp_1.py", line 109, in <module>
    r = p.solve('ALGENCAN')
  File "/usr/lib/python2.5/site-packages/scikits/openopt/Kernel/
BaseProblem.py", line 230, in solve
    return runProbSolver(self, solvers, *args, **kwargs)
  File "/usr/lib/python2.5/site-packages/scikits/openopt/Kernel/
runProbSolver.py", line 225, in runProbSolver
    solver(p)
  File "/usr/lib/python2.5/site-packages/scikits/openopt/solvers/
BrasilOpt/ALGENCAN_oo.py", line 586, in __solver__
 
algencan.solvers(evalf,evalg,evalh,evalc,evaljac,evalhc,evalhlp,inip,endp)
TypeError: solvers() takes exactly 12 arguments (9 given)

Any ideas?

Thanks in advance,

Paul



From phhs80 at gmail.com  Sun Jun  1 17:25:43 2008
From: phhs80 at gmail.com (Paul Smith)
Date: Sun, 1 Jun 2008 14:25:43 -0700 (PDT)
Subject: [SciPy-user] OpenOpt errors with ALGENCAN
In-Reply-To: <cb4c5879-48b8-4831-80d8-7d6b4d92f69c@m45g2000hsb.googlegroups.com>
References: <cb4c5879-48b8-4831-80d8-7d6b4d92f69c@m45g2000hsb.googlegroups.com>
Message-ID: <25912a3c-07e6-47b7-ad62-c02288fce0e2@59g2000hsb.googlegroups.com>

On Jun 1, 10:15?pm, Paul Smith <phh... at gmail.com> wrote:
> I am trying to run the nlp_1.py example of OpenOpt on F9 with ALGENCAN
> solver, but getting these errors:
>
> ========================
> starting solver ALGENCAN (license: GPL) ?with problem ?unnamed
> Traceback (most recent call last):
> ? File "nlp_1.py", line 109, in <module>
> ? ? r = p.solve('ALGENCAN')
> ? File "/usr/lib/python2.5/site-packages/scikits/openopt/Kernel/
> BaseProblem.py", line 230, in solve
> ? ? return runProbSolver(self, solvers, *args, **kwargs)
> ? File "/usr/lib/python2.5/site-packages/scikits/openopt/Kernel/
> runProbSolver.py", line 225, in runProbSolver
> ? ? solver(p)
> ? File "/usr/lib/python2.5/site-packages/scikits/openopt/solvers/
> BrasilOpt/ALGENCAN_oo.py", line 586, in __solver__
>
> algencan.solvers(evalf,evalg,evalh,evalc,evaljac,evalhc,evalhlp,inip,endp)
> TypeError: solvers() takes exactly 12 arguments (9 given)
>
> Any ideas?

Let me add that I have

$ echo $PYTHONPATH
/home/psmith/algencan-2.0-beta/bin/py
$

and therein:

$ dir
algencan.py  algencan.pyc  pywrapper.so
$

Paul


From dmitrey.kroshko at scipy.org  Mon Jun  2 02:51:48 2008
From: dmitrey.kroshko at scipy.org (dmitrey)
Date: Mon, 02 Jun 2008 09:51:48 +0300
Subject: [SciPy-user] OpenOpt errors with ALGENCAN
In-Reply-To: <25912a3c-07e6-47b7-ad62-c02288fce0e2@59g2000hsb.googlegroups.com>
References: <cb4c5879-48b8-4831-80d8-7d6b4d92f69c@m45g2000hsb.googlegroups.com>
	<25912a3c-07e6-47b7-ad62-c02288fce0e2@59g2000hsb.googlegroups.com>
Message-ID: <48439884.3000606@scipy.org>

OpenOpt has no connection to ALGENCAN 2.0 beta yet, as it is mentioned 
in http://scipy.org/scipy/scikits/wiki/OpenOptTODO

AFAIK 2.0 beta has no Python API. So currently you should use 1.0 
instead. If this one is already unavailable I could attach tarball with 
1.0 to an openopt webpage.
D.

Paul Smith wrote:
> On Jun 1, 10:15 pm, Paul Smith <phh... at gmail.com> wrote:
>   
>> I am trying to run the nlp_1.py example of OpenOpt on F9 with ALGENCAN
>> solver, but getting these errors:
>>
>> ========================
>> starting solver ALGENCAN (license: GPL)  with problem  unnamed
>> Traceback (most recent call last):
>>   File "nlp_1.py", line 109, in <module>
>>     r = p.solve('ALGENCAN')
>>   File "/usr/lib/python2.5/site-packages/scikits/openopt/Kernel/
>> BaseProblem.py", line 230, in solve
>>     return runProbSolver(self, solvers, *args, **kwargs)
>>   File "/usr/lib/python2.5/site-packages/scikits/openopt/Kernel/
>> runProbSolver.py", line 225, in runProbSolver
>>     solver(p)
>>   File "/usr/lib/python2.5/site-packages/scikits/openopt/solvers/
>> BrasilOpt/ALGENCAN_oo.py", line 586, in __solver__
>>
>> algencan.solvers(evalf,evalg,evalh,evalc,evaljac,evalhc,evalhlp,inip,endp)
>> TypeError: solvers() takes exactly 12 arguments (9 given)
>>
>> Any ideas?
>>     
>
> Let me add that I have
>
> $ echo $PYTHONPATH
> /home/psmith/algencan-2.0-beta/bin/py
> $
>
> and therein:
>
> $ dir
> algencan.py  algencan.pyc  pywrapper.so
> $
>
> Paul
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
>
>
>   



From phhs80 at gmail.com  Mon Jun  2 04:47:04 2008
From: phhs80 at gmail.com (Paul Smith)
Date: Mon, 2 Jun 2008 09:47:04 +0100
Subject: [SciPy-user] OpenOpt errors with ALGENCAN
In-Reply-To: <48439884.3000606@scipy.org>
References: <cb4c5879-48b8-4831-80d8-7d6b4d92f69c@m45g2000hsb.googlegroups.com>
	<25912a3c-07e6-47b7-ad62-c02288fce0e2@59g2000hsb.googlegroups.com>
	<48439884.3000606@scipy.org>
Message-ID: <6ade6f6c0806020147g26144a49i87811cbd5ecaa973@mail.gmail.com>

On Mon, Jun 2, 2008 at 7:51 AM, dmitrey <dmitrey.kroshko at scipy.org> wrote:
> OpenOpt has no connection to ALGENCAN 2.0 beta yet, as it is mentioned
> in http://scipy.org/scipy/scikits/wiki/OpenOptTODO
>
> AFAIK 2.0 beta has no Python API. So currently you should use 1.0
> instead. If this one is already unavailable I could attach tarball with
> 1.0 to an openopt webpage.

Thanks, Dmitrey. I think it is no longer exact what you claim: there
is now a Python interface for Algencan 2.0 beta (update: May 20th,
2008). Could you please confirm this?

Paul


From dmitrey.kroshko at scipy.org  Mon Jun  2 05:54:57 2008
From: dmitrey.kroshko at scipy.org (dmitrey)
Date: Mon, 02 Jun 2008 12:54:57 +0300
Subject: [SciPy-user] OpenOpt errors with ALGENCAN
In-Reply-To: <6ade6f6c0806020147g26144a49i87811cbd5ecaa973@mail.gmail.com>
References: <cb4c5879-48b8-4831-80d8-7d6b4d92f69c@m45g2000hsb.googlegroups.com>	<25912a3c-07e6-47b7-ad62-c02288fce0e2@59g2000hsb.googlegroups.com>	<48439884.3000606@scipy.org>
	<6ade6f6c0806020147g26144a49i87811cbd5ecaa973@mail.gmail.com>
Message-ID: <4843C371.9050008@scipy.org>

I have read README file from algencan2.0beta.tgz. Python binding refers to

$(ALGENCAN)/sources/interfaces/py/runalgencan.py and
$(ALGENCAN)/sources/interfaces/py/toyprob.py

but these files are absent (after all required steps before already done)
the only one py-file is algencan.py.

I had informed algencan developers of the issue.
W/O up-to-date toyprob.py I cannot connect algencan2.0 beta, because I 
don't know those new 12 arguments instead of old-style 9.
Regards, D.


Paul Smith wrote:
> On Mon, Jun 2, 2008 at 7:51 AM, dmitrey <dmitrey.kroshko at scipy.org> wrote:
>   
>> OpenOpt has no connection to ALGENCAN 2.0 beta yet, as it is mentioned
>> in http://scipy.org/scipy/scikits/wiki/OpenOptTODO
>>
>> AFAIK 2.0 beta has no Python API. So currently you should use 1.0
>> instead. If this one is already unavailable I could attach tarball with
>> 1.0 to an openopt webpage.
>>     
>
> Thanks, Dmitrey. I think it is no longer exact what you claim: there
> is now a Python interface for Algencan 2.0 beta (update: May 20th,
> 2008). Could you please confirm this?
>
> Paul
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
>
>
>   



From phhs80 at gmail.com  Mon Jun  2 06:06:01 2008
From: phhs80 at gmail.com (Paul Smith)
Date: Mon, 2 Jun 2008 11:06:01 +0100
Subject: [SciPy-user] OpenOpt errors with ALGENCAN
In-Reply-To: <4843C371.9050008@scipy.org>
References: <cb4c5879-48b8-4831-80d8-7d6b4d92f69c@m45g2000hsb.googlegroups.com>
	<25912a3c-07e6-47b7-ad62-c02288fce0e2@59g2000hsb.googlegroups.com>
	<48439884.3000606@scipy.org>
	<6ade6f6c0806020147g26144a49i87811cbd5ecaa973@mail.gmail.com>
	<4843C371.9050008@scipy.org>
Message-ID: <6ade6f6c0806020306r7d718a6cn2773efbf4ab57a89@mail.gmail.com>

On Mon, Jun 2, 2008 at 10:54 AM, dmitrey <dmitrey.kroshko at scipy.org> wrote:
> I have read README file from algencan2.0beta.tgz. Python binding refers to
>
> $(ALGENCAN)/sources/interfaces/py/runalgencan.py and
> $(ALGENCAN)/sources/interfaces/py/toyprob.py
>
> but these files are absent (after all required steps before already done)
> the only one py-file is algencan.py.
>
> I had informed algencan developers of the issue.
> W/O up-to-date toyprob.py I cannot connect algencan2.0 beta, because I
> don't know those new 12 arguments instead of old-style 9.

Thanks again, Dmitrey. I had also noticed yesterday the problem that
you describe: none of the two files (runalgencan.py and toyprob.py)
are created. I sent an e-mail to one of the developers, telling him
about the issue, but I have not received any reply so far.

Paul


From phhs80 at gmail.com  Mon Jun  2 10:33:29 2008
From: phhs80 at gmail.com (Paul Smith)
Date: Mon, 2 Jun 2008 15:33:29 +0100
Subject: [SciPy-user] OpenOpt errors with ALGENCAN
In-Reply-To: <6ade6f6c0806020306r7d718a6cn2773efbf4ab57a89@mail.gmail.com>
References: <cb4c5879-48b8-4831-80d8-7d6b4d92f69c@m45g2000hsb.googlegroups.com>
	<25912a3c-07e6-47b7-ad62-c02288fce0e2@59g2000hsb.googlegroups.com>
	<48439884.3000606@scipy.org>
	<6ade6f6c0806020147g26144a49i87811cbd5ecaa973@mail.gmail.com>
	<4843C371.9050008@scipy.org>
	<6ade6f6c0806020306r7d718a6cn2773efbf4ab57a89@mail.gmail.com>
Message-ID: <6ade6f6c0806020733w3d3b5244x5e7869b5919ba95c@mail.gmail.com>

On Mon, Jun 2, 2008 at 11:06 AM, Paul Smith <phhs80 at gmail.com> wrote:
>> I have read README file from algencan2.0beta.tgz. Python binding refers to
>>
>> $(ALGENCAN)/sources/interfaces/py/runalgencan.py and
>> $(ALGENCAN)/sources/interfaces/py/toyprob.py
>>
>> but these files are absent (after all required steps before already done)
>> the only one py-file is algencan.py.
>>
>> I had informed algencan developers of the issue.
>> W/O up-to-date toyprob.py I cannot connect algencan2.0 beta, because I
>> don't know those new 12 arguments instead of old-style 9.
>
> Thanks again, Dmitrey. I had also noticed yesterday the problem that
> you describe: none of the two files (runalgencan.py and toyprob.py)
> are created. I sent an e-mail to one of the developers, telling him
> about the issue, but I have not received any reply so far.

Ernest Birgin has just written to me telling me that there is now
available a new version of Algencan with the missing files. The
problem is solved now.

Paul


From silva at lma.cnrs-mrs.fr  Mon Jun  2 10:35:17 2008
From: silva at lma.cnrs-mrs.fr (Fabrice Silva)
Date: Mon, 02 Jun 2008 16:35:17 +0200
Subject: [SciPy-user] OdrPack and complex model
Message-ID: <1212417317.3242.12.camel@Portable-s2m.cnrs-mrs.fr>

Hi, 
I would like to use odr to fit a complex model on reference values as
causality issues occur when trying to adjust separately modulus and
argument.
Is it possible with odr ?

Fabricio



From timmichelsen at gmx-topmail.de  Mon Jun  2 13:35:54 2008
From: timmichelsen at gmx-topmail.de (Tim Michelsen)
Date: Mon, 2 Jun 2008 17:35:54 +0000 (UTC)
Subject: [SciPy-user] building and installing timeseries fails on Windows
Message-ID: <loom.20080602T170514-482@post.gmane.org>

Hello,
following the (updated) instructions at:
http://www.scipy.org/Cookbook/CompilingExtensionsOnWindowsWithMinGW

I tried to build a windows installer package and install the timeseries scikit
on a Windows XP computer with a recent MinGW.

It fails with the following message:
Microsoft Windows XP [Version 5.1.2600]
(C) Copyright 1985-2001 Microsoft Corp.

C:temp\SVNcheckouts\timeseries>python setup.py bdist_wininst
running bdist_wininst
running build
running scons
Traceback (most recent call last):
  File "setup.py", line 42, in <module>
    setup_package()
  File "setup.py", line 36, in setup_package
    configuration = configuration,
  File "C:\python\Lib\site-packages\numpy\distutils\core.py", line 184, in setup
    return old_setup(**new_attr)
  File "C:\python\lib\distutils\core.py", line 151, in setup
    dist.run_commands()
  File "C:\python\lib\distutils\dist.py", line 974, in run_commands
    self.run_command(cmd)
  File "C:\python\lib\distutils\dist.py", line 994, in run_command
    cmd_obj.run()
  File "C:\python\lib\site-packages\setuptools\command\bdist_wininst.py", line
37, in run
    _bdist_wininst.run(self)
  File "C:\python\lib\distutils\command\bdist_wininst.py", line 107, in run
    self.run_command('build')
  File "C:\python\lib\distutils\cmd.py", line 333, in run_command
    self.distribution.run_command(command)
  File "C:\python\lib\distutils\dist.py", line 994, in run_command
    cmd_obj.run()
  File "C:\python\Lib\site-packages\numpy\distutils\command\build.py", line 38,
in run
    self.run_command('scons')
  File "C:\python\lib\distutils\cmd.py", line 333, in run_command
    self.distribution.run_command(command)
  File "C:\python\lib\distutils\dist.py", line 993, in run_command
    cmd_obj.ensure_finalized()
  File "C:\python\lib\distutils\cmd.py", line 117, in ensure_finalized
    self.finalize_options()
  File "C:\python\Lib\site-packages\numpy\distutils\command\scons.py", line 258,
in finalize_options
    force=self.force)
  File "C:\python\Lib\site-packages\numpy\distutils\ccompiler.py", line 366, in
new_compiler
    compiler = klass(None, dry_run, force)
  File "C:\python\Lib\site-packages\numpy\distutils\mingw32ccompiler.py", line
46, in __init__
    verbose,dry_run, force)
  File "C:\python\lib\distutils\cygwinccompiler.py", line 84, in __init__
    get_versions()
  File "C:\python\lib\distutils\cygwinccompiler.py", line 424, in get_versions
    ld_version = StrictVersion(result.group(1))
  File ".\distutils\version.py", line 40, in __init__
  File ".\distutils\version.py", line 107, in parse
ValueError: invalid version number '2.18.50.20080109'

What can I do now?

Kind regards,
Timmie



From nwagner at iam.uni-stuttgart.de  Mon Jun  2 13:42:55 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Mon, 02 Jun 2008 19:42:55 +0200
Subject: [SciPy-user] OpenOpt errors with ALGENCAN
In-Reply-To: <6ade6f6c0806020733w3d3b5244x5e7869b5919ba95c@mail.gmail.com>
References: <cb4c5879-48b8-4831-80d8-7d6b4d92f69c@m45g2000hsb.googlegroups.com>
	<25912a3c-07e6-47b7-ad62-c02288fce0e2@59g2000hsb.googlegroups.com>
	<48439884.3000606@scipy.org>
	<6ade6f6c0806020147g26144a49i87811cbd5ecaa973@mail.gmail.com>
	<4843C371.9050008@scipy.org>
	<6ade6f6c0806020306r7d718a6cn2773efbf4ab57a89@mail.gmail.com>
	<6ade6f6c0806020733w3d3b5244x5e7869b5919ba95c@mail.gmail.com>
Message-ID: <web-102636032@uni-stuttgart.de>

On Mon, 2 Jun 2008 15:33:29 +0100
  "Paul Smith" <phhs80 at gmail.com> wrote:
> On Mon, Jun 2, 2008 at 11:06 AM, Paul Smith 
><phhs80 at gmail.com> wrote:
>>> I have read README file from algencan2.0beta.tgz. Python 
>>>binding refers to
>>>
>>> $(ALGENCAN)/sources/interfaces/py/runalgencan.py and
>>> $(ALGENCAN)/sources/interfaces/py/toyprob.py
>>>
>>> but these files are absent (after all required steps 
>>>before already done)
>>> the only one py-file is algencan.py.
>>>
>>> I had informed algencan developers of the issue.
>>> W/O up-to-date toyprob.py I cannot connect algencan2.0 
>>>beta, because I
>>> don't know those new 12 arguments instead of old-style 
>>>9.
>>
>> Thanks again, Dmitrey. I had also noticed yesterday the 
>>problem that
>> you describe: none of the two files (runalgencan.py and 
>>toyprob.py)
>> are created. I sent an e-mail to one of the developers, 
>>telling him
>> about the issue, but I have not received any reply so 
>>far.
> 
> Ernest Birgin has just written to me telling me that 
>there is now
> available a new version of Algencan with the missing 
>files. The
> problem is solved now.
> 
> Paul
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
  
Hi,

I have installed the new beta release.
I am using g77 and python2.4
  
./runalgencan.py
Traceback (most recent call last):
   File "./runalgencan.py", line 30, in ?
     import algencan
   File "/home/nwagner/algencan.py", line 25, in ?
     from pywrapper import solver
ImportError: /home/nwagner/pywrapper.so: cannot map 
zero-fill pages: Cannot allocate memory

Any idea ?

Nils


From phhs80 at gmail.com  Mon Jun  2 13:46:49 2008
From: phhs80 at gmail.com (Paul Smith)
Date: Mon, 2 Jun 2008 18:46:49 +0100
Subject: [SciPy-user] OpenOpt errors with ALGENCAN
In-Reply-To: <web-102636032@uni-stuttgart.de>
References: <cb4c5879-48b8-4831-80d8-7d6b4d92f69c@m45g2000hsb.googlegroups.com>
	<25912a3c-07e6-47b7-ad62-c02288fce0e2@59g2000hsb.googlegroups.com>
	<48439884.3000606@scipy.org>
	<6ade6f6c0806020147g26144a49i87811cbd5ecaa973@mail.gmail.com>
	<4843C371.9050008@scipy.org>
	<6ade6f6c0806020306r7d718a6cn2773efbf4ab57a89@mail.gmail.com>
	<6ade6f6c0806020733w3d3b5244x5e7869b5919ba95c@mail.gmail.com>
	<web-102636032@uni-stuttgart.de>
Message-ID: <6ade6f6c0806021046g7d0a7653i7802706930bc8f11@mail.gmail.com>

On Mon, Jun 2, 2008 at 6:42 PM, Nils Wagner
<nwagner at iam.uni-stuttgart.de> wrote:
> Hi,
>
> I have installed the new beta release.
> I am using g77 and python2.4
>
> ./runalgencan.py
> Traceback (most recent call last):
>   File "./runalgencan.py", line 30, in ?
>     import algencan
>   File "/home/nwagner/algencan.py", line 25, in ?
>     from pywrapper import solver
> ImportError: /home/nwagner/pywrapper.so: cannot map
> zero-fill pages: Cannot allocate memory
>
> Any idea ?

The interface Algencan expects Python 2.5 and not Python 2.4, I guess.
It works fine here with Python 2.5.

Paul


From contact at pythonxy.com  Mon Jun  2 14:26:05 2008
From: contact at pythonxy.com (Pierre Raybaut)
Date: Mon, 02 Jun 2008 20:26:05 +0200
Subject: [SciPy-user] [ Python(x,y) ] New release : 1.2.4
Message-ID: <48443B3D.1090206@pythonxy.com>

Hi all,

Python(x,y) 1.2.4 is now available on http://www.pythonxy.com.
Hopefully, updates should now be released at a lower rate.

Changes history
06 -02 -2008 - Version 1.2.4 :

     * Updated:
           o Matplotlib 0.98 - Important change for Python(x,y) users: 
better integration in Qt GUIs (see release notes)
           o IPython 0.8.4 (see release notes)
     * Corrected:
           o Windows installer bug fix: installation for all users did 
not work properly

Regards,
Pierre Raybaut






From gav451 at gmail.com  Mon Jun  2 14:52:29 2008
From: gav451 at gmail.com (Gerard Vermeulen)
Date: Mon, 2 Jun 2008 20:52:29 +0200
Subject: [SciPy-user] OdrPack and complex model
In-Reply-To: <1212417317.3242.12.camel@Portable-s2m.cnrs-mrs.fr>
References: <1212417317.3242.12.camel@Portable-s2m.cnrs-mrs.fr>
Message-ID: <20080602205229.0561d20e@jupiter.rozan.fr>

On Mon, 02 Jun 2008 16:35:17 +0200
Fabrice Silva <silva at lma.cnrs-mrs.fr> wrote:

> Hi, 
> I would like to use odr to fit a complex model on reference values as
> causality issues occur when trying to adjust separately modulus and
> argument.
> Is it possible with odr ?

I have adjusted real and imaginary values at the same time.
Pseudo-code looks like:

import numpy as np

def f(parameters, omega, data=0, sigma=1.0):
    # len(data) = 2*len(omega)
    # data[:len(omega)] contains the "real" experimental data
    # data[len(omega):] contains the "imaginary" experimental data
    zs = np.zeros(len(omega), np.complex_)
    # calculate zs here, using parameters and omega
    xsys = np.zeros(2*len(omega), np.float64)
    xsys[:len(omega)] = zs.real
    xsys[len(omega):] = ys.real
    return (xsys-data)/sigma

# f()

Regards -- Gerard

PS: if data can be complex, then f() can return (zs-data)/sigma.
I have forgotten whether I have tried that.

PPS: same idea works with leastsq()



From calogero.colletto at gmx.de  Tue Jun  3 14:49:43 2008
From: calogero.colletto at gmx.de (Calogero Colletto)
Date: Tue, 3 Jun 2008 18:49:43 +0000 (UTC)
Subject: [SciPy-user] Simpson Bug?
Message-ID: <loom.20080603T182125-280@post.gmane.org>

Hi,

recently I tested the simpson-rule from the scipy-package and I figured out an
inconsistency in the result. Following rule:

S[n] = h/3 * ( f(a) + f(b) + 4 * Sum( f(a+i*h), for all odd i, n) + 2 * Sum(
f(a+i*h), for all even i, n) )

integration from a to b
n: number of intervals. must be even
h = (b-a)/n ... width of an interval

My Script (simple implementation of the simpson rule):
______________________________________________
from math import pi, sin, cos
from numpy import arange
from scipy import integrate


def simps(y_seq, x_seq):
    h = x_seq[1] - x_seq[0]
    val = y_seq[0] + y_seq[-1]

    for i, y in enumerate(y_seq):
        if (i % 2) == 1:
            val += 4 * y
        elif (i % 2) == 0:
            val += 2 * y
        print "loop ", i, h*val/3.

    return h*val/3.


if __name__ == '__main__':
    s = 11   # Anzahl der St?tzstellen: s = n+1
                # Je mehr St?tzstellen, umso genauer das Ergebnis
    f = lambda x: sin(x)

    x = arange(0., pi/2., pi/(2.*s))
    y = tuple([f(i) for i in x])

    result = simps(y, x)

    print "My own implementation: ", result
    print "Scipy implementation: ", integrate.simps(y, x) 
_______________________________________________________

I tested the script with the sin-function and integrate this from 0 to pi/2. The
analytical result is 1.

Following results gives me the script:

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
loop 0 0.0471153904573
loop 1 0.0742120723007
loop 2 0.101032948993
loop 3 0.180127782511
loop 4 0.231596667976
loop 5 0.356281860151
loop 6 0.428229051385
loop 7 0.588403349479
loop 8 0.675000112936
loop 9 0.85768714791
loop 10 0.951917928825

My own implementation: 0.951917928825
Scipy implementation: 0.85768714791
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Maybe its my fault. But I think that the scipy-implementation is loosing the
last increment.

Calo




From berthe.loic at gmail.com  Tue Jun  3 15:05:34 2008
From: berthe.loic at gmail.com (LB)
Date: Tue, 3 Jun 2008 12:05:34 -0700 (PDT)
Subject: [SciPy-user] Scipy.test() fails when building scipy-0.6 against
	numpy 1.1
Message-ID: <61097727-a8d4-42d0-b772-81a341c5d7a5@d77g2000hsb.googlegroups.com>

     hi,

I'm trying to build scipy 0.6 with numpy 1.1 on a linux (32bits )and
I'm seeing errors when I run the test suite :

####################  ERROR MESSAGES
#####################################
.....................................................................................................
======================================================================
ERROR: check_string_add_speed (test_scxx_sequence.test_list)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/tests/test_scxx_sequence.py", line 405, in
check_string_add_speed
    inline_tools.inline(code,['a','b'])
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 339, in inline
    **kw)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 447, in compile_function
    verbose=verbose, **kw)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/ext_tools.py", line 365, in compile
    verbose = verbose, **kw)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/build_tools.py", line 269, in build_extension
    setup(name = module_name, ext_modules = [ext],verbose=verb)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/numpy/
distutils/core.py", line 184, in setup
    return old_setup(**new_attr)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/distutils/core.py",
line 168, in setup
    raise SystemExit, "error: " + str(msg)
CompileError: error: Command "g++ -pthread -fno-strict-aliasing -
DNDEBUG -g -O3 -Wall -fPIC -I/home/loic/tmp/bluelagoon/lib/python2.5/
site-packages/scipy/weave -I/home/loic/tmp/bluelagoon/lib/python2.5/
site-packages/scipy/weave/scxx -I/home/loic/tmp/bluelagoon/lib/
python2.5/site-packages/numpy/core/include -I/home/loic/tmp/bluelagoon/
include/python2.5 -c /home/loic/.python25_compiled/
sc_30bcdfe77c16ba01711b3f9b4f2d38df0.cpp -o /tmp/loic/
python25_intermediate/compiler_e0195f93cbce8aefbd6efe084f3c364b/home/
loic/.python25_compiled/sc_30bcdfe77c16ba01711b3f9b4f2d38df0.o" failed
with exit status 1

======================================================================
ERROR: check_false (test_scxx_object.test_object_is_true)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/tests/test_scxx_object.py", line 656, in check_false
    res = inline_tools.inline('return_val = a.mcall("not");',['a'])
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 342, in inline
    results = attempt_function_call(code,local_dict,global_dict)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 354, in attempt_function_call
    results = apply(func,(local_dict,global_dict))
AttributeError: foo instance has no attribute 'not'

======================================================================
ERROR: check_true (test_scxx_object.test_object_is_true)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/tests/test_scxx_object.py", line 660, in check_true
    res = inline_tools.inline('return_val = a.mcall("not");',['a'])
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 342, in inline
    results = attempt_function_call(code,local_dict,global_dict)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 354, in attempt_function_call
    results = apply(func,(local_dict,global_dict))
AttributeError: 'NoneType' object has no attribute 'not'

======================================================================
ERROR: check_set_from_member
(test_scxx_object.test_object_set_item_op_key)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/tests/test_scxx_object.py", line 820, in check_set_from_member
    inline_tools.inline('a["first"] = a["second"];',['a'])
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 339, in inline
    **kw)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 447, in compile_function
    verbose=verbose, **kw)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/ext_tools.py", line 365, in compile
    verbose = verbose, **kw)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/build_tools.py", line 269, in build_extension
    setup(name = module_name, ext_modules = [ext],verbose=verb)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/numpy/
distutils/core.py", line 184, in setup
    return old_setup(**new_attr)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/distutils/core.py",
line 168, in setup
    raise SystemExit, "error: " + str(msg)
CompileError: error: Command "g++ -pthread -fno-strict-aliasing -
DNDEBUG -g -O3 -Wall -fPIC -I/home/loic/tmp/bluelagoon/lib/python2.5/
site-packages/scipy/weave -I/home/loic/tmp/bluelagoon/lib/python2.5/
site-packages/scipy/weave/scxx -I/home/loic/tmp/bluelagoon/lib/
python2.5/site-packages/numpy/core/include -I/home/loic/tmp/bluelagoon/
include/python2.5 -c /home/loic/.python25_compiled/
sc_a35665b241c24d6c9d3fd21e957285660.cpp -o /tmp/loic/
python25_intermediate/compiler_e0195f93cbce8aefbd6efe084f3c364b/home/
loic/.python25_compiled/sc_a35665b241c24d6c9d3fd21e957285660.o" failed
with exit status 1

======================================================================
ERROR: check_string_add_speed
(scipy.weave.tests.test_scxx_sequence.test_list)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/tests/test_scxx_sequence.py", line 405, in
check_string_add_speed
    inline_tools.inline(code,['a','b'])
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 339, in inline
    **kw)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 447, in compile_function
    verbose=verbose, **kw)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/ext_tools.py", line 365, in compile
    verbose = verbose, **kw)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/build_tools.py", line 269, in build_extension
    setup(name = module_name, ext_modules = [ext],verbose=verb)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/numpy/
distutils/core.py", line 184, in setup
    return old_setup(**new_attr)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/distutils/core.py",
line 168, in setup
    raise SystemExit, "error: " + str(msg)
CompileError: error: Command "g++ -pthread -fno-strict-aliasing -
DNDEBUG -g -O3 -Wall -fPIC -I/home/loic/tmp/bluelagoon/lib/python2.5/
site-packages/scipy/weave -I/home/loic/tmp/bluelagoon/lib/python2.5/
site-packages/scipy/weave/scxx -I/home/loic/tmp/bluelagoon/lib/
python2.5/site-packages/numpy/core/include -I/home/loic/tmp/bluelagoon/
include/python2.5 -c /home/loic/.python25_compiled/
sc_30bcdfe77c16ba01711b3f9b4f2d38df1.cpp -o /tmp/loic/
python25_intermediate/compiler_e0195f93cbce8aefbd6efe084f3c364b/home/
loic/.python25_compiled/sc_30bcdfe77c16ba01711b3f9b4f2d38df1.o" failed
with exit status 1

======================================================================
ERROR: check_false
(scipy.weave.tests.test_scxx_object.test_object_is_true)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/tests/test_scxx_object.py", line 656, in check_false
    res = inline_tools.inline('return_val = a.mcall("not");',['a'])
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 325, in inline
    results = attempt_function_call(code,local_dict,global_dict)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 354, in attempt_function_call
    results = apply(func,(local_dict,global_dict))
AttributeError: foo instance has no attribute 'not'

======================================================================
ERROR: check_true
(scipy.weave.tests.test_scxx_object.test_object_is_true)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/tests/test_scxx_object.py", line 660, in check_true
    res = inline_tools.inline('return_val = a.mcall("not");',['a'])
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 325, in inline
    results = attempt_function_call(code,local_dict,global_dict)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 354, in attempt_function_call
    results = apply(func,(local_dict,global_dict))
AttributeError: 'NoneType' object has no attribute 'not'

======================================================================
ERROR: check_set_from_member
(scipy.weave.tests.test_scxx_object.test_object_set_item_op_key)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/tests/test_scxx_object.py", line 820, in check_set_from_member
    inline_tools.inline('a["first"] = a["second"];',['a'])
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 339, in inline
    **kw)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/inline_tools.py", line 447, in compile_function
    verbose=verbose, **kw)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/ext_tools.py", line 365, in compile
    verbose = verbose, **kw)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/build_tools.py", line 269, in build_extension
    setup(name = module_name, ext_modules = [ext],verbose=verb)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/numpy/
distutils/core.py", line 184, in setup
    return old_setup(**new_attr)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/distutils/core.py",
line 168, in setup
    raise SystemExit, "error: " + str(msg)
CompileError: error: Command "g++ -pthread -fno-strict-aliasing -
DNDEBUG -g -O3 -Wall -fPIC -I/home/loic/tmp/bluelagoon/lib/python2.5/
site-packages/scipy/weave -I/home/loic/tmp/bluelagoon/lib/python2.5/
site-packages/scipy/weave/scxx -I/home/loic/tmp/bluelagoon/lib/
python2.5/site-packages/numpy/core/include -I/home/loic/tmp/bluelagoon/
include/python2.5 -c /home/loic/.python25_compiled/
sc_a35665b241c24d6c9d3fd21e957285661.cpp -o /tmp/loic/
python25_intermediate/compiler_e0195f93cbce8aefbd6efe084f3c364b/home/
loic/.python25_compiled/sc_a35665b241c24d6c9d3fd21e957285661.o" failed
with exit status 1

======================================================================
FAIL: check_syevr
(scipy.lib.lapack.tests.test_lapack.test_flapack_float)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
lib/lapack/tests/esv_tests.py", line 41, in check_syevr
    assert_array_almost_equal(w,exact_w)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/numpy/
testing/utils.py", line 255, in assert_array_almost_equal
    header='Arrays are not almost equal')
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/numpy/
testing/utils.py", line 240, in assert_array_compare
    assert cond, msg
AssertionError:
Arrays are not almost equal

(mismatch 33.3333333333%)
 x: array([-0.66992444,  0.48769444,  9.18222713], dtype=float32)
 y: array([-0.66992434,  0.48769389,  9.18223045])

======================================================================
FAIL: check_syevr_irange
(scipy.lib.lapack.tests.test_lapack.test_flapack_float)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
lib/lapack/tests/esv_tests.py", line 66, in check_syevr_irange
    assert_array_almost_equal(w,exact_w[rslice])
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/numpy/
testing/utils.py", line 255, in assert_array_almost_equal
    header='Arrays are not almost equal')
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/numpy/
testing/utils.py", line 240, in assert_array_compare
    assert cond, msg
AssertionError:
Arrays are not almost equal

(mismatch 33.3333333333%)
 x: array([-0.66992444,  0.48769444,  9.18222713], dtype=float32)
 y: array([-0.66992434,  0.48769389,  9.18223045])

======================================================================
FAIL: check_noargs (test_scxx_object.test_object_call)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/tests/test_scxx_object.py", line 485, in check_noargs
    assert_equal(sys.getrefcount(res),2)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/numpy/
testing/utils.py", line 145, in assert_equal
    assert desired == actual, msg
AssertionError:
Items are not equal:
 ACTUAL: 3
 DESIRED: 2

======================================================================
FAIL: check_set_complex (test_scxx_object.test_object_set_item_op_key)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/tests/test_scxx_object.py", line 789, in check_set_complex
    assert_equal(sys.getrefcount(key),3)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/numpy/
testing/utils.py", line 145, in assert_equal
    assert desired == actual, msg
AssertionError:
Items are not equal:
 ACTUAL: 4
 DESIRED: 3

======================================================================
FAIL: check_noargs
(scipy.weave.tests.test_scxx_object.test_object_call)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/tests/test_scxx_object.py", line 485, in check_noargs
    assert_equal(sys.getrefcount(res),2)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/numpy/
testing/utils.py", line 145, in assert_equal
    assert desired == actual, msg
AssertionError:
Items are not equal:
 ACTUAL: 3
 DESIRED: 2

======================================================================
FAIL: check_set_complex
(scipy.weave.tests.test_scxx_object.test_object_set_item_op_key)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/scipy/
weave/tests/test_scxx_object.py", line 789, in check_set_complex
    assert_equal(sys.getrefcount(key),3)
  File "/home/loic/tmp/bluelagoon/lib/python2.5/site-packages/numpy/
testing/utils.py", line 145, in assert_equal
    assert desired == actual, msg
AssertionError:
Items are not equal:
 ACTUAL: 4
 DESIRED: 3

----------------------------------------------------------------------
Ran 2248 tests in 335.652s

FAILED (failures=6, errors=8)

###################################################################################

I didn't have theses errors with numpy 0.6.
I didn't have any problem when compiling, installing or testing numpy
1.1.


Besides, I see a lot of lines like
"""
/home/loic/.python25_compiled/sc_2b6af25d0035555e6a7579be09c9693c0.cpp:
34: attention : deprecated conversion from string constant to a char*
"""
With theses lines, my installation log is 2.6Mo big.


 Is this related with scipy's ticket  #490 ?
Have you got any clue to fix this ?

Thanks,

--
LB




From peridot.faceted at gmail.com  Tue Jun  3 18:54:51 2008
From: peridot.faceted at gmail.com (Anne Archibald)
Date: Tue, 3 Jun 2008 18:54:51 -0400
Subject: [SciPy-user] Simpson Bug?
In-Reply-To: <loom.20080603T182125-280@post.gmane.org>
References: <loom.20080603T182125-280@post.gmane.org>
Message-ID: <ce557a360806031554q5479625fjcd8971a3e5ef819e@mail.gmail.com>

2008/6/3 Calogero Colletto <calogero.colletto at gmx.de>:
> Hi,
>
> recently I tested the simpson-rule from the scipy-package and I figured out an
> inconsistency in the result. Following rule:
>
> S[n] = h/3 * ( f(a) + f(b) + 4 * Sum( f(a+i*h), for all odd i, n) + 2 * Sum(
> f(a+i*h), for all even i, n) )

This is not quite the right formula. The problem is (if there are an
odd number of points) that it counts the first and last points three
times, rather than once. If you start with -(f(a)+f(b)) it works out
all right. As a cross-check, try using some functions whose integral
is known to be exact when computed with Simpson's rule: constants,
cubic polynomials. You will see that, for example, your code returns
1.1333 rather than 1 as the integral of 1 from 0 to 1.

>    x = arange(0., pi/2., pi/(2.*s))

This use of arange is highly error-prone. Depending on round-off
error, you could get either s or s+1 points:

In [10]: len(arange(0,pi/2,pi/(2*61)))
Out[10]: 62

When you care how many points you get, use linspace:

x = linspace(0,pi/2,s)

It's clearer, and it's guaranteed to give you the right number of
points. arange is appropriate when you are working with integers, or
when what you care about is the size of the step.

Anne


From lopmart at gmail.com  Tue Jun  3 19:45:24 2008
From: lopmart at gmail.com (Jose Lopez)
Date: Tue, 3 Jun 2008 16:45:24 -0700
Subject: [SciPy-user] problem with optimize.leastsq
Message-ID: <4eeef9d40806031645t20a180ffi9c6b13c71d971866@mail.gmail.com>

hi

i have a problem with "optimize.leastsq", give it the next error:
MemoryError:
File: "c:\python25\lib....\minpack.py",line 268,in leastsq
retval =
_minpack._lmdif(func,x0,args,full_output,ftol,xtol,gtol,maxfev,epsfcn,factor,diag)

if i work whit the same code with some parameters work it!
but i work with more parameters the error above show me.
My os is windows

 someone happened the same error?

thanks
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080603/13cf9dd4/attachment.html>

From glen.shennan at gmail.com  Tue Jun  3 22:23:13 2008
From: glen.shennan at gmail.com (Glen Shennan)
Date: Wed, 4 Jun 2008 12:23:13 +1000
Subject: [SciPy-user] LabVIEW and LabPython
Message-ID: <ff223aae0806031923u1dd2c613k7ee8e5f27c1753e8@mail.gmail.com>

Hi,

I am trying to use a Python program in LabVIEW using LabPython.  My program
works fine outside of LabVIEW but not when compiled and run in LabVIEW via
LabPython.  I am using Scipy and Numpy functions in the program but I think
they are causing the problems.  Does anyone know if it is possible to use
Scipy with LabPython?

Thanks,
Glen
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080604/6efde1df/attachment.html>

From robert.kern at gmail.com  Tue Jun  3 22:31:43 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Tue, 3 Jun 2008 21:31:43 -0500
Subject: [SciPy-user] LabVIEW and LabPython
In-Reply-To: <ff223aae0806031923u1dd2c613k7ee8e5f27c1753e8@mail.gmail.com>
References: <ff223aae0806031923u1dd2c613k7ee8e5f27c1753e8@mail.gmail.com>
Message-ID: <3d375d730806031931p6603f4bbw237c2a699f6e21a6@mail.gmail.com>

On Tue, Jun 3, 2008 at 9:23 PM, Glen Shennan <glen.shennan at gmail.com> wrote:
> Hi,
>
> I am trying to use a Python program in LabVIEW using LabPython.  My program
> works fine outside of LabVIEW but not when compiled and run in LabVIEW via
> LabPython.  I am using Scipy and Numpy functions in the program but I think
> they are causing the problems.  Does anyone know if it is possible to use
> Scipy with LabPython?

Probably, but you will at least need to install numpy and scipy for
the Python executable that LabPython uses. Since there does not appear
to be any documentation, I doubt any of us here can help you beyond
that. Hopefully, the LabPython author will respond to your email on
the LabPython list.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From glen.shennan at gmail.com  Tue Jun  3 22:37:32 2008
From: glen.shennan at gmail.com (Glen Shennan)
Date: Wed, 4 Jun 2008 12:37:32 +1000
Subject: [SciPy-user] LabVIEW and LabPython
In-Reply-To: <3d375d730806031931p6603f4bbw237c2a699f6e21a6@mail.gmail.com>
References: <ff223aae0806031923u1dd2c613k7ee8e5f27c1753e8@mail.gmail.com>
	<3d375d730806031931p6603f4bbw237c2a699f6e21a6@mail.gmail.com>
Message-ID: <ff223aae0806031937r75178a3fj4182ceb4e2cfad17@mail.gmail.com>

"Probably, but you will at least need to install numpy and scipy for
the Python executable that LabPython uses."

Thanks!  Any idea how I would do that?  As far as I know I have one Python
executable installed, and numpy and scipy are installed for that.



2008/6/4 Robert Kern <robert.kern at gmail.com>:

> On Tue, Jun 3, 2008 at 9:23 PM, Glen Shennan <glen.shennan at gmail.com>
> wrote:
> > Hi,
> >
> > I am trying to use a Python program in LabVIEW using LabPython.  My
> program
> > works fine outside of LabVIEW but not when compiled and run in LabVIEW
> via
> > LabPython.  I am using Scipy and Numpy functions in the program but I
> think
> > they are causing the problems.  Does anyone know if it is possible to use
> > Scipy with LabPython?
>
> Probably, but you will at least need to install numpy and scipy for
> the Python executable that LabPython uses. Since there does not appear
> to be any documentation, I doubt any of us here can help you beyond
> that. Hopefully, the LabPython author will respond to your email on
> the LabPython list.
>
> --
> Robert Kern
>
> "I have come to believe that the whole world is an enigma, a harmless
> enigma that is made terrible by our own mad attempt to interpret it as
> though it had an underlying truth."
>  -- Umberto Eco
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080604/1868212f/attachment.html>

From robert.kern at gmail.com  Tue Jun  3 22:40:59 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Tue, 3 Jun 2008 21:40:59 -0500
Subject: [SciPy-user] LabVIEW and LabPython
In-Reply-To: <ff223aae0806031937r75178a3fj4182ceb4e2cfad17@mail.gmail.com>
References: <ff223aae0806031923u1dd2c613k7ee8e5f27c1753e8@mail.gmail.com>
	<3d375d730806031931p6603f4bbw237c2a699f6e21a6@mail.gmail.com>
	<ff223aae0806031937r75178a3fj4182ceb4e2cfad17@mail.gmail.com>
Message-ID: <3d375d730806031940i762dd493t85bdf97276c8acaa@mail.gmail.com>

On Tue, Jun 3, 2008 at 9:37 PM, Glen Shennan <glen.shennan at gmail.com> wrote:
> "Probably, but you will at least need to install numpy and scipy for
> the Python executable that LabPython uses."
>
> Thanks!  Any idea how I would do that?  As far as I know I have one Python
> executable installed, and numpy and scipy are installed for that.

I'm afraid not. Like I said, I can't find any documentation on how
LabPython is set up.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From cohen at slac.stanford.edu  Wed Jun  4 04:21:49 2008
From: cohen at slac.stanford.edu (Johann Cohen-Tanugi)
Date: Wed, 04 Jun 2008 10:21:49 +0200
Subject: [SciPy-user] problem with optimize.leastsq
In-Reply-To: <4eeef9d40806031645t20a180ffi9c6b13c71d971866@mail.gmail.com>
References: <4eeef9d40806031645t20a180ffi9c6b13c71d971866@mail.gmail.com>
Message-ID: <4846509D.2030606@slac.stanford.edu>

hi Jose,
can you please provide a test example that fails? There is no way one 
can do anything to help you with the information you are providing below.
best,
JCT

Jose Lopez wrote:
>
> hi
>
> i have a problem with "optimize.leastsq", give it the next error:
> MemoryError:
> File: "c:\python25\lib....\minpack.py",line 268,in leastsq
> retval = 
> _minpack._lmdif(func,x0,args,full_output,ftol,xtol,gtol,maxfev,epsfcn,factor,diag)
>
> if i work whit the same code with some parameters work it!
>
> but i work with more parameters the error above show me.
> My os is windows
>  
> someone happened the same error?
>  
> thanks
> ------------------------------------------------------------------------
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>   


From lbolla at gmail.com  Wed Jun  4 10:47:15 2008
From: lbolla at gmail.com (lorenzo bolla)
Date: Wed, 4 Jun 2008 16:47:15 +0200
Subject: [SciPy-user] interpn
Message-ID: <80c99e790806040747j236ec655xac82a720b0acc593@mail.gmail.com>

Hello all!

Is there, in scipy or numpy, a function similar to Matlab's interpn?
>From Matlab's help on interpn:
"VI = interpn(X1,X2,X3,...,V,Y1,Y2,Y3,...) interpolates to find VI,
the values of the underlying multidimensional function V at the points
in the arrays Y1, Y2, Y3, etc. For an n-dimensional array V, interpn
is called with 2*N+1 arguments. Arrays X1, X2, X3, etc. specify the
points at which the data V is given. Out of range values are returned
as NaNs. ..."

Basically, I've got a function from R^n to R. I know its value in a
set of points in R^n. I want to interpolate them so as to be able to
evaluate the function in any R^n point.

Thank you in advance!
L.

-- 
"Whereof one cannot speak, thereof one must be silent." -- Ludwig Wittgenstein


From oliphant at enthought.com  Wed Jun  4 11:58:39 2008
From: oliphant at enthought.com (Travis E. Oliphant)
Date: Wed, 04 Jun 2008 10:58:39 -0500
Subject: [SciPy-user] interpn
In-Reply-To: <80c99e790806040747j236ec655xac82a720b0acc593@mail.gmail.com>
References: <80c99e790806040747j236ec655xac82a720b0acc593@mail.gmail.com>
Message-ID: <4846BBAF.9080003@enthought.com>

lorenzo bolla wrote:
> Hello all!
>
> Is there, in scipy or numpy, a function similar to Matlab's interpn?
> >From Matlab's help on interpn:
> "VI = interpn(X1,X2,X3,...,V,Y1,Y2,Y3,...) interpolates to find VI,
> the values of the underlying multidimensional function V at the points
> in the arrays Y1, Y2, Y3, etc. For an n-dimensional array V, interpn
> is called with 2*N+1 arguments. Arrays X1, X2, X3, etc. specify the
> points at which the data V is given. Out of range values are returned
> as NaNs. ..."
>
> Basically, I've got a function from R^n to R. I know its value in a
> set of points in R^n. I want to interpolate them so as to be able to
> evaluate the function in any R^n point.
>   

scipy.ndimage has something called map_coordinates which (I think) can 
do what you want, but it's interface is a bit more complicated than 
interpn.

-Travis




From orest.kozyar at gmail.com  Wed Jun  4 12:01:58 2008
From: orest.kozyar at gmail.com (Orest Kozyar)
Date: Wed, 4 Jun 2008 09:01:58 -0700 (PDT)
Subject: [SciPy-user] Debugging scipy.weave?
Message-ID: <931e7a5c-e26a-47f9-ae52-96297b4e4b18@m36g2000hse.googlegroups.com>

I have written some C++ code that is integrated into my python scripts
using scipy.weave.  In general, it runs pretty smoothly when I run it
over a small portion of my dataset.  When I run my script over the
entire dataset, it crashes the Python interpreter.  I suspect this is
a memory issue, but cannot figure out what is going on because it just
refuses to crash when running under a debugger.  When I run python
under the GNU debugger (gdb), the script runs just fine.  Any advice
or pointers on how to debug this issue?


From berthe.loic at gmail.com  Wed Jun  4 12:12:34 2008
From: berthe.loic at gmail.com (LB)
Date: Wed, 4 Jun 2008 09:12:34 -0700 (PDT)
Subject: [SciPy-user] How to build scipy 0.6 with numpy 1.1 ?
In-Reply-To: <61097727-a8d4-42d0-b772-81a341c5d7a5@d77g2000hsb.googlegroups.com>
References: <61097727-a8d4-42d0-b772-81a341c5d7a5@d77g2000hsb.googlegroups.com>
Message-ID: <07b8c0c5-0bd9-48b0-9b24-e73c05f5122e@j22g2000hsf.googlegroups.com>

I have tested this on another computer which has a  64bits processor
and I got the same problem.

So is there a way have a clean installation of both numpy 1.1 and
scipy 0.6 on linux or do we have to wait for the next release of
scipy ?

Regards,

--
LB


From zunzun at zunzun.com  Wed Jun  4 13:00:32 2008
From: zunzun at zunzun.com (James Phillips)
Date: Wed, 4 Jun 2008 12:00:32 -0500
Subject: [SciPy-user] Debugging scipy.weave?
In-Reply-To: <931e7a5c-e26a-47f9-ae52-96297b4e4b18@m36g2000hse.googlegroups.com>
References: <931e7a5c-e26a-47f9-ae52-96297b4e4b18@m36g2000hse.googlegroups.com>
Message-ID: <268756d30806041000j1e1f4c90i64a5a882c1403b9a@mail.gmail.com>

In your C++, use good old-fashioned:

#include <stdio.h>
std::cout << "name of something: " << value_of_something << std::endl;

Now *that* brings back memories of youth...  P.U.

     James Phillips
     http://zunzun.com


On Wed, Jun 4, 2008 at 11:01 AM, Orest Kozyar <orest.kozyar at gmail.com>
wrote:

> I have written some C++ code that is integrated into my python scripts
> using scipy.weave.  In general, it runs pretty smoothly when I run it
> over a small portion of my dataset.  When I run my script over the
> entire dataset, it crashes the Python interpreter.  I suspect this is
> a memory issue, but cannot figure out what is going on because it just
> refuses to crash when running under a debugger.  When I run python
> under the GNU debugger (gdb), the script runs just fine.  Any advice
> or pointers on how to debug this issue?
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080604/1cf9da1c/attachment.html>

From orest.kozyar at gmail.com  Wed Jun  4 15:03:38 2008
From: orest.kozyar at gmail.com (Orest Kozyar)
Date: Wed, 4 Jun 2008 15:03:38 -0400
Subject: [SciPy-user] Debugging scipy.weave?
In-Reply-To: <268756d30806041000j1e1f4c90i64a5a882c1403b9a@mail.gmail.com>
References: <931e7a5c-e26a-47f9-ae52-96297b4e4b18@m36g2000hse.googlegroups.com>
	<268756d30806041000j1e1f4c90i64a5a882c1403b9a@mail.gmail.com>
Message-ID: <d635fdd00806041203o353324bei3f26f33fe4242f56@mail.gmail.com>

Perfect.  Old fashioned "print" debugging helped me figure out the
problem.  Thanks!

On Wed, Jun 4, 2008 at 1:00 PM, James Phillips <zunzun at zunzun.com> wrote:
> In your C++, use good old-fashioned:
>
> #include <stdio.h>
> std::cout << "name of something: " << value_of_something << std::endl;
>
> Now *that* brings back memories of youth...  P.U.
>
>      James Phillips
>      http://zunzun.com
>
>
> On Wed, Jun 4, 2008 at 11:01 AM, Orest Kozyar <orest.kozyar at gmail.com>
> wrote:
>>
>> I have written some C++ code that is integrated into my python scripts
>> using scipy.weave.  In general, it runs pretty smoothly when I run it
>> over a small portion of my dataset.  When I run my script over the
>> entire dataset, it crashes the Python interpreter.  I suspect this is
>> a memory issue, but cannot figure out what is going on because it just
>> refuses to crash when running under a debugger.  When I run python
>> under the GNU debugger (gdb), the script runs just fine.  Any advice
>> or pointers on how to debug this issue?
>> _______________________________________________
>> SciPy-user mailing list
>> SciPy-user at scipy.org
>> http://projects.scipy.org/mailman/listinfo/scipy-user
>
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
>


From topengineer at gmail.com  Wed Jun  4 20:49:34 2008
From: topengineer at gmail.com (HuiChang MOON)
Date: Thu, 5 Jun 2008 09:49:34 +0900
Subject: [SciPy-user] Does Scipy or Numpy have contemporary physical
	constants?
Message-ID: <296323b50806041749o15424c44ucf43bd4f17350eab@mail.gmail.com>

Hello, users.
A lot of physical constants such as elementary charge and Planck's constant
is used to make some physical calculations.
It is annoying to define the constants so often.
Does Numpy or Scipy include some physical constants?
Thank you.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080605/c4fa4221/attachment.html>

From peridot.faceted at gmail.com  Wed Jun  4 21:04:05 2008
From: peridot.faceted at gmail.com (Anne Archibald)
Date: Wed, 4 Jun 2008 21:04:05 -0400
Subject: [SciPy-user] Does Scipy or Numpy have contemporary physical
	constants?
In-Reply-To: <296323b50806041749o15424c44ucf43bd4f17350eab@mail.gmail.com>
References: <296323b50806041749o15424c44ucf43bd4f17350eab@mail.gmail.com>
Message-ID: <ce557a360806041804n3337e791icab21d59f8673ea5@mail.gmail.com>

2008/6/4 HuiChang MOON <topengineer at gmail.com>:
> Hello, users.
> A lot of physical constants such as elementary charge and Planck's constant
> is used to make some physical calculations.
> It is annoying to define the constants so often.
> Does Numpy or Scipy include some physical constants?

No, unfortunately. I has been discussed from time to time, and the
conclusion is that they're not very useful without an automatic
unit-tracking system. Such a system does exist, and it provides
standard constants; it's in Scientific (a slightly dusty package
somewhat complementary to scipy). There is even some support for units
built into ipython, though personally I don't use it.

The biggest limitation of the unit-tracking system is that it doesn't
really work with arrays. There are definitely applications for a
subclass of ndarray that keeps track of units, but I don't know of a
publicly-released one. For pocket-calculator type calculations,
Scientific is pretty handy.

Anne


From hasslerjc at comcast.net  Wed Jun  4 21:09:06 2008
From: hasslerjc at comcast.net (John Hassler)
Date: Wed, 04 Jun 2008 21:09:06 -0400
Subject: [SciPy-user] Does Scipy or Numpy have contemporary
	physical	constants?
In-Reply-To: <296323b50806041749o15424c44ucf43bd4f17350eab@mail.gmail.com>
References: <296323b50806041749o15424c44ucf43bd4f17350eab@mail.gmail.com>
Message-ID: <48473CB2.8010308@comcast.net>

An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080604/94519855/attachment.html>

From hasslerjc at comcast.net  Wed Jun  4 21:24:34 2008
From: hasslerjc at comcast.net (John Hassler)
Date: Wed, 04 Jun 2008 21:24:34 -0400
Subject: [SciPy-user] Does Scipy or Numpy have contemporary
	physical	constants?
In-Reply-To: <ce557a360806041804n3337e791icab21d59f8673ea5@mail.gmail.com>
References: <296323b50806041749o15424c44ucf43bd4f17350eab@mail.gmail.com>
	<ce557a360806041804n3337e791icab21d59f8673ea5@mail.gmail.com>
Message-ID: <48474052.40700@comcast.net>

An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080604/d32f064f/attachment.html>

From peridot.faceted at gmail.com  Wed Jun  4 21:40:54 2008
From: peridot.faceted at gmail.com (Anne Archibald)
Date: Wed, 4 Jun 2008 21:40:54 -0400
Subject: [SciPy-user] Does Scipy or Numpy have contemporary physical
	constants?
In-Reply-To: <48474052.40700@comcast.net>
References: <296323b50806041749o15424c44ucf43bd4f17350eab@mail.gmail.com>
	<ce557a360806041804n3337e791icab21d59f8673ea5@mail.gmail.com>
	<48474052.40700@comcast.net>
Message-ID: <ce557a360806041840r5efea072o382dd8fb7de7d69d@mail.gmail.com>

2008/6/4 John Hassler <hasslerjc at comcast.net>:
> Ah.  I interpreted his question differently.  I used to teach Chemical
> Engineering.  We used Mathcad, which has a built-in set of units,
> conversions, and tracking.  I found it to be a pain in the coccyx, but the
> students adored it.  They would throw in whatever units they happened to
> find, and hope that M'cad would figure it all out for them.  It usually did,
> although the units in their answers sometimes didn't have (obvious) physical
> meanings.  M'cad would usually collapse everything as far as possible,
> rather than recognizing meaningful collections like force (mass length /
> time^2), etc.
>
> But I'm just an old fogy who grew up with a slide rule.  Why, back in MY
> day, we even had to keep track of our own DECIMAL POINTS ... barefoot in the
> snow .... uphill all the way ...
> john

Interesting. Automatic units tracking has definitely kept me from
shooting myself in the foot many times, but I agree there's a
temptation to assume that if the units are right the answer must be
right too.

Anne


From millman at berkeley.edu  Wed Jun  4 21:56:29 2008
From: millman at berkeley.edu (Jarrod Millman)
Date: Wed, 4 Jun 2008 18:56:29 -0700
Subject: [SciPy-user] Does Scipy or Numpy have contemporary physical
	constants?
In-Reply-To: <296323b50806041749o15424c44ucf43bd4f17350eab@mail.gmail.com>
References: <296323b50806041749o15424c44ucf43bd4f17350eab@mail.gmail.com>
Message-ID: <c7009a550806041856p6b9d2dd9u900af1b2bf14ba9@mail.gmail.com>

On Wed, Jun 4, 2008 at 5:49 PM, HuiChang MOON <topengineer at gmail.com> wrote:
> A lot of physical constants such as elementary charge and Planck's constant
> is used to make some physical calculations.
> It is annoying to define the constants so often.
> Does Numpy or Scipy include some physical constants?

SciPy trunk has this:
http://projects.scipy.org/scipy/scipy/browser/trunk/scipy/constants

-- 
Jarrod Millman
Computational Infrastructure for Research Labs
10 Giannini Hall, UC Berkeley
phone: 510.643.4014
http://cirl.berkeley.edu/


From haase at msg.ucsf.edu  Thu Jun  5 07:23:43 2008
From: haase at msg.ucsf.edu (Sebastian Haase)
Date: Thu, 5 Jun 2008 13:23:43 +0200
Subject: [SciPy-user] easy to fix bug in ndimage.center_of_mass
Message-ID: <bc657ead0806050423w7ee122a1nd40ede2d5c140f35@mail.gmail.com>

Hi,
I found this code in measurements.py
def center_of_mass(input, labels = None, index = None):
    """Calculate the center of mass of of the array.

    The index parameter is a single label number or a sequence of
    label numbers of the objects to be measured. If index is None, all
    values are used where labels is larger than zero.
    """
    input = numpy.asarray(input)
    if numpy.iscomplexobj(input):
        raise TypeError, 'Complex type not supported'
    if labels != None:
        labels = numpy.asarray(labels)
        labels = _broadcast(labels, input.shape)

        if labels.shape != input.shape:
            raise RuntimeError, 'input and labels shape are not equal'
    return _nd_image.center_of_mass(input, labels, index)


but the if part:
if labels.shape != input.shape:

is probably indented one level too much !?

Regards,
Sebastian


From stefan at sun.ac.za  Thu Jun  5 08:11:40 2008
From: stefan at sun.ac.za (=?ISO-8859-1?Q?St=E9fan_van_der_Walt?=)
Date: Thu, 5 Jun 2008 14:11:40 +0200
Subject: [SciPy-user] easy to fix bug in ndimage.center_of_mass
In-Reply-To: <bc657ead0806050423w7ee122a1nd40ede2d5c140f35@mail.gmail.com>
References: <bc657ead0806050423w7ee122a1nd40ede2d5c140f35@mail.gmail.com>
Message-ID: <9457e7c80806050511v33705620s641453c31e15eedf@mail.gmail.com>

2008/6/5 Sebastian Haase <haase at msg.ucsf.edu>:
> Hi,
> I found this code in measurements.py
> def center_of_mass(input, labels = None, index = None):
>    """Calculate the center of mass of of the array.
>
>    The index parameter is a single label number or a sequence of
>    label numbers of the objects to be measured. If index is None, all
>    values are used where labels is larger than zero.
>    """
>    input = numpy.asarray(input)
>    if numpy.iscomplexobj(input):
>        raise TypeError, 'Complex type not supported'
>    if labels != None:
>        labels = numpy.asarray(labels)
>        labels = _broadcast(labels, input.shape)
>
>        if labels.shape != input.shape:
>            raise RuntimeError, 'input and labels shape are not equal'
>    return _nd_image.center_of_mass(input, labels, index)
>
>
> but the if part:
> if labels.shape != input.shape:
>
> is probably indented one level too much !?

That looks right.  You don't want to access labels.shape if labels is
None.  The "!=" is wrong, though.  It should be "is not"

Regards
St?fan


From matthieu.brucher at gmail.com  Thu Jun  5 12:39:01 2008
From: matthieu.brucher at gmail.com (Matthieu Brucher)
Date: Thu, 5 Jun 2008 18:39:01 +0200
Subject: [SciPy-user] Using VariantDir
Message-ID: <e76aa17f0806050939y3dcbcea6j90085ad951d7837b@mail.gmail.com>

Hi,

I'm trying to build some files out of place, in a subfolder named "build".
The sources are in the root folder.

I've tried the obvious :
env.VariantDir("build", ".", duplicate=0)

but the files are built in the current folder instead of the "build" one. If
I build some files in a subfolder with :
env.SConscript('folder/SConstruct', variant_dir='build/folder', duplicate=0)

the files are built in "build/folder", as expected. But I do not understand
why it is not the same as using VariantDir.

If someone has a clue, I'm all ears ;)

Matthieu
-- 
French PhD student
Website : http://matthieu-brucher.developpez.com/
Blogs : http://matt.eifelle.com and http://blog.developpez.com/?blog=92
LinkedIn : http://www.linkedin.com/in/matthieubrucher
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080605/40ae0858/attachment.html>

From travis at enthought.com  Thu Jun  5 13:59:18 2008
From: travis at enthought.com (Travis Vaught)
Date: Thu, 5 Jun 2008 12:59:18 -0500
Subject: [SciPy-user] Does Scipy or Numpy have contemporary physical
	constants?
In-Reply-To: <ce557a360806041804n3337e791icab21d59f8673ea5@mail.gmail.com>
References: <296323b50806041749o15424c44ucf43bd4f17350eab@mail.gmail.com>
	<ce557a360806041804n3337e791icab21d59f8673ea5@mail.gmail.com>
Message-ID: <42E77762-CFAB-4A91-8A9A-4614451AFEC4@enthought.com>


On Jun 4, 2008, at 8:04 PM, Anne Archibald wrote:

> 2008/6/4 HuiChang MOON <topengineer at gmail.com>:
>> Hello, users.
>> A lot of physical constants such as elementary charge and Planck's  
>> constant
>> is used to make some physical calculations.
>> It is annoying to define the constants so often.
>> Does Numpy or Scipy include some physical constants?
>
> No, unfortunately. I has been discussed from time to time, and the
> conclusion is that they're not very useful without an automatic
> unit-tracking system. Such a system does exist, and it provides
> standard constants; it's in Scientific (a slightly dusty package
> somewhat complementary to scipy). There is even some support for units
> built into ipython, though personally I don't use it.
>
> The biggest limitation of the unit-tracking system is that it doesn't
> really work with arrays. There are definitely applications for a
> subclass of ndarray that keeps track of units, but I don't know of a
> publicly-released one. For pocket-calculator type calculations,
> Scientific is pretty handy.
>

FWIW I'll mention the units package in ETS:

https://svn.enthought.com/enthought/browser/SciMath/trunk/enthought/units

It's based on some good work that Michael Avaizis at Caltech did.

The upper layers are not very mature and could do with a refactor to  
make them more general (the notion of a unit manager and unit families  
is a nice one, but the current examples are very specific).

You can do things like this:

In [1]: # This allows me to use strings and it finds the symbols in  
the namespace for me

In [2]: from enthought.units.unit_parser import unit_parser

In [3]: import numpy

In [4]: a = numpy.arange(0., 1., 0.01)

In [5]: b = unit_parser.parse_unit("m/s")

In [6]: c = b * a

In [7]: c
Out[7]: <array>*m*s**-1

In [8]: d = unit_parser.parse_unit("minutes") * 15.

In [9]: d # didn't work (minutes not in namespace)
Out[9]: 15.0

In [10]: d = unit_parser.parse_unit("seconds") * 15. * 60. # 15 minutes

In [11]: d
Out[11]: 900.0*s

In [12]: d * c
Out[12]: <array>*m

In [13]: e = d*c

In [14]: e.value
Out[14]:
array([   0.,    9.,   18.,   27.,   36.,   45.,   54.,   63.,   72.,
          81.,   90.,   99.,  108.,  117.,  126.,  135.,  144.,  153.,
         162.,  171.,  180.,  189.,  198.,  207.,  216.,  225.,  234.,
         243.,  252.,  261.,  270.,  279.,  288.,  297.,  306.,  315.,
         324.,  333.,  342.,  351.,  360.,  369.,  378.,  387.,  396.,
         405.,  414.,  423.,  432.,  441.,  450.,  459.,  468.,  477.,
         486.,  495.,  504.,  513.,  522.,  531.,  540.,  549.,  558.,
         567.,  576.,  585.,  594.,  603.,  612.,  621.,  630.,  639.,
         648.,  657.,  666.,  675.,  684.,  693.,  702.,  711.,  720.,
         729.,  738.,  747.,  756.,  765.,  774.,  783.,  792.,  801.,
         810.,  819.,  828.,  837.,  846.,  855.,  864.,  873.,   
882.,  891.])



This doesn't work quite right if you reverse the multiplication  
because of NumPy's aggressive casting, but after talking with Travis  
O. it seems like I can add an __rmul__ method to fix this.

Best,

Travis

> Anne
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>



From timmichelsen at gmx-topmail.de  Thu Jun  5 14:50:48 2008
From: timmichelsen at gmx-topmail.de (Tim Michelsen)
Date: Thu, 05 Jun 2008 20:50:48 +0200
Subject: [SciPy-user] [SOLVED] Re: building and installing timeseries fails
	on Windows
In-Reply-To: <loom.20080602T170514-482@post.gmane.org>
References: <loom.20080602T170514-482@post.gmane.org>
Message-ID: <g29cia$6iq$1@ger.gmane.org>

> following the (updated) instructions at:
> http://www.scipy.org/Cookbook/CompilingExtensionsOnWindowsWithMinGW

> 
> What can I do now?
After upgrading my SVN checkout it all worked well.
Thanks again to Pierre and Matt for the very useful package. I may 
contribute more to the Cookbook some time.

Kind regards,
Timmie



From fperez.net at gmail.com  Thu Jun  5 20:21:34 2008
From: fperez.net at gmail.com (Fernando Perez)
Date: Thu, 5 Jun 2008 17:21:34 -0700
Subject: [SciPy-user] [ Python(x,y) ] New release : 1.2.4
In-Reply-To: <48443B3D.1090206@pythonxy.com>
References: <48443B3D.1090206@pythonxy.com>
Message-ID: <db6b5ecc0806051721q1e40e632vbbf50c57f16dd38f@mail.gmail.com>

On Mon, Jun 2, 2008 at 11:26 AM, Pierre Raybaut <contact at pythonxy.com> wrote:
> Hi all,
>
> Python(x,y) 1.2.4 is now available on http://www.pythonxy.com.
> Hopefully, updates should now be released at a lower rate.
>
> Changes history
> 06 -02 -2008 - Version 1.2.4 :
>
>     * Updated:
>           o Matplotlib 0.98 - Important change for Python(x,y) users:
> better integration in Qt GUIs (see release notes)
>           o IPython 0.8.4 (see release notes)

Many thanks!  This is really great.

One feature request, if I might: any chance ofgetting scons in?

http://www.scons.org/

The reason is that ctypes is one of the nicest and easiest ways of
quickly prototyping access to libraries and extension code, but
writing a Makefile to build a shared library portably
(osx/win32/linux) isn't completely trivial. Scons knows a fair bit
about shared library building, so having scons in would help users who
want to test the ctypes/numpy support such as:

http://www.scipy.org/Cookbook/Ctypes

Best regards,

f


From fperez.net at gmail.com  Thu Jun  5 20:26:30 2008
From: fperez.net at gmail.com (Fernando Perez)
Date: Thu, 5 Jun 2008 17:26:30 -0700
Subject: [SciPy-user] [ Python(x,y) ] New release : 1.2.4
In-Reply-To: <db6b5ecc0806051721q1e40e632vbbf50c57f16dd38f@mail.gmail.com>
References: <48443B3D.1090206@pythonxy.com>
	<db6b5ecc0806051721q1e40e632vbbf50c57f16dd38f@mail.gmail.com>
Message-ID: <db6b5ecc0806051726l4a323c3fx507ee94dd9cda699@mail.gmail.com>

On Thu, Jun 5, 2008 at 5:21 PM, Fernando Perez <fperez.net at gmail.com> wrote:
> On Mon, Jun 2, 2008 at 11:26 AM, Pierre Raybaut <contact at pythonxy.com> wrote:
>> Hi all,
>>
>> Python(x,y) 1.2.4 is now available on http://www.pythonxy.com.
>> Hopefully, updates should now be released at a lower rate.

Ooops, and I forgot: do you ship nose?  Since it is now required for
scipy testing (and will soon be for numpy, as the tests transition),
it would be nice to have it too :)

http://code.google.com/p/python-nose/

Again, many thanks and sorry for asking without offering much in
return.  Your project is too nice not to :)

Regards,

f


From bemclaugh at gmail.com  Thu Jun  5 21:55:04 2008
From: bemclaugh at gmail.com (BEM)
Date: Thu, 5 Jun 2008 18:55:04 -0700 (PDT)
Subject: [SciPy-user] scikits.timeseries AttributeError
Message-ID: <1f481223-faa2-4675-98c0-099d86956a69@z16g2000prn.googlegroups.com>

Hi,
I am trying to generate a plot of data sampled at about 1/5Hz...
(ts.__version__ = 1.0)

import matplotlib.pyplot as plt
import scikits.timeseries as ts
import scikits.timeseries.lib.plotlib as tpl


dates = ts.date_array([q[1] for q in data], freq='S')
val = [q[4] for q in data]
raw_series = ts.time_series(val, dates)
series = ts.fill_missing_dates(raw_series)

fig = tpl.tsfigure()
fsp = fig.add_tsplot(111)
fsp.tsplot(series, '-')

I get an:
AttributeError: TimeSeries_DateLocator instance has no attribute
'finder'

I can, however, run the Yahoo financial example and produce a report
of the data I am attempting to plot that looks correct:
import scikits.timeseries.lib.reportlib as rl
basicReport = rl.Report(series)
print basicReport()

Any ideas?
thanks,
Brian


From pgmdevlist at gmail.com  Thu Jun  5 22:18:08 2008
From: pgmdevlist at gmail.com (Pierre GM)
Date: Thu, 5 Jun 2008 22:18:08 -0400
Subject: [SciPy-user] scikits.timeseries AttributeError
In-Reply-To: <1f481223-faa2-4675-98c0-099d86956a69@z16g2000prn.googlegroups.com>
References: <1f481223-faa2-4675-98c0-099d86956a69@z16g2000prn.googlegroups.com>
Message-ID: <200806052218.08419.pgmdevlist@gmail.com>

Brian,
What's the SVN version (ts.__version__ is far from informative, sorry, I need 
to check that) ?
What version of matplotlib do you use (not that really matters here) ?

Good call anyway. As stated on the doc page, support for series with a 
frequency higher than daily is shaky at best. In particular, plotting 
capabilities with timeseries.lib.plotlib are restricted to daily or less. But 
you're right, we should at least try to issue a warning (or raise an 
exception).
Meanwhile, you can always use the standard matplotlib functions on 
series.dates.tolist() for abscissae and series.series for ordinates.


From mattknox.ca at gmail.com  Thu Jun  5 22:34:41 2008
From: mattknox.ca at gmail.com (Matt Knox)
Date: Fri, 6 Jun 2008 02:34:41 +0000 (UTC)
Subject: [SciPy-user] scikits.timeseries AttributeError
References: <1f481223-faa2-4675-98c0-099d86956a69@z16g2000prn.googlegroups.com>
Message-ID: <loom.20080606T022704-444@post.gmane.org>

>> dates = ts.date_array([q[1] for q in data], freq='S')

This is one of the known limitations of the timeseries module right now
unfortunately. Plotting for frequencies higher than daily is not currently
supported. It actually shouldn't be too much work to implement it, but it just
hasn't been a priority for me and I haven't had much free time lately. I hope to
be able to get to this sometime this month... but contributions are always
welcome if someone beats me to it :)

Pierre and I have been discussing doing a first official release of the
timeseries module in the near future, but there are a few remaining outstanding
issues we want to clear up first.



From lists at vrbka.net  Fri Jun  6 03:10:54 2008
From: lists at vrbka.net (Lubos Vrbka)
Date: Fri, 06 Jun 2008 09:10:54 +0200
Subject: [SciPy-user] Fourier-sine transform with SciPy
Message-ID: <4848E2FE.9070402@vrbka.net>

hi guys,

i'm a new (and so far very happy) user of the scipy bundle. however, i 
have a problem at the moment, and i don't know how to solve it (apart 
from somehow writing slow and unoptimized transform routine myself).

i need to perform a discrete fourier transform of a radially symmetric 
function, dependent only on the |r| and |k| in real and fourier space, 
respectively. after some straightforward math, i arrived at the 
expression for the sine transform (continuous transform), namely
F(k) = A \int_0^\infty f(r) sin(kr) dr (FT)
f(r) = B \int_0^\infty f(k) sin(kr) dk (iFT)
with A and B being constant coefficients. naturally this has got the 
advantage of just one function being needed to perform the transform 
(just with different coefficient every time).

the problem is, that i don't know how (and whether it actually is 
posible) to perform this task (discrete sine transform) with scipy. 
searching the google and lists wasn't of much help, either.

i know that for example fftw includes the function for sine and cosine 
transforms, but i didn't find any connection between the fftw functions 
and scipy.fft, that would be younger than several years.

i would be grateful for any help or hint.

with best regards,

-- 
Lubos _ at _"
http://www.lubos.vrbka.net


From peridot.faceted at gmail.com  Fri Jun  6 04:44:07 2008
From: peridot.faceted at gmail.com (Anne Archibald)
Date: Fri, 6 Jun 2008 04:44:07 -0400
Subject: [SciPy-user] Fourier-sine transform with SciPy
In-Reply-To: <4848E2FE.9070402@vrbka.net>
References: <4848E2FE.9070402@vrbka.net>
Message-ID: <ce557a360806060144i17a4243eu8c328b0c43b4e99f@mail.gmail.com>

2008/6/6 Lubos Vrbka <lists at vrbka.net>:

> i need to perform a discrete fourier transform of a radially symmetric
> function, dependent only on the |r| and |k| in real and fourier space,
> respectively. after some straightforward math, i arrived at the
> expression for the sine transform (continuous transform), namely
> F(k) = A \int_0^\infty f(r) sin(kr) dr (FT)
> f(r) = B \int_0^\infty f(k) sin(kr) dk (iFT)
> with A and B being constant coefficients. naturally this has got the
> advantage of just one function being needed to perform the transform
> (just with different coefficient every time).
>
> the problem is, that i don't know how (and whether it actually is
> posible) to perform this task (discrete sine transform) with scipy.
> searching the google and lists wasn't of much help, either.
>
> i know that for example fftw includes the function for sine and cosine
> transforms, but i didn't find any connection between the fftw functions
> and scipy.fft, that would be younger than several years.

I don't think we have a function that computes discrete sine or cosine
transforms, but if f is real, you can get the sine transform you wrote
as the imaginary part of a complex Fourier transform.

Anne


From lists at vrbka.net  Fri Jun  6 05:17:44 2008
From: lists at vrbka.net (Lubos Vrbka)
Date: Fri, 06 Jun 2008 11:17:44 +0200
Subject: [SciPy-user] Fourier-sine transform with SciPy
In-Reply-To: <ce557a360806060144i17a4243eu8c328b0c43b4e99f@mail.gmail.com>
References: <4848E2FE.9070402@vrbka.net>
	<ce557a360806060144i17a4243eu8c328b0c43b4e99f@mail.gmail.com>
Message-ID: <484900B8.6080006@vrbka.net>

hi,

thanks for quick reply!

> I don't think we have a function that computes discrete sine or cosine
> transforms, but if f is real, you can get the sine transform you wrote
> as the imaginary part of a complex Fourier transform.
by this you mean using scipy.fftpack.rfft()? i must admit i am a bit 
lost in the various possibilities that are available.

scipy.fftpack.*fft* functions are clear (is this the netlib implementation?)

scipy.ifft?? tells me (among others)
File:             /usr/lib/python2.4/site-packages/numpy/fft/fftpack.py

or should i use numpy.fft? what's the difference in fftpack in scipy and 
numpy?

is there any place where these things are documented, i.e. what code 
does actually lie behind these functions?

and just out of curiosity - from what i've seen during my search for the 
  information on google, fftw was (at least in the past) used in scipy. 
the scipy package on my debian box depends on fftw2, but the names refer 
to fftpack, that shouldn't be fftw but netlib implementation instead (if 
i am not wrong). could anybody clarify this for me, please? sorry if 
this was discussed before, but i wasn't able to find the relevant thread.

thanks once more!
lubos

-- 
Lubos _ at _"
http://www.lubos.vrbka.net


From peridot.faceted at gmail.com  Fri Jun  6 05:42:29 2008
From: peridot.faceted at gmail.com (Anne Archibald)
Date: Fri, 6 Jun 2008 05:42:29 -0400
Subject: [SciPy-user] Fourier-sine transform with SciPy
In-Reply-To: <484900B8.6080006@vrbka.net>
References: <4848E2FE.9070402@vrbka.net>
	<ce557a360806060144i17a4243eu8c328b0c43b4e99f@mail.gmail.com>
	<484900B8.6080006@vrbka.net>
Message-ID: <ce557a360806060242l37941515m2bff328ebe20e07d@mail.gmail.com>

2008/6/6 Lubos Vrbka <lists at vrbka.net>:

>> I don't think we have a function that computes discrete sine or cosine
>> transforms, but if f is real, you can get the sine transform you wrote
>> as the imaginary part of a complex Fourier transform.
> by this you mean using scipy.fftpack.rfft()? i must admit i am a bit
> lost in the various possibilities that are available.

Yes, there are rather a lot.

A complex fft - numpy.fft.fft - computes:

F_m = A \sum_{k=0}^{N-1} f_k (cos(2\pi m k/N)+i*sin(2\pi m k/N))

and an ifft computes:

f_k = B \sum_{m=0}^{N-1} F_m (cos(2\pi m k/N)-i*sin(2\pi m k/N))

So if you want

\sum_{k=0}^{N-1} f_k sin(2\pi m k/N)

all you have to do is take a complex FFT of your (real) data, then
take the imaginary part. This is somewhat wasteful, which is why FFTW
implements the sine and cosine transforms, but it's only a factor of
two or four (i.e. still far better than coding it by hand).

To save a little time, you can use a "real" FFT, which computes F_m as
above, but (1) takes advantage of the fact that f_k is known to be
real and (2) doesn't bother computing F_m for m>N/2 (or so), since
when f_k is real, F_m = conjugate(F_{N-m}) IIRC. So you may want to do
this and take the imaginary part.

Note that the sine transform only works on odd functions, so either
your input data should have f_k = f_{N-k} or you are representing only
half the values, i.e., N is twice the number of data points you have.
This too is somewhere a proper sine transform would help, if we had
it.

> scipy.fftpack.*fft* functions are clear (is this the netlib implementation?)
>
> scipy.ifft?? tells me (among others)
> File:             /usr/lib/python2.4/site-packages/numpy/fft/fftpack.py
>
> or should i use numpy.fft? what's the difference in fftpack in scipy and
> numpy?
>
> is there any place where these things are documented, i.e. what code
> does actually lie behind these functions?
>
> and just out of curiosity - from what i've seen during my search for the
>  information on google, fftw was (at least in the past) used in scipy.
> the scipy package on my debian box depends on fftw2, but the names refer
> to fftpack, that shouldn't be fftw but netlib implementation instead (if
> i am not wrong). could anybody clarify this for me, please? sorry if
> this was discussed before, but i wasn't able to find the relevant thread.

There are several reasons this is complicated. numpy's fft is designed
to use one of several external libraries depending on how numpy is
compiled. FFTPACK is always available, but is not especially fast.
FFTW is distributed under the GPL, so numpy cannot depend on it
without also being distributed under the GPL. But if it's an optional
plug-in, numpy can make use if it where it's available, since it's
much faster than FFTPACK (in spite of the fact that we use its API in
a not-very-efficient manner). numpy also supports several other
libraries, most notably Intel's proprietary and hardware-specific but
very fast implementation. At a python level, the user need not know
which backend is being used, and FFTPACK is mentioned a few times.

Anne


From oliphant at enthought.com  Fri Jun  6 12:10:23 2008
From: oliphant at enthought.com (Travis E. Oliphant)
Date: Fri, 06 Jun 2008 11:10:23 -0500
Subject: [SciPy-user] Fourier-sine transform with SciPy
In-Reply-To: <4848E2FE.9070402@vrbka.net>
References: <4848E2FE.9070402@vrbka.net>
Message-ID: <4849616F.3050009@enthought.com>

Lubos Vrbka wrote:
> hi guys,
>
>
> the problem is, that i don't know how (and whether it actually is 
> posible) to perform this task (discrete sine transform) with scipy. 
> searching the google and lists wasn't of much help, either.
>   

In the scipy sandbox under image you can find a transforms.py file which 
contains an implementation of the DST (discrete sine transform).

http://projects.scipy.org/scipy/scipy/browser/trunk/scipy/sandbox/image/transforms.py

-Travis



From Dharhas.Pothina at twdb.state.tx.us  Fri Jun  6 12:32:01 2008
From: Dharhas.Pothina at twdb.state.tx.us (Dharhas Pothina)
Date: Fri, 06 Jun 2008 11:32:01 -0500
Subject: [SciPy-user] Adding new columns to 2D array & populating them from
	another	array.
Message-ID: <484920310200009B00013308@GWWEB.twdb.state.tx.us>

Hi,

I have an array 'a'

a = 
array([[2006,    1,    3,    9,   40],
       [2006,    1,    3,   10,    9],
       [2006,    1,    3,   10,   40],
       ..., 
       [2008,    3,   20,   10,   27],
       [2008,    3,   20,   10,   51],
       [2008,    3,   20,   12,    2]])

where a.shape = (420, 5)

I have other arrays b,c etc of the shape (420,) (say b contains all 99's and c contains all 50's)

how to I add the new arrays to 'a' to form an array 

  array([[2006,    1,    3,    9,   40 , 99, 50],
       [2006,    1,    3,   10,    9, 99 , 50],
       [2006,    1,    3,   10,   40, 99, 50],
       ...,
       [2008,    3,   20,   10,   27, 99, 50 ],
       [2008,    3,   20,   10,   51, 99, 50],
       [2008,    3,   20,   12,    2, 99, 50]])

I've tried hstack but I'm not sure how to use it with 2D arrays.

thanks

- dharhas


From ivo.maljevic at gmail.com  Fri Jun  6 12:38:44 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Fri, 6 Jun 2008 12:38:44 -0400
Subject: [SciPy-user] Adding new columns to 2D array & populating them
	from another array.
In-Reply-To: <484920310200009B00013308@GWWEB.twdb.state.tx.us>
References: <484920310200009B00013308@GWWEB.twdb.state.tx.us>
Message-ID: <826c64da0806060938l1bc0a88dj5cdddd90a5d71346@mail.gmail.com>

The simplest thing is to use c_[ ] operator.

Example:
>>> a=ones([2,2])
>>> a
array([[ 1.,  1.],
       [ 1.,  1.]])
>>> b=array([4,4])
>>> c=array([9,9])
>>> d=c_[a,b,c]
>>> d
array([[ 1.,  1.,  4.,  9.],
       [ 1.,  1.,  4.,  9.]])

Hope this helps,
Ivo


2008/6/6 Dharhas Pothina <Dharhas.Pothina at twdb.state.tx.us>:

> Hi,
>
> I have an array 'a'
>
> a =
> array([[2006,    1,    3,    9,   40],
>       [2006,    1,    3,   10,    9],
>       [2006,    1,    3,   10,   40],
>       ...,
>       [2008,    3,   20,   10,   27],
>       [2008,    3,   20,   10,   51],
>       [2008,    3,   20,   12,    2]])
>
> where a.shape = (420, 5)
>
> I have other arrays b,c etc of the shape (420,) (say b contains all 99's
> and c contains all 50's)
>
> how to I add the new arrays to 'a' to form an array
>
>  array([[2006,    1,    3,    9,   40 , 99, 50],
>       [2006,    1,    3,   10,    9, 99 , 50],
>       [2006,    1,    3,   10,   40, 99, 50],
>       ...,
>       [2008,    3,   20,   10,   27, 99, 50 ],
>       [2008,    3,   20,   10,   51, 99, 50],
>       [2008,    3,   20,   12,    2, 99, 50]])
>
> I've tried hstack but I'm not sure how to use it with 2D arrays.
>
> thanks
>
> - dharhas
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080606/5f215aa3/attachment.html>

From aisaac at american.edu  Fri Jun  6 12:46:36 2008
From: aisaac at american.edu (Alan G Isaac)
Date: Fri, 6 Jun 2008 12:46:36 -0400
Subject: [SciPy-user] Adding new columns to 2D array & populating them
	from another array.
In-Reply-To: <484920310200009B00013308@GWWEB.twdb.state.tx.us>
References: <484920310200009B00013308@GWWEB.twdb.state.tx.us>
Message-ID: <Mahogany-0.67.0-892-20080606-124636.00@american.edu>

On Fri, 06 Jun 2008, Dharhas Pothina apparently wrote:
> I have an array 'a'

> a = 
> array([[2006,    1,    3,    9,   40], 
>        [2006,    1,    3,   10,    9],
>        [2006,    1,    3,   10,   40],
>        ...,
>        [2008,    3,   20,   10,   27],
>        [2008,    3,   20,   10,   51],
>        [2008,    3,   20,   12,    2]])

> where a.shape = (420, 5)

> I have other arrays b,c etc of the shape (420,) (say b contains all 99's and c contains all 50's)

> how to I add the new arrays to 'a' to form an array 

>   array([[2006,    1,    3,    9,   40 , 99, 50],
>        [2006,    1,    3,   10,    9, 99 , 50],
>        [2006,    1,    3,   10,   40, 99, 50],
>        ...,
>        [2008,    3,   20,   10,   27, 99, 50 ],
>        [2008,    3,   20,   10,   51, 99, 50],
>        [2008,    3,   20,   12,    2, 99, 50]])

>>> import numpy as np
>>> x = np.ones((10,3))
>>> y = np.ones((10,))*2
>>> z = np.ones((10,))*3
>>> np.hstack([x,y[:,None],z[:,None]])
array([[ 1.,  1.,  1.,  2.,  3.],
       [ 1.,  1.,  1.,  2.,  3.],
       [ 1.,  1.,  1.,  2.,  3.],
       [ 1.,  1.,  1.,  2.,  3.],
       [ 1.,  1.,  1.,  2.,  3.],
       [ 1.,  1.,  1.,  2.,  3.],
       [ 1.,  1.,  1.,  2.,  3.],
       [ 1.,  1.,  1.,  2.,  3.],
       [ 1.,  1.,  1.,  2.,  3.],
       [ 1.,  1.,  1.,  2.,  3.]])
>>>

hth,
Alan Isaac



















































From mhearne at usgs.gov  Fri Jun  6 12:44:44 2008
From: mhearne at usgs.gov (Michael Hearne)
Date: Fri, 06 Jun 2008 10:44:44 -0600
Subject: [SciPy-user] Adding new columns to 2D array & populating them
 from	another	array.
In-Reply-To: <484920310200009B00013308@GWWEB.twdb.state.tx.us>
References: <484920310200009B00013308@GWWEB.twdb.state.tx.us>
Message-ID: <4849697C.3010607@usgs.gov>

Here's a simple example that I think demonstrates what you want to do:
a = zeros((4,5))
b = ones((4,1))
c = hstack((a,b))

--Mike

Dharhas Pothina wrote:
> Hi,
>
> I have an array 'a'
>
> a = 
> array([[2006,    1,    3,    9,   40],
>        [2006,    1,    3,   10,    9],
>        [2006,    1,    3,   10,   40],
>        ..., 
>        [2008,    3,   20,   10,   27],
>        [2008,    3,   20,   10,   51],
>        [2008,    3,   20,   12,    2]])
>
> where a.shape = (420, 5)
>
> I have other arrays b,c etc of the shape (420,) (say b contains all 99's and c contains all 50's)
>
> how to I add the new arrays to 'a' to form an array 
>
>   array([[2006,    1,    3,    9,   40 , 99, 50],
>        [2006,    1,    3,   10,    9, 99 , 50],
>        [2006,    1,    3,   10,   40, 99, 50],
>        ...,
>        [2008,    3,   20,   10,   27, 99, 50 ],
>        [2008,    3,   20,   10,   51, 99, 50],
>        [2008,    3,   20,   12,    2, 99, 50]])
>
> I've tried hstack but I'm not sure how to use it with 2D arrays.
>
> thanks
>
> - dharhas
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>   

-- 
------------------------------------------------------
Michael Hearne
mhearne at usgs.gov
(303) 273-8620
USGS National Earthquake Information Center
1711 Illinois St. Golden CO 80401
Senior Software Engineer
Synergetics, Inc.
------------------------------------------------------



From strawman at astraw.com  Fri Jun  6 12:45:08 2008
From: strawman at astraw.com (Andrew Straw)
Date: Fri, 06 Jun 2008 09:45:08 -0700
Subject: [SciPy-user] Call for contributions: Python in Neuroscience
Message-ID: <48496994.3030301@astraw.com>

>From http://www.neuroinf.org/pipermail/comp-neuro/2008-June/000858.html

CALL FOR CONTRIBUTIONS

Special Section: "Python in Neuroscience"
Journal: Frontiers in Neuroinformatics
[www.frontiersin.org/neuroinformatics]

Associate Editor:
Rolf K?tter (Radboud University Medical Centre Nijmegen, Netherlands)

Guest Editors:

James A. Bednar (University of Edinburgh, UK)
Andrew Davison (UNIC, CNRS, France)
Markus Diesmann (RIKEN Brain Science Institute, Japan)
Marc-Oliver Gewaltig (Honda Research Institute Europe GmbH, Germany)
Michael Hines (Yale University, USA)
Eilif Muller (LCN-EPFL, Switzerland)


Important Dates:

Abstract/outline submission deadline: June 14th, 2008.
Invitations for full paper submissions sent by June 21st, 2008.
Invited full paper submission deadline: September 14th, 2008.


Special Section Abstract:

Python is rapidly becoming the de facto standard language for systems
integration.  Python has a large user and developer-base external to
the neuroscience community, and a vast module library that facilitates
rapid and maintainable development of complex and intricate systems.

In this special section, we highlight recent efforts to develop Python
modules for the domain of neuroscience software and neuroinformatics:

- simulators and simulator interfaces
- data collection and analysis
- sharing, re-use, storage and databasing of models and data
- stimulus generation
- parameter search and optimization
- visualization
- VLSI hardware interfacing
- ...

Moreover, we seek to provide a representative overview of existing
mature Python modules for neuroscience and neuroinformatics, to
demonstrate a critical mass and show that Python is an appropriate
choice of interpreter interface for future neuroscience software
development.

Submission Procedure:

Researchers and practitioners are invited to submit on or before June
14th, 2008 a max. 1 page abstract/outline of work related to the focus
of the special section to eilif.mueller at epfl.ch, CC'd to rk at
cns.umcn.nl
for consideration for inclusion as an elaborated full article in the
special section.

Please include a provisional title, a full author list, and format the
subject of your email as follows: "[python SI] outline - Your Name".

Authors will be notified whether their contribution has been accepted
by June 21st, 2008.


Full Article Information:

* Full articles will be solicited by invitation only, based on the
abstracts/outlines we receive by June 14th, 2008.

* The deadline for submission of invited full articles will be
September 14th, 2008.

* Article formatting will be as for standard Frontiers "Original
Research Articles".  Guidelines and instructions for their preparation
can be found at
http://frontiersin.org/authorinstructions#manuscriptGuidelines.

* General author instructions for Frontiers in Neuroinformatics can be
found at http://frontiersin.org/authorinstructions/.

* Frontiers in Neuroinformatics is an open access journal, following a
pay-for-publication model. Details of the publication fees can be found
at http://www.frontiersin.org/publicationfees/.

* Further details will be provided to authors of accepted abstracts by
June 21st, 2008.


From tgray at protozoic.com  Fri Jun  6 13:15:53 2008
From: tgray at protozoic.com (Tim Gray)
Date: Fri, 6 Jun 2008 13:15:53 -0400 (EDT)
Subject: [SciPy-user]  Adding new columns to 2D array & populating them
In-Reply-To: <mailman.11.1212771669.26596.scipy-user@scipy.org>
References: <mailman.11.1212771669.26596.scipy-user@scipy.org>
Message-ID: <alpine.OSX.1.10.0806061312250.3174@d228.scdc1.swarthmore.edu>

So what is the difference between the c_ operator and hstack?  Or r_ and vstack?  I'm not too clear on the c_ and r_ operators.  Is there an efficiency gain related to using them?  Do the do something special?


From Dharhas.Pothina at twdb.state.tx.us  Fri Jun  6 13:16:10 2008
From: Dharhas.Pothina at twdb.state.tx.us (Dharhas Pothina)
Date: Fri, 06 Jun 2008 12:16:10 -0500
Subject: [SciPy-user] Adding new columns to 2D array &
	populating	themfrom	another	array.
In-Reply-To: <4849697C.3010607@usgs.gov>
References: <484920310200009B00013308@GWWEB.twdb.state.tx.us>
	<4849697C.3010607@usgs.gov>
Message-ID: <48492A8A.63BA.009B.0@twdb.state.tx.us>


Thanks to all of you. I used the c_[] technique.

- dharhas

>>> Michael Hearne <mhearne at usgs.gov> 6/6/2008 11:44 AM >>>
Here's a simple example that I think demonstrates what you want to do:
a = zeros((4,5))
b = ones((4,1))
c = hstack((a,b))

--Mike

Dharhas Pothina wrote:
> Hi,
>
> I have an array 'a'
>
> a = 
> array([[2006,    1,    3,    9,   40],
>        [2006,    1,    3,   10,    9],
>        [2006,    1,    3,   10,   40],
>        ..., 
>        [2008,    3,   20,   10,   27],
>        [2008,    3,   20,   10,   51],
>        [2008,    3,   20,   12,    2]])
>
> where a.shape = (420, 5)
>
> I have other arrays b,c etc of the shape (420,) (say b contains all 99's and c contains all 50's)
>
> how to I add the new arrays to 'a' to form an array 
>
>   array([[2006,    1,    3,    9,   40 , 99, 50],
>        [2006,    1,    3,   10,    9, 99 , 50],
>        [2006,    1,    3,   10,   40, 99, 50],
>        ...,
>        [2008,    3,   20,   10,   27, 99, 50 ],
>        [2008,    3,   20,   10,   51, 99, 50],
>        [2008,    3,   20,   12,    2, 99, 50]])
>
> I've tried hstack but I'm not sure how to use it with 2D arrays.
>
> thanks
>
> - dharhas
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org 
> http://projects.scipy.org/mailman/listinfo/scipy-user 
>   

-- 
------------------------------------------------------
Michael Hearne
mhearne at usgs.gov 
(303) 273-8620
USGS National Earthquake Information Center
1711 Illinois St. Golden CO 80401
Senior Software Engineer
Synergetics, Inc.
------------------------------------------------------

_______________________________________________
SciPy-user mailing list
SciPy-user at scipy.org 
http://projects.scipy.org/mailman/listinfo/scipy-user



From lists at vrbka.net  Fri Jun  6 13:47:58 2008
From: lists at vrbka.net (Lubos Vrbka)
Date: Fri, 06 Jun 2008 19:47:58 +0200
Subject: [SciPy-user] Fourier-sine transform with SciPy
In-Reply-To: <ce557a360806060242l37941515m2bff328ebe20e07d@mail.gmail.com>
References: <4848E2FE.9070402@vrbka.net>	
	<ce557a360806060144i17a4243eu8c328b0c43b4e99f@mail.gmail.com>	
	<484900B8.6080006@vrbka.net>
	<ce557a360806060242l37941515m2bff328ebe20e07d@mail.gmail.com>
Message-ID: <4849784E.1020306@vrbka.net>

anne, thanks for detailed discussion (and sorry for accidentally 
replying directly to you instead to list)!

 > Note that the sine transform only works on odd functions, so either
 > your input data should have f_k = f_{N-k} or you are representing only
 > half the values, i.e., N is twice the number of data points you have.
 > This too is somewhere a proper sine transform would help, if we had
 > it.
well, the original function itself is neither even nor odd, since it is 
defined only in the interval <0, +inf>. at certain point in the 
derivation, the original function is multiplied by the r or k variable, 
so it might actually be odd. the sine comes from the fact that the 
original function is actually radially symmetric. i think i will have to 
have a look into that an check the result.

hopefully the last question. how is it with the 'normalization 
constants' for the dft? from theory it should be T/N for FT and 1/T for 
iFT, but different packages treat this differently (for example fftw 
does not use them and it has to be done afterwards 'manually') - how is 
it in case of scipy.fft?

best,
lubos


-- 
Lubos _ at _"
http://www.lubos.vrbka.net


From ivo.maljevic at gmail.com  Fri Jun  6 13:49:23 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Fri, 6 Jun 2008 13:49:23 -0400
Subject: [SciPy-user] Adding new columns to 2D array & populating them
In-Reply-To: <alpine.OSX.1.10.0806061312250.3174@d228.scdc1.swarthmore.edu>
References: <mailman.11.1212771669.26596.scipy-user@scipy.org>
	<alpine.OSX.1.10.0806061312250.3174@d228.scdc1.swarthmore.edu>
Message-ID: <826c64da0806061049t678cedf0iff71417b6046ebcb@mail.gmail.com>

The short answer is: they do not appear to be exactly the same.

The longer answer (there are may who will know this way better than I do),
can be found if you look at the source code in :

/usr/lib/python2.5/site-packages/numpy/lib

and then open file

index_tricks.py

You will find the concatenator class that does a whole bunch of things, and
then r_class and c_class which inherit concatenator. c_ and r_ are just
"attached" to those classes.


2008/6/6 Tim Gray <tgray at protozoic.com>:

> So what is the difference between the c_ operator and hstack?  Or r_ and
> vstack?  I'm not too clear on the c_ and r_ operators.  Is there an
> efficiency gain related to using them?  Do the do something special?
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080606/7a4cb147/attachment.html>

From lists at vrbka.net  Sat Jun  7 09:54:53 2008
From: lists at vrbka.net (Lubos Vrbka)
Date: Sat, 07 Jun 2008 15:54:53 +0200
Subject: [SciPy-user] Fourier-sine transform with SciPy
In-Reply-To: <4849616F.3050009@enthought.com>
References: <4848E2FE.9070402@vrbka.net> <4849616F.3050009@enthought.com>
Message-ID: <484A932D.7000500@vrbka.net>

hi everybody!

> In the scipy sandbox under image you can find a transforms.py file which 
> contains an implementation of the DST (discrete sine transform).
travis, thanks for the code you sent me. actually, under my version of 
scipy/numpy (0.5.2/1.0.1) it doesn't work, since
Xt = np.fft(xtilde,axis=axis)
isn't a callable object. i had to use
Xt = np.fft.fft(xtilde,axis=axis)
and its equivalent in the inverse sinus transform.

regarding the implementation - it seems to me that to get optimal 
performance, the length of the original array should be 2^m - 1. if i 
have, e.g., 4096 points in the array, then newly constructed one will 
have 8194 points and fft of the length of 16194 will have to be done. is 
this correct?

i know that the following question isn't particularly related to scipy 
itself, but if anybody would be willing to explain these things to me, 
i'd be really grateful.

let's assume the distance (r)/reciprocal distance (k) pair. having my 
odd real function f(r) discretized on <0,L>, i have dr = L/N. the value 
of dk should be then dk = 2*pi/L = 2*pi/(N*dr).

actually, where does the value of N+1 come from (it's mentioned on the 
wiki page http://en.wikipedia.org/wiki/Discrete_sine_transform )?

according to the sampling theorem, the wave vectors corresponding to 0, 
1, ... N/2 should represent the function f in the fourier space, the 
rest of the wave vectors are redundant.

what happens in the case of the dst() routine? is the interval extended 
to <-L,L>, with the dk = 2*pi/2*L? but then, for N points of original 
array, i get N non-redundant values. how does this compare to the 
sampling theorem?

since i need to operate on the ft'ed function before transforming it 
back, and this operation involves the value of dk. i need to know what 
the value of dk really is.

in one implementation of my problem (not using fft), the original 
function was paddded with zeros to the length of 2N and the sum over the 
products f(r)sin(kr) was performed... how does then this compare to the 
dst() above?

any comments/pointers to relevant literature are more than welcome.

with best regards,
lubos

-- 
Lubos _ at _"
http://www.lubos.vrbka.net


From dmitrey.kroshko at scipy.org  Sat Jun  7 10:58:55 2008
From: dmitrey.kroshko at scipy.org (dmitrey)
Date: Sat, 07 Jun 2008 17:58:55 +0300
Subject: [SciPy-user] "from scipy.linalg.flapack import dgelss" fails
Message-ID: <484AA22F.7070508@scipy.org>

I have problems with the statement
from scipy.linalg.flapack import dgelss

Traceback (innermost last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python2.5/site-packages/scipy/linalg/__init__.py", line 
13, in <module>
    from iterative import *
  File "/usr/lib/python2.5/site-packages/scipy/linalg/iterative.py", 
line 5, in <module>
    from scipy.sparse.linalg import isolve
  File "/usr/lib/python2.5/site-packages/scipy/sparse/__init__.py", line 
5, in <module>
    from base import *
  File "/usr/lib/python2.5/site-packages/scipy/sparse/base.py", line 45, 
in <module>
    class spmatrix(object):
  File "/usr/lib/python2.5/site-packages/scipy/sparse/base.py", line 
139, in spmatrix
    @deprecate
TypeError: deprecate() takes exactly 3 arguments (1 given)

at least with latest scipy from svn

 >>> scipy.__version__
'0.7.0.dev4416'

Do you have the same?
This one was working correctly in previous scipy versions

Regards, D.


From nwagner at iam.uni-stuttgart.de  Sat Jun  7 11:11:10 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Sat, 07 Jun 2008 17:11:10 +0200
Subject: [SciPy-user] "from scipy.linalg.flapack import dgelss" fails
In-Reply-To: <484AA22F.7070508@scipy.org>
References: <484AA22F.7070508@scipy.org>
Message-ID: <web-102972971@uni-stuttgart.de>

On Sat, 07 Jun 2008 17:58:55 +0300
  dmitrey <dmitrey.kroshko at scipy.org> wrote:
> I have problems with the statement
> from scipy.linalg.flapack import dgelss
> 
> Traceback (innermost last):
>  File "<stdin>", line 1, in <module>
>  File 
>"/usr/lib/python2.5/site-packages/scipy/linalg/__init__.py", 
>line 
> 13, in <module>
>    from iterative import *
>  File 
>"/usr/lib/python2.5/site-packages/scipy/linalg/iterative.py", 
> line 5, in <module>
>    from scipy.sparse.linalg import isolve
>  File 
>"/usr/lib/python2.5/site-packages/scipy/sparse/__init__.py", 
>line 
> 5, in <module>
>    from base import *
>  File 
>"/usr/lib/python2.5/site-packages/scipy/sparse/base.py", 
>line 45, 
> in <module>
>    class spmatrix(object):
>  File 
>"/usr/lib/python2.5/site-packages/scipy/sparse/base.py", 
>line 
> 139, in spmatrix
>    @deprecate
> TypeError: deprecate() takes exactly 3 arguments (1 
>given)
> 
> at least with latest scipy from svn
> 
> >>> scipy.__version__
> '0.7.0.dev4416'
> 
> Do you have the same?
> This one was working correctly in previous scipy 
>versions
> 
> Regards, D.
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user

No problem here

/usr/bin/python
Python 2.4 (#1, Oct 13 2006, 17:13:31)
[GCC 3.3.5 20050117 (prerelease) (SUSE Linux)] on linux2
Type "help", "copyright", "credits" or "license" for more 
information.
>>> from scipy.linalg.flapack import dgelss
>>> import scipy
>>> scipy.__version__
'0.7.0.dev4416'
>>> help (dgelss)
  


From dmitrey.kroshko at scipy.org  Sat Jun  7 11:26:25 2008
From: dmitrey.kroshko at scipy.org (dmitrey)
Date: Sat, 07 Jun 2008 18:26:25 +0300
Subject: [SciPy-user] "from scipy.linalg.flapack import dgelss" fails
In-Reply-To: <web-102972971@uni-stuttgart.de>
References: <484AA22F.7070508@scipy.org> <web-102972971@uni-stuttgart.de>
Message-ID: <484AA8A1.5050901@scipy.org>

IIRC Python 2.4 have no decorators (like the "@deprecate" line), they 
are available since 2.5 (that I have).
Could anyone else try  "from scipy.linalg.flapack import dgelss" from 
python2.5?
Regards, D.

Nils Wagner wrote:

> On Sat, 07 Jun 2008 17:58:55 +0300
>   dmitrey <dmitrey.kroshko at scipy.org> wrote:
>   
>> "/usr/lib/python2.5/site-packages/scipy/sparse/base.py", 
>> line 
>> 139, in spmatrix
>>    @deprecate
>> TypeError: deprecate() takes exactly 3 arguments (1 
>> given)
>>
>> at least with latest scipy from svn
>>
>>     
>>>>> scipy.__version__
>>>>>           
>> '0.7.0.dev4416'
>>
>> Do you have the same?
>> This one was working correctly in previous scipy 
>> versions
>>
>> Regards, D.
>> _______________________________________________
>> SciPy-user mailing list
>> SciPy-user at scipy.org
>> http://projects.scipy.org/mailman/listinfo/scipy-user
>>     
>
> No problem here
>
> /usr/bin/python
> Python 2.4 (#1, Oct 13 2006, 17:13:31)
> [GCC 3.3.5 20050117 (prerelease) (SUSE Linux)] on linux2
> Type "help", "copyright", "credits" or "license" for more 
> information.
>   
>>>> from scipy.linalg.flapack import dgelss
>>>> import scipy
>>>> scipy.__version__
>>>>         
> '0.7.0.dev4416'
>   
>>>> help (dgelss)
>>>>         
>   
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
>
>
>   



From matthieu.brucher at gmail.com  Sat Jun  7 11:31:21 2008
From: matthieu.brucher at gmail.com (Matthieu Brucher)
Date: Sat, 7 Jun 2008 17:31:21 +0200
Subject: [SciPy-user] "from scipy.linalg.flapack import dgelss" fails
In-Reply-To: <484AA22F.7070508@scipy.org>
References: <484AA22F.7070508@scipy.org>
Message-ID: <e76aa17f0806070831q166fb19eh483749d26d6bbe0f@mail.gmail.com>

Hi,

Please update to the latest Numpy (this error showed up several times in
this ML).

Matthieu

2008/6/7 dmitrey <dmitrey.kroshko at scipy.org>:

> I have problems with the statement
> from scipy.linalg.flapack import dgelss
>
> Traceback (innermost last):
>  File "<stdin>", line 1, in <module>
>  File "/usr/lib/python2.5/site-packages/scipy/linalg/__init__.py", line
> 13, in <module>
>    from iterative import *
>  File "/usr/lib/python2.5/site-packages/scipy/linalg/iterative.py",
> line 5, in <module>
>    from scipy.sparse.linalg import isolve
>  File "/usr/lib/python2.5/site-packages/scipy/sparse/__init__.py", line
> 5, in <module>
>    from base import *
>  File "/usr/lib/python2.5/site-packages/scipy/sparse/base.py", line 45,
> in <module>
>    class spmatrix(object):
>  File "/usr/lib/python2.5/site-packages/scipy/sparse/base.py", line
> 139, in spmatrix
>    @deprecate
> TypeError: deprecate() takes exactly 3 arguments (1 given)
>
> at least with latest scipy from svn
>
>  >>> scipy.__version__
> '0.7.0.dev4416'
>
> Do you have the same?
> This one was working correctly in previous scipy versions
>
> Regards, D.
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>



-- 
French PhD student
Website : http://matthieu-brucher.developpez.com/
Blogs : http://matt.eifelle.com and http://blog.developpez.com/?blog=92
LinkedIn : http://www.linkedin.com/in/matthieubrucher
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080607/50b100ac/attachment.html>

From dmitrey.kroshko at scipy.org  Sat Jun  7 11:38:17 2008
From: dmitrey.kroshko at scipy.org (dmitrey)
Date: Sat, 07 Jun 2008 18:38:17 +0300
Subject: [SciPy-user] "from scipy.linalg.flapack import dgelss" fails
In-Reply-To: <e76aa17f0806070831q166fb19eh483749d26d6bbe0f@mail.gmail.com>
References: <484AA22F.7070508@scipy.org>
	<e76aa17f0806070831q166fb19eh483749d26d6bbe0f@mail.gmail.com>
Message-ID: <484AAB69.4030403@scipy.org>

Matthieu Brucher wrote:
> Hi,
>
> Please update to the latest Numpy (this error showed up several times 
> in this ML).
so this will require numpy download & installation from scratch, because 
currently I use numpy 1.0.4 from KUBUNTU 8.04 distribution. Ok, I'll try.
D.


From pav at iki.fi  Sat Jun  7 12:13:38 2008
From: pav at iki.fi (Pauli Virtanen)
Date: Sat, 7 Jun 2008 16:13:38 +0000 (UTC)
Subject: [SciPy-user] "from scipy.linalg.flapack import dgelss" fails
References: <484AA22F.7070508@scipy.org> <web-102972971@uni-stuttgart.de>
	<484AA8A1.5050901@scipy.org>
Message-ID: <g2ec3i$6h2$1@ger.gmane.org>

Sat, 07 Jun 2008 18:26:25 +0300, dmitrey wrote:

> IIRC Python 2.4 have no decorators (like the "@deprecate" line), they
> are available since 2.5 (that I have). Could anyone else try  "from
> scipy.linalg.flapack import dgelss" from python2.5?

Decorators were introduced already in Python 2.4.

The problem you see is the changed semantics in @deprecate between numpy 
1.0.4 and 1.1.0. (Scipy 0.7.dev uses the 1.1.0 semantics). It can be 
fixed by upgrading to numpy 1.1.0.

-- 
Pauli Virtanen



From contact at pythonxy.com  Sat Jun  7 14:52:18 2008
From: contact at pythonxy.com (Pierre Raybaut)
Date: Sat, 07 Jun 2008 20:52:18 +0200
Subject: [SciPy-user] [ Python(x,y) ] New release : 1.2.5
Message-ID: <484AD8E2.4080302@pythonxy.com>

Hi all,

Python(x,y) 1.2.5 is now available on http://www.pythonxy.com.

Changes history
06 -07 -2008 - Version 1.2.5 :

     * Updated:
           o Parallel Python 1.5.4
     * Added:
           o MDP 2.3 - Modular toolkit for Data Processing (MDP): data 
processing framework (see included algorithms on website)
           o Automatic logging management for IPython/Matplotlib 
consoles ("logs" folder added to the Python(x,y) base directory, 
associated shortcuts in start menu and "Welcome to Python(x,y)" GUI) - 
See screenshots
           o The (lightweight) default editor in IPython consoles 
(syntax: "edit script.py" or "ed script.py") is now set to Notepad++ - 
See screenshots
           o Windows explorer integration: "Edit with Notepad++" has 
been added to .py and .pyw context menu
           o StartExplorer Eclipse Plug-in
     * Corrected:
           o Some NumPy and SciPy unit tests were not successfull

Regards,
Pierre Raybaut







From mattknox.ca at gmail.com  Sun Jun  8 16:50:43 2008
From: mattknox.ca at gmail.com (Matt Knox)
Date: Sun, 8 Jun 2008 20:50:43 +0000 (UTC)
Subject: [SciPy-user] scikits.timeseries AttributeError
References: <1f481223-faa2-4675-98c0-099d86956a69@z16g2000prn.googlegroups.com>
	<loom.20080606T022704-444@post.gmane.org>
Message-ID: <loom.20080608T204934-639@post.gmane.org>

Matt Knox <mattknox.ca <at> gmail.com> writes:

> 
> >> dates = ts.date_array([q[1] for q in data], freq='S')
> 
> This is one of the known limitations of the timeseries module right now
> unfortunately. Plotting for frequencies higher than daily is not currently
> supported.

I just added support for this today in the timeseries module if you want to try
it again.

- Matt



From gary.pajer at gmail.com  Sun Jun  8 18:21:00 2008
From: gary.pajer at gmail.com (Gary Pajer)
Date: Sun, 8 Jun 2008 18:21:00 -0400
Subject: [SciPy-user] LabVIEW and LabPython
In-Reply-To: <ff223aae0806031923u1dd2c613k7ee8e5f27c1753e8@mail.gmail.com>
References: <ff223aae0806031923u1dd2c613k7ee8e5f27c1753e8@mail.gmail.com>
Message-ID: <88fe22a0806081521j2db5b1eblb68692af2e3beb43@mail.gmail.com>

I was unaware of LabPython until this post.  But I can report that I've been
using National Instruments' DAQmx library through ctypes, and it seems to
work perfectly.  I don't know how far along you are in development,  but
there's an option for you.

I've married DAQmx to scipy/numpy to Traits UI (
http://code.enthought.com/projects/traits ) and I have a beautiful GUI based
data acquisition/display/analysis system.  I'll admit that I've been too
scared to even try to implement callbacks (used by a small number of DAQmx C
functions) but I've achieved the same functionality using python threads.
Aside from that caveat, all features have worked as they should.

-gary



On Tue, Jun 3, 2008 at 10:23 PM, Glen Shennan <glen.shennan at gmail.com>
wrote:

> Hi,
>
> I am trying to use a Python program in LabVIEW using LabPython.  My program
> works fine outside of LabVIEW but not when compiled and run in LabVIEW via
> LabPython.  I am using Scipy and Numpy functions in the program but I think
> they are causing the problems.  Does anyone know if it is possible to use
> Scipy with LabPython?
>
> Thanks,
> Glen
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080608/ac0d2228/attachment.html>

From bernardo.rocha at meduni-graz.at  Mon Jun  9 04:51:57 2008
From: bernardo.rocha at meduni-graz.at (Bernardo Martins Rocha)
Date: Mon, 09 Jun 2008 10:51:57 +0200
Subject: [SciPy-user] Reading binary files
Message-ID: <484D0B4D0200002A0001558D@si062.meduni-graz.at>

Hello,

I was trying to write a script in order to read a binary file. Then, I started looking for a python function to do it. First, I came up with scipy.io.fopen, which was OK. The only problem was that whenever I ran the script I got a message warning me that this class is deprecated, which is really annoying. 

Then I had to change to the new class for reading binary files: scipy.io.npfile

And it took me some time to change the script properly...so I'm posting what I did in order to describe my experience...and I'm open for suggestions that could improve this script.

----- fopen -------------------------------
from scipy.io.fopen import fopen

fh = fopen (filename,'r',fmt) # fmt is 'l' or 'B` (little or big endian)
header = fh.read(1024,'char')
slice_buf = fh.read(slice_size, dtype)
fh.close()

----- npfile -------------------------------
from scipy.io.npfile import npfile

fh = npfile (filename,'r',fmt)
dt = numpy.dtype('uint8')
header = fh.read_array(dt, 1024)
dt = numpy.dtype(dtype) # dtype = 'float32'
slice_buf = fh.read_array(dt, slice_size)
fh.close()

Bye!
Bernardo M. Rocha


From s.mientki at ru.nl  Mon Jun  9 06:51:24 2008
From: s.mientki at ru.nl (Stef Mientki)
Date: Mon, 09 Jun 2008 12:51:24 +0200
Subject: [SciPy-user] LabVIEW and LabPython
In-Reply-To: <88fe22a0806081521j2db5b1eblb68692af2e3beb43@mail.gmail.com>
References: <ff223aae0806031923u1dd2c613k7ee8e5f27c1753e8@mail.gmail.com>
	<88fe22a0806081521j2db5b1eblb68692af2e3beb43@mail.gmail.com>
Message-ID: <484D0B2C.8080505@ru.nl>



Gary Pajer wrote:
> I was unaware of LabPython until this post.  But I can report that 
> I've been using National Instruments' DAQmx library through ctypes, 
> and it seems to work perfectly.  I don't know how far along you are in 
> development,  but there's an option for you.
>
> I've married DAQmx to scipy/numpy to Traits UI ( 
> http://code.enthought.com/projects/traits ) and I have a beautiful GUI 
> based data acquisition/display/analysis system.  I'll admit that I've 
> been too scared to even try to implement callbacks (used by a small 
> number of DAQmx C functions) but I've achieved the same functionality 
> using python threads.  Aside from that caveat, all features have 
> worked as they should.
hi Gary,
This sounds very very interesting.
As I'm writing an opensource replacement for LabView / MatLab in Python 
(called PyLab_Works),
and DAQmx is one of the major AD-systems in this package,
I'm anxious to get DAQmx libraries (as till now I have some 
clumsy-windows-only DLL).
Any change I could use your code ?

thanks,
Stef Mientki

>
> -gary
>
>
>
> On Tue, Jun 3, 2008 at 10:23 PM, Glen Shennan <glen.shennan at gmail.com 
> <mailto:glen.shennan at gmail.com>> wrote:
>
>     Hi,
>
>     I am trying to use a Python program in LabVIEW using LabPython. 
>     My program works fine outside of LabVIEW but not when compiled and
>     run in LabVIEW via LabPython.  I am using Scipy and Numpy
>     functions in the program but I think they are causing the
>     problems.  Does anyone know if it is possible to use Scipy with
>     LabPython?
>
>     Thanks,
>     Glen
>
>     _______________________________________________
>     SciPy-user mailing list
>     SciPy-user at scipy.org <mailto:SciPy-user at scipy.org>
>     http://projects.scipy.org/mailman/listinfo/scipy-user
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>   


Het UMC St Radboud staat geregistreerd bij de Kamer van Koophandel in het handelsregister onder nummer 41055629.
The Radboud University Nijmegen Medical Centre is listed in the Commercial Register of the Chamber of Commerce under file number 41055629.




From mwojc at p.lodz.pl  Mon Jun  9 09:50:53 2008
From: mwojc at p.lodz.pl (Marek Wojciechowski)
Date: Mon, 9 Jun 2008 15:50:53 +0200
Subject: [SciPy-user] optimize.fmin_l_bfgs_b problem
Message-ID: <200806091550.53753.mwojc@p.lodz.pl>

Hi!
The following command:
    optimize.fmin_l_bfgs_b(lambda x: x[0]**2, [-1.], iprint=1)
causes an error and breaks the python session with the following output:

RUNNING THE L-BFGS-B CODE

           * * *

At line 2647 of file scipy/optimize/lbfgsb/routines.f
Internal Error: printf is broken
Machine precision =

This occurs on scipy-0.6.0 and python 2.4 under Gentoo Linux. Is this the 
known bug?

Greetings,
-- 
Marek Wojciechowski


From nwagner at iam.uni-stuttgart.de  Mon Jun  9 10:12:04 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Mon, 09 Jun 2008 16:12:04 +0200
Subject: [SciPy-user] optimize.fmin_l_bfgs_b problem
In-Reply-To: <200806091550.53753.mwojc@p.lodz.pl>
References: <200806091550.53753.mwojc@p.lodz.pl>
Message-ID: <web-103041015@uni-stuttgart.de>

On Mon, 9 Jun 2008 15:50:53 +0200
  Marek Wojciechowski <mwojc at p.lodz.pl> wrote:
> Hi!
> The following command:
>    optimize.fmin_l_bfgs_b(lambda x: x[0]**2, [-1.], 
>iprint=1)
> causes an error and breaks the python session with the 
>following output:
> 
> RUNNING THE L-BFGS-B CODE
> 
>           * * *
> 
> At line 2647 of file scipy/optimize/lbfgsb/routines.f
> Internal Error: printf is broken
> Machine precision =
> 
> This occurs on scipy-0.6.0 and python 2.4 under Gentoo 
>Linux. Is this the 
> known bug?
> 
> Greetings,
> -- 
> Marek Wojciechowski

I get

>>> optimize.fmin_l_bfgs_b(lambda x: x[0]**2, [-1.], iprint=1)
RUNNING THE L-BFGS-B CODE

            * * *

Machine precision = 2.220E-16
  N =  1    M =  10
  This problem is unconstrained.

At X0         0 variables are exactly at the bounds
Traceback (most recent call last):
   File "<stdin>", line 1, in <module>
   File 
"/data/home/nwagner/local/lib/python2.5/site-packages/scipy/optimize/lbfgsb.py", 
line 205, in fmin_l_bfgs_b
     f, g = func_and_grad(x)
   File 
"/data/home/nwagner/local/lib/python2.5/site-packages/scipy/optimize/lbfgsb.py", 
line 156, in func_and_grad
     f, g = func(x, *args)
TypeError: 'numpy.float64' object is not iterable

>>> scipy.__version__
'0.7.0.dev4420'


  


From gary.pajer at gmail.com  Mon Jun  9 22:41:41 2008
From: gary.pajer at gmail.com (Gary Pajer)
Date: Mon, 9 Jun 2008 22:41:41 -0400
Subject: [SciPy-user] LabVIEW and LabPython
In-Reply-To: <484D0B2C.8080505@ru.nl>
References: <ff223aae0806031923u1dd2c613k7ee8e5f27c1753e8@mail.gmail.com>
	<88fe22a0806081521j2db5b1eblb68692af2e3beb43@mail.gmail.com>
	<484D0B2C.8080505@ru.nl>
Message-ID: <88fe22a0806091941o4b61750cn32dff2d311c5381e@mail.gmail.com>

On Mon, Jun 9, 2008 at 6:51 AM, Stef Mientki <s.mientki at ru.nl> wrote:

>
>
> Gary Pajer wrote:
> > I was unaware of LabPython until this post.  But I can report that
> > I've been using National Instruments' DAQmx library through ctypes,
> > and it seems to work perfectly.  I don't know how far along you are in
> > development,  but there's an option for you.
> >
> > I've married DAQmx to scipy/numpy to Traits UI (
> > http://code.enthought.com/projects/traits ) and I have a beautiful GUI
> > based data acquisition/display/analysis system.  I'll admit that I've
> > been too scared to even try to implement callbacks (used by a small
> > number of DAQmx C functions) but I've achieved the same functionality
> > using python threads.  Aside from that caveat, all features have
> > worked as they should.
> hi Gary,
> This sounds very very interesting.
> As I'm writing an opensource replacement for LabView / MatLab in Python
> (called PyLab_Works),
> and DAQmx is one of the major AD-systems in this package,
> I'm anxious to get DAQmx libraries (as till now I have some
> clumsy-windows-only DLL).
> Any change I could use your code ?


I'd be happy to share what I've done, but let me make clear that I am using
NI's DAQmx Windows dll.    I'm just calling the functions in there using
ctypes.   I have a feeling that you are after something different.

I'm not sure what you have and what you want, but if I read between your
lines, I think you might be looking for Linux libraries.  NI does have Linux
version with reduced functionality available for download.  License may be
an issue.  Would Comedi help?

If we get any deeper, we might want to take the discussion off-line.



>
>
> thanks,
> Stef Mientki


-gary


>
>
> >
> > -gary
> >
> >
> >
> > On Tue, Jun 3, 2008 at 10:23 PM, Glen Shennan <glen.shennan at gmail.com
> > <mailto:glen.shennan at gmail.com>> wrote:
> >
> >     Hi,
> >
> >     I am trying to use a Python program in LabVIEW using LabPython.
> >     My program works fine outside of LabVIEW but not when compiled and
> >     run in LabVIEW via LabPython.  I am using Scipy and Numpy
> >     functions in the program but I think they are causing the
> >     problems.  Does anyone know if it is possible to use Scipy with
> >     LabPython?
> >
> >     Thanks,
> >     Glen
> >
> >     _______________________________________________
> >     SciPy-user mailing list
> >     SciPy-user at scipy.org <mailto:SciPy-user at scipy.org>
> >     http://projects.scipy.org/mailman/listinfo/scipy-user
> >
> >
> > ------------------------------------------------------------------------
> >
> > _______________________________________________
> > SciPy-user mailing list
> > SciPy-user at scipy.org
> > http://projects.scipy.org/mailman/listinfo/scipy-user
> >
>
>
> Het UMC St Radboud staat geregistreerd bij de Kamer van Koophandel in het
> handelsregister onder nummer 41055629.
> The Radboud University Nijmegen Medical Centre is listed in the Commercial
> Register of the Chamber of Commerce under file number 41055629.
>
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080609/1966940c/attachment.html>

From strawman at astraw.com  Tue Jun 10 00:01:33 2008
From: strawman at astraw.com (Andrew Straw)
Date: Mon, 09 Jun 2008 21:01:33 -0700
Subject: [SciPy-user] LabVIEW and LabPython
In-Reply-To: <88fe22a0806091941o4b61750cn32dff2d311c5381e@mail.gmail.com>
References: <ff223aae0806031923u1dd2c613k7ee8e5f27c1753e8@mail.gmail.com>	<88fe22a0806081521j2db5b1eblb68692af2e3beb43@mail.gmail.com>	<484D0B2C.8080505@ru.nl>
	<88fe22a0806091941o4b61750cn32dff2d311c5381e@mail.gmail.com>
Message-ID: <484DFC9D.4000305@astraw.com>

Gary Pajer wrote:
> On Mon, Jun 9, 2008 at 6:51 AM, Stef Mientki <s.mientki at ru.nl
> <mailto:s.mientki at ru.nl>> wrote:
>
>
>
>     Gary Pajer wrote:
>     > I was unaware of LabPython until this post.  But I can report that
>     > I've been using National Instruments' DAQmx library through ctypes,
>     > and it seems to work perfectly.  I don't know how far along you
>     are in
>     > development,  but there's an option for you.
>     >
>     > I've married DAQmx to scipy/numpy to Traits UI (
>     > http://code.enthought.com/projects/traits ) and I have a
>     beautiful GUI
>     > based data acquisition/display/analysis system.  I'll admit that
>     I've
>     > been too scared to even try to implement callbacks (used by a small
>     > number of DAQmx C functions) but I've achieved the same
>     functionality
>     > using python threads.  Aside from that caveat, all features have
>     > worked as they should.
>     hi Gary,
>     This sounds very very interesting.
>     As I'm writing an opensource replacement for LabView / MatLab in
>     Python
>     (called PyLab_Works),
>     and DAQmx is one of the major AD-systems in this package,
>     I'm anxious to get DAQmx libraries (as till now I have some
>     clumsy-windows-only DLL).
>     Any change I could use your code ?
>
>
> I'd be happy to share what I've done, but let me make clear that I am
> using NI's DAQmx Windows dll.    I'm just calling the functions in
> there using ctypes.   I have a feeling that you are after something
> different. 
>
> I'm not sure what you have and what you want, but if I read between
> your lines, I think you might be looking for Linux libraries.  NI does
> have Linux version with reduced functionality available for download. 
> License may be an issue.  Would Comedi help?
>
> If we get any deeper, we might want to take the discussion off-line.
Please don't, I (for one!) think this is a very interesting discussion.


From spmcinerney at hotmail.com  Tue Jun 10 02:28:33 2008
From: spmcinerney at hotmail.com (Stephen McInerney)
Date: Mon, 9 Jun 2008 23:28:33 -0700
Subject: [SciPy-user] Python fn for Pascal's coefficient?
Message-ID: <BAY111-W40568684C10DA2D63C5D8A0B30@phx.gbl>


Sorry to ask such a basic question, but I searched and I can't find
a function for Pascal's coefficient, either in standard Python, SciPy or NumPy.
 
Thanks,
Stephen
_________________________________________________________________
Enjoy 5 GB of free, password-protected online storage.
http://www.windowslive.com/skydrive/overview.html?ocid=TXT_TAGLM_WL_Refresh_skydrive_062008
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080609/bb50ed59/attachment.html>

From robert.kern at gmail.com  Tue Jun 10 02:39:45 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Tue, 10 Jun 2008 01:39:45 -0500
Subject: [SciPy-user] Python fn for Pascal's coefficient?
In-Reply-To: <BAY111-W40568684C10DA2D63C5D8A0B30@phx.gbl>
References: <BAY111-W40568684C10DA2D63C5D8A0B30@phx.gbl>
Message-ID: <3d375d730806092339w26abc337q11b782f40d234161@mail.gmail.com>

On Tue, Jun 10, 2008 at 01:28, Stephen McInerney
<spmcinerney at hotmail.com> wrote:
> Sorry to ask such a basic question, but I searched and I can't find
> a function for Pascal's coefficient, either in standard Python, SciPy or
> NumPy.

I presume you mean a function for the binomial coefficients ("Pascal's
coefficient" appears to be a fairly rare term, at least in English
usage).

http://en.wikipedia.org/wiki/Binomial_coefficient

In [3]: from scipy.misc import comb

In [4]: comb?
Type:           function
Base Class:     <type 'function'>
String Form:    <function comb at 0x1b33d70>
Namespace:      Interactive
File:           /Users/rkern/svn/scipy/scipy/misc/common.py
Definition:     comb(N, k, exact=0)
Docstring:
    Combinations of N things taken k at a time.

    If exact==0, then floating point precision is used, otherwise
    exact long integer is computed.

    Notes:
      - Array arguments accepted only for exact=0 case.
      - If k > N, N < 0, or k < 0, then a 0 is returned.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From contact at pythonxy.com  Tue Jun 10 05:08:30 2008
From: contact at pythonxy.com (Pierre Raybaut)
Date: Tue, 10 Jun 2008 11:08:30 +0200
Subject: [SciPy-user] [ Python(x,y) ] Why a linux version?
Message-ID: <629b08a40806100208i7e071f0bm17aaec8a31ce2e47@mail.gmail.com>

Hi all,

Regarding the Python(x,y) linux version, we would be happy to debate on the
following post:
http://groups.google.com/group/pythonxy/browse_thread/thread/e5c707a1ac8bd4a4?hl=en
#

Thanks,
Regards,
Pierre Raybaut
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080610/eb5a7c5f/attachment.html>

From stefan at sun.ac.za  Tue Jun 10 05:38:08 2008
From: stefan at sun.ac.za (=?ISO-8859-1?Q?St=E9fan_van_der_Walt?=)
Date: Tue, 10 Jun 2008 11:38:08 +0200
Subject: [SciPy-user] Python fn for Pascal's coefficient?
In-Reply-To: <BAY111-W40568684C10DA2D63C5D8A0B30@phx.gbl>
References: <BAY111-W40568684C10DA2D63C5D8A0B30@phx.gbl>
Message-ID: <9457e7c80806100238qa54ea50w657e4f31ce146e7@mail.gmail.com>

Hey Stephen

2008/6/10 Stephen McInerney <spmcinerney at hotmail.com>:
> Sorry to ask such a basic question, but I searched and I can't find
> a function for Pascal's coefficient, either in standard Python, SciPy or
> NumPy.

The Python cookbook is normally a good place to find this sort of thing:

http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/392153
http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/528913

In the last one, you can replace comb with the function Robert mentioned.

Cheers
St?fan


From michells at ims.uni-stuttgart.de  Tue Jun 10 13:21:14 2008
From: michells at ims.uni-stuttgart.de (Lukas Michelbacher)
Date: Tue, 10 Jun 2008 19:21:14 +0200
Subject: [SciPy-user] Sparse Boolean matrix
Message-ID: <86faf560806101021v63fa86bub04e8c3633084037@mail.gmail.com>

As the title says I'd like to use a matrix with Boolean values in some
sparse format.
My problem is that even though initialization seems to works fine, the
matrix doesn't
contain Boolean values but the default float type.

In [18]: A = sparse.csr_matrix((1000,1000), dtype=bool)

In [19]: A
Out[19]:
<1000x1000 sparse matrix of type '<type 'numpy.float64'>'
        with 0 stored elements (space for 100)
        in Compressed Sparse Row format>

The same happens for CSC, LIL and COO formats.

I use SciPy 0.6.0


From orest.kozyar at gmail.com  Tue Jun 10 13:26:18 2008
From: orest.kozyar at gmail.com (Orest Kozyar)
Date: Tue, 10 Jun 2008 10:26:18 -0700 (PDT)
Subject: [SciPy-user] Getting PyArray_Resize to work under weave
Message-ID: <f53c006c-2987-4b0f-b23c-4e8e5efca447@e53g2000hsa.googlegroups.com>

I originally posted this on Numpy, but did not get any feedback.  I've
set up a temporary workaround by ensuring the Numpy array I pass to
weave is much, much larger than I really expect it needs to be.
However, I'd like to try and figure out why resize is not working.
Any guidance?

The following code fails:

from scipy import weave
from numpy import zeros

arr = zeros((10,2))
code = """
PyArray_Dims dims;
dims.len = 2;
dims.ptr = Narr;
dims.ptr[0] += 10;
PyArray_Resize(arr_array, &dims, 1);
"""
weave.inline(code, ['arr'], verbose=1)

The error message is:
In function 'PyObject* compiled_func(PyObject*, PyObject*)':
<filename>:678: error: too few arguments to function

678 is the line number for PyArray_Resize.  According to the NumPy
Handbook, PyArray_Resize requires three arguments, which I am
providing. Am I missing something obvious here?  There are times when
I need to be able to resize the array in C++ because I cannot predict
exactly how big the array needs to be before I pass it to weave.  Any
advice or pointers greatly appreciated!

Thanks,
Orest


From nwagner at iam.uni-stuttgart.de  Tue Jun 10 13:41:53 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Tue, 10 Jun 2008 19:41:53 +0200
Subject: [SciPy-user] Sparse Boolean matrix
In-Reply-To: <86faf560806101021v63fa86bub04e8c3633084037@mail.gmail.com>
References: <86faf560806101021v63fa86bub04e8c3633084037@mail.gmail.com>
Message-ID: <web-103109553@uni-stuttgart.de>

On Tue, 10 Jun 2008 19:21:14 +0200
  "Lukas Michelbacher" <michells at ims.uni-stuttgart.de> 
wrote:
> As the title says I'd like to use a matrix with Boolean 
>values in some
> sparse format.
> My problem is that even though initialization seems to 
>works fine, the
> matrix doesn't
> contain Boolean values but the default float type.
> 
> In [18]: A = sparse.csr_matrix((1000,1000), dtype=bool)
> 
> In [19]: A
> Out[19]:
> <1000x1000 sparse matrix of type '<type 
>'numpy.float64'>'
>        with 0 stored elements (space for 100)
>        in Compressed Sparse Row format>
> 
> The same happens for CSC, LIL and COO formats.
> 
> I use SciPy 0.6.0



>>> from scipy import *
>>> A = sparse.csr_matrix((1000,1000), dtype=bool)
>>> A
<1000x1000 sparse matrix of type '<type 'numpy.bool_'>'
         with 0 stored elements in Compressed Sparse Row 
format>
>>> scipy.__version__
'0.7.0.dev4423'

  >>> A = sparse.coo_matrix((1000,1000), dtype=bool)
>>> A
<1000x1000 sparse matrix of type '<type 'numpy.bool_'>'
         with 0 stored elements in COOrdinate format>
>>> A = sparse.lil_matrix((1000,1000), dtype=bool)
>>> A
<1000x1000 sparse matrix of type '<type 'numpy.bool_'>'
         with 0 stored elements in LInked List format>
>>> A = sparse.dok_matrix((1000,1000), dtype=bool)
>>> A
<1000x1000 sparse matrix of type '<type 'numpy.bool_'>'
         with 0 stored elements in Dictionary Of Keys 
format>



Nils



From wnbell at gmail.com  Tue Jun 10 14:46:39 2008
From: wnbell at gmail.com (Nathan Bell)
Date: Tue, 10 Jun 2008 13:46:39 -0500
Subject: [SciPy-user] Sparse Boolean matrix
In-Reply-To: <86faf560806101021v63fa86bub04e8c3633084037@mail.gmail.com>
References: <86faf560806101021v63fa86bub04e8c3633084037@mail.gmail.com>
Message-ID: <d05265cb0806101146n648845a5n20cb38dbf463732a@mail.gmail.com>

On Tue, Jun 10, 2008 at 12:21 PM, Lukas Michelbacher
<michells at ims.uni-stuttgart.de> wrote:
> As the title says I'd like to use a matrix with Boolean values in some
> sparse format.
> My problem is that even though initialization seems to works fine, the
> matrix doesn't
> contain Boolean values but the default float type.

Lukas, SciPy 0.6.x doesn't support sparse boolean matrices, or even
integer matrices for that matter.

The current SVN version of SciPy and the next release (0.7.0) do
support integer matrices (including int8) but don't properly support
booleans.  I believe they will get converted to int8.

-- 
Nathan Bell wnbell at gmail.com
http://graphics.cs.uiuc.edu/~wnbell/


From fperez.net at gmail.com  Tue Jun 10 14:57:31 2008
From: fperez.net at gmail.com (Fernando Perez)
Date: Tue, 10 Jun 2008 11:57:31 -0700
Subject: [SciPy-user] Plans for Scipy Tutorials
Message-ID: <db6b5ecc0806101157q67713dfp8b6206249933714@mail.gmail.com>

Hi all,

I've now put up the near-final tutorial plans for SciPy 2008 here:

http://conference.scipy.org/tutorials

If your name is listed there and you disagree/can't make it, please
let me and Travis Oliphant know as soon as possible.

As the various presenters fine-tune their plan, we'll update the
details on each tutorial and provide links to pre-requisites,
installation pages, etc.  But the main topics are probably not going
to change now, barring any unforeseen event.

Regards,

f


From didier.rano at gmail.com  Tue Jun 10 15:09:27 2008
From: didier.rano at gmail.com (didier rano)
Date: Tue, 10 Jun 2008 15:09:27 -0400
Subject: [SciPy-user] Some mathematics/statisctics books
Message-ID: <da09baf80806101209q7b0b63acs8fa053b93d7dcfce@mail.gmail.com>

Hi all,

I am using Time Series moduled provided for scipy. I am very impressive by
scipy in general. But I don't have enough backgrounds to understand all
mathematics/statistics models inside scipy.

Could you help to find some books, articles, courses to improve my
scientific backgrounds ? In particular, I need some knowledges to generate
graphs to show pertinent information (trends...).

Thank you
Didier Rano
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080610/161ea69b/attachment.html>

From cohen at slac.stanford.edu  Tue Jun 10 15:36:10 2008
From: cohen at slac.stanford.edu (Johann Cohen-Tanugi)
Date: Tue, 10 Jun 2008 21:36:10 +0200
Subject: [SciPy-user] Some mathematics/statisctics books
In-Reply-To: <da09baf80806101209q7b0b63acs8fa053b93d7dcfce@mail.gmail.com>
References: <da09baf80806101209q7b0b63acs8fa053b93d7dcfce@mail.gmail.com>
Message-ID: <484ED7AA.2040907@slac.stanford.edu>

best is to start with the tutorials and examples distributed with scipy 
or available on the website. Then you are most welcome to ask questions 
about them in this forum, and only then might it make sense to look for 
references, because scipy covers a huge class of problems and thus there 
is no way one can possibly give you a reference for an overall picture 
of its possibilities. The only way is to dive in, starting with some 
specific little examples or problems you would like to solve.
Hope that helps,
Johann

didier rano wrote:
> Hi all,
>  
> I am using Time Series moduled provided for scipy. I am very 
> impressive by scipy in general. But I don't have enough backgrounds to 
> understand all mathematics/statistics models inside scipy.
>  
> Could you help to find some books, articles, courses to improve my 
> scientific backgrounds ? In particular, I need some knowledges to 
> generate graphs to show pertinent information (trends...).
>  
> Thank you
> Didier Rano
> ------------------------------------------------------------------------
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>   


From Karl.Young at ucsf.edu  Tue Jun 10 15:35:55 2008
From: Karl.Young at ucsf.edu (Karl Young)
Date: Tue, 10 Jun 2008 12:35:55 -0700
Subject: [SciPy-user] Some mathematics/statisctics books
In-Reply-To: <484ED7AA.2040907@slac.stanford.edu>
References: <da09baf80806101209q7b0b63acs8fa053b93d7dcfce@mail.gmail.com>
	<484ED7AA.2040907@slac.stanford.edu>
Message-ID: <484ED79B.8000607@ucsf.edu>


I completely agree with Johann that the best way to start is to just 
dive into the tutorials and examples but there are a few books around 
that might not be bad to have at your side when doing so. The reason I 
say this is that a particular book came to mind when I saw the original 
post. Though like anything else it has it's pluses and minuses it seems 
to me that Neil Gershenfeld's book "The Nature of Mathematical Modeling" 
would be pretty useful in this context in having a broad enough scope to 
discuss a lot of the types of problem that one would be working on re. 
using SciPy (including some bits on time series analysis).

>best is to start with the tutorials and examples distributed with scipy 
>or available on the website. Then you are most welcome to ask questions 
>about them in this forum, and only then might it make sense to look for 
>references, because scipy covers a huge class of problems and thus there 
>is no way one can possibly give you a reference for an overall picture 
>of its possibilities. The only way is to dive in, starting with some 
>specific little examples or problems you would like to solve.
>Hope that helps,
>Johann
>
>didier rano wrote:
>  
>
>>Hi all,
>> 
>>I am using Time Series moduled provided for scipy. I am very 
>>impressive by scipy in general. But I don't have enough backgrounds to 
>>understand all mathematics/statistics models inside scipy.
>> 
>>Could you help to find some books, articles, courses to improve my 
>>scientific backgrounds ? In particular, I need some knowledges to 
>>generate graphs to show pertinent information (trends...).
>> 
>>Thank you
>>Didier Rano
>>------------------------------------------------------------------------
>>
>>_______________________________________________
>>SciPy-user mailing list
>>SciPy-user at scipy.org
>>http://projects.scipy.org/mailman/listinfo/scipy-user
>>  
>>    
>>
>_______________________________________________
>SciPy-user mailing list
>SciPy-user at scipy.org
>http://projects.scipy.org/mailman/listinfo/scipy-user
>
>  
>


-- 

Karl Young
Center for Imaging of Neurodegenerative Diseases, UCSF          
VA Medical Center (114M)              Phone:  (415) 221-4810 x3114  lab        
4150 Clement Street                   FAX:    (415) 668-2864
San Francisco, CA 94121               Email:  karl young at ucsf edu



From oliphant at enthought.com  Tue Jun 10 15:56:47 2008
From: oliphant at enthought.com (Travis E. Oliphant)
Date: Tue, 10 Jun 2008 14:56:47 -0500
Subject: [SciPy-user] Getting PyArray_Resize to work under weave
In-Reply-To: <f53c006c-2987-4b0f-b23c-4e8e5efca447@e53g2000hsa.googlegroups.com>
References: <f53c006c-2987-4b0f-b23c-4e8e5efca447@e53g2000hsa.googlegroups.com>
Message-ID: <484EDC7F.9090505@enthought.com>

Orest Kozyar wrote:
> I originally posted this on Numpy, but did not get any feedback.  I've
> set up a temporary workaround by ensuring the Numpy array I pass to
> weave is much, much larger than I really expect it needs to be.
> However, I'd like to try and figure out why resize is not working.
> Any guidance?
>
> The following code fails:
>
> from scipy import weave
> from numpy import zeros
>
> arr = zeros((10,2))
> code = """
> PyArray_Dims dims;
> dims.len = 2;
> dims.ptr = Narr;
> dims.ptr[0] += 10;
> PyArray_Resize(arr_array, &dims, 1);
> """
> weave.inline(code, ['arr'], verbose=1)
>
> The error message is:
> In function 'PyObject* compiled_func(PyObject*, PyObject*)':
> <filename>:678: error: too few arguments to function
>
> 678 is the line number for PyArray_Resize.  According to the NumPy
> Handbook, PyArray_Resize requires three arguments, which I am
> providing. Am I missing something obvious here?  There are times when
> I need to be able to resize the array in C++ because I cannot predict
> exactly how big the array needs to be before I pass it to weave.  Any
> advice or pointers greatly appreciated!
>   
The function was updated ahead of the book.   There is a NPY_order 
fortran argument at the end that must be supplied to PyArray_Resize.    
Use this for the last line of your C-code.

PyArray_Resize(arr_array, &dims, 1, NPY_ANYORDER);


-Travis




From didier.rano at gmail.com  Tue Jun 10 16:00:07 2008
From: didier.rano at gmail.com (didier rano)
Date: Tue, 10 Jun 2008 16:00:07 -0400
Subject: [SciPy-user] Some mathematics/statisctics books
In-Reply-To: <484ED79B.8000607@ucsf.edu>
References: <da09baf80806101209q7b0b63acs8fa053b93d7dcfce@mail.gmail.com>
	<484ED7AA.2040907@slac.stanford.edu> <484ED79B.8000607@ucsf.edu>
Message-ID: <da09baf80806101300k3fbddf0fuae84a2dc9970d2a1@mail.gmail.com>

I tried some weeks ago to use the tutorial, but I was lost about my poor
mathematics background.
Then thank you to provide me the name of this book.


2008/6/10, Karl Young <Karl.Young at ucsf.edu>:
>
>
> I completely agree with Johann that the best way to start is to just
> dive into the tutorials and examples but there are a few books around
> that might not be bad to have at your side when doing so. The reason I
> say this is that a particular book came to mind when I saw the original
> post. Though like anything else it has it's pluses and minuses it seems
> to me that Neil Gershenfeld's book "The Nature of Mathematical Modeling"
> would be pretty useful in this context in having a broad enough scope to
> discuss a lot of the types of problem that one would be working on re.
> using SciPy (including some bits on time series analysis).
>
> >best is to start with the tutorials and examples distributed with scipy
> >or available on the website. Then you are most welcome to ask questions
> >about them in this forum, and only then might it make sense to look for
> >references, because scipy covers a huge class of problems and thus there
> >is no way one can possibly give you a reference for an overall picture
> >of its possibilities. The only way is to dive in, starting with some
> >specific little examples or problems you would like to solve.
> >Hope that helps,
> >Johann
> >
> >didier rano wrote:
> >
> >
> >>Hi all,
> >>
> >>I am using Time Series moduled provided for scipy. I am very
> >>impressive by scipy in general. But I don't have enough backgrounds to
> >>understand all mathematics/statistics models inside scipy.
> >>
> >>Could you help to find some books, articles, courses to improve my
> >>scientific backgrounds ? In particular, I need some knowledges to
> >>generate graphs to show pertinent information (trends...).
> >>
> >>Thank you
> >>Didier Rano
> >>------------------------------------------------------------------------
> >>
> >>_______________________________________________
> >>SciPy-user mailing list
> >>SciPy-user at scipy.org
> >>http://projects.scipy.org/mailman/listinfo/scipy-user
> >>
> >>
> >>
> >_______________________________________________
> >SciPy-user mailing list
> >SciPy-user at scipy.org
> >http://projects.scipy.org/mailman/listinfo/scipy-user
> >
> >
> >
>
>
> --
>
> Karl Young
> Center for Imaging of Neurodegenerative Diseases, UCSF
> VA Medical Center (114M)              Phone:  (415) 221-4810 x3114  lab
> 4150 Clement Street                   FAX:    (415) 668-2864
> San Francisco, CA 94121               Email:  karl young at ucsf edu
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>



-- 
Didier Rano
didier.rano at gmail.com
http://www.jaxtr.com/didierrano
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080610/bc327593/attachment.html>

From cohen at slac.stanford.edu  Tue Jun 10 16:29:44 2008
From: cohen at slac.stanford.edu (Johann Cohen-Tanugi)
Date: Tue, 10 Jun 2008 22:29:44 +0200
Subject: [SciPy-user] Some mathematics/statisctics books
In-Reply-To: <da09baf80806101300k3fbddf0fuae84a2dc9970d2a1@mail.gmail.com>
References: <da09baf80806101209q7b0b63acs8fa053b93d7dcfce@mail.gmail.com>	<484ED7AA.2040907@slac.stanford.edu>
	<484ED79B.8000607@ucsf.edu>
	<da09baf80806101300k3fbddf0fuae84a2dc9970d2a1@mail.gmail.com>
Message-ID: <484EE438.9020802@slac.stanford.edu>

well, I browsed through this book some time ago, and it is really nice, 
but does involve some mathematics. I reiterate that picking up one 
example in the tutorial, and asking about it might send you on a 
fruitful track already.
best,
Johann

didier rano wrote:
> I tried some weeks ago to use the tutorial, but I was lost about my 
> poor mathematics background.
> Then thank you to provide me the name of this book.
>
>  
> 2008/6/10, Karl Young <Karl.Young at ucsf.edu <mailto:Karl.Young at ucsf.edu>>:
>
>
>     I completely agree with Johann that the best way to start is to just
>     dive into the tutorials and examples but there are a few books around
>     that might not be bad to have at your side when doing so. The reason I
>     say this is that a particular book came to mind when I saw the
>     original
>     post. Though like anything else it has it's pluses and minuses it
>     seems
>     to me that Neil Gershenfeld's book "The Nature of Mathematical
>     Modeling"
>     would be pretty useful in this context in having a broad enough
>     scope to
>     discuss a lot of the types of problem that one would be working on re.
>     using SciPy (including some bits on time series analysis).
>
>     >best is to start with the tutorials and examples distributed with
>     scipy
>     >or available on the website. Then you are most welcome to ask
>     questions
>     >about them in this forum, and only then might it make sense to
>     look for
>     >references, because scipy covers a huge class of problems and
>     thus there
>     >is no way one can possibly give you a reference for an overall
>     picture
>     >of its possibilities. The only way is to dive in, starting with some
>     >specific little examples or problems you would like to solve.
>     >Hope that helps,
>     >Johann
>     >
>     >didier rano wrote:
>     >
>     >
>     >>Hi all,
>     >>
>     >>I am using Time Series moduled provided for scipy. I am very
>     >>impressive by scipy in general. But I don't have enough
>     backgrounds to
>     >>understand all mathematics/statistics models inside scipy.
>     >>
>     >>Could you help to find some books, articles, courses to improve my
>     >>scientific backgrounds ? In particular, I need some knowledges to
>     >>generate graphs to show pertinent information (trends...).
>     >>
>     >>Thank you
>     >>Didier Rano
>     >>------------------------------------------------------------------------
>     >>
>     >>_______________________________________________
>     >>SciPy-user mailing list
>     >>SciPy-user at scipy.org <mailto:SciPy-user at scipy.org>
>     >>http://projects.scipy.org/mailman/listinfo/scipy-user
>     >>
>     >>
>     >>
>     >_______________________________________________
>     >SciPy-user mailing list
>     >SciPy-user at scipy.org <mailto:SciPy-user at scipy.org>
>     >http://projects.scipy.org/mailman/listinfo/scipy-user
>     >
>     >
>     >
>
>
>     --
>
>     Karl Young
>     Center for Imaging of Neurodegenerative Diseases, UCSF
>     VA Medical Center (114M)              Phone:  (415) 221-4810
>     x3114  lab
>     4150 Clement Street                   FAX:    (415) 668-2864
>     San Francisco, CA 94121               Email:  karl young at ucsf edu
>
>     _______________________________________________
>     SciPy-user mailing list
>     SciPy-user at scipy.org <mailto:SciPy-user at scipy.org>
>     http://projects.scipy.org/mailman/listinfo/scipy-user
>
>
>
>
> -- 
> Didier Rano
> didier.rano at gmail.com <mailto:didier.rano at gmail.com>
> http://www.jaxtr.com/didierrano
> ------------------------------------------------------------------------
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>   


From millman at berkeley.edu  Tue Jun 10 17:10:42 2008
From: millman at berkeley.edu (Jarrod Millman)
Date: Tue, 10 Jun 2008 14:10:42 -0700
Subject: [SciPy-user] ANN: SciPy 2008 Conference
Message-ID: <c7009a550806101410h4725b984j6c9cce651501bfd4@mail.gmail.com>

Greetings,

The SciPy 2008 Conference website is now open:  http://conference.scipy.org

This year's conference will be at Caltech from August 19-24:

Tutorials:        August 19-20 (Tuesday and Wednesday)
Conference:  August 21-22 (Thursday and Friday)
Sprints:           August 23-24 (Saturday and Sunday)

Exciting things are happening in the Python community, and the SciPy
2008 Conference is an excellent opportunity to exchange ideas, learn
techniques, contribute code and affect the direction of scientific
computing (or just to learn what all the fuss is about).  We'll be
announcing the Keynote Speaker and providing a detailed schedule in
the coming weeks.

This year we are asking presenters to submit short papers to be included
in the conference proceedings:  http://conference.scipy.org/call_for_papers

Cheers,

-- 
Jarrod Millman
Computational Infrastructure for Research Labs
10 Giannini Hall, UC Berkeley
phone: 510.643.4014
http://cirl.berkeley.edu/


From bryan at cole.uklinux.net  Tue Jun 10 17:20:36 2008
From: bryan at cole.uklinux.net (Bryan Cole)
Date: Tue, 10 Jun 2008 22:20:36 +0100
Subject: [SciPy-user] LabVIEW and LabPython
In-Reply-To: <88fe22a0806081521j2db5b1eblb68692af2e3beb43@mail.gmail.com>
References: <ff223aae0806031923u1dd2c613k7ee8e5f27c1753e8@mail.gmail.com>
	<88fe22a0806081521j2db5b1eblb68692af2e3beb43@mail.gmail.com>
Message-ID: <1213132836.25044.13.camel@pc2.cole.uklinux.net>


> 
> I've married DAQmx to scipy/numpy to Traits UI
> ( http://code.enthought.com/projects/traits ) and I have a beautiful
> GUI based data acquisition/display/analysis system.  I'll admit that
> I've been too scared to even try to implement callbacks (used by a
> small number of DAQmx C functions) but I've achieved the same
> functionality using python threads.

FWIW Python callbacks with DAQmx work great. ctypes makes these easy to
implement.

I'm also considering Trait'ification of my primary data-acquisition
application. I'd like to see what you've done in this respect. Are you
using Chaco?

BC






From dwf at cs.toronto.edu  Tue Jun 10 19:26:28 2008
From: dwf at cs.toronto.edu (David Warde-Farley)
Date: Tue, 10 Jun 2008 19:26:28 -0400
Subject: [SciPy-user] Some mathematics/statisctics books
In-Reply-To: <484ED79B.8000607@ucsf.edu>
References: <da09baf80806101209q7b0b63acs8fa053b93d7dcfce@mail.gmail.com>
	<484ED7AA.2040907@slac.stanford.edu> <484ED79B.8000607@ucsf.edu>
Message-ID: <0BAB70CC-5B85-4CB5-9355-195B537B08EA@cs.toronto.edu>

On 10-Jun-08, at 3:35 PM, Karl Young wrote:

> I completely agree with Johann that the best way to start is to just
> dive into the tutorials and examples but there are a few books around
> that might not be bad to have at your side when doing so.

I just looked through the Gershenfeld book's table of contents, and  
while it's full of tons of useful stuff, it may be the wrong place to  
start. Certainly, don't try to read the book in sequence from start  
to finish if you don't have the requisite background.

What background? Probably some linear algebra at least, and some  
single and multivariable calculus. The OP didn't specify what level  
of education he's had in these matters, so just in case I'll include  
a few books.

I'm not too familiar with books on single-variable calculus but Tom  
Apostol's book on the subject appears to be quite well reviewed. It's  
also been around since the 1960's so it should be possible to find an  
inexpensive used copy.

As for multivariable calculus and linear algebra, a book that I'm  
fond of that takes an integrated approach to these two subjects is  
"Multivariable Mathematics" by Theodore Shifrin, ISBN 047152638X.  It  
goes through a lot of examples but doesn't sacrifice rigour. It won't  
help you much if you haven't done any single variable calculus.

My supervisor introduced me to Gilbert Strang's "Linear Algebra and  
its Applications", which I quite like. People seem to either love or  
hate this book, but it endeavours to help you develop intuition for  
linear algebraic concepts, and is quite light on theory, heavy on  
application.

Hope that helps,

David


From patrick.m.bouffard at gmail.com  Tue Jun 10 20:24:15 2008
From: patrick.m.bouffard at gmail.com (Patrick Bouffard)
Date: Wed, 11 Jun 2008 00:24:15 +0000 (UTC)
Subject: [SciPy-user] ANN: NumPy/SciPy Documentation Marathon 2008
References: <wl8zlqpo54z.fsf@glup.physics.ucf.edu>
Message-ID: <loom.20080611T002205-900@post.gmane.org>

Hi,

I'm registered on the wiki as PatrickBouffard. I'd like to chip away at
reviewing and proofing (I have a good eye for speling misstakes) and I can
probably write some docstrings too.

Cheers

Pat




From stefan at sun.ac.za  Tue Jun 10 20:41:04 2008
From: stefan at sun.ac.za (=?ISO-8859-1?Q?St=E9fan_van_der_Walt?=)
Date: Wed, 11 Jun 2008 02:41:04 +0200
Subject: [SciPy-user] ANN: NumPy/SciPy Documentation Marathon 2008
In-Reply-To: <loom.20080611T002205-900@post.gmane.org>
References: <wl8zlqpo54z.fsf@glup.physics.ucf.edu>
	<loom.20080611T002205-900@post.gmane.org>
Message-ID: <9457e7c80806101741j393b81c2k255ecebd485c1ec6@mail.gmail.com>

Hi Pat

2008/6/11 Patrick Bouffard <patrick.m.bouffard at gmail.com>:
> I'm registered on the wiki as PatrickBouffard. I'd like to chip away at
> reviewing and proofing (I have a good eye for speling misstakes) and I can
> probably write some docstrings too.

Thank you for helping!  Your account is now active.

Regards
St?fan


From ivo.maljevic at gmail.com  Tue Jun 10 20:49:59 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Tue, 10 Jun 2008 20:49:59 -0400
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
Message-ID: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>

Hi,
I am relatively new SciPy/NumPy user and I must say I like it. I don't know
if this is something that is well known, but I run some simple tests,
and I found out that, depending on whether one imports NumPy or SciPy,
programs executes at quite different speeds. I'm guessing the problem has
something to do with how it cycles through loops.

My initial goal was to compare python scripts with octave/matlab ones, but
then I noticed the speed difference between NumPy and SciPy.

To cut the story short, here are results:

Execution times in seconds, test 1:

Python              Octave       Fortran            C
========================================================
NumPy: 23.9         36.05        6.7                6.8
SciPy: 44.5


Execution times in seconds, test 2:

Python              Octave       Fortran            C
========================================================
NumPy: 9.4          5.5          6.7                6.8
SciPy: 10.1


Test cases 1 and 2 refer to different parameter values for the same scipts
(test 1: FrameSize=100, test 2: FrameSize=10000).

If anybody would like to experiment, I've copied the source code for the
simple BPSK simulation that determines the raw bit error rate.

To calculate erfc function with NumPy, I used a simple fotran funciton and
comiled it into .so library:

subroutine erfc(x,y)
  real, intent(in)  :: x
  real, intent(out) :: y
  y = 1-erf(x)
  end

# Python script

# For some reason, numpy is much faster than scipy, Try to comment out one
or the other line
#from numpy import *
from scipy import *
import erfc

SNR_MIN   = -1
SNR_MAX   = 10
FrameSize = 100   # <-- change me for test 2

Eb_No_dB  = arange(SNR_MIN,SNR_MAX+1)

# signal vector
s = ones(FrameSize)
a = random.rand(FrameSize)
a_i = where(a < 0.5)
s[a_i] = -1

for snr in Eb_No_dB:

  No        = 10**(-snr/10.0)
  Pe        = 0.5*erfc.erfc(sqrt(1.0/No))
  nFrames   = ceil(200.0/FrameSize/Pe)
  error_sum = 0

  for frame in arange(nFrames):

    # noise
    #n = sqrt(No/2)*random.randn(FrameSize)
    n = random.normal(scale=sqrt(No/2), size=FrameSize)
    x = s + n

    # detection
    y = sign(x)

    # error counting
    err = where (y != s)
    error_sum += len(err[0])

  print 'Eb_No_dB=%2d, BER=%10.4e, Pe=%10.4e' % \
         (snr, error_sum/(FrameSize*nFrames), Pe)


Octave m-file:
% bpsk_sim.m - octave version
clear

tic

SNR_MIN    = -1;
SNR_MAX    = 10;
FrameSize  = 100;
Eb_No_dB = SNR_MIN:SNR_MAX;

% signal vector
s = ones(1,FrameSize);
a = rand(1,FrameSize);
a_i = find(a < 0.5);
s(a_i) = -1;

for snr=SNR_MIN:SNR_MAX

  No = 10^(-snr/10.0);
  Pe = 0.5*erfc(sqrt(1.0/No));
  nFrames = ceil(200/FrameSize/Pe);
  error_sum = 0;

  for frame=1:nFrames

    % noise
    n = sqrt(No/2)*randn(1,FrameSize);
    x = s + n;

    y = sign(x);

    err = find( y ~= s);
    error_sum = error_sum + length(err);
  end

  fprintf('Eb_No_dB=%2d, BER=%10.4e, Pe=%10.4e\n', ...
           snr, error_sum/(FrameSize*nFrames), Pe)
end
toc
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080610/062c0d86/attachment.html>

From alan at ajackson.org  Tue Jun 10 21:10:05 2008
From: alan at ajackson.org (Alan Jackson)
Date: Tue, 10 Jun 2008 20:10:05 -0500
Subject: [SciPy-user] Some mathematics/statisctics books
In-Reply-To: <da09baf80806101209q7b0b63acs8fa053b93d7dcfce@mail.gmail.com>
References: <da09baf80806101209q7b0b63acs8fa053b93d7dcfce@mail.gmail.com>
Message-ID: <20080610201005.43296f34@ajackson.org>

For learning statistics, sorry they aren't python, but I really like :

Introductory Statistics with R - Peter Dalgaard
An R and S-plus Companion to Applied Regression - John Fox
Primer of Biostatistics - Stanton Glantz
Modern Applied Statistics with S - Venebles and Ripley
Practical Time Series - Gareth Janacek

For what you want, the first two in particular are probably relevant.
Personally I switch between scipy and R frequently, depending on which
one seems best for the problem at hand. They both have their strengths.

On Tue, 10 Jun 2008 15:09:27 -0400
"didier rano" <didier.rano at gmail.com> wrote:

> Hi all,
> 
> I am using Time Series moduled provided for scipy. I am very impressive by
> scipy in general. But I don't have enough backgrounds to understand all
> mathematics/statistics models inside scipy.
> 
> Could you help to find some books, articles, courses to improve my
> scientific backgrounds ? In particular, I need some knowledges to generate
> graphs to show pertinent information (trends...).
> 
> Thank you
> Didier Rano


-- 
-----------------------------------------------------------------------
| Alan K. Jackson            | To see a World in a Grain of Sand      |
| alan at ajackson.org          | And a Heaven in a Wild Flower,         |
| www.ajackson.org           | Hold Infinity in the palm of your hand |
| Houston, Texas             | And Eternity in an hour. - Blake       |
-----------------------------------------------------------------------


From peter.skomoroch at gmail.com  Tue Jun 10 22:00:52 2008
From: peter.skomoroch at gmail.com (Peter Skomoroch)
Date: Tue, 10 Jun 2008 19:00:52 -0700
Subject: [SciPy-user] Some mathematics/statisctics books
In-Reply-To: <20080610201005.43296f34@ajackson.org>
References: <da09baf80806101209q7b0b63acs8fa053b93d7dcfce@mail.gmail.com>
	<20080610201005.43296f34@ajackson.org>
Message-ID: <e4fc0d2a0806101900r4e91ef42m157af3d7b8384ae9@mail.gmail.com>

Strang and the R books are good places to start....

for stats, check out:

*"All of Statistics*. A Concise Course in Statistical Inference". by Larry
Wasserman
http://www.stat.cmu.edu/~larry/all-of-statistics/index.html

On Tue, Jun 10, 2008 at 6:10 PM, Alan Jackson <alan at ajackson.org> wrote:

> For learning statistics, sorry they aren't python, but I really like :
>
> Introductory Statistics with R - Peter Dalgaard
> An R and S-plus Companion to Applied Regression - John Fox
> Primer of Biostatistics - Stanton Glantz
> Modern Applied Statistics with S - Venebles and Ripley
> Practical Time Series - Gareth Janacek
>
> For what you want, the first two in particular are probably relevant.
> Personally I switch between scipy and R frequently, depending on which
> one seems best for the problem at hand. They both have their strengths.
>
> On Tue, 10 Jun 2008 15:09:27 -0400
> "didier rano" <didier.rano at gmail.com> wrote:
>
> > Hi all,
> >
> > I am using Time Series moduled provided for scipy. I am very impressive
> by
> > scipy in general. But I don't have enough backgrounds to understand all
> > mathematics/statistics models inside scipy.
> >
> > Could you help to find some books, articles, courses to improve my
> > scientific backgrounds ? In particular, I need some knowledges to
> generate
> > graphs to show pertinent information (trends...).
> >
> > Thank you
> > Didier Rano
>
>
> --
> -----------------------------------------------------------------------
> | Alan K. Jackson            | To see a World in a Grain of Sand      |
> | alan at ajackson.org          | And a Heaven in a Wild Flower,         |
> | www.ajackson.org           | Hold Infinity in the palm of your hand |
> | Houston, Texas             | And Eternity in an hour. - Blake       |
> -----------------------------------------------------------------------
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>



-- 
Peter N. Skomoroch
peter.skomoroch at gmail.com
http://www.datawrangling.com
http://del.icio.us/pskomoroch
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080610/497de0d0/attachment.html>

From didier.rano at gmail.com  Tue Jun 10 22:07:23 2008
From: didier.rano at gmail.com (didier rano)
Date: Tue, 10 Jun 2008 22:07:23 -0400
Subject: [SciPy-user] Some mathematics/statisctics books
In-Reply-To: <20080610201005.43296f34@ajackson.org>
References: <da09baf80806101209q7b0b63acs8fa053b93d7dcfce@mail.gmail.com>
	<20080610201005.43296f34@ajackson.org>
Message-ID: <da09baf80806101907k1b27fb4g302829dd41dd228e@mail.gmail.com>

Wahh so many books to read.

But to analysis graph related to time-series, I don't know if I need more a
statistic approach or pure mathematic approach. Maybe that I could use both
approaches.

For example, I need to analysis the trend of a graph, or remove some no
relevants data in a graph.

Thanks for all your help, and sorry for my poor background in mathematics (I
need to learn linear algebre too !)



2008/6/10 Alan Jackson <alan at ajackson.org>:

> For learning statistics, sorry they aren't python, but I really like :
>
> Introductory Statistics with R - Peter Dalgaard
> An R and S-plus Companion to Applied Regression - John Fox
> Primer of Biostatistics - Stanton Glantz
> Modern Applied Statistics with S - Venebles and Ripley
> Practical Time Series - Gareth Janacek
>
> For what you want, the first two in particular are probably relevant.
> Personally I switch between scipy and R frequently, depending on which
> one seems best for the problem at hand. They both have their strengths.
>
> On Tue, 10 Jun 2008 15:09:27 -0400
> "didier rano" <didier.rano at gmail.com> wrote:
>
> > Hi all,
> >
> > I am using Time Series moduled provided for scipy. I am very impressive
> by
> > scipy in general. But I don't have enough backgrounds to understand all
> > mathematics/statistics models inside scipy.
> >
> > Could you help to find some books, articles, courses to improve my
> > scientific backgrounds ? In particular, I need some knowledges to
> generate
> > graphs to show pertinent information (trends...).
> >
> > Thank you
> > Didier Rano
>
>
> --
> -----------------------------------------------------------------------
> | Alan K. Jackson            | To see a World in a Grain of Sand      |
> | alan at ajackson.org          | And a Heaven in a Wild Flower,         |
> | www.ajackson.org           | Hold Infinity in the palm of your hand |
> | Houston, Texas             | And Eternity in an hour. - Blake       |
> -----------------------------------------------------------------------
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>



-- 
Didier Rano
didier.rano at gmail.com
http://www.jaxtr.com/didierrano
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080610/df1423e3/attachment.html>

From matthieu.brucher at gmail.com  Wed Jun 11 03:15:44 2008
From: matthieu.brucher at gmail.com (Matthieu Brucher)
Date: Wed, 11 Jun 2008 09:15:44 +0200
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
Message-ID: <e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>

Hi,

The difference in speed may be only due to the fact that the scipy modle has
much more functions than numpy's.

Import only what you need and try again.

Matthieu

2008/6/11 Ivo Maljevic <ivo.maljevic at gmail.com>:

> Hi,
> I am relatively new SciPy/NumPy user and I must say I like it. I don't know
> if this is something that is well known, but I run some simple tests,
> and I found out that, depending on whether one imports NumPy or SciPy,
> programs executes at quite different speeds. I'm guessing the problem has
> something to do with how it cycles through loops.
>
> My initial goal was to compare python scripts with octave/matlab ones, but
> then I noticed the speed difference between NumPy and SciPy.
>
> To cut the story short, here are results:
>
> Execution times in seconds, test 1:
>
> Python              Octave       Fortran            C
> ========================================================
> NumPy: 23.9         36.05        6.7                6.8
> SciPy: 44.5
>
>
> Execution times in seconds, test 2:
>
> Python              Octave       Fortran            C
> ========================================================
> NumPy: 9.4          5.5          6.7                6.8
> SciPy: 10.1
>
>
> Test cases 1 and 2 refer to different parameter values for the same scipts
> (test 1: FrameSize=100, test 2: FrameSize=10000).
>
> If anybody would like to experiment, I've copied the source code for the
> simple BPSK simulation that determines the raw bit error rate.
>
> To calculate erfc function with NumPy, I used a simple fotran funciton and
> comiled it into .so library:
>
> subroutine erfc(x,y)
>   real, intent(in)  :: x
>   real, intent(out) :: y
>   y = 1-erf(x)
>   end
>
> # Python script
>
> # For some reason, numpy is much faster than scipy, Try to comment out one
> or the other line
> #from numpy import *
> from scipy import *
> import erfc
>
> SNR_MIN   = -1
> SNR_MAX   = 10
> FrameSize = 100   # <-- change me for test 2
>
> Eb_No_dB  = arange(SNR_MIN,SNR_MAX+1)
>
> # signal vector
> s = ones(FrameSize)
> a = random.rand(FrameSize)
> a_i = where(a < 0.5)
> s[a_i] = -1
>
> for snr in Eb_No_dB:
>
>   No        = 10**(-snr/10.0)
>   Pe        = 0.5*erfc.erfc(sqrt(1.0/No))
>   nFrames   = ceil(200.0/FrameSize/Pe)
>   error_sum = 0
>
>   for frame in arange(nFrames):
>
>     # noise
>     #n = sqrt(No/2)*random.randn(FrameSize)
>     n = random.normal(scale=sqrt(No/2), size=FrameSize)
>     x = s + n
>
>     # detection
>     y = sign(x)
>
>     # error counting
>     err = where (y != s)
>     error_sum += len(err[0])
>
>   print 'Eb_No_dB=%2d, BER=%10.4e, Pe=%10.4e' % \
>          (snr, error_sum/(FrameSize*nFrames), Pe)
>
>
> Octave m-file:
> % bpsk_sim.m - octave version
> clear
>
> tic
>
> SNR_MIN    = -1;
> SNR_MAX    = 10;
> FrameSize  = 100;
> Eb_No_dB = SNR_MIN:SNR_MAX;
>
> % signal vector
> s = ones(1,FrameSize);
> a = rand(1,FrameSize);
> a_i = find(a < 0.5);
> s(a_i) = -1;
>
> for snr=SNR_MIN:SNR_MAX
>
>   No = 10^(-snr/10.0);
>   Pe = 0.5*erfc(sqrt(1.0/No));
>   nFrames = ceil(200/FrameSize/Pe);
>   error_sum = 0;
>
>   for frame=1:nFrames
>
>     % noise
>     n = sqrt(No/2)*randn(1,FrameSize);
>     x = s + n;
>
>     y = sign(x);
>
>     err = find( y ~= s);
>     error_sum = error_sum + length(err);
>   end
>
>   fprintf('Eb_No_dB=%2d, BER=%10.4e, Pe=%10.4e\n', ...
>            snr, error_sum/(FrameSize*nFrames), Pe)
> end
> toc
>
>
>
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
>


-- 
French PhD student
Website : http://matthieu-brucher.developpez.com/
Blogs : http://matt.eifelle.com and http://blog.developpez.com/?blog=92
LinkedIn : http://www.linkedin.com/in/matthieubrucher
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080611/6af3b924/attachment.html>

From david at ar.media.kyoto-u.ac.jp  Wed Jun 11 04:03:32 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Wed, 11 Jun 2008 17:03:32 +0900
Subject: [SciPy-user] Some mathematics/statisctics books
In-Reply-To: <da09baf80806101907k1b27fb4g302829dd41dd228e@mail.gmail.com>
References: <da09baf80806101209q7b0b63acs8fa053b93d7dcfce@mail.gmail.com>	<20080610201005.43296f34@ajackson.org>
	<da09baf80806101907k1b27fb4g302829dd41dd228e@mail.gmail.com>
Message-ID: <484F86D4.8030408@ar.media.kyoto-u.ac.jp>

didier rano wrote:
> Wahh so many books to read.
>
> But to analysis graph related to time-series, I don't know if I need 
> more a statistic approach or pure mathematic approach. Maybe that I 
> could use both approaches.

It really depends on what you mean by pure mathematic approach. Purely 
mathematic approach to probabilities and statistics are mostly just 
that: purely mathematical. Don't get me wrong, maths is great, and 
probabilities/statistics are interesting mathematics topics on their 
own, but if you want to handle graphs, time series and all that, I don't 
think it will help you much.

I second the book by Wasserman, although it does not treat a lot of time 
series stuff. But it is concise and precise (it is written with a 
relatively practical POV by someone who is definitely familiar with the 
theory; in particular, there are a lot of subtle examples and counter 
examples which are well explained, contrary to many other books).

Another book. which I have not read entirely yet, but looks related to 
what you are looking for, is the book by Gelman et al.:

"Bayesian Data Analysis", by Gelman A., John B. Carlin 
<http://www.rch.org.au/cebu/staff.cfm?doc_id=5690>, Hal S. Stern 
<http://www.ics.uci.edu/%7Esternh/>, and Donald B. Rubin.
http://www.stat.columbia.edu/~gelman/book/

Not much theory there, but is really oriented toward data analysis as 
the title suggests :)
>
> Thanks for all your help, and sorry for my poor background in 
> mathematics (I need to learn linear algebre too !)

If you do multivariate analysis, you need to be more than familiar with 
linear algebra, I think. I don't know any good reference on this, but 
some open courseware may be nice (they have some video, too):

http://ocw.mit.edu/OcwWeb/Mathematics/index.htm

cheers,

David


From gael.varoquaux at normalesup.org  Wed Jun 11 04:39:27 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Wed, 11 Jun 2008 10:39:27 +0200
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>
Message-ID: <20080611083927.GA3926@phare.normalesup.org>

On Wed, Jun 11, 2008 at 09:15:44AM +0200, Matthieu Brucher wrote:
>    The difference in speed may be only due to the fact that the scipy modle
>    has much more functions than numpy's.

Matthieux does pipoint an interesting point: import time of scipy will be
much larger than import time of numpy. You could try to run the tests in
order to get rid of import time, eg by importing once, and running the
code many times.

However, if I recall correctly, there are other differences in fonctions
between scipy and numpy. Some functions in scipy do more than the
corresponding fonctions in numpy. Here is an example:

In [1]: from scipy import sqrt as ssqrt

In [2]: from numpy import sqrt as nsqrt

In [3]: ssqrt(-1)
Out[3]: 1j

In [4]: nsqrt(-1)
Out[4]: nan

In [5]: %timeit ssqrt(-1)
10000 loops, best of 3: 37.7 ?s per loop

In [6]: %timeit nsqrt(-1)
100000 loops, best of 3: 6.14 ?s per loop

Maybe this is what you are seeing.

Ga?l


From haase at msg.ucsf.edu  Wed Jun 11 04:50:14 2008
From: haase at msg.ucsf.edu (Sebastian Haase)
Date: Wed, 11 Jun 2008 10:50:14 +0200
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <20080611083927.GA3926@phare.normalesup.org>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>
	<20080611083927.GA3926@phare.normalesup.org>
Message-ID: <bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>

On Wed, Jun 11, 2008 at 10:39 AM, Gael Varoquaux
<gael.varoquaux at normalesup.org> wrote:
> On Wed, Jun 11, 2008 at 09:15:44AM +0200, Matthieu Brucher wrote:
>>    The difference in speed may be only due to the fact that the scipy modle
>>    has much more functions than numpy's.
>
> Matthieux does pipoint an interesting point: import time of scipy will be
> much larger than import time of numpy. You could try to run the tests in
> order to get rid of import time, eg by importing once, and running the
> code many times.
>
> However, if I recall correctly, there are other differences in fonctions
> between scipy and numpy. Some functions in scipy do more than the
> corresponding fonctions in numpy. Here is an example:
>
> In [1]: from scipy import sqrt as ssqrt
>
> In [2]: from numpy import sqrt as nsqrt
>
> In [3]: ssqrt(-1)
> Out[3]: 1j
>
> In [4]: nsqrt(-1)
> Out[4]: nan
>
> In [5]: %timeit ssqrt(-1)
> 10000 loops, best of 3: 37.7 ?s per loop
>
> In [6]: %timeit nsqrt(-1)
> 100000 loops, best of 3: 6.14 ?s per loop
>
> Maybe this is what you are seeing.
>

Not to say that this is scary, that two functions, same name, give
different results ....

But is there a list of these differences ? (Maybe a wiki page, like
ScipyNumpyDifferences !?)

Thanks,
Sebastian Haase


From gael.varoquaux at normalesup.org  Wed Jun 11 04:55:55 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Wed, 11 Jun 2008 10:55:55 +0200
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>
	<20080611083927.GA3926@phare.normalesup.org>
	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>
Message-ID: <20080611085555.GC3926@phare.normalesup.org>

On Wed, Jun 11, 2008 at 10:50:14AM +0200, Sebastian Haase wrote:
> Not to say that this is scary, that two functions, same name, give
> different results ....

I don't agree. That's why you have namespaces. numpy.sum and
__builtin__.sum don't give the same results, and I find that expected.

Ga?l


From haase at msg.ucsf.edu  Wed Jun 11 05:18:31 2008
From: haase at msg.ucsf.edu (Sebastian Haase)
Date: Wed, 11 Jun 2008 11:18:31 +0200
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <20080611085555.GC3926@phare.normalesup.org>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>
	<20080611083927.GA3926@phare.normalesup.org>
	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>
	<20080611085555.GC3926@phare.normalesup.org>
Message-ID: <bc657ead0806110218m28888a95l77962828ee1076cf@mail.gmail.com>

On Wed, Jun 11, 2008 at 10:55 AM, Gael Varoquaux
<gael.varoquaux at normalesup.org> wrote:
> On Wed, Jun 11, 2008 at 10:50:14AM +0200, Sebastian Haase wrote:
>> Not to say that this is scary, that two functions, same name, give
>> different results ....
>
> I don't agree. That's why you have namespaces. numpy.sum and
> __builtin__.sum don't give the same results, and I find that expected.
>
Of course -- but scipy.sum and numpy.sum  *I* would intuitively
presume to be "alike".
But I really did not want to bring up a new discussion about this, I
*do* see the point for having a "more user friendly"  version of sqrt
(even if it is slower)
What I said is:
there should be a list of all such differences, to minimize surprises.

-Sebastian


From gaedol at gmail.com  Wed Jun 11 05:20:17 2008
From: gaedol at gmail.com (Marco)
Date: Wed, 11 Jun 2008 09:20:17 +0000
Subject: [SciPy-user] Some mathematics/statisctics books
In-Reply-To: <484F86D4.8030408@ar.media.kyoto-u.ac.jp>
References: <da09baf80806101209q7b0b63acs8fa053b93d7dcfce@mail.gmail.com>
	<20080610201005.43296f34@ajackson.org>
	<da09baf80806101907k1b27fb4g302829dd41dd228e@mail.gmail.com>
	<484F86D4.8030408@ar.media.kyoto-u.ac.jp>
Message-ID: <f53b69070806110220q39fa782fm49dd3094c544259c@mail.gmail.com>

On Wed, Jun 11, 2008 at 8:03 AM, David Cournapeau
<david at ar.media.kyoto-u.ac.jp> wrote:
> didier rano wrote:
>> Wahh so many books to read.
>>
>> But to analysis graph related to time-series, I don't know if I need
>> more a statistic approach or pure mathematic approach. Maybe that I
>> could use both approaches.
>
> It really depends on what you mean by pure mathematic approach. Purely
> mathematic approach to probabilities and statistics are mostly just
> that: purely mathematical. Don't get me wrong, maths is great, and
> probabilities/statistics are interesting mathematics topics on their
> own, but if you want to handle graphs, time series and all that, I don't
> think it will help you much.
>
> I second the book by Wasserman, although it does not treat a lot of time
> series stuff. But it is concise and precise (it is written with a
> relatively practical POV by someone who is definitely familiar with the
> theory; in particular, there are a lot of subtle examples and counter
> examples which are well explained, contrary to many other books).
>
> Another book. which I have not read entirely yet, but looks related to
> what you are looking for, is the book by Gelman et al.:
>
> "Bayesian Data Analysis", by Gelman A., John B. Carlin
> <http://www.rch.org.au/cebu/staff.cfm?doc_id=5690>, Hal S. Stern
> <http://www.ics.uci.edu/%7Esternh/>, and Donald B. Rubin.
> http://www.stat.columbia.edu/~gelman/book/
>
> Not much theory there, but is really oriented toward data analysis as
> the title suggests :)
>>
>> Thanks for all your help, and sorry for my poor background in
>> mathematics (I need to learn linear algebre too !)
>
> If you do multivariate analysis, you need to be more than familiar with
> linear algebra, I think. I don't know any good reference on this, but
> some open courseware may be nice (they have some video, too):
>
> http://ocw.mit.edu/OcwWeb/Mathematics/index.htm
>
> cheers,
>
> David
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>



-- 
Fred Allen  - "Imitation is the sincerest form of television."


From robert.kern at gmail.com  Wed Jun 11 05:20:30 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Wed, 11 Jun 2008 04:20:30 -0500
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <20080611085555.GC3926@phare.normalesup.org>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>
	<20080611083927.GA3926@phare.normalesup.org>
	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>
	<20080611085555.GC3926@phare.normalesup.org>
Message-ID: <3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>

On Wed, Jun 11, 2008 at 03:55, Gael Varoquaux
<gael.varoquaux at normalesup.org> wrote:
> On Wed, Jun 11, 2008 at 10:50:14AM +0200, Sebastian Haase wrote:
>> Not to say that this is scary, that two functions, same name, give
>> different results ....
>
> I don't agree. That's why you have namespaces. numpy.sum and
> __builtin__.sum don't give the same results, and I find that expected.

While I don't *particularly* want to open this can of worms again, the
power of the Internet compels me. To my mind, the big fault with the
current arrangement is not that two functions with the same name have
different functionality, but that scipy.* is *almost* the same
namespace as numpy.* without any clear signposts as to the
differences. It's one thing to have the single common name between
numpy.* and __builtin__.* have different functions in each namespace.
It's another to say that 9 of 489 common names do.

For what it's worth, the "scipy versions" of the 9 functions are all
exposed from numpy.lib.scimath.

What really annoys me is that "from scipy import *" imports all of the
subpackages. Again. I don't know how many times I thought I removed
that nonsense, but like a bloody vampire, it just ... keeps ... coming
... back.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From cohen at slac.stanford.edu  Wed Jun 11 05:31:11 2008
From: cohen at slac.stanford.edu (Johann Cohen-Tanugi)
Date: Wed, 11 Jun 2008 11:31:11 +0200
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>	<20080611083927.GA3926@phare.normalesup.org>	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>	<20080611085555.GC3926@phare.normalesup.org>
	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>
Message-ID: <484F9B5F.80608@slac.stanford.edu>

FWIW:

In [28]: numpy.complex(0,1)**2
Out[28]: (-1+0j)

In [29]: numpy.sqrt(numpy.complex(0,1)**2)
Out[29]: (0.0+1.0j)

In [30]: numpy.sqrt(-1)
Out[30]: nan

So numpy knows about complex, but numpy.sqrt does not cast -1 into a 
complex number. Beyond Gael's valid statement about namespacing,
I find this slightly surprising and maybe even a tad inconsistent, but I 
will certainly not fight a war over it :)
JCT

Robert Kern wrote:
> On Wed, Jun 11, 2008 at 03:55, Gael Varoquaux
> <gael.varoquaux at normalesup.org> wrote:
>   
>> On Wed, Jun 11, 2008 at 10:50:14AM +0200, Sebastian Haase wrote:
>>     
>>> Not to say that this is scary, that two functions, same name, give
>>> different results ....
>>>       
>> I don't agree. That's why you have namespaces. numpy.sum and
>> __builtin__.sum don't give the same results, and I find that expected.
>>     
>
> While I don't *particularly* want to open this can of worms again, the
> power of the Internet compels me. To my mind, the big fault with the
> current arrangement is not that two functions with the same name have
> different functionality, but that scipy.* is *almost* the same
> namespace as numpy.* without any clear signposts as to the
> differences. It's one thing to have the single common name between
> numpy.* and __builtin__.* have different functions in each namespace.
> It's another to say that 9 of 489 common names do.
>
> For what it's worth, the "scipy versions" of the 9 functions are all
> exposed from numpy.lib.scimath.
>
> What really annoys me is that "from scipy import *" imports all of the
> subpackages. Again. I don't know how many times I thought I removed
> that nonsense, but like a bloody vampire, it just ... keeps ... coming
> ... back.
>
>   


From silva at lma.cnrs-mrs.fr  Wed Jun 11 05:43:57 2008
From: silva at lma.cnrs-mrs.fr (Fabrice Silva)
Date: Wed, 11 Jun 2008 11:43:57 +0200
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <484F9B5F.80608@slac.stanford.edu>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>
	<20080611083927.GA3926@phare.normalesup.org>
	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>
	<20080611085555.GC3926@phare.normalesup.org>
	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>
	<484F9B5F.80608@slac.stanford.edu>
Message-ID: <1213177437.3201.20.camel@Portable-s2m.cnrs-mrs.fr>

Python 2.5.2 (r252:60911, May 28 2008, 08:35:32) 
[GCC 4.2.4 (Debian 4.2.4-1)] on linux2
Type "help", "copyright", "credits" or "license" for more information.
>>> import numpy as np
>>> np.sqrt(-1.)
nan
>>> np.sqrt(-1.+0.j)
1j
>>> np.__version__
'1.1.0'

-- 
Fabrice Silva <silva at lma.cnrs-mrs.fr>
LMA UPR CNRS 7051 - ?quipe S2M



From cohen at slac.stanford.edu  Wed Jun 11 05:50:07 2008
From: cohen at slac.stanford.edu (Johann Cohen-Tanugi)
Date: Wed, 11 Jun 2008 11:50:07 +0200
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <1213177437.3201.20.camel@Portable-s2m.cnrs-mrs.fr>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>	<20080611083927.GA3926@phare.normalesup.org>	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>	<20080611085555.GC3926@phare.normalesup.org>	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>	<484F9B5F.80608@slac.stanford.edu>
	<1213177437.3201.20.camel@Portable-s2m.cnrs-mrs.fr>
Message-ID: <484F9FCF.1030203@slac.stanford.edu>

heuh... oui, c'est exactement mon point.
JCT

Fabrice Silva wrote:
> Python 2.5.2 (r252:60911, May 28 2008, 08:35:32) 
> [GCC 4.2.4 (Debian 4.2.4-1)] on linux2
> Type "help", "copyright", "credits" or "license" for more information.
>   
>>>> import numpy as np
>>>> np.sqrt(-1.)
>>>>         
> nan
>   
>>>> np.sqrt(-1.+0.j)
>>>>         
> 1j
>   
>>>> np.__version__
>>>>         
> '1.1.0'
>
>   


From pearu at cens.ioc.ee  Wed Jun 11 05:52:15 2008
From: pearu at cens.ioc.ee (Pearu Peterson)
Date: Wed, 11 Jun 2008 11:52:15 +0200
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>	<20080611083927.GA3926@phare.normalesup.org>	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>	<20080611085555.GC3926@phare.normalesup.org>
	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>
Message-ID: <484FA04F.1000808@cens.ioc.ee>

Robert Kern wrote:

> What really annoys me is that "from scipy import *" imports all of the
> subpackages. Again. I don't know how many times I thought I removed
> that nonsense, but like a bloody vampire, it just ... keeps ... coming
> ... back.

There should be a unittest for that..

Eventhough scipy.__version__ == 0.7.0.dev4124 (that I have at the 
current computer) does not import scipy subpackages, it does import
lots of other stuff that seems unnecessary. For example, importing
nose takes 1/3 of the scipy import time:

pearu at pearu-laptop:~$ python
 >>> %time import scipy
CPU times: user 0.22 s, sys: 0.08 s, total: 0.30 s
Wall time: 0.29
 >>>
pearu at pearu-laptop:~$ python
 >>> %time import nose
CPU times: user 0.07 s, sys: 0.02 s, total: 0.10 s
Wall time: 0.10
 >>> %time import scipy
CPU times: user 0.13 s, sys: 0.04 s, total: 0.17 s
Wall time: 0.18

Another 1/3 of the time goes to importing numpy:
pearu at pearu-laptop:~$ python
 >>> %time import nose
CPU times: user 0.08 s, sys: 0.02 s, total: 0.10 s
Wall time: 0.10
 >>> %time import numpy
CPU times: user 0.07 s, sys: 0.02 s, total: 0.09 s
Wall time: 0.10
 >>> %time import scipy
CPU times: user 0.08 s, sys: 0.04 s, total: 0.11 s
Wall time: 0.11

Pearu


From robert.kern at gmail.com  Wed Jun 11 05:55:37 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Wed, 11 Jun 2008 04:55:37 -0500
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <484FA04F.1000808@cens.ioc.ee>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>
	<20080611083927.GA3926@phare.normalesup.org>
	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>
	<20080611085555.GC3926@phare.normalesup.org>
	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>
	<484FA04F.1000808@cens.ioc.ee>
Message-ID: <3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>

On Wed, Jun 11, 2008 at 04:52, Pearu Peterson <pearu at cens.ioc.ee> wrote:
> Robert Kern wrote:
>
>> What really annoys me is that "from scipy import *" imports all of the
>> subpackages. Again. I don't know how many times I thought I removed
>> that nonsense, but like a bloody vampire, it just ... keeps ... coming
>> ... back.
>
> There should be a unittest for that..
>
> Eventhough scipy.__version__ == 0.7.0.dev4124 (that I have at the
> current computer) does not import scipy subpackages,

Specifically, it is "from scipy import *" that imports the
subpackages. "import scipy" does not.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From pearu at cens.ioc.ee  Wed Jun 11 06:06:15 2008
From: pearu at cens.ioc.ee (Pearu Peterson)
Date: Wed, 11 Jun 2008 12:06:15 +0200
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>	<20080611083927.GA3926@phare.normalesup.org>	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>	<20080611085555.GC3926@phare.normalesup.org>	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>	<484FA04F.1000808@cens.ioc.ee>
	<3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>
Message-ID: <484FA397.5070704@cens.ioc.ee>



Robert Kern wrote:
> On Wed, Jun 11, 2008 at 04:52, Pearu Peterson <pearu at cens.ioc.ee> wrote:
>> Robert Kern wrote:
>>
>>> What really annoys me is that "from scipy import *" imports all of the
>>> subpackages. Again. I don't know how many times I thought I removed
>>> that nonsense, but like a bloody vampire, it just ... keeps ... coming
>>> ... back.
>> There should be a unittest for that..
>>
>> Eventhough scipy.__version__ == 0.7.0.dev4124 (that I have at the
>> current computer) does not import scipy subpackages,
> 
> Specifically, it is "from scipy import *" that imports the
> subpackages. "import scipy" does not.

Ahh, right.

We used to have postponed import hooks for that in past
but afaik we dropped these because they were hackish and at some moment 
the time of importing scipy improved (I think it was in Python 2.4 or 
2.5) considerably.

Nevertheless, the test functions could import nose only when required.
That would save some import time.

Pearu


From fredmfp at gmail.com  Wed Jun 11 06:30:45 2008
From: fredmfp at gmail.com (fred)
Date: Wed, 11 Jun 2008 12:30:45 +0200
Subject: [SciPy-user] array mean issue...
Message-ID: <484FA955.3000009@gmail.com>

Hi,

I get the following issue I don't understand:

marsu:~/{1}/> a=rand(400,400,400)

marsu:~/{2}/> a.mean()
Out[2]: 0.500002086829

marsu:~/{3}/> b=asarray(a, dtype='f')

marsu:~/{4}/> b.mean()
Out[4]: 0.262144

What's going on ?

How can I compute the mean on "big" float arrays ?

By "big", I mean that for array of 300x300x300, I get the same results 
for float32 and float64 arrays.

TIA.

Cheers,

-- 
Fred


From robert.kern at gmail.com  Wed Jun 11 06:33:30 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Wed, 11 Jun 2008 05:33:30 -0500
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <484FA397.5070704@cens.ioc.ee>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>
	<20080611083927.GA3926@phare.normalesup.org>
	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>
	<20080611085555.GC3926@phare.normalesup.org>
	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>
	<484FA04F.1000808@cens.ioc.ee>
	<3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>
	<484FA397.5070704@cens.ioc.ee>
Message-ID: <3d375d730806110333r43316877ne507befa6021bbe3@mail.gmail.com>

On Wed, Jun 11, 2008 at 05:06, Pearu Peterson <pearu at cens.ioc.ee> wrote:
>
> Robert Kern wrote:
>> On Wed, Jun 11, 2008 at 04:52, Pearu Peterson <pearu at cens.ioc.ee> wrote:
>>> Robert Kern wrote:
>>>
>>>> What really annoys me is that "from scipy import *" imports all of the
>>>> subpackages. Again. I don't know how many times I thought I removed
>>>> that nonsense, but like a bloody vampire, it just ... keeps ... coming
>>>> ... back.
>>> There should be a unittest for that..
>>>
>>> Eventhough scipy.__version__ == 0.7.0.dev4124 (that I have at the
>>> current computer) does not import scipy subpackages,
>>
>> Specifically, it is "from scipy import *" that imports the
>> subpackages. "import scipy" does not.
>
> Ahh, right.
>
> We used to have postponed import hooks for that in past
> but afaik we dropped these because they were hackish and at some moment
> the time of importing scipy improved (I think it was in Python 2.4 or
> 2.5) considerably.

The problem is that we left scipy.pkgload in, and creating that from
numpy._import_tools.PackageLoader implicitly adds all of the
subpackages to scipy.__all__.

> Nevertheless, the test functions could import nose only when required.
> That would save some import time.

Yes, I have a patch for that.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From robert.kern at gmail.com  Wed Jun 11 06:44:34 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Wed, 11 Jun 2008 05:44:34 -0500
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <3d375d730806110333r43316877ne507befa6021bbe3@mail.gmail.com>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>
	<20080611083927.GA3926@phare.normalesup.org>
	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>
	<20080611085555.GC3926@phare.normalesup.org>
	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>
	<484FA04F.1000808@cens.ioc.ee>
	<3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>
	<484FA397.5070704@cens.ioc.ee>
	<3d375d730806110333r43316877ne507befa6021bbe3@mail.gmail.com>
Message-ID: <3d375d730806110344h2cf14d51h722589e8606ca2aa@mail.gmail.com>

On Wed, Jun 11, 2008 at 05:33, Robert Kern <robert.kern at gmail.com> wrote:
> On Wed, Jun 11, 2008 at 05:06, Pearu Peterson <pearu at cens.ioc.ee> wrote:

>> We used to have postponed import hooks for that in past
>> but afaik we dropped these because they were hackish and at some moment
>> the time of importing scipy improved (I think it was in Python 2.4 or
>> 2.5) considerably.
>
> The problem is that we left scipy.pkgload in, and creating that from
> numpy._import_tools.PackageLoader implicitly adds all of the
> subpackages to scipy.__all__.

Correction: we actually do call pkgload(postpone=True). However, even
with the postponed import (which does not add any proxy objects), it
still appends to __all__. I believe the following patch to numpy fixes
the problem, but I'm not sure if leaving the if clause alone is
correct in all cases:

Index: numpy/_import_tools.py
===================================================================
--- numpy/_import_tools.py      (revision 5245)
+++ numpy/_import_tools.py      (working copy)
@@ -183,9 +183,6 @@
             postpone_import = getattr(info_module,'postpone_import',False)
             if (postpone and not global_symbols) \
                    or (postpone_import and postpone is not None):
-                self.log('__all__.append(%r)' % (package_name))
-                if '.' not in package_name:
-                    self.parent_export_names.append(package_name)
                 continue

             old_object = frame.f_locals.get(package_name,None)


This can probably go into numpy 1.1.1 as a bugfix, so I don't think
it's critical to work around it in scipy/__init__.py.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From pearu at cens.ioc.ee  Wed Jun 11 07:12:40 2008
From: pearu at cens.ioc.ee (Pearu Peterson)
Date: Wed, 11 Jun 2008 13:12:40 +0200
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <3d375d730806110344h2cf14d51h722589e8606ca2aa@mail.gmail.com>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>	<20080611083927.GA3926@phare.normalesup.org>	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>	<20080611085555.GC3926@phare.normalesup.org>	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>	<484FA04F.1000808@cens.ioc.ee>	<3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>	<484FA397.5070704@cens.ioc.ee>	<3d375d730806110333r43316877ne507befa6021bbe3@mail.gmail.com>
	<3d375d730806110344h2cf14d51h722589e8606ca2aa@mail.gmail.com>
Message-ID: <484FB328.3090405@cens.ioc.ee>



Robert Kern wrote:
> On Wed, Jun 11, 2008 at 05:33, Robert Kern <robert.kern at gmail.com> wrote:
>> On Wed, Jun 11, 2008 at 05:06, Pearu Peterson <pearu at cens.ioc.ee> wrote:
> 
>>> We used to have postponed import hooks for that in past
>>> but afaik we dropped these because they were hackish and at some moment
>>> the time of importing scipy improved (I think it was in Python 2.4 or
>>> 2.5) considerably.
>> The problem is that we left scipy.pkgload in, and creating that from
>> numpy._import_tools.PackageLoader implicitly adds all of the
>> subpackages to scipy.__all__.
> 
> Correction: we actually do call pkgload(postpone=True). However, even
> with the postponed import (which does not add any proxy objects), it
> still appends to __all__. I believe the following patch to numpy fixes
> the problem, but I'm not sure if leaving the if clause alone is
> correct in all cases:
> 
> Index: numpy/_import_tools.py
> ===================================================================
> --- numpy/_import_tools.py      (revision 5245)
> +++ numpy/_import_tools.py      (working copy)
> @@ -183,9 +183,6 @@
>              postpone_import = getattr(info_module,'postpone_import',False)
>              if (postpone and not global_symbols) \
>                     or (postpone_import and postpone is not None):
> -                self.log('__all__.append(%r)' % (package_name))
> -                if '.' not in package_name:
> -                    self.parent_export_names.append(package_name)
>                  continue
> 
>              old_object = frame.f_locals.get(package_name,None)
> 
> 
> This can probably go into numpy 1.1.1 as a bugfix, so I don't think
> it's critical to work around it in scipy/__init__.py.

The patch does not affect the 'from scipy import *' time much.
In my computer it is around 0.62secs in both cases when having
names in __all__ or not.

The long import seems to be due to scipy/linalg/iterative.py that
imports scipy.sparse which takes most of the import time. And 
scipy.linalg should not be imported when importing scipy.
So, I think the real reason is hiding somewhere else.. I am looking into 
it...

Pearu


From ivo.maljevic at gmail.com  Wed Jun 11 07:17:08 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Wed, 11 Jun 2008 07:17:08 -0400
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>
	<20080611083927.GA3926@phare.normalesup.org>
	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>
	<20080611085555.GC3926@phare.normalesup.org>
	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>
	<484FA04F.1000808@cens.ioc.ee>
	<3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>
Message-ID: <826c64da0806110417s3b87fb95w497267004ed7e842@mail.gmail.com>

2008/6/11 Robert Kern <robert.kern at gmail.com>:

Specifically, it is "from scipy import *" that imports the
subpackages. "import scipy" does not.

Sorry, but I am not sure I understand everything here.

1. "import scipy as sp" vs "import numpy as np", and then I added prefixes
in front of all the functions. The times are the same, 44 and 23 seconds,
respectively. There goes that theory.

2. I tried explicitly importing only the required functions, e.g,  from
numpy import sqrt, arange, ones, zeros, random, where, ceil, sign
Agian, the same results.

Would the loading time of the two packages account for over 20 seconds
difference in execution time?

Robert, could you actually try the code I've sent?

Thanks,
Ivo




>
> --
> Robert Kern
>
> "I have come to believe that the whole world is an enigma, a harmless
> enigma that is made terrible by our own mad attempt to interpret it as
> though it had an underlying truth."
>  -- Umberto Eco
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080611/53befef7/attachment.html>

From david at ar.media.kyoto-u.ac.jp  Wed Jun 11 07:13:08 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Wed, 11 Jun 2008 20:13:08 +0900
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <826c64da0806110417s3b87fb95w497267004ed7e842@mail.gmail.com>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>	<20080611083927.GA3926@phare.normalesup.org>	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>	<20080611085555.GC3926@phare.normalesup.org>	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>	<484FA04F.1000808@cens.ioc.ee>	<3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>
	<826c64da0806110417s3b87fb95w497267004ed7e842@mail.gmail.com>
Message-ID: <484FB344.4050905@ar.media.kyoto-u.ac.jp>

Ivo Maljevic wrote:
> 2. I tried explicitly importing only the required functions, e.g,  
> from numpy import sqrt, arange, ones, zeros, random, where, ceil, sign
> Agian, the same results.
>
> Would the loading time of the two packages account for over 20 seconds 
> difference in execution time?

No, I think the discussion carried away from your initial problem. 
Although importing scipy is slow, it certainly  does not take 20 
seconds, unless you are running an ancient computer.

Let me check your problem, which may just show that one function is 
scipy is much slower than a similar one in numpy.

cheers,

David


From pearu at cens.ioc.ee  Wed Jun 11 07:27:22 2008
From: pearu at cens.ioc.ee (Pearu Peterson)
Date: Wed, 11 Jun 2008 13:27:22 +0200
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <484FB328.3090405@cens.ioc.ee>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>	<20080611083927.GA3926@phare.normalesup.org>	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>	<20080611085555.GC3926@phare.normalesup.org>	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>	<484FA04F.1000808@cens.ioc.ee>	<3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>	<484FA397.5070704@cens.ioc.ee>	<3d375d730806110333r43316877ne507befa6021bbe3@mail.gmail.com>	<3d375d730806110344h2cf14d51h722589e8606ca2aa@mail.gmail.com>
	<484FB328.3090405@cens.ioc.ee>
Message-ID: <484FB69A.80207@cens.ioc.ee>

Ok, found it. To fix the long scipy import time one needs to apply
Roberts patch to numpy/_import_tools.py and the following patch to
scipy:

Index: scipy/__init__.py
===================================================================
--- scipy/__init__.py   (revision 4424)
+++ scipy/__init__.py   (working copy)
@@ -31,9 +31,11 @@
  _num.seterr(all='ignore')

  __all__ += ['oldnumeric']+_num.__all__

  __all__ += ['randn', 'rand', 'fft', 'ifft']

+if 'linalg' in __all__:
+    __all__.remove('linalg')
+
  __doc__ += """
  Contents
  --------

With these patches, the 'from scipy import *' time drops from
0.6 to 0.2 in my computer.

Pearu


Pearu Peterson wrote:
> 
> Robert Kern wrote:
>> On Wed, Jun 11, 2008 at 05:33, Robert Kern <robert.kern at gmail.com> wrote:
>>> On Wed, Jun 11, 2008 at 05:06, Pearu Peterson <pearu at cens.ioc.ee> wrote:
>>>> We used to have postponed import hooks for that in past
>>>> but afaik we dropped these because they were hackish and at some moment
>>>> the time of importing scipy improved (I think it was in Python 2.4 or
>>>> 2.5) considerably.
>>> The problem is that we left scipy.pkgload in, and creating that from
>>> numpy._import_tools.PackageLoader implicitly adds all of the
>>> subpackages to scipy.__all__.
>> Correction: we actually do call pkgload(postpone=True). However, even
>> with the postponed import (which does not add any proxy objects), it
>> still appends to __all__. I believe the following patch to numpy fixes
>> the problem, but I'm not sure if leaving the if clause alone is
>> correct in all cases:
>>
>> Index: numpy/_import_tools.py
>> ===================================================================
>> --- numpy/_import_tools.py      (revision 5245)
>> +++ numpy/_import_tools.py      (working copy)
>> @@ -183,9 +183,6 @@
>>              postpone_import = getattr(info_module,'postpone_import',False)
>>              if (postpone and not global_symbols) \
>>                     or (postpone_import and postpone is not None):
>> -                self.log('__all__.append(%r)' % (package_name))
>> -                if '.' not in package_name:
>> -                    self.parent_export_names.append(package_name)
>>                  continue
>>
>>              old_object = frame.f_locals.get(package_name,None)
>>
>>
>> This can probably go into numpy 1.1.1 as a bugfix, so I don't think
>> it's critical to work around it in scipy/__init__.py.
> 
> The patch does not affect the 'from scipy import *' time much.
> In my computer it is around 0.62secs in both cases when having
> names in __all__ or not.
> 
> The long import seems to be due to scipy/linalg/iterative.py that
> imports scipy.sparse which takes most of the import time. And 
> scipy.linalg should not be imported when importing scipy.
> So, I think the real reason is hiding somewhere else.. I am looking into 
> it...
> 
> Pearu
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user


From se.berg at stud.uni-goettingen.de  Wed Jun 11 07:55:59 2008
From: se.berg at stud.uni-goettingen.de (Sebastian Stephan Berg)
Date: Wed, 11 Jun 2008 13:55:59 +0200
Subject: [SciPy-user] array mean issue...
In-Reply-To: <484FA955.3000009@gmail.com>
References: <484FA955.3000009@gmail.com>
Message-ID: <1213185359.5968.8.camel@sebook>

Hello,

This is an overflow problem in the way the mean is computed I believe --
it would start to add nothing after a while. In any case you can get the
correct/better result with b.mean(dtype='float64').
Maybe someone else can give some more info on why. Just to throw it in,
maybe a warning could be issued by checking something using
log2(size*mean) and float precision?

Sebastian



From david at ar.media.kyoto-u.ac.jp  Wed Jun 11 07:45:09 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Wed, 11 Jun 2008 20:45:09 +0900
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <826c64da0806110417s3b87fb95w497267004ed7e842@mail.gmail.com>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>	<20080611083927.GA3926@phare.normalesup.org>	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>	<20080611085555.GC3926@phare.normalesup.org>	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>	<484FA04F.1000808@cens.ioc.ee>	<3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>
	<826c64da0806110417s3b87fb95w497267004ed7e842@mail.gmail.com>
Message-ID: <484FBAC5.9@ar.media.kyoto-u.ac.jp>

Ivo Maljevic wrote:
> 2008/6/11 Robert Kern <robert.kern at gmail.com 
> <mailto:robert.kern at gmail.com>>:
>
> Specifically, it is "from scipy import *" that imports the
> subpackages. "import scipy" does not.
>
> Sorry, but I am not sure I understand everything here.
>
> 1. "import scipy as sp" vs "import numpy as np", and then I added 
> prefixes in front of all the functions. The times are the same, 44 and 
> 23 seconds, respectively. There goes that theory.
>
> 2. I tried explicitly importing only the required functions, e.g,  
> from numpy import sqrt, arange, ones, zeros, random, where, ceil, sign
> Agian, the same results.
>
> Would the loading time of the two packages account for over 20 seconds 
> difference in execution time? 

Ok, the bad guy is..... sqrt. scipy.sqrt is much slower than numpy.sqrt 
(note that in your script, you could avoid computing the scale of the 
normal in the loop).

cheers,

David


From Jean-Pascal.Mercier at inrialpes.fr  Wed Jun 11 08:03:03 2008
From: Jean-Pascal.Mercier at inrialpes.fr (Jean-Pascal Mercier)
Date: Wed, 11 Jun 2008 14:03:03 +0200
Subject: [SciPy-user] array mean issue...
In-Reply-To: <484FA955.3000009@gmail.com>
References: <484FA955.3000009@gmail.com>
Message-ID: <484FBEF7.80908@inrialpes.fr>

Hi Fred,

The problem is the container used to calculate the mean. The default 
behavior is to use the same container as the passed array dtype. When 
the array is really big, the float32 container simply become very 
inefficient which account for the error you've encounter. This can be 
easily solved by providing the dtype as a function parameter.

In [1]: from scipy import *

In [2]: a = rand(400,400,400)

In [3]: b = asarray(a, dtype=float32)

In [4]: a.mean()
Out[4]: 0.500014522905

In [6]: b.mean()
Out[6]: 0.262144

In [6]: b.mean(dtype=float64)
Out[6]: 0.500014522902


Cheers,

/J-Pascal MERCIER
Projet PRIMA - Laboratoire LIG
INRIA Grenoble Rhone-Alpes Research Centre
/


fred wrote:
> Hi,
>
> I get the following issue I don't understand:
>
> marsu:~/{1}/> a=rand(400,400,400)
>
> marsu:~/{2}/> a.mean()
> Out[2]: 0.500002086829
>
> marsu:~/{3}/> b=asarray(a, dtype='f')
>
> marsu:~/{4}/> b.mean()
> Out[4]: 0.262144
>
> What's going on ?
>
> How can I compute the mean on "big" float arrays ?
>
> By "big", I mean that for array of 300x300x300, I get the same results 
> for float32 and float64 arrays.
>
> TIA.
>
> Cheers,
>
>   


From fredmfp at gmail.com  Wed Jun 11 08:08:20 2008
From: fredmfp at gmail.com (fred)
Date: Wed, 11 Jun 2008 14:08:20 +0200
Subject: [SciPy-user] array mean issue...
In-Reply-To: <1213185359.5968.8.camel@sebook>
References: <484FA955.3000009@gmail.com> <1213185359.5968.8.camel@sebook>
Message-ID: <484FC034.2010904@gmail.com>

Sebastian Stephan Berg a ?crit :
> Hello,
> 
> This is an overflow problem in the way the mean is computed I believe --
> it would start to add nothing after a while. In any case you can get the
> correct/better result with b.mean(dtype='float64').
Thanks, Sebastian.

However, I use nansum to compute the mean.
nansum/size gives the same wrong result as mean(), and it seems that is 
not possible to set the dtype with nansum.

Any clue ?


TIA.

Cheers,

-- 
Fred


From fredmfp at gmail.com  Wed Jun 11 08:11:34 2008
From: fredmfp at gmail.com (fred)
Date: Wed, 11 Jun 2008 14:11:34 +0200
Subject: [SciPy-user] array mean issue...
In-Reply-To: <484FBEF7.80908@inrialpes.fr>
References: <484FA955.3000009@gmail.com> <484FBEF7.80908@inrialpes.fr>
Message-ID: <484FC0F6.9090201@gmail.com>

Jean-Pascal Mercier a ?crit :
> Hi Fred,
> 
> The problem is the container used to calculate the mean. The default 
> behavior is to use the same container as the passed array dtype. When 
> the array is really big, the float32 container simply become very 
> inefficient which account for the error you've encounter. This can be 
> easily solved by providing the dtype as a function parameter.
Thanks to you too, Jean-Pascal.

But I still have an issue I have just posted, with nansum.


Cheers,

-- 
Fred


From fredmfp at gmail.com  Wed Jun 11 08:18:12 2008
From: fredmfp at gmail.com (fred)
Date: Wed, 11 Jun 2008 14:18:12 +0200
Subject: [SciPy-user] specify NaN...
Message-ID: <484FC284.7010308@gmail.com>

Hi again,

In my set of data, a few has "really" NaN, others has "special" value
for it (say -9999 for positive integer arrays).

So, I don't want to take in account these values to compute, min, max,
mean and so on.

How could I do this ?
I recall that the arrays are quite "big", ie ~500x500x500 but <
1000x1000x1000.

Any clue ?

TIA.

Cheers,

-- 
Fred



From david at ar.media.kyoto-u.ac.jp  Wed Jun 11 08:23:13 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Wed, 11 Jun 2008 21:23:13 +0900
Subject: [SciPy-user] specify NaN...
In-Reply-To: <484FC284.7010308@gmail.com>
References: <484FC284.7010308@gmail.com>
Message-ID: <484FC3B1.3000102@ar.media.kyoto-u.ac.jp>

fred wrote:
> Hi again,
>
> In my set of data, a few has "really" NaN, others has "special" value
> for it (say -9999 for positive integer arrays).
>
> So, I don't want to take in account these values to compute, min, max,
> mean and so on.
>   

If you do not want to take into accout both special and nan, why not 
using masked array ?

David


From bsouthey at gmail.com  Wed Jun 11 09:14:50 2008
From: bsouthey at gmail.com (Bruce Southey)
Date: Wed, 11 Jun 2008 08:14:50 -0500
Subject: [SciPy-user] array mean issue...
In-Reply-To: <484FC0F6.9090201@gmail.com>
References: <484FA955.3000009@gmail.com> <484FBEF7.80908@inrialpes.fr>
	<484FC0F6.9090201@gmail.com>
Message-ID: <484FCFCA.30807@gmail.com>

fred wrote:
> Jean-Pascal Mercier a ?crit :
>   
>> Hi Fred,
>>
>> The problem is the container used to calculate the mean. The default 
>> behavior is to use the same container as the passed array dtype. When 
>> the array is really big, the float32 container simply become very 
>> inefficient which account for the error you've encounter. This can be 
>> easily solved by providing the dtype as a function parameter.
>>     
> Thanks to you too, Jean-Pascal.
>
> But I still have an issue I have just posted, with nansum.
>
>
> Cheers,
>
>   
The issue is the numerical precision (as indicated) and algorithm. See 
also the  Numpy-discussion thread 'calculating the mean and variance of 
a large float vector' 
(http://projects.scipy.org/pipermail/numpy-discussion/2008-June/034766.html).

Perhaps there is a case to promote everything to float128.

Bruce


From se.berg at stud.uni-goettingen.de  Wed Jun 11 09:36:41 2008
From: se.berg at stud.uni-goettingen.de (Sebastian Stephan Berg)
Date: Wed, 11 Jun 2008 15:36:41 +0200
Subject: [SciPy-user] array mean issue...
In-Reply-To: <484FC034.2010904@gmail.com>
References: <484FA955.3000009@gmail.com> <1213185359.5968.8.camel@sebook>
	<484FC034.2010904@gmail.com>
Message-ID: <1213191401.5968.13.camel@sebook>

I am not familiar with things really, but looking at ??nansum code:

def nansum(a, axis=None):
    """Sum the array over the given axis, treating NaNs as 0.
    """
    y = array(a,subok=True)
    if not issubclass(y.dtype.type, _nx.integer):
        y[isnan(a)] = 0
    return y.sum(axis)

If this is all there is to nansum, you can just do this small
replacement with y[isnan[a]] = 0 and your special values including the
isnan, and then again use sum with a more precise dtype.

Regards,

Sebastian



From ivo.maljevic at gmail.com  Wed Jun 11 09:40:27 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Wed, 11 Jun 2008 09:40:27 -0400
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <484FB344.4050905@ar.media.kyoto-u.ac.jp>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>
	<20080611083927.GA3926@phare.normalesup.org>
	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>
	<20080611085555.GC3926@phare.normalesup.org>
	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>
	<484FA04F.1000808@cens.ioc.ee>
	<3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>
	<826c64da0806110417s3b87fb95w497267004ed7e842@mail.gmail.com>
	<484FB344.4050905@ar.media.kyoto-u.ac.jp>
Message-ID: <826c64da0806110640n4e3c7322l349b98310c5c469c@mail.gmail.com>

Thank you David. Few pointers:

1. You can use scipy's implementation of erfc() function. It will not make
the difference in relative results. That is:

from scipy.special import erfc

and than replace erfc.erfc with erfc

2. Robert discounted the speed difference observation as a nonsense that
keeps coming back because he saw "from scipy import *" statement. I believe
it would have been better if he looked at the numbers. When I change/reduce
the number of inner loops (test 2 case, where I make the vector longer), the
results for scipy and numpy are almost the same, which does not fit well
with his comments about how to import scipy/numpy.

You can easily exagarate the problem by reducing the FrameSize parameter to
10. The execution time difference becomes huge.

Then again, it is my mistake that I did not elaborate the
observation/"problem" more precisely.

Another interesting thing is that octave beats even C and Fortran when there
is effectively only one (snr) loop in the script. Could be that I am not
optimizing the C and Fortran code properly. I did use -O3 option, but octave
probably has randn function optimized, whereas I have a function that does a
lot of things just to produce a gaussian random number.

Thanks,
Ivo Maljevic



2008/6/11 David Cournapeau <david at ar.media.kyoto-u.ac.jp>:

> Ivo Maljevic wrote:
> > 2. I tried explicitly importing only the required functions, e.g,
> > from numpy import sqrt, arange, ones, zeros, random, where, ceil, sign
> > Agian, the same results.
> >
> > Would the loading time of the two packages account for over 20 seconds
> > difference in execution time?
>
> No, I think the discussion carried away from your initial problem.
> Although importing scipy is slow, it certainly  does not take 20
> seconds, unless you are running an ancient computer.
>
> Let me check your problem, which may just show that one function is
> scipy is much slower than a similar one in numpy.
>
> cheers,
>
> David
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080611/8989bd7f/attachment.html>

From ivo.maljevic at gmail.com  Wed Jun 11 09:42:46 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Wed, 11 Jun 2008 09:42:46 -0400
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <484FBAC5.9@ar.media.kyoto-u.ac.jp>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>
	<20080611083927.GA3926@phare.normalesup.org>
	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>
	<20080611085555.GC3926@phare.normalesup.org>
	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>
	<484FA04F.1000808@cens.ioc.ee>
	<3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>
	<826c64da0806110417s3b87fb95w497267004ed7e842@mail.gmail.com>
	<484FBAC5.9@ar.media.kyoto-u.ac.jp>
Message-ID: <826c64da0806110642y28da67a8g1186e0b82a9443fc@mail.gmail.com>

Wow, I was typing a message and when I hit "Send" I found all these replies.
Robert, my apologies for my "counterattack" and thank you for the
explanations.

Thanks,
Ivo

2008/6/11 David Cournapeau <david at ar.media.kyoto-u.ac.jp>:

> Ivo Maljevic wrote:
> > 2008/6/11 Robert Kern <robert.kern at gmail.com
> > <mailto:robert.kern at gmail.com>>:
> >
> > Specifically, it is "from scipy import *" that imports the
> > subpackages. "import scipy" does not.
> >
> > Sorry, but I am not sure I understand everything here.
> >
> > 1. "import scipy as sp" vs "import numpy as np", and then I added
> > prefixes in front of all the functions. The times are the same, 44 and
> > 23 seconds, respectively. There goes that theory.
> >
> > 2. I tried explicitly importing only the required functions, e.g,
> > from numpy import sqrt, arange, ones, zeros, random, where, ceil, sign
> > Agian, the same results.
> >
> > Would the loading time of the two packages account for over 20 seconds
> > difference in execution time?
>
> Ok, the bad guy is..... sqrt. scipy.sqrt is much slower than numpy.sqrt
> (note that in your script, you could avoid computing the scale of the
> normal in the loop).
>
> cheers,
>
> David
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080611/d26d3710/attachment.html>

From ivo.maljevic at gmail.com  Wed Jun 11 09:44:29 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Wed, 11 Jun 2008 09:44:29 -0400
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <826c64da0806110642y28da67a8g1186e0b82a9443fc@mail.gmail.com>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>
	<20080611083927.GA3926@phare.normalesup.org>
	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>
	<20080611085555.GC3926@phare.normalesup.org>
	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>
	<484FA04F.1000808@cens.ioc.ee>
	<3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>
	<826c64da0806110417s3b87fb95w497267004ed7e842@mail.gmail.com>
	<484FBAC5.9@ar.media.kyoto-u.ac.jp>
	<826c64da0806110642y28da67a8g1186e0b82a9443fc@mail.gmail.com>
Message-ID: <826c64da0806110644g36d60550ud74c1d35973d11f9@mail.gmail.com>

Actually, it was David who responded, so thanks David. My apology for Robert
still holds though.

Ivo

2008/6/11 Ivo Maljevic <ivo.maljevic at gmail.com>:

> Wow, I was typing a message and when I hit "Send" I found all these
> replies. Robert, my apologies for my "counterattack" and thank you for the
> explanations.
>
> Thanks,
> Ivo
>
> 2008/6/11 David Cournapeau <david at ar.media.kyoto-u.ac.jp>:
>
>> Ivo Maljevic wrote:
>>
>> > 2008/6/11 Robert Kern <robert.kern at gmail.com
>> > <mailto:robert.kern at gmail.com>>:
>> >
>> > Specifically, it is "from scipy import *" that imports the
>> > subpackages. "import scipy" does not.
>> >
>> > Sorry, but I am not sure I understand everything here.
>> >
>> > 1. "import scipy as sp" vs "import numpy as np", and then I added
>> > prefixes in front of all the functions. The times are the same, 44 and
>> > 23 seconds, respectively. There goes that theory.
>> >
>> > 2. I tried explicitly importing only the required functions, e.g,
>> > from numpy import sqrt, arange, ones, zeros, random, where, ceil, sign
>> > Agian, the same results.
>> >
>> > Would the loading time of the two packages account for over 20 seconds
>> > difference in execution time?
>>
>> Ok, the bad guy is..... sqrt. scipy.sqrt is much slower than numpy.sqrt
>> (note that in your script, you could avoid computing the scale of the
>> normal in the loop).
>>
>> cheers,
>>
>> David
>> _______________________________________________
>> SciPy-user mailing list
>> SciPy-user at scipy.org
>> http://projects.scipy.org/mailman/listinfo/scipy-user
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080611/ee3135ec/attachment.html>

From david at ar.media.kyoto-u.ac.jp  Wed Jun 11 10:20:23 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Wed, 11 Jun 2008 23:20:23 +0900
Subject: [SciPy-user] NumPy vs. SciPy and other speed comparisons
In-Reply-To: <826c64da0806110640n4e3c7322l349b98310c5c469c@mail.gmail.com>
References: <826c64da0806101749w4cc8084h27edff6b47160859@mail.gmail.com>	<e76aa17f0806110015t16335bbav8da179aeff70492@mail.gmail.com>	<20080611083927.GA3926@phare.normalesup.org>	<bc657ead0806110150j159c578fja93c7b61a3d30a34@mail.gmail.com>	<20080611085555.GC3926@phare.normalesup.org>	<3d375d730806110220u432e808fkf292bd8b9765b91a@mail.gmail.com>	<484FA04F.1000808@cens.ioc.ee>	<3d375d730806110255icfd6ebdv5b02ccd04d2a02a2@mail.gmail.com>	<826c64da0806110417s3b87fb95w497267004ed7e842@mail.gmail.com>	<484FB344.4050905@ar.media.kyoto-u.ac.jp>
	<826c64da0806110640n4e3c7322l349b98310c5c469c@mail.gmail.com>
Message-ID: <484FDF27.7030007@ar.media.kyoto-u.ac.jp>

Ivo Maljevic wrote:
>
> 2. Robert discounted the speed difference observation as a nonsense 
> that keeps coming back because he saw "from scipy import *" statement. 
> I believe it would have been better if he looked at the numbers.

I think Robert just pointed out that he was tired of seeing erroneous 
import in scipy coming back when he fixed it some time ago.

David


From fredmfp at gmail.com  Wed Jun 11 10:47:05 2008
From: fredmfp at gmail.com (fred)
Date: Wed, 11 Jun 2008 16:47:05 +0200
Subject: [SciPy-user] array mean issue...
In-Reply-To: <1213191401.5968.13.camel@sebook>
References: <484FA955.3000009@gmail.com>
	<1213185359.5968.8.camel@sebook>	<484FC034.2010904@gmail.com>
	<1213191401.5968.13.camel@sebook>
Message-ID: <484FE569.70206@gmail.com>

Sebastian Stephan Berg a ?crit :
> I am not familiar with things really, but looking at ??nansum code:
> 
> def nansum(a, axis=None):
>     """Sum the array over the given axis, treating NaNs as 0.
>     """
>     y = array(a,subok=True)
>     if not issubclass(y.dtype.type, _nx.integer):
>         y[isnan(a)] = 0
>     return y.sum(axis)
> 
> If this is all there is to nansum, you can just do this small
> replacement with y[isnan[a]] = 0 and your special values including the
> isnan, and then again use sum with a more precise dtype.
Ok, thanks to all of you.

I'll look at it asap.

However, I wonder why there is no nan*** methods related to arrays,
like array.mean(), array.min(), etc.
And also why there is no nanmean & nanstd method...


Cheers,

-- 
Fred


From gary.pajer at gmail.com  Wed Jun 11 10:47:48 2008
From: gary.pajer at gmail.com (Gary Pajer)
Date: Wed, 11 Jun 2008 10:47:48 -0400
Subject: [SciPy-user] LabVIEW and LabPython
In-Reply-To: <1213132836.25044.13.camel@pc2.cole.uklinux.net>
References: <ff223aae0806031923u1dd2c613k7ee8e5f27c1753e8@mail.gmail.com>
	<88fe22a0806081521j2db5b1eblb68692af2e3beb43@mail.gmail.com>
	<1213132836.25044.13.camel@pc2.cole.uklinux.net>
Message-ID: <88fe22a0806110747p408f6c85nb9db9ec9492c7995@mail.gmail.com>

On Tue, Jun 10, 2008 at 5:20 PM, Bryan Cole <bryan at cole.uklinux.net> wrote:

>
> >
> > I've married DAQmx to scipy/numpy to Traits UI
> > ( http://code.enthought.com/projects/traits ) and I have a beautiful
> > GUI based data acquisition/display/analysis system.  I'll admit that
> > I've been too scared to even try to implement callbacks (used by a
> > small number of DAQmx C functions) but I've achieved the same
> > functionality using python threads.
>
> FWIW Python callbacks with DAQmx work great. ctypes makes these easy to
> implement.
>
> I'm also considering Trait'ification of my primary data-acquisition
> application. I'd like to see what you've done in this respect. Are you
> using Chaco?


Yes, I'm using Chaco. I've attached a screenshot.  It's a work in
progress.   All essential functionality works, and I use it daily.   When it
runs it looks more or less like an oscilloscope (but it does more than
that).  On the whole, it's about half done.

The hard parts were not scipy/numpy/Traits related.  The hard parts were
understanding the timing, synchronization, read, and write mechanisms in
DAQmx; I was starting from scratch with no knowledge.  The NI docs are quite
spare.  The NI forum helped with that, although it took several days to get
replies.  Andrew Straw has translated one of the NI examples to
python/ctypes.  That helped me get started.

The chaco part is just a little sticky because the docs just aren't there
yet.  But the response to questions on the enthought forum is almost
instantaneous.

The Traits/Traits UI part is *so easy*.   I've made GUIs in Matlab and WX
and Tkinter, but making them in Traits UI is easier and more sensible and
easier to maintain and upgrade.  The learning curve is much shorter than WX
or Tkinter.  For those who care: I'm not using Traits 3, but rather Traits
2.

Question on callbacks:  if I write a callback routine in python, do I simply
use the python name of the routine as the callback parameter in the call
(via ctypes) to the DAQmx routine in the dll?

-gary



>
>
> BC
>
>
>
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080611/72af952b/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: acquire_77.png
Type: image/png
Size: 36494 bytes
Desc: not available
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080611/72af952b/attachment.png>

From fredmfp at gmail.com  Wed Jun 11 10:48:00 2008
From: fredmfp at gmail.com (fred)
Date: Wed, 11 Jun 2008 16:48:00 +0200
Subject: [SciPy-user] specify NaN...
In-Reply-To: <484FC3B1.3000102@ar.media.kyoto-u.ac.jp>
References: <484FC284.7010308@gmail.com>
	<484FC3B1.3000102@ar.media.kyoto-u.ac.jp>
Message-ID: <484FE5A0.50604@gmail.com>

David Cournapeau a ?crit :

> If you do not want to take into accout both special and nan, why not 
> using masked array ?
Because I did not event think about it ;-)

Thanks for the hint, David.

Cheers,

-- 
Fred


From david at ar.media.kyoto-u.ac.jp  Wed Jun 11 10:44:39 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Wed, 11 Jun 2008 23:44:39 +0900
Subject: [SciPy-user] array mean issue...
In-Reply-To: <484FE569.70206@gmail.com>
References: <484FA955.3000009@gmail.com>	<1213185359.5968.8.camel@sebook>	<484FC034.2010904@gmail.com>	<1213191401.5968.13.camel@sebook>
	<484FE569.70206@gmail.com>
Message-ID: <484FE4D7.5080006@ar.media.kyoto-u.ac.jp>

fred wrote:
> I'll look at it asap.
>
> However, I wonder why there is no nan*** methods related to arrays,
> like array.mean(), array.min(), etc.
> And also why there is no nanmean & nanstd method...
>   

there is:

 >> from scipy.stats.stats import nanmean
 >> import numpy as np
 >> nanmean([np.Nan, 1])
1.0

If you are about to say "how the hell would I know that", you may well 
be right. I am fixing this to put them into scipy.stats instead of 
scipy.stats.stats

cheers,

David


From fredmfp at gmail.com  Wed Jun 11 11:04:05 2008
From: fredmfp at gmail.com (fred)
Date: Wed, 11 Jun 2008 17:04:05 +0200
Subject: [SciPy-user] array mean issue...
In-Reply-To: <484FE4D7.5080006@ar.media.kyoto-u.ac.jp>
References: <484FA955.3000009@gmail.com>	<1213185359.5968.8.camel@sebook>	<484FC034.2010904@gmail.com>	<1213191401.5968.13.camel@sebook>	<484FE569.70206@gmail.com>
	<484FE4D7.5080006@ar.media.kyoto-u.ac.jp>
Message-ID: <484FE965.6080501@gmail.com>

David Cournapeau a ?crit :

> If you are about to say "how the hell would I know that", you may well 
> be right. I am fixing this to put them into scipy.stats instead of 
> scipy.stats.stats
Thanks !! :-))

Cheers,

-- 
Fred


From lists at vrbka.net  Wed Jun 11 11:15:45 2008
From: lists at vrbka.net (Lubos Vrbka)
Date: Wed, 11 Jun 2008 17:15:45 +0200
Subject: [SciPy-user] optimal way to solve large amount of sets of linear
	equations
Message-ID: <484FEC21.3090306@vrbka.net>

hi guys,

i have to solve a set of equations, that can be in matrix notation 
written as
   {E-C}H = C
where E is identity matrix. actually as a result, i want to have 
function G, defined as H - C, i.e.
   G = H - C = {E-C}^(-1)C - C
the problem is, that each of the matrices H, C, G represent set of 
discretized functions. then, i have to solve this problem for every step
   G(1) = {E-C(1)}^(-1)C(1) - C(1)
   G(2) = {E-C(2)}^(-1)C(2) - C(2)
   ...
the matrices themselves are usually small (2x2, 3x3), but the number 
these equations is quite large (1024, 2048, 4096, or 8192; usually not 
larger)

i tried to use the following pieces of code, but don't know (since i'm a 
newbie to scipy), whether it cannot be done in a better or (less stupid) 
way.

i'd be really very grateful for any comments.

G_f_ij is scipy.array((n,n,npoints))
E_ij is scipy.eye(npoints)
c_f_ij is scipy.array((n,n,npoints))

the slow way:
===============
for i in range(npoints):
   # solve the matrix problem for all dr
   G_f_ij[:,:,i] = scipy.mat((E_ij - c_f_ij[:,:,i])).I * 
scipy.mat(c_f_ij[:,:,i]) - scipy.mat(c_f_ij[:,:,i])

the faster way:
===============
for i in range(npoints):
   G_f_ij[:,:,dr] = numpy.linalg.solve(scipy.mat((E_ij - 
c_f_ij[:,:,dr])), scipy.mat(c_f_ij[:,:,dr])) - scipy.mat(c_f_ij[:,:,dr])

the fastest way:
===============
forget the scipy.mat as it's not needed for linalg
for i in range(npoints):
   G_f_ij[:,:,dr] = numpy.linalg.solve(E_ij - c_f_ij[:,:,dr], 
c_f_ij[:,:,dr]) - c_f_ij[:,:,dr]

the last snippet of code is roughly 2x faster than the first one. for 
2x2x1024 points it takes 0.2, 1.3, 0.7 seconds on my computer. alone, 
that's not that bad - but when i need to repeat that 100-1000x during 
the execution of the program, then it starts to take a lot of time...

best regards,
lubos

-- 
Lubos _ at _"
http://www.lubos.vrbka.net


From david at ar.media.kyoto-u.ac.jp  Wed Jun 11 11:25:36 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Thu, 12 Jun 2008 00:25:36 +0900
Subject: [SciPy-user] array mean issue...
In-Reply-To: <484FE965.6080501@gmail.com>
References: <484FA955.3000009@gmail.com>	<1213185359.5968.8.camel@sebook>	<484FC034.2010904@gmail.com>	<1213191401.5968.13.camel@sebook>	<484FE569.70206@gmail.com>	<484FE4D7.5080006@ar.media.kyoto-u.ac.jp>
	<484FE965.6080501@gmail.com>
Message-ID: <484FEE70.1040205@ar.media.kyoto-u.ac.jp>

fred wrote:
>
> Thanks !! :-))
>   

Fixed in trunk (r4426). I also noticed that for some reason, test_stats 
was not picked up by nosetests, which should be fixed now.

cheers,

David


From ivo.maljevic at gmail.com  Wed Jun 11 11:48:32 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Wed, 11 Jun 2008 11:48:32 -0400
Subject: [SciPy-user] Speed differences in sqrt calculation: what is good to
	know
Message-ID: <826c64da0806110848r24248337jaf3caa2819e6906d@mail.gmail.com>

Based on comments from Gael Varoquaux and David Cournapeau , I did the
execution time test.
At least for me, it is clear that if the number is a real scalar, AND the
expected result is also real,
the best way is to call the math version of sqrt() function. The differences
are more than significant, as you can see:

In [3]: from math import sqrt as msqrt
In [4]: from numpy import sqrt as nsqrt
In [5]: from scipy import sqrt as ssqrt

In [6]: %timeit msqrt(3.14)
1000000 loops, best of 3: 479 ns per loop

In [7]: %timeit nsqrt(3.14)
100000 loops, best of 3: 10.8 ?s per loop

In [8]: %timeit ssqrt(3.14)
10000 loops, best of 3: 74.5 ?s per loop
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080611/b1d1fca0/attachment.html>

From david at ar.media.kyoto-u.ac.jp  Wed Jun 11 11:54:43 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Thu, 12 Jun 2008 00:54:43 +0900
Subject: [SciPy-user] Speed differences in sqrt calculation: what is
 good to	know
In-Reply-To: <826c64da0806110848r24248337jaf3caa2819e6906d@mail.gmail.com>
References: <826c64da0806110848r24248337jaf3caa2819e6906d@mail.gmail.com>
Message-ID: <484FF543.10605@ar.media.kyoto-u.ac.jp>

Ivo Maljevic wrote:
> Based on comments from Gael Varoquaux and David Cournapeau , I did the 
> execution time test.
> At least for me, it is clear that if the number is a real scalar, AND 
> the expected result is also real,
> the best way is to call the math version of sqrt() function. The 
> differences are more than significant, as you can see:
>

It is expected for numpy/scipy functions to be much slower than python 
*for scalar*. Since they are optimized to be used with arrays, you are 
paying the cost to initialize the machinery to handle arrays (ufunc), 
without the benefit.

The problem really is the difference between numpy and scipy. The funny 
thing is that for arrays, scipy.sqrt (that is, numpy.lib.scimath.sqrt) 
is faster than numpy.sqrt, which does not quite make sense to me since 
scipy.sqrt calls numpy.sqrt... But it is 1 a.m, so maybe I should just 
get some sleep

cheers,

David


From Karl.Young at ucsf.edu  Wed Jun 11 14:07:18 2008
From: Karl.Young at ucsf.edu (Karl Young)
Date: Wed, 11 Jun 2008 11:07:18 -0700
Subject: [SciPy-user] ANN: SciPy 2008 Conference
In-Reply-To: <c7009a550806101410h4725b984j6c9cce651501bfd4@mail.gmail.com>
References: <c7009a550806101410h4725b984j6c9cce651501bfd4@mail.gmail.com>
Message-ID: <48501456.4030306@ucsf.edu>


Hey Jarrod,

How are you doing ? Will you be in Melbourne ? I was thinking of 
submitting something for SciPy 2008 but wanted to get your take on it 
before doing so. The content is basically estimating pattern complexity 
from brain images for diagnostic purposes (the general content of a 
couple of papers in Neuroimage and Human Brain Mapping). I did the 
analysis using SciPy (and a little Rpy) but I'm sure the code is pretty 
brain dead and wouldn't teach the SciPy community anything - so can 
these submissions be more about content, e.g. gee look what I did using 
SciPy ?

>Greetings,
>
>The SciPy 2008 Conference website is now open:  http://conference.scipy.org
>
>This year's conference will be at Caltech from August 19-24:
>
>Tutorials:        August 19-20 (Tuesday and Wednesday)
>Conference:  August 21-22 (Thursday and Friday)
>Sprints:           August 23-24 (Saturday and Sunday)
>
>Exciting things are happening in the Python community, and the SciPy
>2008 Conference is an excellent opportunity to exchange ideas, learn
>techniques, contribute code and affect the direction of scientific
>computing (or just to learn what all the fuss is about).  We'll be
>announcing the Keynote Speaker and providing a detailed schedule in
>the coming weeks.
>
>This year we are asking presenters to submit short papers to be included
>in the conference proceedings:  http://conference.scipy.org/call_for_papers
>
>Cheers,
>
>  
>


-- 

Karl Young
Center for Imaging of Neurodegenerative Diseases, UCSF          
VA Medical Center (114M)              Phone:  (415) 221-4810 x3114  lab        
4150 Clement Street                   FAX:    (415) 668-2864
San Francisco, CA 94121               Email:  karl young at ucsf edu



From stefan at sun.ac.za  Wed Jun 11 14:42:02 2008
From: stefan at sun.ac.za (=?ISO-8859-1?Q?St=E9fan_van_der_Walt?=)
Date: Wed, 11 Jun 2008 20:42:02 +0200
Subject: [SciPy-user] ANN: SciPy 2008 Conference
In-Reply-To: <48501456.4030306@ucsf.edu>
References: <c7009a550806101410h4725b984j6c9cce651501bfd4@mail.gmail.com>
	<48501456.4030306@ucsf.edu>
Message-ID: <9457e7c80806111142o2518d6e4odf6265609190a369@mail.gmail.com>

2008/6/11 Karl Young <Karl.Young at ucsf.edu>:
> How are you doing ? Will you be in Melbourne ? I was thinking of
> submitting something for SciPy 2008 but wanted to get your take on it
> before doing so. The content is basically estimating pattern complexity
> from brain images for diagnostic purposes (the general content of a
> couple of papers in Neuroimage and Human Brain Mapping). I did the
> analysis using SciPy (and a little Rpy) but I'm sure the code is pretty
> brain dead and wouldn't teach the SciPy community anything - so can
> these submissions be more about content, e.g. gee look what I did using
> SciPy ?

I don't know if you intended this e-mail for such a wide audience, but
it sounds very interesting!  I look forward to seeing your talk.

Regards
St?fan


From matthieu.brucher at gmail.com  Wed Jun 11 14:54:48 2008
From: matthieu.brucher at gmail.com (Matthieu Brucher)
Date: Wed, 11 Jun 2008 20:54:48 +0200
Subject: [SciPy-user] Speed differences in sqrt calculation: what is
	good to know
In-Reply-To: <826c64da0806110848r24248337jaf3caa2819e6906d@mail.gmail.com>
References: <826c64da0806110848r24248337jaf3caa2819e6906d@mail.gmail.com>
Message-ID: <e76aa17f0806111154y7c0343a8sbee06a0d85c00c55@mail.gmail.com>

Hi,

Don't forget that numpy's sqrt and scipy's sqrt are not optimized towards
single computations.
If you really want to compare them, time the square root of 10000 elements.

Matthieu

2008/6/11 Ivo Maljevic <ivo.maljevic at gmail.com>:

> Based on comments from Gael Varoquaux and David Cournapeau , I did the
> execution time test.
> At least for me, it is clear that if the number is a real scalar, AND the
> expected result is also real,
> the best way is to call the math version of sqrt() function. The
> differences are more than significant, as you can see:
>
> In [3]: from math import sqrt as msqrt
> In [4]: from numpy import sqrt as nsqrt
> In [5]: from scipy import sqrt as ssqrt
>
> In [6]: %timeit msqrt(3.14)
> 1000000 loops, best of 3: 479 ns per loop
>
> In [7]: %timeit nsqrt(3.14)
> 100000 loops, best of 3: 10.8 ?s per loop
>
> In [8]: %timeit ssqrt(3.14)
> 10000 loops, best of 3: 74.5 ?s per loop
>
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
>


-- 
French PhD student
Website : http://matthieu-brucher.developpez.com/
Blogs : http://matt.eifelle.com and http://blog.developpez.com/?blog=92
LinkedIn : http://www.linkedin.com/in/matthieubrucher
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080611/f5c95488/attachment.html>

From Karl.Young at ucsf.edu  Wed Jun 11 15:25:46 2008
From: Karl.Young at ucsf.edu (Karl Young)
Date: Wed, 11 Jun 2008 12:25:46 -0700
Subject: [SciPy-user] ANN: SciPy 2008 Conference
In-Reply-To: <9457e7c80806111142o2518d6e4odf6265609190a369@mail.gmail.com>
References: <c7009a550806101410h4725b984j6c9cce651501bfd4@mail.gmail.com>	<48501456.4030306@ucsf.edu>
	<9457e7c80806111142o2518d6e4odf6265609190a369@mail.gmail.com>
Message-ID: <485026BA.1000405@ucsf.edu>


Yeah, thanks Stefan, this was a case of hitting return before putting 
Jarrod's address in the To field - but coincidentally it might also 
pertain to others on the list thinking about submitting something so I 
guess  it wasn't the worst case of operator error...

>2008/6/11 Karl Young <Karl.Young at ucsf.edu>:
>  
>
>>How are you doing ? Will you be in Melbourne ? I was thinking of
>>submitting something for SciPy 2008 but wanted to get your take on it
>>before doing so. The content is basically estimating pattern complexity
>>from brain images for diagnostic purposes (the general content of a
>>couple of papers in Neuroimage and Human Brain Mapping). I did the
>>analysis using SciPy (and a little Rpy) but I'm sure the code is pretty
>>brain dead and wouldn't teach the SciPy community anything - so can
>>these submissions be more about content, e.g. gee look what I did using
>>SciPy ?
>>    
>>
>
>I don't know if you intended this e-mail for such a wide audience, but
>it sounds very interesting!  I look forward to seeing your talk.
>
>Regards
>St?fan
>_______________________________________________
>SciPy-user mailing list
>SciPy-user at scipy.org
>http://projects.scipy.org/mailman/listinfo/scipy-user
>
>  
>


-- 

Karl Young
Center for Imaging of Neurodegenerative Diseases, UCSF          
VA Medical Center (114M)              Phone:  (415) 221-4810 x3114  lab        
4150 Clement Street                   FAX:    (415) 668-2864
San Francisco, CA 94121               Email:  karl young at ucsf edu



From gael.varoquaux at normalesup.org  Wed Jun 11 15:54:51 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Wed, 11 Jun 2008 21:54:51 +0200
Subject: [SciPy-user] ANN: SciPy 2008 Conference
In-Reply-To: <485026BA.1000405@ucsf.edu>
References: <c7009a550806101410h4725b984j6c9cce651501bfd4@mail.gmail.com>
	<48501456.4030306@ucsf.edu>
	<9457e7c80806111142o2518d6e4odf6265609190a369@mail.gmail.com>
	<485026BA.1000405@ucsf.edu>
Message-ID: <20080611195451.GH2924@phare.normalesup.org>

On Wed, Jun 11, 2008 at 12:25:46PM -0700, Karl Young wrote:

> Yeah, thanks Stefan, this was a case of hitting return before putting 
> Jarrod's address in the To field - but coincidentally it might also 
> pertain to others on the list thinking about submitting something so I 
> guess  it wasn't the worst case of operator error...

Sound interesting. For a 35 minutes talk I think you have enough content,
especially if you give a good introduction to the field and the methods
that are normaly used in the field. I invite you to submit, and we'll see
what the selection commitee decides. We can also wait for Jarrod's
feedback.

Ga?l


From mwojc at p.lodz.pl  Wed Jun 11 16:00:05 2008
From: mwojc at p.lodz.pl (Marek Wojciechowski)
Date: Wed, 11 Jun 2008 22:00:05 +0200
Subject: [SciPy-user] optimize.fmin_l_bfgs_b problem
References: <200806091550.53753.mwojc@p.lodz.pl>
	<web-103041015@uni-stuttgart.de>
Message-ID: <g2panl$ofd$1@ger.gmane.org>

Nils Wagner wrote:

> On Mon, 9 Jun 2008 15:50:53 +0200
>   Marek Wojciechowski <mwojc at p.lodz.pl> wrote:
>> Hi!
>> The following command:
>>    optimize.fmin_l_bfgs_b(lambda x: x[0]**2, [-1.],
>>iprint=1)
>> causes an error and breaks the python session with the
>>following output:
>> 
>> RUNNING THE L-BFGS-B CODE
>> 
>>           * * *
>> 
>> At line 2647 of file scipy/optimize/lbfgsb/routines.f
>> Internal Error: printf is broken
>> Machine precision =
>> 
>> This occurs on scipy-0.6.0 and python 2.4 under Gentoo
>>Linux. Is this the
>> known bug?
>> 
>> Greetings,
>> --
>> Marek Wojciechowski
> 
> I get
> 
>>>> optimize.fmin_l_bfgs_b(lambda x: x[0]**2, [-1.], iprint=1)
> RUNNING THE L-BFGS-B CODE
> 
>             * * *
> 
> Machine precision = 2.220E-16
>   N =  1    M =  10
>   This problem is unconstrained.
> 
> At X0         0 variables are exactly at the bounds
> Traceback (most recent call last):
>    File "<stdin>", line 1, in <module>
>    File
> "/data/home/nwagner/local/lib/python2.5/site-packages/scipy/optimize/lbfgsb.py",
> line 205, in fmin_l_bfgs_b
>      f, g = func_and_grad(x)
>    File
> "/data/home/nwagner/local/lib/python2.5/site-packages/scipy/optimize/lbfgsb.py",
> line 156, in func_and_grad
>      f, g = func(x, *args)
> TypeError: 'numpy.float64' object is not iterable
> 
>>>> scipy.__version__
> '0.7.0.dev4420'

It seems that my bug is corrected in development versions of scipy.
Exception you obtained results probably from new layout of lbfgsb
optimizer. I think the following should work:

optimize.fmin_l_bfgs_b(lambda x: (x[0]**2, 2*x[0]), [-1.], iprint=1)

Greetings,
-- 
Marek Wojciechowski



From cohen at slac.stanford.edu  Wed Jun 11 15:59:08 2008
From: cohen at slac.stanford.edu (Johann Cohen-Tanugi)
Date: Wed, 11 Jun 2008 21:59:08 +0200
Subject: [SciPy-user] Speed differences in sqrt calculation: what
 is	good to know
In-Reply-To: <e76aa17f0806111154y7c0343a8sbee06a0d85c00c55@mail.gmail.com>
References: <826c64da0806110848r24248337jaf3caa2819e6906d@mail.gmail.com>
	<e76aa17f0806111154y7c0343a8sbee06a0d85c00c55@mail.gmail.com>
Message-ID: <48502E8C.9030309@slac.stanford.edu>

In [10]: rndvals=random.rand(10000)

In [11]: %timeit ssqrt(rndvals)
1000 loops, best of 3: 598 ?s per loop

In [12]: %timeit nsqrt(rndvals)
1000 loops, best of 3: 362 ?s per loop

JCT

Matthieu Brucher wrote:
> Hi,
>
> Don't forget that numpy's sqrt and scipy's sqrt are not optimized 
> towards single computations.
> If you really want to compare them, time the square root of 10000 
> elements.
>
> Matthieu
>
> 2008/6/11 Ivo Maljevic <ivo.maljevic at gmail.com 
> <mailto:ivo.maljevic at gmail.com>>:
>
>     Based on comments from Gael Varoquaux and David Cournapeau , I did
>     the execution time test.
>     At least for me, it is clear that if the number is a real scalar,
>     AND the expected result is also real,
>     the best way is to call the math version of sqrt() function. The
>     differences are more than significant, as you can see:
>
>     In [3]: from math import sqrt as msqrt
>     In [4]: from numpy import sqrt as nsqrt
>     In [5]: from scipy import sqrt as ssqrt
>
>     In [6]: %timeit msqrt(3.14)
>     1000000 loops, best of 3: 479 ns per loop
>
>     In [7]: %timeit nsqrt(3.14)
>     100000 loops, best of 3: 10.8 ?s per loop
>
>     In [8]: %timeit ssqrt(3.14)
>     10000 loops, best of 3: 74.5 ?s per loop
>
>
>     _______________________________________________
>     SciPy-user mailing list
>     SciPy-user at scipy.org <mailto:SciPy-user at scipy.org>
>     http://projects.scipy.org/mailman/listinfo/scipy-user
>
>
>
>
> -- 
> French PhD student
> Website : http://matthieu-brucher.developpez.com/
> Blogs : http://matt.eifelle.com and http://blog.developpez.com/?blog=92
> LinkedIn : http://www.linkedin.com/in/matthieubrucher
> ------------------------------------------------------------------------
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>   


From nwagner at iam.uni-stuttgart.de  Wed Jun 11 16:05:10 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Wed, 11 Jun 2008 22:05:10 +0200
Subject: [SciPy-user] optimize.fmin_l_bfgs_b problem
In-Reply-To: <g2panl$ofd$1@ger.gmane.org>
References: <200806091550.53753.mwojc@p.lodz.pl>
	<web-103041015@uni-stuttgart.de> <g2panl$ofd$1@ger.gmane.org>
Message-ID: <web-103171629@uni-stuttgart.de>

On Wed, 11 Jun 2008 22:00:05 +0200
  Marek Wojciechowski <mwojc at p.lodz.pl> wrote:
> Nils Wagner wrote:
> 
>> On Mon, 9 Jun 2008 15:50:53 +0200
>>   Marek Wojciechowski <mwojc at p.lodz.pl> wrote:
>>> Hi!
>>> The following command:
>>>    optimize.fmin_l_bfgs_b(lambda x: x[0]**2, [-1.],
>>>iprint=1)
>>> causes an error and breaks the python session with the
>>>following output:
>>> 
>>> RUNNING THE L-BFGS-B CODE
>>> 
>>>           * * *
>>> 
>>> At line 2647 of file scipy/optimize/lbfgsb/routines.f
>>> Internal Error: printf is broken
>>> Machine precision =
>>> 
>>> This occurs on scipy-0.6.0 and python 2.4 under Gentoo
>>>Linux. Is this the
>>> known bug?
>>> 
>>> Greetings,
>>> --
>>> Marek Wojciechowski
>> 
>> I get
>> 
>>>>> optimize.fmin_l_bfgs_b(lambda x: x[0]**2, [-1.], 
>>>>>iprint=1)
>> RUNNING THE L-BFGS-B CODE
>> 
>>             * * *
>> 
>> Machine precision = 2.220E-16
>>   N =  1    M =  10
>>   This problem is unconstrained.
>> 
>> At X0         0 variables are exactly at the bounds
>> Traceback (most recent call last):
>>    File "<stdin>", line 1, in <module>
>>    File
>> "/data/home/nwagner/local/lib/python2.5/site-packages/scipy/optimize/lbfgsb.py",
>> line 205, in fmin_l_bfgs_b
>>      f, g = func_and_grad(x)
>>    File
>> "/data/home/nwagner/local/lib/python2.5/site-packages/scipy/optimize/lbfgsb.py",
>> line 156, in func_and_grad
>>      f, g = func(x, *args)
>> TypeError: 'numpy.float64' object is not iterable
>> 
>>>>> scipy.__version__
>> '0.7.0.dev4420'
> 
> It seems that my bug is corrected in development 
>versions of scipy.
> Exception you obtained results probably from new layout 
>of lbfgsb
> optimizer. I think the following should work:
> 
> optimize.fmin_l_bfgs_b(lambda x: (x[0]**2, 2*x[0]), 
>[-1.], iprint=1)
> 
No.
>>> optimize.fmin_l_bfgs_b(lambda x: (x[0]**2, 2*x[0]), [-1.], iprint=1)
RUNNING THE L-BFGS-B CODE

            * * *

Machine precision = 2.220E-16
  N =  1    M =  10
  This problem is unconstrained.

At X0         0 variables are exactly at the bounds
Traceback (most recent call last):
   File "<stdin>", line 1, in <module>
   File 
"/usr/local/lib64/python2.5/site-packages/scipy/optimize/lbfgsb.py", 
line 199, in fmin_l_bfgs_b
     isave, dsave)
TypeError: failed to initialize 
intent(inout|inplace|cache) array -- input must be array 
but got <type 'numpy.float64'>
  
Cheers,
            Nils


From ivo.maljevic at gmail.com  Wed Jun 11 16:06:09 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Wed, 11 Jun 2008 16:06:09 -0400
Subject: [SciPy-user] Speed differences in sqrt calculation: what is
	good to know
In-Reply-To: <484FF543.10605@ar.media.kyoto-u.ac.jp>
References: <826c64da0806110848r24248337jaf3caa2819e6906d@mail.gmail.com>
	<484FF543.10605@ar.media.kyoto-u.ac.jp>
Message-ID: <826c64da0806111306x6902fdfbh691666ad713049f5@mail.gmail.com>

2008/6/11 David Cournapeau <david at ar.media.kyoto-u.ac.jp>:

> Ivo Maljevic wrote:
> > Based on comments from Gael Varoquaux and David Cournapeau , I did the
> > execution time test.
> > At least for me, it is clear that if the number is a real scalar, AND
> > the expected result is also real,
> > the best way is to call the math version of sqrt() function. The
> > differences are more than significant, as you can see:
> >
>
> It is expected for numpy/scipy functions to be much slower than python
> *for scalar*. Since they are optimized to be used with arrays, you are
> paying the cost to initialize the machinery to handle arrays (ufunc),
> without the benefit.


No disagreement there. My comment, which I realize was probably unnecessary,
is that
one has to pay attention as to which version of the function is called.
Since I started SciPy I always used everything from it, never even thinking
that math.sqrt(), math.sin(), etc., are faster if I'm working with scalar
values. And there are always coefficients or scaling parameters that do not
need to be in a vector form.


> The problem really is the difference between numpy and scipy. The funny
> thing is that for arrays, scipy.sqrt (that is, numpy.lib.scimath.sqrt)
> is faster than numpy.sqrt, which does not quite make sense to me since
> scipy.sqrt calls numpy.sqrt... But it is 1 a.m, so maybe I should just
> get some sleep
>

I think it is the numpy.sqrt that is faster (about 7 to 10 times) than
scipy.sqrt, but anyway, it would be good to have them at the same speed
level.

Ivo
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080611/0e179be6/attachment.html>

From mwojc at p.lodz.pl  Wed Jun 11 16:15:48 2008
From: mwojc at p.lodz.pl (Marek Wojciechowski)
Date: Wed, 11 Jun 2008 22:15:48 +0200
Subject: [SciPy-user] optimize.fmin_l_bfgs_b problem
References: <200806091550.53753.mwojc@p.lodz.pl>
	<web-103041015@uni-stuttgart.de> <g2panl$ofd$1@ger.gmane.org>
	<web-103171629@uni-stuttgart.de>
Message-ID: <g2pbl3$s47$1@ger.gmane.org>

Nils Wagner wrote:

> On Wed, 11 Jun 2008 22:00:05 +0200
>   Marek Wojciechowski <mwojc at p.lodz.pl> wrote:
>> Nils Wagner wrote:
>> 
>>> On Mon, 9 Jun 2008 15:50:53 +0200
>>>   Marek Wojciechowski <mwojc at p.lodz.pl> wrote:
>>>> Hi!
>>>> The following command:
>>>>    optimize.fmin_l_bfgs_b(lambda x: x[0]**2, [-1.],
>>>>iprint=1)
>>>> causes an error and breaks the python session with the
>>>>following output:
>>>> 
>>>> RUNNING THE L-BFGS-B CODE
>>>> 
>>>>           * * *
>>>> 
>>>> At line 2647 of file scipy/optimize/lbfgsb/routines.f
>>>> Internal Error: printf is broken
>>>> Machine precision =
>>>> 
>>>> This occurs on scipy-0.6.0 and python 2.4 under Gentoo
>>>>Linux. Is this the
>>>> known bug?
>>>> 
>>>> Greetings,
>>>> --
>>>> Marek Wojciechowski
>>> 
>>> I get
>>> 
>>>>>> optimize.fmin_l_bfgs_b(lambda x: x[0]**2, [-1.],
>>>>>>iprint=1)
>>> RUNNING THE L-BFGS-B CODE
>>> 
>>>             * * *
>>> 
>>> Machine precision = 2.220E-16
>>>   N =  1    M =  10
>>>   This problem is unconstrained.
>>> 
>>> At X0         0 variables are exactly at the bounds
>>> Traceback (most recent call last):
>>>    File "<stdin>", line 1, in <module>
>>>    File
>>> "/data/home/nwagner/local/lib/python2.5/site-packages/scipy/optimize/lbfgsb.py",
>>> line 205, in fmin_l_bfgs_b
>>>      f, g = func_and_grad(x)
>>>    File
>>> "/data/home/nwagner/local/lib/python2.5/site-packages/scipy/optimize/lbfgsb.py",
>>> line 156, in func_and_grad
>>>      f, g = func(x, *args)
>>> TypeError: 'numpy.float64' object is not iterable
>>> 
>>>>>> scipy.__version__
>>> '0.7.0.dev4420'
>> 
>> It seems that my bug is corrected in development
>>versions of scipy.
>> Exception you obtained results probably from new layout
>>of lbfgsb
>> optimizer. I think the following should work:
>> 
>> optimize.fmin_l_bfgs_b(lambda x: (x[0]**2, 2*x[0]),
>>[-1.], iprint=1)
>> 
> No.
>>>> optimize.fmin_l_bfgs_b(lambda x: (x[0]**2, 2*x[0]), [-1.], iprint=1)
> RUNNING THE L-BFGS-B CODE
> 
>             * * *
> 
> Machine precision = 2.220E-16
>   N =  1    M =  10
>   This problem is unconstrained.
> 
> At X0         0 variables are exactly at the bounds
> Traceback (most recent call last):
>    File "<stdin>", line 1, in <module>
>    File
> "/usr/local/lib64/python2.5/site-packages/scipy/optimize/lbfgsb.py",
> line 199, in fmin_l_bfgs_b
>      isave, dsave)
> TypeError: failed to initialize
> intent(inout|inplace|cache) array -- input must be array
> but got <type 'numpy.float64'>
>   
> Cheers,
>             Nils

The last chance :) :
optimize.fmin_l_bfgs_b(lambda x: (x[0]**2, 2*x[0]), array([-1.]), iprint=1)

-- 
Marek Wojciechowski



From nwagner at iam.uni-stuttgart.de  Wed Jun 11 16:19:12 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Wed, 11 Jun 2008 22:19:12 +0200
Subject: [SciPy-user] optimize.fmin_l_bfgs_b problem
In-Reply-To: <g2panl$ofd$1@ger.gmane.org>
References: <200806091550.53753.mwojc@p.lodz.pl>
	<web-103041015@uni-stuttgart.de> <g2panl$ofd$1@ger.gmane.org>
Message-ID: <web-103171988@uni-stuttgart.de>

On Wed, 11 Jun 2008 22:00:05 +0200
  Marek Wojciechowski <mwojc at p.lodz.pl> wrote:
> Nils Wagner wrote:
> 
>> On Mon, 9 Jun 2008 15:50:53 +0200
>>   Marek Wojciechowski <mwojc at p.lodz.pl> wrote:
>>> Hi!
>>> The following command:
>>>    optimize.fmin_l_bfgs_b(lambda x: x[0]**2, [-1.],
>>>iprint=1)
>>> causes an error and breaks the python session with the
>>>following output:
>>> 
>>> RUNNING THE L-BFGS-B CODE
>>> 
>>>           * * *
>>> 
>>> At line 2647 of file scipy/optimize/lbfgsb/routines.f
>>> Internal Error: printf is broken
>>> Machine precision =
>>> 
>>> This occurs on scipy-0.6.0 and python 2.4 under Gentoo
>>>Linux. Is this the
>>> known bug?
>>> 
>>> Greetings,
>>> --
>>> Marek Wojciechowski
>> 
>> I get
>> 
>>>>> optimize.fmin_l_bfgs_b(lambda x: x[0]**2, [-1.], 
>>>>>iprint=1)
>> RUNNING THE L-BFGS-B CODE
>> 
>>             * * *
>> 
>> Machine precision = 2.220E-16
>>   N =  1    M =  10
>>   This problem is unconstrained.
>> 
>> At X0         0 variables are exactly at the bounds
>> Traceback (most recent call last):
>>    File "<stdin>", line 1, in <module>
>>    File
>> "/data/home/nwagner/local/lib/python2.5/site-packages/scipy/optimize/lbfgsb.py",
>> line 205, in fmin_l_bfgs_b
>>      f, g = func_and_grad(x)
>>    File
>> "/data/home/nwagner/local/lib/python2.5/site-packages/scipy/optimize/lbfgsb.py",
>> line 156, in func_and_grad
>>      f, g = func(x, *args)
>> TypeError: 'numpy.float64' object is not iterable
>> 
>>>>> scipy.__version__
>> '0.7.0.dev4420'
> 
> It seems that my bug is corrected in development 
>versions of scipy.
> Exception you obtained results probably from new layout 
>of lbfgsb
> optimizer. I think the following should work:
> 
> optimize.fmin_l_bfgs_b(lambda x: (x[0]**2, 2*x[0]), 
>[-1.], iprint=1)
> 
> Greetings,
> -- 
> Marek Wojciechowski
> 
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user

Try

python -i test_l_bfgs_b.py
RUNNING THE L-BFGS-B CODE

            * * *

Machine precision = 2.220E-16
  N =  1    M =  10
  This problem is unconstrained.

At X0         0 variables are exactly at the bounds

At iterate    0    f=  1.00000E+00    |proj g|= 
 2.00000E+00

At iterate    1    f=  0.00000E+00    |proj g|= 
 0.00000E+00

            * * *

Tit   = total number of iterations
Tnf   = total number of function evaluations
Tnint = total number of segments explored during Cauchy 
searches
Skip  = number of BFGS updates skipped
Nact  = number of active bounds at final generalized 
Cauchy point
Projg = norm of the final projected gradient
F     = final function value

            * * *

    N   Tit  Tnf  Tnint  Skip  Nact     Projg        F
     1    1    2      1     0     0   0.000E+00 
  0.000E+00
   F =  0.

CONVERGENCE: NORM OF PROJECTED GRADIENT <= PGTOL

  Cauchy                time 0.000E+00 seconds.
  Subspace minimization time 0.000E+00 seconds.
  Line search           time 0.000E+00 seconds.

  Total User time 4.000E-03 seconds.

Position of the minimum [ 0.]
Value of func(x) [ 0.]
Dictionary {'warnflag': 0, 'task': 'CONVERGENCE: NORM OF 
PROJECTED GRADIENT <= PGTOL', 'grad': array([ 0.]), 
'funcalls': 2}
  
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test_l_bfgs_b.py
Type: text/x-python
Size: 263 bytes
Desc: not available
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080611/2f03fccf/attachment.py>

From millman at berkeley.edu  Wed Jun 11 17:36:45 2008
From: millman at berkeley.edu (Jarrod Millman)
Date: Wed, 11 Jun 2008 14:36:45 -0700
Subject: [SciPy-user] ANN: SciPy 2008 Conference
In-Reply-To: <48501456.4030306@ucsf.edu>
References: <c7009a550806101410h4725b984j6c9cce651501bfd4@mail.gmail.com>
	<48501456.4030306@ucsf.edu>
Message-ID: <c7009a550806111436v6879233cic485db19bb216049@mail.gmail.com>

On Wed, Jun 11, 2008 at 11:07 AM, Karl Young <Karl.Young at ucsf.edu> wrote:
> How are you doing ? Will you be in Melbourne ? I was thinking of
> submitting something for SciPy 2008 but wanted to get your take on it
> before doing so. The content is basically estimating pattern complexity
> from brain images for diagnostic purposes (the general content of a
> couple of papers in Neuroimage and Human Brain Mapping). I did the
> analysis using SciPy (and a little Rpy) but I'm sure the code is pretty
> brain dead and wouldn't teach the SciPy community anything - so can
> these submissions be more about content, e.g. gee look what I did using
> SciPy ?

I agree with Stefan and Gael that this would be an interesting
submission and encourage you to submit a proposal.  I am leaving for
Melbourne in a few hours, so if you are going to HBM as well I will
see you there.

Thanks,

-- 
Jarrod Millman
Computational Infrastructure for Research Labs
10 Giannini Hall, UC Berkeley
phone: 510.643.4014
http://cirl.berkeley.edu/


From kartita at gmail.com  Wed Jun 11 17:57:17 2008
From: kartita at gmail.com (Kimberly Artita)
Date: Wed, 11 Jun 2008 16:57:17 -0500
Subject: [SciPy-user] f2py "Segmentation fault", please help
Message-ID: <ff948bef0806111457q2c005ba9w288c5ecc10e50d33@mail.gmail.com>

I recently updated my system (including python, scipy, numpy) and now get a
bizarre "Segmentation fault" message where I didn't before.

I have problems reading in data.  Here is any example from my fortran file
(*.f90):

     character(len=4) :: title(60)

     open (2,file="file.cio")

     read (2,5100) title

     5100 format (20a4)

The file "file.cio" contains this:
General Input/Output section (file.cio):           Thu Mar 13 17:32:19 2008
AVSWAT2000 - SWAT interface MDL

Previously, it read in the data as :"General Input/Output section
(file.cio):           Thu Mar 1"

But now all I get is: "Segmentation fault" and then my program stops.

Please give me guidance!
Thanks,
Kim
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080611/b827768c/attachment.html>

From robert.kern at gmail.com  Wed Jun 11 18:07:24 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Wed, 11 Jun 2008 17:07:24 -0500
Subject: [SciPy-user] f2py "Segmentation fault", please help
In-Reply-To: <ff948bef0806111457q2c005ba9w288c5ecc10e50d33@mail.gmail.com>
References: <ff948bef0806111457q2c005ba9w288c5ecc10e50d33@mail.gmail.com>
Message-ID: <3d375d730806111507m6eb9e09budf64b39af806ba93@mail.gmail.com>

On Wed, Jun 11, 2008 at 16:57, Kimberly Artita <kartita at gmail.com> wrote:
> I recently updated my system (including python, scipy, numpy) and now get a
> bizarre "Segmentation fault" message where I didn't before.
>
> I have problems reading in data.  Here is any example from my fortran file
> (*.f90):
>
>      character(len=4) :: title(60)
>
>      open (2,file="file.cio")
>
>      read (2,5100) title
>
>      5100 format (20a4)
>
> The file "file.cio" contains this:
> General Input/Output section (file.cio):           Thu Mar 13 17:32:19 2008
> AVSWAT2000 - SWAT interface MDL
>
> Previously, it read in the data as :"General Input/Output section
> (file.cio):           Thu Mar 1"
>
> But now all I get is: "Segmentation fault" and then my program stops.

We need some more information. Can you show us the code? Alternately,
can you narrow the code down to just the part that is crashing? Please
run your program under a C debugger like gdb in order to get a
backtrace which will show us exactly where things are segfaulting.

Here is what that looks like on OS X. The details will be different on
other UNIces, but the "run -c ...", "continue", "backtrace" steps
should be the same.

$ gdb python
GNU gdb 6.3.50-20050815 (Apple version gdb-768) (Tue Oct  2 04:07:49 UTC 2007)
Copyright 2004 Free Software Foundation, Inc.
GDB is free software, covered by the GNU General Public License, and you are
welcome to change it and/or distribute copies of it under certain conditions.
Type "show copying" to see the conditions.
There is absolutely no warranty for GDB.  Type "show warranty" for details.
This GDB was configured as "i386-apple-darwin"...Reading symbols for
shared libraries .. done

(gdb) run -c crashing_script.py any_other_necessary_arguments_for_your_script
Starting program: /usr/local/bin/python -c crashing_script.py
any_other_necessary_arguments_for_your_script
Reading symbols for shared libraries +. done

Program received signal SIGTRAP, Trace/breakpoint trap.
0x8fe01010 in __dyld__dyld_start ()
(gdb) continue
Continuing.
Reading symbols for shared libraries .. done

... segfault happens here ...

(gdb) backtrace

... backtrace output here; this is what we need to see ...

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From kartita at gmail.com  Wed Jun 11 19:41:58 2008
From: kartita at gmail.com (Kimberly Artita)
Date: Wed, 11 Jun 2008 18:41:58 -0500
Subject: [SciPy-user] f2py "Segmentation fault", please help
In-Reply-To: <3d375d730806111507m6eb9e09budf64b39af806ba93@mail.gmail.com>
References: <ff948bef0806111457q2c005ba9w288c5ecc10e50d33@mail.gmail.com>
	<3d375d730806111507m6eb9e09budf64b39af806ba93@mail.gmail.com>
Message-ID: <ff948bef0806111641n2365291cy36e824436d1ccf0b@mail.gmail.com>

Here is a dummy fortran/python code which highlights my problem:

My fortran program readin_f90.f90:
subroutine readin

implicit none

character(len=4) :: title(60)

open (2,file="file.cio")
print *, "file.cio opened"
read (2,5100) title
print *, title

5100 format (20a4)

end subroutine

My python code readin.py:
from psyco import *
full()

from readin_f90 import *

test_readin()

My file file.cio:
General Input/Output section (file.cio):           Wed Jul 18 15:50:10 2007
AVSWAT2000 - SWAT interface MDL


   basins.bsb   basins.sbs   basins.rch   basins.rsv   basins.lqo
basins.wtr
   basins.pso   basins.eve   basins.fig   basins.cod   basins.bsn
basins.wwq
     crop.dat     till.dat     pest.dat     fert.dat    urban.dat
   1   1   3   3   3   3   0   0   0
      pcp.pcp


      tmp.tmp



I generate the *.so file with: f2py -c -m readin_f90 readin_f90.f90

kimi at localhost ~ $ python readin.py
Gives this:
 file.cio opened
Segmentation fault

Here the gdb output:
kimi at localhost ~ $ gdb python

(no debugging symbols found)
(no debugging symbols found)
 file.cio opened

---Type <return> to continue, or q <return> to quit---
Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0xb7d2d6c0 (LWP 8055)]
0xb5c14885 in ?? () from
/usr/lib/gcc/i686-pc-linux-gnu/4.2.3/libgfortran.so.2
(gdb) backtrace
#0  0xb5c14885 in ?? ()
   from /usr/lib/gcc/i686-pc-linux-gnu/4.2.3/libgfortran.so.2
#1  0xb7e5fff4 in ?? () from /lib/libc.so.6
#2  0x00000000 in ?? ()
(gdb)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080611/24f6ce15/attachment.html>

From robert.kern at gmail.com  Wed Jun 11 19:53:48 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Wed, 11 Jun 2008 18:53:48 -0500
Subject: [SciPy-user] f2py "Segmentation fault", please help
In-Reply-To: <ff948bef0806111641n2365291cy36e824436d1ccf0b@mail.gmail.com>
References: <ff948bef0806111457q2c005ba9w288c5ecc10e50d33@mail.gmail.com>
	<3d375d730806111507m6eb9e09budf64b39af806ba93@mail.gmail.com>
	<ff948bef0806111641n2365291cy36e824436d1ccf0b@mail.gmail.com>
Message-ID: <3d375d730806111653y3bcf2affse7585ccd289230bd@mail.gmail.com>

On Wed, Jun 11, 2008 at 18:41, Kimberly Artita <kartita at gmail.com> wrote:
> Here is a dummy fortran/python code which highlights my problem:
>
> My fortran program readin_f90.f90:
> subroutine readin
>
> implicit none
>
> character(len=4) :: title(60)
>
> open (2,file="file.cio")
> print *, "file.cio opened"
> read (2,5100) title
> print *, title
>
> 5100 format (20a4)
>
> end subroutine
>
> My python code readin.py:
> from psyco import *
> full()

Can you try this without psyco? It might be complicating matters.

> from readin_f90 import *
>
> test_readin()
>
> My file file.cio:
> General Input/Output section (file.cio):           Wed Jul 18 15:50:10 2007
> AVSWAT2000 - SWAT interface MDL
>
>
>    basins.bsb   basins.sbs   basins.rch   basins.rsv   basins.lqo
> basins.wtr
>    basins.pso   basins.eve   basins.fig   basins.cod   basins.bsn
> basins.wwq
>      crop.dat     till.dat     pest.dat     fert.dat    urban.dat
>    1   1   3   3   3   3   0   0   0
>       pcp.pcp
>
>
>       tmp.tmp
>
>
>
> I generate the *.so file with: f2py -c -m readin_f90 readin_f90.f90
>
> kimi at localhost ~ $ python readin.py
> Gives this:
>  file.cio opened
> Segmentation fault
>
> Here the gdb output:
> kimi at localhost ~ $ gdb python
>
> (no debugging symbols found)
> (no debugging symbols found)
>  file.cio opened
>
> ---Type <return> to continue, or q <return> to quit---
> Program received signal SIGSEGV, Segmentation fault.
> [Switching to Thread 0xb7d2d6c0 (LWP 8055)]
> 0xb5c14885 in ?? () from
> /usr/lib/gcc/i686-pc-linux-gnu/4.2.3/libgfortran.so.2
> (gdb) backtrace
> #0  0xb5c14885 in ?? ()
>    from /usr/lib/gcc/i686-pc-linux-gnu/4.2.3/libgfortran.so.2
> #1  0xb7e5fff4 in ?? () from /lib/libc.so.6
> #2  0x00000000 in ?? ()
> (gdb)

If psyco isn't to blame, then it looks like a problem in gfortran's
libraries. On OS X with gfortran 4.2.0, I don't get a segfault.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From lorenzo.isella at gmail.com  Thu Jun 12 05:11:05 2008
From: lorenzo.isella at gmail.com (Lorenzo Isella)
Date: Thu, 12 Jun 2008 11:11:05 +0200
Subject: [SciPy-user] How to get rid of nan and Inf
Message-ID: <a2b3004b0806120211n6c4a7f79pf63f7634538ae863@mail.gmail.com>

Dear All,
Say that you have an array which represents the time evolution of
certain quantities as time progresses.
Some entries of this array could be ill-defined (coming from a
division by zero or an indefinite error on a linear regression). How
can you automatically get rid of these?
I am looking for something like:

remove_ill_defined=where(array_ill_defined == is_a_number)
array_well_defined=array_ill_defined[remove_ill_defined]

but I do not know the command to identify a number vs an Inf or a nan
(my fictitious is_a_number condition).
Cheers

Lorenzo


From silva at lma.cnrs-mrs.fr  Thu Jun 12 05:21:18 2008
From: silva at lma.cnrs-mrs.fr (Fabrice Silva)
Date: Thu, 12 Jun 2008 11:21:18 +0200
Subject: [SciPy-user] How to get rid of nan and Inf
In-Reply-To: <a2b3004b0806120211n6c4a7f79pf63f7634538ae863@mail.gmail.com>
References: <a2b3004b0806120211n6c4a7f79pf63f7634538ae863@mail.gmail.com>
Message-ID: <1213262478.3022.2.camel@Portable-s2m.cnrs-mrs.fr>

Le jeudi 12 juin 2008 ? 11:11 +0200, Lorenzo Isella a ?crit :
> I am looking for something like:
> 
> remove_ill_defined=where(array_ill_defined == is_a_number)
> array_well_defined=array_ill_defined[remove_ill_defined]

What about isnan and isinf
import numpy as np
a = np.array([......])
index = (np.isnan(a)==False) & (np.isinf(a)==False)
a = a[index]

-- 
Fabrice Silva <silva at lma.cnrs-mrs.fr>
LMA UPR CNRS 7051 - ?quipe S2M



From david at ar.media.kyoto-u.ac.jp  Thu Jun 12 05:11:30 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Thu, 12 Jun 2008 18:11:30 +0900
Subject: [SciPy-user] How to get rid of nan and Inf
In-Reply-To: <a2b3004b0806120211n6c4a7f79pf63f7634538ae863@mail.gmail.com>
References: <a2b3004b0806120211n6c4a7f79pf63f7634538ae863@mail.gmail.com>
Message-ID: <4850E842.6000005@ar.media.kyoto-u.ac.jp>

Lorenzo Isella wrote:
> but I do not know the command to identify a number vs an Inf or a nan
> (my fictitious is_a_number condition).
>   

What about numpy.isnan and numpy.isfinite ?

David


From peridot.faceted at gmail.com  Thu Jun 12 09:31:54 2008
From: peridot.faceted at gmail.com (Anne Archibald)
Date: Thu, 12 Jun 2008 07:31:54 -0600
Subject: [SciPy-user] How to get rid of nan and Inf
In-Reply-To: <4850E842.6000005@ar.media.kyoto-u.ac.jp>
References: <a2b3004b0806120211n6c4a7f79pf63f7634538ae863@mail.gmail.com>
	<4850E842.6000005@ar.media.kyoto-u.ac.jp>
Message-ID: <ce557a360806120631h1af167f1h8757c5bfeebde9a9@mail.gmail.com>

2008/6/12 David Cournapeau <david at ar.media.kyoto-u.ac.jp>:
> Lorenzo Isella wrote:
>> but I do not know the command to identify a number vs an Inf or a nan
>> (my fictitious is_a_number condition).
>>
>
> What about numpy.isnan and numpy.isfinite ?

In fact np.isfinite(NaN) is False, so it's easy:

A = A[np.isfinite(A)]

Anne


From philbinj at gmail.com  Thu Jun 12 10:24:52 2008
From: philbinj at gmail.com (James Philbin)
Date: Thu, 12 Jun 2008 15:24:52 +0100
Subject: [SciPy-user] Sparse eigenvalues/eigenvectors problem
Message-ID: <2b1c8c4f0806120724n1d159964ubd39eb54922a62b5@mail.gmail.com>

Hi,

I'm trying to find the first few (~50) eigenvectors of a largish
(2130x2130) sparse symmetric real matrix. I've used
scipy.sparse.linalg.eigen_symmetric but have hit upon a few niggles.
If I ask for 10 eigenvectors, with the following command
eigen_symmetric(S,k=10,which='LA'), I get the following eigenvalues:
[ 0.99729875  0.99770773  0.9987255   0.99883746  1.          1.
   1.        1.          1.          1.        ]

Running with k=20 gives:
[ 0.99470154  0.99567495  0.99619173  0.99729875  0.99770773  0.9987255
  0.99883746  1.          1.          1.          1.          1.          1.
  1.          1.          1.          1.          1.          1.
   1.        ]

So it seems that eigen_symmetric is failing to find lots of the
repeated 1 eigenvectors. Is this an inherent problem with ARPACK or
symptomatic of a bug in scipy? Matlab's eigs seems to not suffer from
this nearly as much, returning nearly all ones for both the 10 and 20
cases. I assume this is also using ARPACK under the surface. I can
upload the matrix S somewhere if people are interested.

I also think i've fixed a bug in the current version of arpack.py
relating to the info returned (The warning that maxiters had been
reached was never shown) and added a warning if some eigenvectors
didn't converge:

--- arpack.py.old	2008-06-12 13:28:29.000000000 +0100
+++ arpack.py	2008-06-12 15:22:55.000000000 +0100
@@ -463,8 +463,11 @@
     if  info < -1 :
         raise RuntimeError("Error info=%d in arpack"%info)
         return None
-    if info == -1:
+    if info == 1:
         warnings.warn("Maximum number of iterations taken: %s"%iparam[2])
+
+    if iparam[4]<k:
+      warnings.warn("Only %d/%d eigenvectors converged" % (iparam[4], k))

     # now extract eigenvalues and (optionally) eigenvectors
     rvec = return_eigenvectors


Thanks,
James


From lists at vrbka.net  Thu Jun 12 10:54:52 2008
From: lists at vrbka.net (Lubos Vrbka)
Date: Thu, 12 Jun 2008 16:54:52 +0200
Subject: [SciPy-user] Speed differences in sqrt calculation: what is
 good to	know
In-Reply-To: <826c64da0806110848r24248337jaf3caa2819e6906d@mail.gmail.com>
References: <826c64da0806110848r24248337jaf3caa2819e6906d@mail.gmail.com>
Message-ID: <485138BC.7070506@vrbka.net>

hi,

slightly off-topic, but still regarding the speed differences... what's 
the difference between the following 2 functions?

numpy.linalg.solve
scipy.linalg.solve

the first is almost 2 times faster on the same data (4096 repetitions of 
a A*a = b problem with A = 2x2 float32 matrix) - typically 0.30 vs. 0.57 
sec for the whole operation...

best,

-- 
Lubos _ at _"
http://www.lubos.vrbka.net


From ivo.maljevic at gmail.com  Thu Jun 12 11:05:45 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Thu, 12 Jun 2008 11:05:45 -0400
Subject: [SciPy-user] Speed differences in sqrt calculation: what is
	good to know
In-Reply-To: <485138BC.7070506@vrbka.net>
References: <826c64da0806110848r24248337jaf3caa2819e6906d@mail.gmail.com>
	<485138BC.7070506@vrbka.net>
Message-ID: <826c64da0806120805i56166f5bue2dbbf74c8bc2ff8@mail.gmail.com>

It looks like the same topic, that is, the difference in speed between numpy
and scipy.

There are all sorts of differences that I'm seeing while learning
scipy/numpy/python. Also an interesting one,
but probably with negligible impact:

In [1]: import math
In [2]: %timeit math.sqrt(2.2)
1000000 loops, best of 3: 327 ns per loop
In [3]: %timeit math.sqrt(2.2)
1000000 loops, best of 3: 321 ns per loop
In [4]: %timeit math.sqrt(2.2)
1000000 loops, best of 3: 323 ns per loop

versus:

In [1]: from math import sqrt
In [2]: %timeit sqrt(2.2)
1000000 loops, best of 3: 245 ns per loop
In [3]: %timeit sqrt(2.2)
1000000 loops, best of 3: 248 ns per loop
In [4]: %timeit sqrt(2.2)
1000000 loops, best of 3: 248 ns per loop



2008/6/12 Lubos Vrbka <lists at vrbka.net>:

> hi,
>
> slightly off-topic, but still regarding the speed differences... what's
> the difference between the following 2 functions?
>
> numpy.linalg.solve
> scipy.linalg.solve
>
> the first is almost 2 times faster on the same data (4096 repetitions of
> a A*a = b problem with A = 2x2 float32 matrix) - typically 0.30 vs. 0.57
> sec for the whole operation...
>
> best,
>
> --
> Lubos _ at _"
> http://www.lubos.vrbka.net
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080612/d3c35c67/attachment.html>

From lists at vrbka.net  Thu Jun 12 11:16:25 2008
From: lists at vrbka.net (Lubos Vrbka)
Date: Thu, 12 Jun 2008 17:16:25 +0200
Subject: [SciPy-user] Speed differences in sqrt calculation: what is
 good to know
In-Reply-To: <826c64da0806120805i56166f5bue2dbbf74c8bc2ff8@mail.gmail.com>
References: <826c64da0806110848r24248337jaf3caa2819e6906d@mail.gmail.com>	
	<485138BC.7070506@vrbka.net>
	<826c64da0806120805i56166f5bue2dbbf74c8bc2ff8@mail.gmail.com>
Message-ID: <48513DC9.2050504@vrbka.net>

Ivo Maljevic wrote:
> It looks like the same topic, that is, the difference in speed between numpy
> and scipy.
> 
> There are all sorts of differences that I'm seeing while learning
> scipy/numpy/python. Also an interesting one,
> but probably with negligible impact:
> 
> In [1]: import math
> In [2]: %timeit math.sqrt(2.2)
> 1000000 loops, best of 3: 327 ns per loop
> In [3]: %timeit math.sqrt(2.2)
> 1000000 loops, best of 3: 321 ns per loop
> In [4]: %timeit math.sqrt(2.2)
> 1000000 loops, best of 3: 323 ns per loop
> 
> versus:
> 
> In [1]: from math import sqrt
> In [2]: %timeit sqrt(2.2)
> 1000000 loops, best of 3: 245 ns per loop
> In [3]: %timeit sqrt(2.2)
> 1000000 loops, best of 3: 248 ns per loop
> In [4]: %timeit sqrt(2.2)
> 1000000 loops, best of 3: 248 ns per loop
hm, i see similar thing for the
import numpy; numpy.linalg.solve
and
from numpy import linalg; linalg.solve

but the differences are here relatively small (negligible). it's just 
probably the time involved with the name resolution machinery in 
python... the discrepancy between numpy and scipy is much much larger...

best,

-- 
Lubos _ at _"
http://www.lubos.vrbka.net


From tgray at protozoic.com  Thu Jun 12 14:02:11 2008
From: tgray at protozoic.com (Tim Gray)
Date: Thu, 12 Jun 2008 14:02:11 -0400
Subject: [SciPy-user] How to get rid of nan and Inf
Message-ID: <8f5413a60806121102h4849828dief9fd5387f1b5fee@mail.gmail.com>

nan_to_num() does this. I think it's somewhere in numpy.

Makes Nan's -> 0's and infs -> machine limits.


From nwagner at iam.uni-stuttgart.de  Thu Jun 12 14:03:43 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Thu, 12 Jun 2008 20:03:43 +0200
Subject: [SciPy-user] Sparse eigenvalues/eigenvectors problem
In-Reply-To: <2b1c8c4f0806120724n1d159964ubd39eb54922a62b5@mail.gmail.com>
References: <2b1c8c4f0806120724n1d159964ubd39eb54922a62b5@mail.gmail.com>
Message-ID: <web-103220942@uni-stuttgart.de>

On Thu, 12 Jun 2008 15:24:52 +0100
  "James Philbin" <philbinj at gmail.com> wrote:
> Hi,
> 
> I'm trying to find the first few (~50) eigenvectors of a 
>largish
> (2130x2130) sparse symmetric real matrix. I've used
> scipy.sparse.linalg.eigen_symmetric but have hit upon a 
>few niggles.
> If I ask for 10 eigenvectors, with the following command
> eigen_symmetric(S,k=10,which='LA'), I get the following 
>eigenvalues:
> [ 0.99729875  0.99770773  0.9987255   0.99883746  1. 
>         1.
>   1.        1.          1.          1.        ]
> 
> Running with k=20 gives:
> [ 0.99470154  0.99567495  0.99619173  0.99729875 
> 0.99770773  0.9987255
>  0.99883746  1.          1.          1.          1. 
>         1.          1.
>  1.          1.          1.          1.          1. 
>         1.
>   1.        ]
> 
> So it seems that eigen_symmetric is failing to find lots 
>of the
> repeated 1 eigenvectors. Is this an inherent problem 
>with ARPACK or
> symptomatic of a bug in scipy? Matlab's eigs seems to 
>not suffer from
> this nearly as much, returning nearly all ones for both 
>the 10 and 20
> cases. I assume this is also using ARPACK under the 
>surface. I can
> upload the matrix S somewhere if people are interested.
> 
> I also think i've fixed a bug in the current version of 
>arpack.py
> relating to the info returned (The warning that maxiters 
>had been
> reached was never shown) and added a warning if some 
>eigenvectors
> didn't converge:
> 
> --- arpack.py.old	2008-06-12 13:28:29.000000000 +0100
> +++ arpack.py	2008-06-12 15:22:55.000000000 +0100
> @@ -463,8 +463,11 @@
>     if  info < -1 :
>         raise RuntimeError("Error info=%d in 
>arpack"%info)
>         return None
> -    if info == -1:
> +    if info == 1:
>         warnings.warn("Maximum number of iterations 
>taken: %s"%iparam[2])
> +
> +    if iparam[4]<k:
> +      warnings.warn("Only %d/%d eigenvectors converged" 
>% (iparam[4], k))
> 
>     # now extract eigenvalues and (optionally) 
>eigenvectors
>     rvec = return_eigenvectors
> 
> 
> Thanks,
> James

Hi James,

Please can you send me the matrix off-list (*.mtx).
  

Thanks in advance,
                      Nils


From peridot.faceted at gmail.com  Thu Jun 12 15:06:49 2008
From: peridot.faceted at gmail.com (Anne Archibald)
Date: Thu, 12 Jun 2008 13:06:49 -0600
Subject: [SciPy-user] Speed differences in sqrt calculation: what is
	good to know
In-Reply-To: <826c64da0806111306x6902fdfbh691666ad713049f5@mail.gmail.com>
References: <826c64da0806110848r24248337jaf3caa2819e6906d@mail.gmail.com>
	<484FF543.10605@ar.media.kyoto-u.ac.jp>
	<826c64da0806111306x6902fdfbh691666ad713049f5@mail.gmail.com>
Message-ID: <ce557a360806121206g584ee352i4948a1c2e405daf6@mail.gmail.com>

2008/6/11 Ivo Maljevic <ivo.maljevic at gmail.com>:
> 2008/6/11 David Cournapeau <david at ar.media.kyoto-u.ac.jp>:
>>
>> Ivo Maljevic wrote:
>> > Based on comments from Gael Varoquaux and David Cournapeau , I did the
>> > execution time test.
>> > At least for me, it is clear that if the number is a real scalar, AND
>> > the expected result is also real,
>> > the best way is to call the math version of sqrt() function. The
>> > differences are more than significant, as you can see:
>> >
>>
>> It is expected for numpy/scipy functions to be much slower than python
>> *for scalar*. Since they are optimized to be used with arrays, you are
>> paying the cost to initialize the machinery to handle arrays (ufunc),
>> without the benefit.
>
> No disagreement there. My comment, which I realize was probably unnecessary,
> is that
> one has to pay attention as to which version of the function is called.
> Since I started SciPy I always used everything from it, never even thinking
> that math.sqrt(), math.sin(), etc., are faster if I'm working with scalar
> values. And there are always coefficients or scaling parameters that do not
> need to be in a vector form.

Keep in mind that if you are doing many computations, doing them with
scalars exposes you to a great deal of overhead from the python
interpreter. Put another way, if you're writing a program in which you
compute enough square roots that the speed difference between
math.sqrt and numpy.sqrt is significant, you should almost certainly
rewrite your program so that you are computing vector square roots;
the time you save in python overhead will make your program vastly
faster. In fact it is probably difficult to write anontrivial program
in which the time spent computing scalar square roots is significant.

Anne


From dwf at cs.toronto.edu  Tue Jun 10 19:26:28 2008
From: dwf at cs.toronto.edu (David Warde-Farley)
Date: Tue, 10 Jun 2008 19:26:28 -0400
Subject: [SciPy-user] Some mathematics/statisctics books
In-Reply-To: <484ED79B.8000607@ucsf.edu>
References: <da09baf80806101209q7b0b63acs8fa053b93d7dcfce@mail.gmail.com>
	<484ED7AA.2040907@slac.stanford.edu> <484ED79B.8000607@ucsf.edu>
Message-ID: <0BAB70CC-5B85-4CB5-9355-195B537B08EA@cs.toronto.edu>

On 10-Jun-08, at 3:35 PM, Karl Young wrote:

> I completely agree with Johann that the best way to start is to just
> dive into the tutorials and examples but there are a few books around
> that might not be bad to have at your side when doing so.

I just looked through the Gershenfeld book's table of contents, and  
while it's full of tons of useful stuff, it may be the wrong place to  
start. Certainly, don't try to read the book in sequence from start  
to finish if you don't have the requisite background.

What background? Probably some linear algebra at least, and some  
single and multivariable calculus. The OP didn't specify what level  
of education he's had in these matters, so just in case I'll include  
a few books.

I'm not too familiar with books on single-variable calculus but Tom  
Apostol's book on the subject appears to be quite well reviewed. It's  
also been around since the 1960's so it should be possible to find an  
inexpensive used copy.

As for multivariable calculus and linear algebra, a book that I'm  
fond of that takes an integrated approach to these two subjects is  
"Multivariable Mathematics" by Theodore Shifrin, ISBN 047152638X.  It  
goes through a lot of examples but doesn't sacrifice rigour. It won't  
help you much if you haven't done any single variable calculus.

My supervisor introduced me to Gilbert Strang's "Linear Algebra and  
its Applications", which I quite like. People seem to either love or  
hate this book, but it endeavours to help you develop intuition for  
linear algebraic concepts, and is quite light on theory, heavy on  
application.

Hope that helps,

David


From timmichelsen at gmx-topmail.de  Thu Jun 12 17:14:03 2008
From: timmichelsen at gmx-topmail.de (Tim Michelsen)
Date: Thu, 12 Jun 2008 23:14:03 +0200
Subject: [SciPy-user] iteratively masking timeseries
Message-ID: <g2s3ir$o2f$1@ger.gmane.org>

Hello,
I am making my way forward into timeseries processing with the scikit 
package.

I have a timeseries where NoData values are masked when loading the data 
into the timeseries.

Now I would like to apply some filters on the data like discarding data 
values below measurement device accuracy or above a certain threshold.

Therefore I followed the appraoch outlined in the FAQ [1].
But then I get the pasted below at the end.

Does that mean that one can only mask an array once and cannot mask more 
values later on?

I would like to do something like:

1) create timeseries with NoData values => already can do that
2) apply various filters masking more and more data.
	a) e.g. get a series with masked values below 10.
	b) e.g. get a series with masked values above 100.

I would appreciate any help or hint here.

Kind regards,
Tim


#### pasted from Ipython ####

In [18]: mask[mask<0] = numpy.ma.masked

---------------------------------------------------------------------------

IndexError                                Traceback (most recent call last)



D:\scripts\timeseries.py i

----> 1

       2

       3

       4

       5



C:\python25\lib\site-packages\scikits\timeseries\tseries.pyc in 
__setitem__(se

     522         if self is masked:

     523             raise MAError, 'Cannot alter the masked element.'

--> 524         (sindx, _) = self.__checkindex(indx)

     525         super(TimeSeries, self).__setitem__(sindx, value)

     526     #......................................................



C:\python25\lib\site-packages\scikits\timeseries\tseries.pyc in 
__checkindex(s

     490             msg = "Masked arrays must be filled before they can 
be use

     491                   "as indices!"

--> 492             raise IndexError, msg

     493         return (indx,indx)

     494



IndexError: Masked arrays must be filled before they can be used as indices!

#### end from Ipython ####

[1] 
http://www.scipy.org/Cookbook/TimeSeries/FAQ#head-cfe3617dda0b030f0474a2a773e2dca4da8eaea0



From wnbell at gmail.com  Thu Jun 12 17:52:52 2008
From: wnbell at gmail.com (Nathan Bell)
Date: Thu, 12 Jun 2008 16:52:52 -0500
Subject: [SciPy-user] Sparse eigenvalues/eigenvectors problem
In-Reply-To: <2b1c8c4f0806120724n1d159964ubd39eb54922a62b5@mail.gmail.com>
References: <2b1c8c4f0806120724n1d159964ubd39eb54922a62b5@mail.gmail.com>
Message-ID: <d05265cb0806121452s2e2c2038o31ebab0fd410fa0c@mail.gmail.com>

On Thu, Jun 12, 2008 at 9:24 AM, James Philbin <philbinj at gmail.com> wrote:
> So it seems that eigen_symmetric is failing to find lots of the
> repeated 1 eigenvectors. Is this an inherent problem with ARPACK or
> symptomatic of a bug in scipy? Matlab's eigs seems to not suffer from
> this nearly as much, returning nearly all ones for both the 10 and 20
> cases. I assume this is also using ARPACK under the surface. I can
> upload the matrix S somewhere if people are interested.

Hi James,

While your problem could be caused by a bug in our ARPACK wrappers, it
could also be due to differences in default parameters.  Are you sure
that MATLAB is calling ARPACK with the same parameters as
eigen_symmetric?  For instance, it the same tolerance used in both
cases?  Have you setup eigs() to use the symmetric algorithm?

There may also be a difference in the way MATLAB and SciPy chose the
initial vector of the Krylov subspace (parameter v0 in eigen()).

If you believe this to be a genuine bug, please submit a ticket in
Trac with a short script that demonstrates the error.
http://projects.scipy.org/scipy/scipy

If your matrix is small enough, you could include it with the ticket
as well, perhaps as a compressed MatrixMarket file:
>>> A = #your matrix
>>> from scipy.io import mmwrite
>>> mmwrite('A.mtx', A)
$ gzip A.mtx

> I also think i've fixed a bug in the current version of arpack.py
> relating to the info returned (The warning that maxiters had been
> reached was never shown) and added a warning if some eigenvectors
> didn't converge:
>

Added in r4441:
http://projects.scipy.org/scipy/scipy/changeset/4441


-- 
Nathan Bell wnbell at gmail.com
http://graphics.cs.uiuc.edu/~wnbell/


From philbinj at gmail.com  Thu Jun 12 20:01:59 2008
From: philbinj at gmail.com (James Philbin)
Date: Fri, 13 Jun 2008 01:01:59 +0100
Subject: [SciPy-user] Sparse eigenvalues/eigenvectors problem
In-Reply-To: <d05265cb0806121452s2e2c2038o31ebab0fd410fa0c@mail.gmail.com>
References: <2b1c8c4f0806120724n1d159964ubd39eb54922a62b5@mail.gmail.com>
	<d05265cb0806121452s2e2c2038o31ebab0fd410fa0c@mail.gmail.com>
Message-ID: <2b1c8c4f0806121701l1e91ce99h2236a8ed20a73029@mail.gmail.com>

Hi,

>>>> A = #your matrix
>>>> from scipy.io import mmwrite
>>>> mmwrite('A.mtx', A)

mmwrite seems not to work for me (latest svn checkout):
  File "/usr/lib/python2.5/site-packages/scipy/io/mmio.py", line 67, in mmwrite
    MMFile().write(target, a, comment, field, precision)
  File "/usr/lib/python2.5/site-packages/scipy/io/mmio.py", line 284, in write
    self._write(stream, a, comment, field, precision)
  File "/usr/lib/python2.5/site-packages/scipy/io/mmio.py", line 564, in _write
    IJV = vstack((a.row, a.col, a.data)).T
  File "/usr/lib/python2.5/site-packages/scipy/sparse/csr.py", line
93, in __getattr__
    return _cs_matrix.__getattr__(self, attr)
  File "/usr/lib/python2.5/site-packages/scipy/sparse/base.py", line
316, in __getattr__
    raise AttributeError, attr + " not found"
AttributeError: row not found

I've saved the matrix in (row,col,data) format and uploaded it here:
www.robots.ox.ac.uk/~james/np_temp/S.txt.gz

Thanks,
James


From david at ar.media.kyoto-u.ac.jp  Thu Jun 12 22:49:43 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Fri, 13 Jun 2008 11:49:43 +0900
Subject: [SciPy-user] Speed differences in sqrt calculation: what
 is	good to know
In-Reply-To: <826c64da0806120805i56166f5bue2dbbf74c8bc2ff8@mail.gmail.com>
References: <826c64da0806110848r24248337jaf3caa2819e6906d@mail.gmail.com>	<485138BC.7070506@vrbka.net>
	<826c64da0806120805i56166f5bue2dbbf74c8bc2ff8@mail.gmail.com>
Message-ID: <4851E047.2040704@ar.media.kyoto-u.ac.jp>

Ivo Maljevic wrote:
> It looks like the same topic, that is, the difference in speed between 
> numpy and scipy.
>
> There are all sorts of differences that I'm seeing while learning 
> scipy/numpy/python. Also an interesting one,
> but probably with negligible impact:

Name resolution is slow in python (it is generally slow in interpreted 
language, and python has a rather primitive implementation - no JIT, 
etc...). It is one of the reason why numpy exists in the first place 
(f([1, 2, 3, 4, 5, ....]) is much faster than for i in range([1, 2, 3, 
....]): f(i) - name resolution is not the only slowdown involved here).

cheers,

David


From wnbell at gmail.com  Thu Jun 12 23:11:27 2008
From: wnbell at gmail.com (Nathan Bell)
Date: Thu, 12 Jun 2008 22:11:27 -0500
Subject: [SciPy-user] Sparse eigenvalues/eigenvectors problem
In-Reply-To: <2b1c8c4f0806121701l1e91ce99h2236a8ed20a73029@mail.gmail.com>
References: <2b1c8c4f0806120724n1d159964ubd39eb54922a62b5@mail.gmail.com>
	<d05265cb0806121452s2e2c2038o31ebab0fd410fa0c@mail.gmail.com>
	<2b1c8c4f0806121701l1e91ce99h2236a8ed20a73029@mail.gmail.com>
Message-ID: <d05265cb0806122011s6d740054jf29bb4b00daafc2@mail.gmail.com>

On Thu, Jun 12, 2008 at 7:01 PM, James Philbin <philbinj at gmail.com> wrote:
>
> mmwrite seems not to work for me (latest svn checkout):
>

Should be fixed now (SVN r4444).

-- 
Nathan Bell wnbell at gmail.com
http://graphics.cs.uiuc.edu/~wnbell/


From david at ar.media.kyoto-u.ac.jp  Thu Jun 12 23:23:00 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Fri, 13 Jun 2008 12:23:00 +0900
Subject: [SciPy-user] Speed differences in sqrt calculation: what is
 good to	know
In-Reply-To: <485138BC.7070506@vrbka.net>
References: <826c64da0806110848r24248337jaf3caa2819e6906d@mail.gmail.com>
	<485138BC.7070506@vrbka.net>
Message-ID: <4851E814.4020703@ar.media.kyoto-u.ac.jp>

Lubos Vrbka wrote:
> hi,
>
> slightly off-topic, but still regarding the speed differences... what's 
> the difference between the following 2 functions?
>
> numpy.linalg.solve
> scipy.linalg.solve
>
> the first is almost 2 times faster on the same data (4096 repetitions of 
> a A*a = b problem with A = 2x2 float32 matrix) - typically 0.30 vs. 0.57 
> sec for the whole operation...
>   

I think you will find a lot of those differences with very small data. I 
would not even be surprised if it was faster to do the inversion in 
python. Using numpy and scipy for those kind of data just do not make 
much sense. I am sure you will have exactly the same problem in matlab, 
albeit in a less significant way, because recent matlab engines have JIT 
(and a much simpler language, which certainly helps here).

Difference in functionalities between numpy and scipy are much more 
bothering to me (sqrt(-1) giving a different answer for numpy and scipy, 
for example).

cheers,

David


From waterbug at pangalactic.us  Fri Jun 13 00:49:18 2008
From: waterbug at pangalactic.us (Stephen Waterbury)
Date: Fri, 13 Jun 2008 00:49:18 -0400
Subject: [SciPy-user] SciPy 2008 Registration payment page gives traceback
Message-ID: <4851FC4E.9030100@pangalactic.us>

When I tried to pay, I got a traceback --
something about a key error on "sprints" (which
I didn't sign up for).  I should have copied it,
sorry.

Steve


From travis at enthought.com  Fri Jun 13 08:16:04 2008
From: travis at enthought.com (Travis Vaught)
Date: Fri, 13 Jun 2008 07:16:04 -0500
Subject: [SciPy-user] SciPy 2008 Registration payment page gives
	traceback
In-Reply-To: <4851FC4E.9030100@pangalactic.us>
References: <4851FC4E.9030100@pangalactic.us>
Message-ID: <ED174880-0B53-4D60-AAD1-14007CB831F4@enthought.com>

Steve,

This is fixed now (and tested).  Many apologies.  You will need to try  
again, though :/

Thanks,

Travis


On Jun 12, 2008, at 11:49 PM, Stephen Waterbury wrote:

> When I tried to pay, I got a traceback --
> something about a key error on "sprints" (which
> I didn't sign up for).  I should have copied it,
> sorry.
>
> Steve
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>



From alexander.borghgraef.rma at gmail.com  Fri Jun 13 08:16:08 2008
From: alexander.borghgraef.rma at gmail.com (Alexander Borghgraef)
Date: Fri, 13 Jun 2008 14:16:08 +0200
Subject: [SciPy-user] Mlab doesn't work
Message-ID: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>

Hi all,

 I've been playing a bit with mayavi2/mlab, but I run into a problem
when trying to run the first example at
http://www.scipy.org/Cookbook/MayaVi/mlab:

  import scipy

  # prepare some interesting function:
  def f(x, y):
      return 3.0*scipy.sin(x*y+1e-4)/(x*y+1e-4)

  x = scipy.arange(-7., 7.05, 0.1)
  y = scipy.arange(-5., 5.05, 0.1)

  # 3D visualization of f:
  from enthought.tvtk.tools import mlab
  fig = mlab.figure()
  s = mlab.SurfRegular(x, y, f)
  fig.add(s)

I get the following messages:

  mlabtest.py:3: DeprecationWarning: The wxPython compatibility
package is no longer automatically generated or actively maintained.
Please switch to the wx package as   soon as possible.
    from wxPython.wx import *

Then a blank gtk window pops up, disappears, and then I get:

  (python:4960): Gtk-CRITICAL **: gtk_widget_set_colormap: assertion
`!GTK_WIDGET_REALIZED (widget)' failed

I've built the enthought libraries from source on an Fedora 8 system,
and installed them in a local directory (I have no access to
/usr/lib). Importing the libraries works fine, and so does running the
mayavi2 binary, but I haven't managed to get something plotted from a
python script. Any ideas?

-- 
Alex Borghgraef


From fredmfp at gmail.com  Fri Jun 13 08:25:00 2008
From: fredmfp at gmail.com (fred)
Date: Fri, 13 Jun 2008 14:25:00 +0200
Subject: [SciPy-user] Mlab doesn't work
In-Reply-To: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>
References: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>
Message-ID: <4852671C.9080402@gmail.com>

Alexander Borghgraef a ?crit :

> I've built the enthought libraries from source on an Fedora 8 system,
> and installed them in a local directory (I have no access to
> /usr/lib). Importing the libraries works fine, and so does running the
> mayavi2 binary, but I haven't managed to get something plotted from a
> python script. Any ideas?
Can you tell us how you run your script ?

Cheers,

-- 
Fred


From alexander.borghgraef.rma at gmail.com  Fri Jun 13 08:34:00 2008
From: alexander.borghgraef.rma at gmail.com (Alexander Borghgraef)
Date: Fri, 13 Jun 2008 14:34:00 +0200
Subject: [SciPy-user] Mlab doesn't work
In-Reply-To: <4852671C.9080402@gmail.com>
References: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>
	<4852671C.9080402@gmail.com>
Message-ID: <9e8c52a20806130534x12a6c087jf486daca9a28c6e1@mail.gmail.com>

On Fri, Jun 13, 2008 at 2:25 PM, fred <fredmfp at gmail.com> wrote:
> Alexander Borghgraef a ?crit :
>
>> I've built the enthought libraries from source on an Fedora 8 system,
>> and installed them in a local directory (I have no access to
>> /usr/lib). Importing the libraries works fine, and so does running the
>> mayavi2 binary, but I haven't managed to get something plotted from a
>> python script. Any ideas?
> Can you tell us how you run your script ?

Saved it to mlabtest.py, then ran

 python mlabtest.py

from the command line in a bash shell.

-- 
Alex Borghgraef


From travis at enthought.com  Fri Jun 13 08:35:20 2008
From: travis at enthought.com (Travis Vaught)
Date: Fri, 13 Jun 2008 07:35:20 -0500
Subject: [SciPy-user] EuroSciPy Early Registration Reminder
Message-ID: <26C7F1FC-FE73-4DF2-8AF5-72D15A816D98@enthought.com>

Greetings,

This is a friendly reminder that the Early Registration deadline for  
the EuroSciPy Conference is June 15th.  If you're interested in  
attending, but have not yet registered, please visit:

http://www.scipy.org/EuroSciPy2008

The talks schedule is also now available there.

Also, the keynote speaker this year will be Travis Oliphant, the  
primary author of the recent NumPy rewrite.  For those doing  
scientific computing using Python, this is a conference you'll not  
want to miss.

See you there.

Travis


From fredmfp at gmail.com  Fri Jun 13 08:48:11 2008
From: fredmfp at gmail.com (fred)
Date: Fri, 13 Jun 2008 14:48:11 +0200
Subject: [SciPy-user] Mlab doesn't work
In-Reply-To: <9e8c52a20806130534x12a6c087jf486daca9a28c6e1@mail.gmail.com>
References: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>	<4852671C.9080402@gmail.com>
	<9e8c52a20806130534x12a6c087jf486daca9a28c6e1@mail.gmail.com>
Message-ID: <48526C8B.5040300@gmail.com>

Alexander Borghgraef a ?crit :

> Saved it to mlabtest.py, then ran
> 
>  python mlabtest.py
> 
> from the command line in a bash shell.
Ok.

Note that these cookbook pages are said to be obsolete.

Try it running ipython like this:
ipython -whtread
 > run mlabtest.py

Cheers,

PS: Please don't be offensive with your subject.
Mlab _does_ work and Ga?l is doing a great job on this.

-- 
Fred


From gael.varoquaux at normalesup.org  Fri Jun 13 08:51:50 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Fri, 13 Jun 2008 14:51:50 +0200
Subject: [SciPy-user] Mlab doesn't work
In-Reply-To: <9e8c52a20806130534x12a6c087jf486daca9a28c6e1@mail.gmail.com>
References: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>
	<4852671C.9080402@gmail.com>
	<9e8c52a20806130534x12a6c087jf486daca9a28c6e1@mail.gmail.com>
Message-ID: <20080613125150.GA3573@phare.normalesup.org>

On Fri, Jun 13, 2008 at 02:34:00PM +0200, Alexander Borghgraef wrote:
> On Fri, Jun 13, 2008 at 2:25 PM, fred <fredmfp at gmail.com> wrote:
> > Alexander Borghgraef a ?crit :

> >> I've built the enthought libraries from source on an Fedora 8 system,
> >> and installed them in a local directory (I have no access to
> >> /usr/lib). Importing the libraries works fine, and so does running the
> >> mayavi2 binary, but I haven't managed to get something plotted from a
> >> python script. Any ideas?
> > Can you tell us how you run your script ?

> Saved it to mlabtest.py, then ran

As it is indicated on the top of the page you are looking at, you should
run this in "ipython -wthread", for instance using "%run mlabtest.py".

Just a question, why are you using TVTK's mlab. Mayavi's mlab (developped
by the same people) is more maintained, eventhought it has a bit more
dependancies. You can have a look at
https://svn.enthought.com/enthought/attachment/wiki/MayaVi/user_guide.pdf?format=raw
on section 6 (page 22). One remark, if you are using an oldish version of
mayavi, you need to import mlab from enthought.mayavi.tools", rather than
"enthought.mayavi".

HTH,

Ga?l


From gael.varoquaux at normalesup.org  Fri Jun 13 08:57:49 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Fri, 13 Jun 2008 14:57:49 +0200
Subject: [SciPy-user] Mlab doesn't work
In-Reply-To: <48526C8B.5040300@gmail.com>
References: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>
	<4852671C.9080402@gmail.com>
	<9e8c52a20806130534x12a6c087jf486daca9a28c6e1@mail.gmail.com>
	<48526C8B.5040300@gmail.com>
Message-ID: <20080613125749.GB3573@phare.normalesup.org>

On Fri, Jun 13, 2008 at 02:48:11PM +0200, fred wrote:
> PS: Please don't be offensive with your subject.
> Mlab _does_ work and Ga?l is doing a great job on this.

I didn't find the subject offensive at all. I actually think that this
thread shows that we have a real problem, as many people stumble into
this.

First of all we have a communication problem. I have just edited the wiki
page to hopefully improve this. Second, this limitation of mlab is real a
stumbling block for new users. Prabhu and I are fully aware of it, and we
are looking for a solution. It is hard to find a good solution, as the
technical reasons behind this are fundamental. We need to find an
API/trick to present this in a way that doesn't hurt the user's intuition
(ipython -wthread is already much better then starting and stopping the
GUI mainloop manually).

Cheers,

Ga?l


From alexander.borghgraef.rma at gmail.com  Fri Jun 13 09:27:46 2008
From: alexander.borghgraef.rma at gmail.com (Alexander Borghgraef)
Date: Fri, 13 Jun 2008 15:27:46 +0200
Subject: [SciPy-user] Mlab doesn't work
In-Reply-To: <20080613125150.GA3573@phare.normalesup.org>
References: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>
	<4852671C.9080402@gmail.com>
	<9e8c52a20806130534x12a6c087jf486daca9a28c6e1@mail.gmail.com>
	<20080613125150.GA3573@phare.normalesup.org>
Message-ID: <9e8c52a20806130627ib41e905s3c92a7c10fbf4834@mail.gmail.com>

On Fri, Jun 13, 2008 at 2:51 PM, Gael Varoquaux
<gael.varoquaux at normalesup.org> wrote:
> On Fri, Jun 13, 2008 at 02:34:00PM +0200, Alexander Borghgraef wrote:
>> On Fri, Jun 13, 2008 at 2:25 PM, fred <fredmfp at gmail.com> wrote:
>> > Alexander Borghgraef a ?crit :
>
>> >> I've built the enthought libraries from source on an Fedora 8 system,
>> >> and installed them in a local directory (I have no access to
>> >> /usr/lib). Importing the libraries works fine, and so does running the
>> >> mayavi2 binary, but I haven't managed to get something plotted from a
>> >> python script. Any ideas?
>> > Can you tell us how you run your script ?
>
>> Saved it to mlabtest.py, then ran
>
> As it is indicated on the top of the page you are looking at, you should
> run this in "ipython -wthread", for instance using "%run mlabtest.py".

 Ok. That works, thanks.

> Just a question, why are you using TVTK's mlab. Mayavi's mlab (developped
> by the same people) is more maintained, eventhought it has a bit more
> dependancies.

 Well, because it says tvtk in the example on scipy's cookbook page. I
tend to use matplotlib for data visualization, but since 3D
visualization was hideously sluggish the last time I tried it, I
decided to give mayavi a spin. Download it , build it, try to run the
example code, harass the mailing list... you all know the drill, I
guess. :-) Standard OSS user behaviour.

> You can have a look at
> https://svn.enthought.com/enthought/attachment/wiki/MayaVi/user_guide.pdf?format=raw
> on section 6 (page 22). One remark, if you are using an oldish version of
> mayavi, you need to import mlab from enthought.mayavi.tools", rather than
> "enthought.mayavi".

RTFM, IOW :-D Will do. I just wanted a quick 3D plot of my data,
nothing fancy or interactive, so I skipped the manual and went for the
wiki.

Oh, and about the subject: no offense intended (and none taken
apparently, thanks Ga?l). I just lacked inspiration for the topic
title (and I didn't have much info to share in it) so I kept it
simple. Furthermore, I truly appreciate the work done by the mayavi
people, as well as the scipy people and all open source developers for
that matter. Couldn't do my work without them.

-- 
Alex Borghgraef


From fredmfp at gmail.com  Fri Jun 13 09:32:58 2008
From: fredmfp at gmail.com (fred)
Date: Fri, 13 Jun 2008 15:32:58 +0200
Subject: [SciPy-user] Mlab doesn't work
In-Reply-To: <9e8c52a20806130627ib41e905s3c92a7c10fbf4834@mail.gmail.com>
References: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>	<4852671C.9080402@gmail.com>	<9e8c52a20806130534x12a6c087jf486daca9a28c6e1@mail.gmail.com>	<20080613125150.GA3573@phare.normalesup.org>
	<9e8c52a20806130627ib41e905s3c92a7c10fbf4834@mail.gmail.com>
Message-ID: <4852770A.5010707@gmail.com>

Alexander Borghgraef a ?crit :

> Oh, and about the subject: no offense intended (and none taken
> apparently, thanks Ga?l). I just lacked inspiration for the topic

Sorry.
I'm getting too nervous, too tired, these days ;-)

Cheers,

-- 
Fred


From gael.varoquaux at normalesup.org  Fri Jun 13 09:33:23 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Fri, 13 Jun 2008 15:33:23 +0200
Subject: [SciPy-user] Mlab doesn't work
In-Reply-To: <9e8c52a20806130627ib41e905s3c92a7c10fbf4834@mail.gmail.com>
References: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>
	<4852671C.9080402@gmail.com>
	<9e8c52a20806130534x12a6c087jf486daca9a28c6e1@mail.gmail.com>
	<20080613125150.GA3573@phare.normalesup.org>
	<9e8c52a20806130627ib41e905s3c92a7c10fbf4834@mail.gmail.com>
Message-ID: <20080613133323.GG3573@phare.normalesup.org>

On Fri, Jun 13, 2008 at 03:27:46PM +0200, Alexander Borghgraef wrote:
> > Just a question, why are you using TVTK's mlab. Mayavi's mlab (developped
> > by the same people) is more maintained, eventhought it has a bit more
> > dependancies.

>  Well, because it says tvtk in the example on scipy's cookbook page. I
> tend to use matplotlib for data visualization, but since 3D
> visualization was hideously sluggish the last time I tried it, I
> decided to give mayavi a spin. Download it , build it, try to run the
> example code, harass the mailing list... you all know the drill, I
> guess. :-) Standard OSS user behaviour.

And standard OSS communication problem on our part (that is especially
bad for the Enthought Tools Suite, I will hopefully be working on it
soon). That's a good answer. I have added a paragraph to the wiki page to
help avoid user stumbling on TVTK.

The problem is that these modules are still fairly new, and best practice
is only starting to form, so the great "IntarWeb" isn't as full of
helpful on up to date info as it should. But the mailing lists are
reactive :) (thanks Fred for keeping a vigilant eye, I am often
unreachable currently, and it is good to know that you will be there to
answer these questions).

Cheers,

Ga?l


From ndbecker2 at gmail.com  Fri Jun 13 10:21:53 2008
From: ndbecker2 at gmail.com (Neal Becker)
Date: Fri, 13 Jun 2008 10:21:53 -0400
Subject: [SciPy-user] Using scipy specfunc in integration
Message-ID: <g2tvpv$vf6$1@ger.gmane.org>

Any ideas on this?
from scipy.special import erf
from math import exp, tan
def cot(x):
    return 1/tan(x)

N = 8
esnodB = 10
Rd = 10**(.1 * esnodB)

def F (y):
    return exp (-(y**2)) * erf (y * cot (pi/N))


Pe = float(N-1)/float(N) - 0.5 * erf (sqrt (Rd * sin (pi/N))) - 1/(sqrt(pi))
* quadrature(F, 0, sqrt(Rd) * sin (pi/N))[0]
TypeError: only length-1 arrays can be converted to Python scalars

It seems to be complaining about erf.
erf
Out[33]: <ufunc 'erf'>




From philbinj at gmail.com  Fri Jun 13 10:53:05 2008
From: philbinj at gmail.com (James Philbin)
Date: Fri, 13 Jun 2008 15:53:05 +0100
Subject: [SciPy-user] Sparse eigenvalues/eigenvectors problem
In-Reply-To: <d05265cb0806122011s6d740054jf29bb4b00daafc2@mail.gmail.com>
References: <2b1c8c4f0806120724n1d159964ubd39eb54922a62b5@mail.gmail.com>
	<d05265cb0806121452s2e2c2038o31ebab0fd410fa0c@mail.gmail.com>
	<2b1c8c4f0806121701l1e91ce99h2236a8ed20a73029@mail.gmail.com>
	<d05265cb0806122011s6d740054jf29bb4b00daafc2@mail.gmail.com>
Message-ID: <2b1c8c4f0806130753v4f141cb6x7548eaa55733598c@mail.gmail.com>

> Should be fixed now (SVN r4444).

http://www.robots.ox.ac.uk/~james/np_temp/S.mtx.gz

James


From philbinj at gmail.com  Fri Jun 13 10:57:54 2008
From: philbinj at gmail.com (James Philbin)
Date: Fri, 13 Jun 2008 15:57:54 +0100
Subject: [SciPy-user] Sparse eigenvalues/eigenvectors problem
In-Reply-To: <2b1c8c4f0806130753v4f141cb6x7548eaa55733598c@mail.gmail.com>
References: <2b1c8c4f0806120724n1d159964ubd39eb54922a62b5@mail.gmail.com>
	<d05265cb0806121452s2e2c2038o31ebab0fd410fa0c@mail.gmail.com>
	<2b1c8c4f0806121701l1e91ce99h2236a8ed20a73029@mail.gmail.com>
	<d05265cb0806122011s6d740054jf29bb4b00daafc2@mail.gmail.com>
	<2b1c8c4f0806130753v4f141cb6x7548eaa55733598c@mail.gmail.com>
Message-ID: <2b1c8c4f0806130757j7c3bc168u6ca78e7116addb79@mail.gmail.com>

Hmm... Has something changed in eigen_symmetric? Now, when I run
eigen_symmetric(S,k=10,which='LA') i get the following eigenvalues
returned: [ 1.  1.  1.  1.  1.  1.  1.  1.  1.  1.], which is good!
Curious as to what was changed.

Thanks,
James


From nwagner at iam.uni-stuttgart.de  Fri Jun 13 11:24:01 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Fri, 13 Jun 2008 17:24:01 +0200
Subject: [SciPy-user] Sparse eigenvalues/eigenvectors problem
In-Reply-To: <2b1c8c4f0806130757j7c3bc168u6ca78e7116addb79@mail.gmail.com>
References: <2b1c8c4f0806120724n1d159964ubd39eb54922a62b5@mail.gmail.com>
	<d05265cb0806121452s2e2c2038o31ebab0fd410fa0c@mail.gmail.com>
	<2b1c8c4f0806121701l1e91ce99h2236a8ed20a73029@mail.gmail.com>
	<d05265cb0806122011s6d740054jf29bb4b00daafc2@mail.gmail.com>
	<2b1c8c4f0806130753v4f141cb6x7548eaa55733598c@mail.gmail.com>
	<2b1c8c4f0806130757j7c3bc168u6ca78e7116addb79@mail.gmail.com>
Message-ID: <web-103265944@uni-stuttgart.de>

On Fri, 13 Jun 2008 15:57:54 +0100
  "James Philbin" <philbinj at gmail.com> wrote:
> Hmm... Has something changed in eigen_symmetric? Now, 
>when I run
> eigen_symmetric(S,k=10,which='LA') i get the following 
>eigenvalues
> returned: [ 1.  1.  1.  1.  1.  1.  1.  1.  1.  1.], 
>which is good!
> Curious as to what was changed.
> 
> Thanks,
> James

See

http://projects.scipy.org/scipy/scipy/changeset/4441

Nils
  


From nwagner at iam.uni-stuttgart.de  Fri Jun 13 12:54:13 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Fri, 13 Jun 2008 18:54:13 +0200
Subject: [SciPy-user] Sparse eigenvalues/eigenvectors problem
In-Reply-To: <2b1c8c4f0806130757j7c3bc168u6ca78e7116addb79@mail.gmail.com>
References: <2b1c8c4f0806120724n1d159964ubd39eb54922a62b5@mail.gmail.com>
	<d05265cb0806121452s2e2c2038o31ebab0fd410fa0c@mail.gmail.com>
	<2b1c8c4f0806121701l1e91ce99h2236a8ed20a73029@mail.gmail.com>
	<d05265cb0806122011s6d740054jf29bb4b00daafc2@mail.gmail.com>
	<2b1c8c4f0806130753v4f141cb6x7548eaa55733598c@mail.gmail.com>
	<2b1c8c4f0806130757j7c3bc168u6ca78e7116addb79@mail.gmail.com>
Message-ID: <web-103269268@uni-stuttgart.de>

On Fri, 13 Jun 2008 15:57:54 +0100
  "James Philbin" <philbinj at gmail.com> wrote:
> Hmm... Has something changed in eigen_symmetric? Now, 
>when I run
> eigen_symmetric(S,k=10,which='LA') i get the following 
>eigenvalues
> returned: [ 1.  1.  1.  1.  1.  1.  1.  1.  1.  1.], 
>which is good!
> Curious as to what was changed.
> 
> Thanks,
> James
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user

Is your matrix positive definite ?
Did you try symeig ?
http://mdp-toolkit.sourceforge.net/symeig.html

from scipy import *
from pylab import spy, show
from symeig import symeig
A = io.mmread('S.mtx.gz')
w,v = symeig(A.todense(),range=(1,25))
spy(A.todense())
show()

Nils


From robert.kern at gmail.com  Fri Jun 13 14:57:40 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Fri, 13 Jun 2008 13:57:40 -0500
Subject: [SciPy-user] Using scipy specfunc in integration
In-Reply-To: <g2tvpv$vf6$1@ger.gmane.org>
References: <g2tvpv$vf6$1@ger.gmane.org>
Message-ID: <3d375d730806131157k36ca3535mc2c428c2f3980137@mail.gmail.com>

On Fri, Jun 13, 2008 at 09:21, Neal Becker <ndbecker2 at gmail.com> wrote:
> Any ideas on this?
> from scipy.special import erf
> from math import exp, tan
> def cot(x):
>    return 1/tan(x)
>
> N = 8
> esnodB = 10
> Rd = 10**(.1 * esnodB)
>
> def F (y):
>    return exp (-(y**2)) * erf (y * cot (pi/N))
>
>
> Pe = float(N-1)/float(N) - 0.5 * erf (sqrt (Rd * sin (pi/N))) - 1/(sqrt(pi))
> * quadrature(F, 0, sqrt(Rd) * sin (pi/N))[0]
> TypeError: only length-1 arrays can be converted to Python scalars
>
> It seems to be complaining about erf.
> erf
> Out[33]: <ufunc 'erf'>

I think it's complaining about exp(), actually. quadrature() is going
to pass arrays to F(), not scalars.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From contact at pythonxy.com  Fri Jun 13 17:06:05 2008
From: contact at pythonxy.com (Pierre Raybaut)
Date: Fri, 13 Jun 2008 23:06:05 +0200
Subject: [SciPy-user] [ Python(x,y) ] New release : 1.2.6
Message-ID: <4852E13D.6000405@pythonxy.com>

Hi all,

Python(x,y) 1.2.6 is now available on http://www.pythonxy.com.

Changes history
06 -14 -2008 - Version 1.2.6 :

     * Updated:
           o Cython 0.9.8 (see website)
     * Added:
           o py2exe 0.6.6 - Deployment tool which converts Python 
scripts into stand-alone Windows executables (i.e. target machine does 
not require Python or any other library to be installed) - see website
           o PyDAP 2.2.6.4 - Python implementation of the Data Access 
Protocol, a.k.a. DODS or OPeNDAP (see website)
           o httplib2 0.4 - A comprehensive HTTP client library that 
supports many features left out of other HTTP libraries (see website)
           o Python(x,y) console: some improvements on automatic logging
           o Interactive consoles: default working directory is the 
Eclipse/Python workspace folder (default path: User Documents\Python)
           o Notepad++: tab has been replaced by 4 spaces (better 
compatibility with Python indentation)

Regards,
Pierre Raybaut








From pgmdevlist at gmail.com  Fri Jun 13 19:23:46 2008
From: pgmdevlist at gmail.com (Pierre GM)
Date: Fri, 13 Jun 2008 19:23:46 -0400
Subject: [SciPy-user] iteratively masking timeseries
In-Reply-To: <g2s3ir$o2f$1@ger.gmane.org>
References: <g2s3ir$o2f$1@ger.gmane.org>
Message-ID: <200806131923.46336.pgmdevlist@gmail.com>

On Thursday 12 June 2008 17:14:03 Tim Michelsen wrote:
> Does that mean that one can only mask an array once and cannot mask more
> values later on?

No no, the message
> IndexError: Masked arrays must be filled before they can be used as
> indices!

doesn't mean you can't mask arrays more than once, just that you can't use a 
MaskedArray as index in a TimeSeries without having to fill it first. The 
reason for this behavior is that we can't tell beforehand how you want to 
deal with your masked values: should a masked value be considered True ? 
False ?

So, when you want to do something like:
> mask[mask<0] = numpy.ma.masked

just do:

mask[(mask<0).filled(True)] = numpy.ma.masked

That way, you're masking the values of mask that are negative, and keep the 
masked values as masked.



From peridot.faceted at gmail.com  Fri Jun 13 19:36:08 2008
From: peridot.faceted at gmail.com (Anne Archibald)
Date: Fri, 13 Jun 2008 17:36:08 -0600
Subject: [SciPy-user] Using scipy specfunc in integration
In-Reply-To: <3d375d730806131157k36ca3535mc2c428c2f3980137@mail.gmail.com>
References: <g2tvpv$vf6$1@ger.gmane.org>
	<3d375d730806131157k36ca3535mc2c428c2f3980137@mail.gmail.com>
Message-ID: <ce557a360806131636n22346ea6s80a6abe93dfbb4a7@mail.gmail.com>

2008/6/13 Robert Kern <robert.kern at gmail.com>:
> On Fri, Jun 13, 2008 at 09:21, Neal Becker <ndbecker2 at gmail.com> wrote:
>> Any ideas on this?
>> from scipy.special import erf
>> from math import exp, tan
>> def cot(x):
>>    return 1/tan(x)
>>
>> N = 8
>> esnodB = 10
>> Rd = 10**(.1 * esnodB)
>>
>> def F (y):
>>    return exp (-(y**2)) * erf (y * cot (pi/N))
>>
>>
>> Pe = float(N-1)/float(N) - 0.5 * erf (sqrt (Rd * sin (pi/N))) - 1/(sqrt(pi))
>> * quadrature(F, 0, sqrt(Rd) * sin (pi/N))[0]
>> TypeError: only length-1 arrays can be converted to Python scalars
>>
>> It seems to be complaining about erf.
>> erf
>> Out[33]: <ufunc 'erf'>
>
> I think it's complaining about exp(), actually. quadrature() is going
> to pass arrays to F(), not scalars.

Specifically, math.exp and math.tan do not accept vector arguments;
don't use them. Use numpy.exp and numpy.tan instead.

Anne


From lopmart at gmail.com  Sat Jun 14 05:26:21 2008
From: lopmart at gmail.com (Jose Lopez)
Date: Sat, 14 Jun 2008 02:26:21 -0700
Subject: [SciPy-user] memory error at leastsq
Message-ID: <4eeef9d40806140226w519c2b2do95d8f40067caf108@mail.gmail.com>

hi, help with memory error at leastsq

I am working with a model mathematic of mean squeared error, according to
the least square appproach, a solution is   sought by minimizing

          MSE(I,B) = (S1-(I*B))^2+ (S2-(I*B))^2 + (S3-(I*B))^2

where S1,S2,S3,I and B are matrix of k x l


and i work whit the optimize.leastsq of scipy, well, for k = 5 and l = 5
work fine, but for k=64 and l =64 scipy give  me  the  next error

MemoryError:
File: "c:\python25\lib....\minpack.py",line 268,in leastsq
retval =
_minpack._lmdif(func,x0,args,full_output,ftol,xtol,gtol,maxfev,epsfcn,factor,diag)

 my line code call  is

xopt=optimize.leastsq(funcion,IB,args=(Sgen)) where Sgen is S1,S2,S3 at
order lexicografic and IB is I and B at order lexicografic.
so, my questions are :  why happend it? and how am  I  solve it?
thanks

JL
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080614/e39c1cad/attachment.html>

From david at ar.media.kyoto-u.ac.jp  Sat Jun 14 05:35:11 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Sat, 14 Jun 2008 18:35:11 +0900
Subject: [SciPy-user] memory error at leastsq
In-Reply-To: <4eeef9d40806140226w519c2b2do95d8f40067caf108@mail.gmail.com>
References: <4eeef9d40806140226w519c2b2do95d8f40067caf108@mail.gmail.com>
Message-ID: <485390CF.3000806@ar.media.kyoto-u.ac.jp>

Jose Lopez wrote:
>
> hi, help with memory error at leastsq
>

Hi Jose,

We can't really help you unless you give us more information:
    - Operating System (here it looks like windows, but 32 bits, 64 
bits, vista, xp ?)
    - How did you install numpy / scipy (from sources, binaries, etc...)

cheers,

David


From robince at gmail.com  Sat Jun 14 07:12:57 2008
From: robince at gmail.com (Robin)
Date: Sat, 14 Jun 2008 12:12:57 +0100
Subject: [SciPy-user] memory error at leastsq
In-Reply-To: <4eeef9d40806140226w519c2b2do95d8f40067caf108@mail.gmail.com>
References: <4eeef9d40806140226w519c2b2do95d8f40067caf108@mail.gmail.com>
Message-ID: <b14b627d0806140412l743c2f9aiae3a6efc47b56388@mail.gmail.com>

On Sat, Jun 14, 2008 at 10:26 AM, Jose Lopez <lopmart at gmail.com> wrote:
> hi, help with memory error at leastsq
>
> I am working with a model mathematic of mean squeared error, according to
> the least square appproach, a solution is   sought by minimizing
>
>           MSE(I,B) = (S1-(I*B))^2+ (S2-(I*B))^2 + (S3-(I*B))^2
>
> where S1,S2,S3,I and B are matrix of k x l

I don't know if it could be related to your error, but you should be
careful whether you are using arrays or matrices. I think with
matrices the ^2 above will do matrix product, whereas for mean square
error you probably want this operation element-wise... So perhaps try
using arrays instead of matrices? You would probably want to take the
sum as well. It might be faster to use dot as well to do the squaring
(dot(v,v))...

Cheers

Robin


From lopmart at gmail.com  Sat Jun 14 08:44:50 2008
From: lopmart at gmail.com (Jose Lopez)
Date: Sat, 14 Jun 2008 05:44:50 -0700
Subject: [SciPy-user] memory error at leastsq
In-Reply-To: <485390CF.3000806@ar.media.kyoto-u.ac.jp>
References: <4eeef9d40806140226w519c2b2do95d8f40067caf108@mail.gmail.com>
	<485390CF.3000806@ar.media.kyoto-u.ac.jp>
Message-ID: <4eeef9d40806140544q45ee7869o9dbd1601b96d1547@mail.gmail.com>

the operating system  is windows vista 32 bits
and install scipy from python xy (http://www.pythonxy.com)

On Sat, Jun 14, 2008 at 2:35 AM, David Cournapeau <
david at ar.media.kyoto-u.ac.jp> wrote:

> Jose Lopez wrote:
> >
> > hi, help with memory error at leastsq
> >
>
> Hi Jose,
>
> We can't really help you unless you give us more information:
>    - Operating System (here it looks like windows, but 32 bits, 64
> bits, vista, xp ?)
>    - How did you install numpy / scipy (from sources, binaries, etc...)
>
> cheers,
>
> David
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080614/b253808a/attachment.html>

From gael.varoquaux at normalesup.org  Sat Jun 14 13:21:23 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Sat, 14 Jun 2008 19:21:23 +0200
Subject: [SciPy-user] Alex Martelli giving the SciPy 2008 Keynote
Message-ID: <20080614172123.GB31186@phare.normalesup.org>

On behalf of the SciPy2008 organizing committee, I am happy to announce
that the Keynote at the conference will be given by Alex Martelli.

It is a pleasure for us to receive Alex. He currently works as "Uber Tech
Leader" at Google and is the author of two of the Python classics: 
"Python in a nutshell" and the "Python CookBook". Alex graduated in
electronic engineering from the university of Bologna and worked in chip
design first for Texas Instrument, and later for IBM Research. During the
8 years he spent at IBM, he gradually shifted from hardware design to
software development while winning three Outstanding Technical
Achievement Awards. Then he joined think3 inc., and Italian CAD company,
as Senior Software Consultant where he developed libraries, network
protocols, GUI engines, event frameworks, and web access frontends. After
12 years at think3, he worked for 3 years as a freelance consultant,
mostly doing Python development, before joining Google.

Alex won the 2002 Activators' Choice Award, and the 2006 Frank Willison
award for outstanding contributions to the Python community.

Alex has also taught courses on programming, development methods,
object-oriented design, and numerical computing, at Ferrara University
(Italy) and other venues.  Alex's proudest achievement is the articles
that appeared in Bridge World (January/February 2000), which were hailed
as giant steps towards solving issues that had haunted contract-bridge
game theoreticians for decades.

This biography was loosely adapted from Alex's autobiography
(http://www.aleax.it/bio.txt), more information can be found on his
website http://www.aleax.it .

Ga?l


From ndbecker2 at gmail.com  Sat Jun 14 17:57:07 2008
From: ndbecker2 at gmail.com (Neal Becker)
Date: Sat, 14 Jun 2008 17:57:07 -0400
Subject: [SciPy-user] Using scipy specfunc in integration
References: <g2tvpv$vf6$1@ger.gmane.org>
	<3d375d730806131157k36ca3535mc2c428c2f3980137@mail.gmail.com>
	<ce557a360806131636n22346ea6s80a6abe93dfbb4a7@mail.gmail.com>
Message-ID: <g31ere$vtf$1@ger.gmane.org>

Anne Archibald wrote:

> 2008/6/13 Robert Kern <robert.kern at gmail.com>:
>> On Fri, Jun 13, 2008 at 09:21, Neal Becker <ndbecker2 at gmail.com> wrote:
>>> Any ideas on this?
>>> from scipy.special import erf
>>> from math import exp, tan
>>> def cot(x):
>>>    return 1/tan(x)
>>>
>>> N = 8
>>> esnodB = 10
>>> Rd = 10**(.1 * esnodB)
>>>
>>> def F (y):
>>>    return exp (-(y**2)) * erf (y * cot (pi/N))
>>>
>>>
>>> Pe = float(N-1)/float(N) - 0.5 * erf (sqrt (Rd * sin (pi/N))) -
>>> 1/(sqrt(pi)) * quadrature(F, 0, sqrt(Rd) * sin (pi/N))[0]
>>> TypeError: only length-1 arrays can be converted to Python scalars
>>>
>>> It seems to be complaining about erf.
>>> erf
>>> Out[33]: <ufunc 'erf'>
>>
>> I think it's complaining about exp(), actually. quadrature() is going
>> to pass arrays to F(), not scalars.
> 
> Specifically, math.exp and math.tan do not accept vector arguments;
> don't use them. Use numpy.exp and numpy.tan instead.
> 
> Anne
Thanks, but I'm confused.  The above code is strictly scalar - who's asking
for vectors?



From robert.kern at gmail.com  Sat Jun 14 18:25:14 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Sat, 14 Jun 2008 17:25:14 -0500
Subject: [SciPy-user] Using scipy specfunc in integration
In-Reply-To: <g31ere$vtf$1@ger.gmane.org>
References: <g2tvpv$vf6$1@ger.gmane.org>
	<3d375d730806131157k36ca3535mc2c428c2f3980137@mail.gmail.com>
	<ce557a360806131636n22346ea6s80a6abe93dfbb4a7@mail.gmail.com>
	<g31ere$vtf$1@ger.gmane.org>
Message-ID: <3d375d730806141525r19ae804ek3691a3accf43fb36@mail.gmail.com>

On Sat, Jun 14, 2008 at 16:57, Neal Becker <ndbecker2 at gmail.com> wrote:
> Anne Archibald wrote:
>
>> 2008/6/13 Robert Kern <robert.kern at gmail.com>:
>>> On Fri, Jun 13, 2008 at 09:21, Neal Becker <ndbecker2 at gmail.com> wrote:
>>>> Any ideas on this?
>>>> from scipy.special import erf
>>>> from math import exp, tan
>>>> def cot(x):
>>>>    return 1/tan(x)
>>>>
>>>> N = 8
>>>> esnodB = 10
>>>> Rd = 10**(.1 * esnodB)
>>>>
>>>> def F (y):
>>>>    return exp (-(y**2)) * erf (y * cot (pi/N))
>>>>
>>>>
>>>> Pe = float(N-1)/float(N) - 0.5 * erf (sqrt (Rd * sin (pi/N))) -
>>>> 1/(sqrt(pi)) * quadrature(F, 0, sqrt(Rd) * sin (pi/N))[0]
>>>> TypeError: only length-1 arrays can be converted to Python scalars
>>>>
>>>> It seems to be complaining about erf.
>>>> erf
>>>> Out[33]: <ufunc 'erf'>
>>>
>>> I think it's complaining about exp(), actually. quadrature() is going
>>> to pass arrays to F(), not scalars.
>>
>> Specifically, math.exp and math.tan do not accept vector arguments;
>> don't use them. Use numpy.exp and numpy.tan instead.
>>
>> Anne
> Thanks, but I'm confused.  The above code is strictly scalar - who's asking
> for vectors?

Like I said, quadrature() passes arrays to the integrand function, not
scalars, and expects the integrand function to evaluate itself
elementwise on those arrays. Look at its docstring. Set vec_func=False
if you want the function to only take scalars.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From haase at msg.ucsf.edu  Sun Jun 15 03:00:52 2008
From: haase at msg.ucsf.edu (Sebastian Haase)
Date: Sun, 15 Jun 2008 09:00:52 +0200
Subject: [SciPy-user] The IO library and image file formats -- compare
	with with PIL
In-Reply-To: <3d375d730804211349p4e75ab5cs6603fc13869f519@mail.gmail.com>
References: <bc657ead0804180136x5bf54ebax50cef6c0b5092a84@mail.gmail.com>
	<9457e7c80804210038n2d619393k74cdee0cb481a59b@mail.gmail.com>
	<bc657ead0804210119h4237b6f8o18212a7dde8c0b47@mail.gmail.com>
	<10B72505-ACEB-4EAE-A1B9-39A5038224CD@yale.edu>
	<9457e7c80804210704n348d2129l7228e5fce3ab164d@mail.gmail.com>
	<53E48B1E-91A0-422A-8309-CA58D4BCE486@yale.edu>
	<9457e7c80804211013i27a32020l4d704bf75d318c5f@mail.gmail.com>
	<3d375d730804211020n22ba0d72k8ae86cf8724a4c5b@mail.gmail.com>
	<0D36E713-05E6-43FA-9634-AD8BEA2B8743@yale.edu>
	<3d375d730804211349p4e75ab5cs6603fc13869f519@mail.gmail.com>
Message-ID: <bc657ead0806150000t11818686xfb5ddb4fbc061cab@mail.gmail.com>

Zach et al,

hi, has there been a development on this ... ?
I.e. has anyone gotten in contact with Fredrik Lundh  ?

I recently got more problems with reading a Zeiss Confocal Microscope
(LSM) file -- as far as I can tell, LSM files are TIFF files, and
ImageMagick recognizes the file in question as such ....
The point here is,
that I don't think that the Image-SIG mailing list is very helpful ....
(it'is only a fraction as good as what we are used to over here at
SciPy ....  )

So, (again,) it would be great if we could unite our image-io
interests over here at SciPy...

Thanks,
Sebastian



On Mon, Apr 21, 2008 at 10:49 PM, Robert Kern <robert.kern at gmail.com> wrote:
> On Mon, Apr 21, 2008 at 3:37 PM, Zachary Pincus <zachary.pincus at yale.edu> wrote:
>>  Let me look into whether this idea is at all feasible, and if it is we
>>  can revisit the issue of whether it belongs anywhere near scipy.
>>  (Would getting Fredrik Lundh's OK to use various bits in this way make
>>  things easier? He does seem much more responsive to direct queries
>>  than patch-submissions.)
>
> That would alleviate my concerns, yes.
>
> --
> Robert Kern


From yosefmel at post.tau.ac.il  Sun Jun 15 04:18:59 2008
From: yosefmel at post.tau.ac.il (Yosef Meller)
Date: Sun, 15 Jun 2008 11:18:59 +0300
Subject: [SciPy-user] Solver with n-dimentional steps
Message-ID: <200806151118.59818.yosefmel@post.tau.ac.il>


Hi all,

I'm trying to use scipy.optimize.fsolve(function, x0, args) with a function 
that has an input vector with length in the tens of thousands (about 3*4000). 
This turns out to be impractical, because fsolve runs the function after 
taking a step in each direction in turn. Runing the function takes about 1/10 
second, so in the thousands of runs it's way too much.

What I'm looking for, I guess, is a way to make fsolve take a multidimentional 
step, so the function does not need to run 12000 times for each step. 

Thanks in advance, 
Yosef Meller.


From dwf at cs.toronto.edu  Sun Jun 15 06:19:54 2008
From: dwf at cs.toronto.edu (David Warde-Farley)
Date: Sun, 15 Jun 2008 06:19:54 -0400
Subject: [SciPy-user] Solver with n-dimentional steps
In-Reply-To: <200806151118.59818.yosefmel@post.tau.ac.il>
References: <200806151118.59818.yosefmel@post.tau.ac.il>
Message-ID: <1B9D8F05-27B4-4227-9119-B8EC66BD5C4B@cs.toronto.edu>

On 15-Jun-08, at 4:18 AM, Yosef Meller wrote:

> I'm trying to use scipy.optimize.fsolve(function, x0, args) with a  
> function
> that has an input vector with length in the tens of thousands (about  
> 3*4000).
> This turns out to be impractical, because fsolve runs the function  
> after
> taking a step in each direction in turn. Runing the function takes  
> about 1/10
> second, so in the thousands of runs it's way too much.

It's doing finite differences to estimate the gradient. This is pretty  
much unavoidable if you're only giving it those three arguments. Have  
you thought about supplying fprime as a function to analytically  
compute the gradient?

David


From dmitrey.kroshko at scipy.org  Sun Jun 15 06:16:13 2008
From: dmitrey.kroshko at scipy.org (dmitrey)
Date: Sun, 15 Jun 2008 13:16:13 +0300
Subject: [SciPy-user] Solver with n-dimentional steps
In-Reply-To: <200806151118.59818.yosefmel@post.tau.ac.il>
References: <200806151118.59818.yosefmel@post.tau.ac.il>
Message-ID: <4854EBED.303@scipy.org>

you should provide gradient analytically, there are no other ways to 
speedup fsolve with so much nVars problem.
Regards, D.

Yosef Meller wrote:
> Hi all,
>
> I'm trying to use scipy.optimize.fsolve(function, x0, args) with a function 
> that has an input vector with length in the tens of thousands (about 3*4000). 
> This turns out to be impractical, because fsolve runs the function after 
> taking a step in each direction in turn. Runing the function takes about 1/10 
> second, so in the thousands of runs it's way too much.
>
> What I'm looking for, I guess, is a way to make fsolve take a multidimentional 
> step, so the function does not need to run 12000 times for each step. 
>
> Thanks in advance, 
> Yosef Meller.
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
>
>
>   



From yosefmel at post.tau.ac.il  Sun Jun 15 06:41:16 2008
From: yosefmel at post.tau.ac.il (Yosef Meller)
Date: Sun, 15 Jun 2008 13:41:16 +0300
Subject: [SciPy-user] Solver with n-dimentional steps
In-Reply-To: <1B9D8F05-27B4-4227-9119-B8EC66BD5C4B@cs.toronto.edu>
References: <200806151118.59818.yosefmel@post.tau.ac.il>
	<1B9D8F05-27B4-4227-9119-B8EC66BD5C4B@cs.toronto.edu>
Message-ID: <200806151341.16358.yosefmel@post.tau.ac.il>

On Sunday 15 June 2008 13:19:54 David Warde-Farley wrote:
> It's doing finite differences to estimate the gradient. This is pretty
> much unavoidable if you're only giving it those three arguments. Have
> you thought about supplying fprime as a function to analytically
> compute the gradient?

Havind dug through the fortran code, I realized that I would have to do that. 
Thanks for the answer.



From fredmfp at gmail.com  Sun Jun 15 07:54:22 2008
From: fredmfp at gmail.com (fred)
Date: Sun, 15 Jun 2008 13:54:22 +0200
Subject: [SciPy-user] reading binary files written by a gfortran code...
Message-ID: <485502EE.6030708@gmail.com>

Hi,

Ok, this is a well known issue that binary files written by a fortran 
program have a "special" format.

So my question is : is it possible to write binary file _using gfortran_
(no problem with intel fortran compiler) without this formatting ?

Is there a peculiar syntax to the open() function ?

I did not find any relevant information on the web.

The obvious reason is that I have to read these files with 
scipy.io.numpyio fread method.

TIA.

Cheers,

-- 
Fred


From hoytak at gmail.com  Sun Jun 15 09:55:47 2008
From: hoytak at gmail.com (Hoyt Koepke)
Date: Sun, 15 Jun 2008 16:55:47 +0300
Subject: [SciPy-user] Solver with n-dimentional steps
In-Reply-To: <200806151341.16358.yosefmel@post.tau.ac.il>
References: <200806151118.59818.yosefmel@post.tau.ac.il>
	<1B9D8F05-27B4-4227-9119-B8EC66BD5C4B@cs.toronto.edu>
	<200806151341.16358.yosefmel@post.tau.ac.il>
Message-ID: <4db580fd0806150655v4a13b4b5yd0c49b482d4a8ec1@mail.gmail.com>

Yosef,

I'm suggesting this partly to satisfy my own curiosity, so ignore this
if it is difficult.   Since you are running with so many dimensions,
you might want to try minimizing the square of your function using
fmin_l_bfgs_b.  lbfgs is written for higher dimensions, and I've heard
it can give incredible speed improvements over many other methods.
I'm curious if it would help in your case.

--Hoyt

On Sun, Jun 15, 2008 at 1:41 PM, Yosef Meller <yosefmel at post.tau.ac.il> wrote:
> On Sunday 15 June 2008 13:19:54 David Warde-Farley wrote:
>> It's doing finite differences to estimate the gradient. This is pretty
>> much unavoidable if you're only giving it those three arguments. Have
>> you thought about supplying fprime as a function to analytically
>> compute the gradient?
>
> Havind dug through the fortran code, I realized that I would have to do that.
> Thanks for the answer.
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>



-- 
+++++++++++++++++++++++++++++++++++
Hoyt Koepke
UBC Department of Computer Science
http://www.cs.ubc.ca/~hoytak/
hoytak at gmail.com
+++++++++++++++++++++++++++++++++++


From zachary.pincus at yale.edu  Sun Jun 15 11:33:57 2008
From: zachary.pincus at yale.edu (Zachary Pincus)
Date: Sun, 15 Jun 2008 11:33:57 -0400
Subject: [SciPy-user] The IO library and image file formats -- compare
	with with PIL
In-Reply-To: <bc657ead0806150000t11818686xfb5ddb4fbc061cab@mail.gmail.com>
References: <bc657ead0804180136x5bf54ebax50cef6c0b5092a84@mail.gmail.com>
	<9457e7c80804210038n2d619393k74cdee0cb481a59b@mail.gmail.com>
	<bc657ead0804210119h4237b6f8o18212a7dde8c0b47@mail.gmail.com>
	<10B72505-ACEB-4EAE-A1B9-39A5038224CD@yale.edu>
	<9457e7c80804210704n348d2129l7228e5fce3ab164d@mail.gmail.com>
	<53E48B1E-91A0-422A-8309-CA58D4BCE486@yale.edu>
	<9457e7c80804211013i27a32020l4d704bf75d318c5f@mail.gmail.com>
	<3d375d730804211020n22ba0d72k8ae86cf8724a4c5b@mail.gmail.com>
	<0D36E713-05E6-43FA-9634-AD8BEA2B8743@yale.edu>
	<3d375d730804211349p4e75ab5cs6603fc13869f519@mail.gmail.com>
	<bc657ead0806150000t11818686xfb5ddb4fbc061cab@mail.gmail.com>
Message-ID: <53B94D38-4F29-4FB6-9AED-A9647FE3A905@yale.edu>

Hi,

I've made little progress on this as I've been really quite busy with  
my research, I'm afraid. Sorry.

I will try to contact Fredrik today, though.

> I recently got more problems with reading a Zeiss Confocal Microscope
> (LSM) file -- as far as I can tell, LSM files are TIFF files, and
> ImageMagick recognizes the file in question as such ....

I recently have been looking at some numpy bindings for ImageMagick  
that were sent to me by an acquaintance at CMU. I can send them over  
(once I make sure there's no problem re-distributing them), if that  
would be helpful for anyone before we get a dependency-free solution.

Also on the microscopy-file-format question (a most vexed issue, to be  
sure), perhaps the tools at http://www.loci.wisc.edu/ome/formats.html  
will be useful. I've been using the bfconvert tool to turn Zeiss ZVI  
files into normal tiffs, and it handles a ton of other microscopy  
formats. (The bad news is that it's all in Java, so direct python  
bindings are rather unlikely.)

Zach



On Jun 15, 2008, at 3:00 AM, Sebastian Haase wrote:

> Zach et al,
>
> hi, has there been a development on this ... ?
> I.e. has anyone gotten in contact with Fredrik Lundh  ?
>
> I recently got more problems with reading a Zeiss Confocal Microscope
> (LSM) file -- as far as I can tell, LSM files are TIFF files, and
> ImageMagick recognizes the file in question as such ....
> The point here is,
> that I don't think that the Image-SIG mailing list is very  
> helpful ....
> (it'is only a fraction as good as what we are used to over here at
> SciPy ....  )
>
> So, (again,) it would be great if we could unite our image-io
> interests over here at SciPy...
>
> Thanks,
> Sebastian
>
>
>
> On Mon, Apr 21, 2008 at 10:49 PM, Robert Kern  
> <robert.kern at gmail.com> wrote:
>> On Mon, Apr 21, 2008 at 3:37 PM, Zachary Pincus <zachary.pincus at yale.edu 
>> > wrote:
>>> Let me look into whether this idea is at all feasible, and if it  
>>> is we
>>> can revisit the issue of whether it belongs anywhere near scipy.
>>> (Would getting Fredrik Lundh's OK to use various bits in this way  
>>> make
>>> things easier? He does seem much more responsive to direct queries
>>> than patch-submissions.)
>>
>> That would alleviate my concerns, yes.
>>
>> --
>> Robert Kern
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user



From phaustin at gmail.com  Sun Jun 15 12:10:30 2008
From: phaustin at gmail.com (Phil Austin)
Date: Sun, 15 Jun 2008 09:10:30 -0700
Subject: [SciPy-user] reading binary files written by a gfortran code...
In-Reply-To: <485502EE.6030708@gmail.com>
References: <485502EE.6030708@gmail.com>
Message-ID: <48553EF6.1060001@gmail.com>

fred wrote:
> Hi,
> 
> Ok, this is a well known issue that binary files written by a fortran 
> program have a "special" format.
> 
> So my question is : is it possible to write binary file _using gfortran_
> (no problem with intel fortran compiler) without this formatting ?

You want to write/read direct access files, which are just the bytes
without the compiler-specific recordlength information.  Here is
an example of a some tutorial code I wrote for myself exercising
memmap and a fortran direct-access read -- just replace read with
write in the fortran code to get output

1) write a test binary  data file in python with fortran layout and make
sure you can read it back in using mmemap

import numpy as np
xdim=4
ydim=5
zdim=6
theData=np.empty([xdim,ydim,zdim],dtype=np.int32)
for i in np.arange(xdim):
     for j in np.arange(ydim):
         for k in np.arange(zdim):
             theData[i,j,k]=100*(i+1) + 10*(j+1) + (k+1)
print theData
fortout=open('fortout.dat','w')
Cout=open('Cout.dat','w')
theData.tofile(fortout)
theData.tofile(Cout)
fortout.close()
Cout.close()
theFout = np.memmap('fortout.dat', dtype=np.int32,\
                     shape=(xdim,ydim,zdim), order='F')
theCout = np.memmap('Cout.dat', dtype=np.int32, \
                     shape=(xdim,ydim,zdim), order='C')
for i in np.arange(xdim):
     for j in np.arange(ydim):
         for k in np.arange(zdim):
             theFout[i,j,k]=theData[i,j,k]
             theCout[i,j,k]=theData[i,j,k]
theFout.sync()
theCout.sync()


2) and now read fortout.dat using standard f95 direct access:

program readfile
integer,parameter :: xdim=4,ydim=5,zdim=6
integer(kind=4):: thevar(xdim,ydim,zdim)
integer :: i,j,k,recnum
open(unit=12,file="fortout.dat",access="direct",action="read",recl=4)
recnum=0
do i=1,xdim
   do j=1,ydim
      do k=1,zdim
         recnum=recnum+1
         read(12,rec=recnum) thevar(i,j,k)
         print *,i,j,k,thevar(i,j,k)
      enddo
   enddo
enddo
close(unit=12)
end


From haase at msg.ucsf.edu  Sun Jun 15 12:50:18 2008
From: haase at msg.ucsf.edu (Sebastian Haase)
Date: Sun, 15 Jun 2008 18:50:18 +0200
Subject: [SciPy-user] The IO library and image file formats -- compare
	with with PIL
In-Reply-To: <53B94D38-4F29-4FB6-9AED-A9647FE3A905@yale.edu>
References: <bc657ead0804180136x5bf54ebax50cef6c0b5092a84@mail.gmail.com>
	<10B72505-ACEB-4EAE-A1B9-39A5038224CD@yale.edu>
	<9457e7c80804210704n348d2129l7228e5fce3ab164d@mail.gmail.com>
	<53E48B1E-91A0-422A-8309-CA58D4BCE486@yale.edu>
	<9457e7c80804211013i27a32020l4d704bf75d318c5f@mail.gmail.com>
	<3d375d730804211020n22ba0d72k8ae86cf8724a4c5b@mail.gmail.com>
	<0D36E713-05E6-43FA-9634-AD8BEA2B8743@yale.edu>
	<3d375d730804211349p4e75ab5cs6603fc13869f519@mail.gmail.com>
	<bc657ead0806150000t11818686xfb5ddb4fbc061cab@mail.gmail.com>
	<53B94D38-4F29-4FB6-9AED-A9647FE3A905@yale.edu>
Message-ID: <bc657ead0806150950q1dbecacfud799c96436792c8f@mail.gmail.com>

On Sun, Jun 15, 2008 at 5:33 PM, Zachary Pincus <zachary.pincus at yale.edu> wrote:
> Hi,
>
> I've made little progress on this as I've been really quite busy with
> my research, I'm afraid. Sorry.
>
> I will try to contact Fredrik today, though.
>
great - thanks.

>> I recently got more problems with reading a Zeiss Confocal Microscope
>> (LSM) file -- as far as I can tell, LSM files are TIFF files, and
>> ImageMagick recognizes the file in question as such ....
>
> I recently have been looking at some numpy bindings for ImageMagick
> that were sent to me by an acquaintance at CMU. I can send them over
> (once I make sure there's no problem re-distributing them), if that
> would be helpful for anyone before we get a dependency-free solution.
>
What I have seen of imagemagick in terms of binding was very discouraging ....
( Mostly abandoned stuff because I.M. seems to be moving a target;
with every release a new API -- maybe that has changed by now. ) Main
point though: PIL is really close to what we need....

> Also on the microscopy-file-format question (a most vexed issue, to be
> sure), perhaps the tools at http://www.loci.wisc.edu/ome/formats.html
> will be useful. I've been using the bfconvert tool to turn Zeiss ZVI
> files into normal tiffs, and it handles a ton of other microscopy
> formats. (The bad news is that it's all in Java, so direct python
> bindings are rather unlikely.)

Yeah I know about this. I actually tried some Jython on it: and I made
a nice script which converts "whole directory trees full of "
LOCI-readable files into my preferred (memory mappable) file format
(it's close to the Deltavision format, a derivative of the MRC format)

- Sebastian

>
> Zach
>
>
>
> On Jun 15, 2008, at 3:00 AM, Sebastian Haase wrote:
>
>> Zach et al,
>>
>> hi, has there been a development on this ... ?
>> I.e. has anyone gotten in contact with Fredrik Lundh  ?
>>
>> I recently got more problems with reading a Zeiss Confocal Microscope
>> (LSM) file -- as far as I can tell, LSM files are TIFF files, and
>> ImageMagick recognizes the file in question as such ....
>> The point here is,
>> that I don't think that the Image-SIG mailing list is very
>> helpful ....
>> (it'is only a fraction as good as what we are used to over here at
>> SciPy ....  )
>>
>> So, (again,) it would be great if we could unite our image-io
>> interests over here at SciPy...
>>
>> Thanks,
>> Sebastian
>>
>>
>>
>> On Mon, Apr 21, 2008 at 10:49 PM, Robert Kern
>> <robert.kern at gmail.com> wrote:
>>> On Mon, Apr 21, 2008 at 3:37 PM, Zachary Pincus <zachary.pincus at yale.edu
>>> > wrote:
>>>> Let me look into whether this idea is at all feasible, and if it
>>>> is we
>>>> can revisit the issue of whether it belongs anywhere near scipy.
>>>> (Would getting Fredrik Lundh's OK to use various bits in this way
>>>> make
>>>> things easier? He does seem much more responsive to direct queries
>>>> than patch-submissions.)
>>>
>>> That would alleviate my concerns, yes.
>>>
>>> --
>>> Robert Kern


From zachary.pincus at yale.edu  Sun Jun 15 13:14:45 2008
From: zachary.pincus at yale.edu (Zachary Pincus)
Date: Sun, 15 Jun 2008 13:14:45 -0400
Subject: [SciPy-user] The IO library and image file formats -- compare
	with with PIL
In-Reply-To: <bc657ead0806150950q1dbecacfud799c96436792c8f@mail.gmail.com>
References: <bc657ead0804180136x5bf54ebax50cef6c0b5092a84@mail.gmail.com>
	<10B72505-ACEB-4EAE-A1B9-39A5038224CD@yale.edu>
	<9457e7c80804210704n348d2129l7228e5fce3ab164d@mail.gmail.com>
	<53E48B1E-91A0-422A-8309-CA58D4BCE486@yale.edu>
	<9457e7c80804211013i27a32020l4d704bf75d318c5f@mail.gmail.com>
	<3d375d730804211020n22ba0d72k8ae86cf8724a4c5b@mail.gmail.com>
	<0D36E713-05E6-43FA-9634-AD8BEA2B8743@yale.edu>
	<3d375d730804211349p4e75ab5cs6603fc13869f519@mail.gmail.com>
	<bc657ead0806150000t11818686xfb5ddb4fbc061cab@mail.gmail.com>
	<53B94D38-4F29-4FB6-9AED-A9647FE3A905@yale.edu>
	<bc657ead0806150950q1dbecacfud799c96436792c8f@mail.gmail.com>
Message-ID: <32CCE7B7-F876-41E8-AD5B-0347BFC756DF@yale.edu>

Hi all,

>> I will try to contact Fredrik today, though.
>>
> great - thanks.

I'll let you know what I learn.

Anyhow, I think that I can't really take the lead on any project right  
now because I have a ton of other stuff on my plate for the next month  
or two. If I do hear back from Fredrik in the positive, what I can do  
is send an interested party what I have so far, which is called "PIL- 
Lite" and is a private fork of PIL. From there, it should be possible  
to tear everything out except the image header IO (in the *ImagePlugin  
files) and graft that on to numpy/python decompression/pixel decoding.  
(At that point, what we'd have is less a fork of PIL, which rightly  
concerned Robert, and more a separate entity that happens to share a  
bit of code with PIL.)

But, it won't be super-easy -- image formats are a real bear, and  
there are things like palette-modes to consider (do we want to support  
them?) as well as the issue of JPEG decoding.

Zach


From cohen at slac.stanford.edu  Sun Jun 15 13:36:04 2008
From: cohen at slac.stanford.edu (Johann Cohen-Tanugi)
Date: Sun, 15 Jun 2008 19:36:04 +0200
Subject: [SciPy-user] How to get rid of nan and Inf
In-Reply-To: <8f5413a60806121102h4849828dief9fd5387f1b5fee@mail.gmail.com>
References: <8f5413a60806121102h4849828dief9fd5387f1b5fee@mail.gmail.com>
Message-ID: <48555304.4080301@slac.stanford.edu>

In [1]: import numpy

In [2]: numpy.nan_to_num?
Type:           function
Base Class:     <type 'function'>
String Form:    <function nan_to_num at 0xb79a1844>
Namespace:      Interactive
File:           /usr/lib/python2.5/site-packages/numpy/lib/type_check.py
Definition:     numpy.nan_to_num(x)
Docstring:
    Returns a copy of replacing NaN's with 0 and Infs with large numbers
   
    The following mappings are applied:
        NaN -> 0
        Inf -> limits.double_max
       -Inf -> limits.double_min

JCT

Tim Gray wrote:
> nan_to_num() does this. I think it's somewhere in numpy.
>
> Makes Nan's -> 0's and infs -> machine limits.
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>   


From fredmfp at gmail.com  Sun Jun 15 13:38:32 2008
From: fredmfp at gmail.com (fred)
Date: Sun, 15 Jun 2008 19:38:32 +0200
Subject: [SciPy-user] reading binary files written by a gfortran code...
In-Reply-To: <48553EF6.1060001@gmail.com>
References: <485502EE.6030708@gmail.com> <48553EF6.1060001@gmail.com>
Message-ID: <48555398.70100@gmail.com>

Phil Austin a ?crit :
>
> You want to write/read direct access files, which are just the bytes
> without the compiler-specific recordlength information.  Here is
> an example of a some tutorial code I wrote for myself exercising
> memmap and a fortran direct-access read -- just replace read with
> write in the fortran code to get output
Thanks Phil.

But I would rather prefer the "magic line" fortran code (if it exists) 
to write directly the binary file with the right format.

For instance, as seen on the web,

open(unit=20, file='foo.dat', form='unformatted', access='stream')
does not work:

  Fortran runtime error: Bad ACCESS parameter in OPEN statement

In fact, I would like to know all access parameters for the open 
statement. Where could I get them ?

I did not find relevant info, in gfortran manpage, gfortran info, etc.


Cheers,

-- 
Fred


From gnurser at googlemail.com  Sun Jun 15 14:31:14 2008
From: gnurser at googlemail.com (George Nurser)
Date: Sun, 15 Jun 2008 19:31:14 +0100
Subject: [SciPy-user] reading binary files written by a gfortran code...
In-Reply-To: <485502EE.6030708@gmail.com>
References: <485502EE.6030708@gmail.com>
Message-ID: <1d1e6ea70806151131n6ed98424ye47e0da40ab24a05@mail.gmail.com>

Hi Fred,

Fortran ( and this includes gfortran) seems to add stuff at the
beginning and end of a file.

If the unformatted fortran file just holds one array of real*8 data
the following works for me (and I think it works for real*4 as well).
It may need some modernizing...:

from numpy import *
def readbin(file_in,swap=False,f8=True):
    if f8:
        htot =fromfile(file = file_in ,dtype=float)
        c = htot.view(single)
        hc = c[1:-1].view(double)
    else:
        htot =fromfile(file = file_in ,dtype=float32)
        hc = c[1:-1]

    if swap:
        hc = hc.byteswap()
    return hc


import readbin
a1 = readbin.readbin('unf.dat')

2008/6/15 fred <fredmfp at gmail.com>:
> Hi,
>
> Ok, this is a well known issue that binary files written by a fortran
> program have a "special" format.
>
> So my question is : is it possible to write binary file _using gfortran_
> (no problem with intel fortran compiler) without this formatting ?
>
> Is there a peculiar syntax to the open() function ?
>
> I did not find any relevant information on the web.
>
> The obvious reason is that I have to read these files with
> scipy.io.numpyio fread method.
>
> TIA.
>
> Cheers,
>
> --
> Fred
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>


From meesters at uni-mainz.de  Sun Jun 15 15:07:23 2008
From: meesters at uni-mainz.de (Christian Meesters)
Date: Sun, 15 Jun 2008 21:07:23 +0200
Subject: [SciPy-user] NNLS in scipy
Message-ID: <1213556843.5734.10.camel@meesters.biologie.uni-mainz.de>

Hi,

I'd like to perform some "non-negatively constraint least squares"
algorithm to fit my data, like this:

Signal = SUM a_i C_i

where C_i is some simulated signal and a_i the amplitude contributed by
that simulated signal. Or in terms of arrays, I will have one reference
array and several arrays of simulated signals. How can find the
(non-negative) coefficients a_i for each simulated signal array? (All
negative contributions should be discarded.)

Is there anything like that in scipy (which I couldn't find)? Or any
other code doing that?

Else I could write it myself and contribute, but having some working
code would be nice, of course.

TIA
Christian



From mforbes at physics.ubc.ca  Sun Jun 15 15:28:19 2008
From: mforbes at physics.ubc.ca (Michael McNeil Forbes)
Date: Sun, 15 Jun 2008 12:28:19 -0700
Subject: [SciPy-user] Solver with n-dimentional steps
In-Reply-To: <4db580fd0806150655v4a13b4b5yd0c49b482d4a8ec1@mail.gmail.com>
References: <200806151118.59818.yosefmel@post.tau.ac.il>
	<1B9D8F05-27B4-4227-9119-B8EC66BD5C4B@cs.toronto.edu>
	<200806151341.16358.yosefmel@post.tau.ac.il>
	<4db580fd0806150655v4a13b4b5yd0c49b482d4a8ec1@mail.gmail.com>
Message-ID: <0177B04F-BD46-4DE0-9E85-FDF9530573B4@physics.ubc.ca>

Yosef,

You can also try using the Brodyen method (the b in bfgs) directly to  
solve your problem.  You could try some of the methods  
scipy.optimize.broyden*, or look at my attached code.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: broyden.py
Type: text/x-python-script
Size: 9221 bytes
Desc: not available
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080615/06b19b44/attachment.bin>
-------------- next part --------------


The Broyden method is a generalized secant method, so the Jacobian is  
slowly updated as the iteration proceeds.  It is useful if you cannot  
compute the Jacobian.  It starts as an iteration, so if you can  
arrange your equations so that they are expressed as an iteration x->F 
(x) that sort of converges, then this will work best, but it often  
works even if the iteration does not converge.

Michael.

On 15 Jun 2008, at 6:55 AM, Hoyt Koepke wrote:

> Yosef,
>
> I'm suggesting this partly to satisfy my own curiosity, so ignore this
> if it is difficult.   Since you are running with so many dimensions,
> you might want to try minimizing the square of your function using
> fmin_l_bfgs_b.  lbfgs is written for higher dimensions, and I've heard
> it can give incredible speed improvements over many other methods.
> I'm curious if it would help in your case.
>
> --Hoyt
>
> On Sun, Jun 15, 2008 at 1:41 PM, Yosef Meller  
> <yosefmel at post.tau.ac.il> wrote:
>> On Sunday 15 June 2008 13:19:54 David Warde-Farley wrote:
>>> It's doing finite differences to estimate the gradient. This is  
>>> pretty
>>> much unavoidable if you're only giving it those three arguments.  
>>> Have
>>> you thought about supplying fprime as a function to analytically
>>> compute the gradient?
>>

----------------------
Mailing address:

Michael McNeil Forbes
UW Dept. of Physics
Box 351560
Seattle, WA, 98195-1560

For couriers:
Physics/Astronomy Building, Room C121
3910 15th Ave NE
Seattle, WA, 98195-1560
Font Desk: (206)-543-2770

If you would like to visit me personally:
Room B482 (Fourth floor)
(206) 543-9754

From dmitrey.kroshko at scipy.org  Sun Jun 15 16:02:23 2008
From: dmitrey.kroshko at scipy.org (dmitrey)
Date: Sun, 15 Jun 2008 23:02:23 +0300
Subject: [SciPy-user] [optimization] OpenOpt release v 0.18
Message-ID: <4855754F.8090207@scipy.org>

Hi all,
I'm glad to inform you about new OpenOpt release: v 0.18.

OpenOpt is free (license: BSD) optimization framework (written in Python 
language) with connections to lots of solvers (some are C- or 
Fortran-written) and some native ones.

Changes since previous release 0.17 (March 15, 2008):

    * connection to glpk MILP solver (requires cvxopt v >= 1.0)

    * connection to NLP <http://scipy.org/scipy/scikits/wiki/NLP> solver
      IPOPT (requires python-ipopt wrapper (made by Eric Xu You)
      installation, that is currently available for Linux only, see
      openopt NLP webpage for more details)

    * major changes for NLP/NSP solver ralg

    * splitting non-linear constraints can benefit
      <http://openopt.blogspot.com/2008/05/major-changes-for-non-linear.html>
      for some solvers

    * unified text output for NLP solvers

    * handling of maximization problems
      <http://openopt.blogspot.com/2008/06/maximization-problems.html>
      (via p.goal = 'max' or 'maximum')

    * some bugfixes, lots of code cleanup

more details here:
http://openopt.blogspot.com/2008/06/openopt-018.html

Regards,
Dmitrey.


From peter.skomoroch at gmail.com  Sun Jun 15 17:45:08 2008
From: peter.skomoroch at gmail.com (Peter Skomoroch)
Date: Sun, 15 Jun 2008 14:45:08 -0700
Subject: [SciPy-user] NNLS in scipy
In-Reply-To: <1213556843.5734.10.camel@meesters.biologie.uni-mainz.de>
References: <1213556843.5734.10.camel@meesters.biologie.uni-mainz.de>
Message-ID: <e4fc0d2a0806151445m23c3b77apa1a54c3046fcdb72@mail.gmail.com>

Christian,

I was just going to do a blog post on this same topic next week.  I have a
few different versions of python NMF code I used a few years ago, including
sparse, parallel, and memmap versions.

Until I post my examples, you can find a basic implementation based on scipy
here (from Toby Segaran):

http://examples.oreilly.com/9780596529321/

These guys also did an implementation in python:

http://www.ma.utexas.edu/users/zmccoy/plsinmf.html
http://www.ma.utexas.edu/users/zmccoy/nmf.py

Some more NMF related links here:

http://del.icio.us/pskomoroch/nmf

-Pete


On Sun, Jun 15, 2008 at 12:07 PM, Christian Meesters <meesters at uni-mainz.de>
wrote:

> Hi,
>
> I'd like to perform some "non-negatively constraint least squares"
> algorithm to fit my data, like this:
>
> Signal = SUM a_i C_i
>
> where C_i is some simulated signal and a_i the amplitude contributed by
> that simulated signal. Or in terms of arrays, I will have one reference
> array and several arrays of simulated signals. How can find the
> (non-negative) coefficients a_i for each simulated signal array? (All
> negative contributions should be discarded.)
>
> Is there anything like that in scipy (which I couldn't find)? Or any
> other code doing that?
>
> Else I could write it myself and contribute, but having some working
> code would be nice, of course.
>
> TIA
> Christian
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>



-- 
Peter N. Skomoroch
peter.skomoroch at gmail.com
http://www.datawrangling.com
http://del.icio.us/pskomoroch
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080615/69911a62/attachment.html>

From graeme.okeefe at petnm.unimelb.edu.au  Sun Jun 15 18:56:37 2008
From: graeme.okeefe at petnm.unimelb.edu.au (Graeme O'Keefe)
Date: Mon, 16 Jun 2008 08:56:37 +1000
Subject: [SciPy-user] NNLS in scipy
In-Reply-To: <1213556843.5734.10.camel@meesters.biologie.uni-mainz.de>
References: <1213556843.5734.10.camel@meesters.biologie.uni-mainz.de>
Message-ID: <AB82F8D3-31A7-4B35-A50C-CE39FF7294D8@petnm.unimelb.edu.au>

Here is an implementation I put together to solve for x non-negatively  
the system: y = X.x

-------------- next part --------------
A non-text attachment was scrubbed...
Name: fnnls.py
Type: text/x-python-script
Size: 4667 bytes
Desc: not available
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080616/436558a4/attachment.bin>
-------------- next part --------------




regards,

Graeme


On 16/06/2008, at 5:07 AM, Christian Meesters wrote:

> Hi,
>
> I'd like to perform some "non-negatively constraint least squares"
> algorithm to fit my data, like this:
>
> Signal = SUM a_i C_i
>
> where C_i is some simulated signal and a_i the amplitude contributed  
> by
> that simulated signal. Or in terms of arrays, I will have one  
> reference
> array and several arrays of simulated signals. How can find the
> (non-negative) coefficients a_i for each simulated signal array? (All
> negative contributions should be discarded.)
>
> Is there anything like that in scipy (which I couldn't find)? Or any
> other code doing that?
>
> Else I could write it myself and contribute, but having some working
> code would be nice, of course.
>
> TIA
> Christian
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>



From meesters at uni-mainz.de  Mon Jun 16 02:45:54 2008
From: meesters at uni-mainz.de (Christian Meesters)
Date: Mon, 16 Jun 2008 08:45:54 +0200
Subject: [SciPy-user] NNLS in scipy
In-Reply-To: <1213556843.5734.10.camel@meesters.biologie.uni-mainz.de>
References: <1213556843.5734.10.camel@meesters.biologie.uni-mainz.de>
Message-ID: <1213598754.5734.13.camel@meesters.biologie.uni-mainz.de>

Hi,

Thanks Peter & Graeme!
I'm curious how the examples will look like, Peter. At least you'll have
a willing tester ;-).


Christian



From lists at benair.net  Mon Jun 16 02:57:30 2008
From: lists at benair.net (Benedikt Koenig)
Date: Mon, 16 Jun 2008 08:57:30 +0200
Subject: [SciPy-user] reading binary files written by a gfortran code...
In-Reply-To: <485502EE.6030708@gmail.com>
References: <485502EE.6030708@gmail.com>
Message-ID: <1213599451.20630.7.camel@iagpc71.iag.uni-stuttgart.de>

Hi Fred,

I am not sure whether I get you question right, but I once had the
problem, that a program complied with gfortran could not read binaries
written by the same version but compiled with ifort or g77.

This was caused by the marker length Fortran uses in binary files. Ifort
and g77 used a length of 4 bytes, whereas gfortran uses whatever is
stored in "off_t" on the parcticular system. So adding the following
compile option to gfortran did the job for me:
FFLOAT = -frecord-marker=4

HTH,
bene



Am Sonntag, den 15.06.2008, 13:54 +0200 schrieb fred:
> Hi,
> 
> Ok, this is a well known issue that binary files written by a fortran 
> program have a "special" format.
> 
> So my question is : is it possible to write binary file _using gfortran_
> (no problem with intel fortran compiler) without this formatting ?
> 
> Is there a peculiar syntax to the open() function ?
> 
> I did not find any relevant information on the web.
> 
> The obvious reason is that I have to read these files with 
> scipy.io.numpyio fread method.
> 
> TIA.
> 
> Cheers,
> 



From fredmfp at gmail.com  Mon Jun 16 04:11:54 2008
From: fredmfp at gmail.com (fred)
Date: Mon, 16 Jun 2008 10:11:54 +0200
Subject: [SciPy-user] reading binary files written by a gfortran code...
In-Reply-To: <1213599451.20630.7.camel@iagpc71.iag.uni-stuttgart.de>
References: <485502EE.6030708@gmail.com>
	<1213599451.20630.7.camel@iagpc71.iag.uni-stuttgart.de>
Message-ID: <4856204A.4040503@gmail.com>

Benedikt Koenig a ?crit :
> Hi Fred,
Hi Benedikt,

> I am not sure whether I get you question right, but I once had the
> problem, that a program complied with gfortran could not read binaries
> written by the same version but compiled with ifort or g77.
One can say it like this, yes ;-)
I want my program compiled with gfortran to write binary files with the 
_right_ size, say, 1 float -> 4 bytes.
No more, no less ;-)

> This was caused by the marker length Fortran uses in binary files. Ifort
> and g77 used a length of 4 bytes, whereas gfortran uses whatever is
> stored in "off_t" on the parcticular system. So adding the following
> compile option to gfortran did the job for me:
> FFLOAT = -frecord-marker=4
I guess you mean to put this flag in a Makefile, right?
However, how do you open and write your file in your fortran code?
I use this:

   real :: x

   x = 0

   open(unit=20, file='a.dat', form='unformatted')
   write(20) x
   close(20)

With FFLOAT set as above, I still have a 20 bytes file size for one float.
In fact, whatever I set in FFLOT (even negative values), I get the same 
result.

Cheers,

-- 
Fred


From lists at benair.net  Mon Jun 16 04:38:55 2008
From: lists at benair.net (Benedikt Koenig)
Date: Mon, 16 Jun 2008 10:38:55 +0200
Subject: [SciPy-user] reading binary files written by a gfortran code...
In-Reply-To: <4856204A.4040503@gmail.com>
References: <485502EE.6030708@gmail.com>
	<1213599451.20630.7.camel@iagpc71.iag.uni-stuttgart.de>
	<4856204A.4040503@gmail.com>
Message-ID: <1213605538.21575.11.camel@iagpc71.iag.uni-stuttgart.de>

Hi Fred,



> > This was caused by the marker length Fortran uses in binary files. Ifort
> > and g77 used a length of 4 bytes, whereas gfortran uses whatever is
> > stored in "off_t" on the parcticular system. So adding the following
> > compile option to gfortran did the job for me:
> > FFLOAT = -frecord-marker=4
> I guess you mean to put this flag in a Makefile, right?
yeah, the -frecord-marker option is in my case used in the Makefile.
Actually I am using '-fdefault-real-8 -frecord-marker=4' together with
the usual OPT settings and library stuff as compile options to gfortran.

> However, how do you open and write your file in your fortran code?
> I use this:
> 
>    real :: x
> 
>    x = 0
> 
>    open(unit=20, file='a.dat', form='unformatted')
>    write(20) x
>    close(20)
> 
> With FFLOAT set as above, I still have a 20 bytes file size for one float.
> In fact, whatever I set in FFLOT (even negative values), I get the same 
> result.
Unfortunately, I am not really familiar with fortran. The code I am
using was not written by myself, I am just using the source code to
compile on different platforms I am working on. So I am not the right
one to ask about ways of Fortran.

However, IIRC Fortran writes a marker before and after every record. If
you write one real (4 byte) and you end up with 20 byte output this
seems to me like you still have the markers of 8 byte length rather than
the 4 bytes as should be the case using  -frecord-marker=4. Maybe you
want to check the hex code, that is acutally written? The record markers
should be 4 bytes length and contain the length of the record (in your
case 1 real)

BTW, which version of gfortran are you using? 

cheers,
bene


> 
> Cheers,
> 



From yosefmel at post.tau.ac.il  Mon Jun 16 06:53:15 2008
From: yosefmel at post.tau.ac.il (Yosef Meller)
Date: Mon, 16 Jun 2008 13:53:15 +0300
Subject: [SciPy-user] Solver with n-dimentional steps
In-Reply-To: <4db580fd0806150655v4a13b4b5yd0c49b482d4a8ec1@mail.gmail.com>
References: <200806151118.59818.yosefmel@post.tau.ac.il>
	<200806151341.16358.yosefmel@post.tau.ac.il>
	<4db580fd0806150655v4a13b4b5yd0c49b482d4a8ec1@mail.gmail.com>
Message-ID: <200806161353.15378.yosefmel@post.tau.ac.il>

On Sunday 15 June 2008 16:55:47 Hoyt Koepke wrote:
> I'm suggesting this partly to satisfy my own curiosity, so ignore this
> if it is difficult.   Since you are running with so many dimensions,
> you might want to try minimizing the square of your function using
> fmin_l_bfgs_b.  lbfgs is written for higher dimensions, and I've heard
> it can give incredible speed improvements over many other methods.
> I'm curious if it would help in your case.

Thanks for the tip. After trying half a day to figure out why a jacobian that 
should take 1GiB (8 byte floats) clogs the memory in a 2GiB machine, I 
decided to go for  fmin_l_bfgs_b(). I can't tell you if it improves the 
speed, but the low memory requirements make it work for me, as opposed to 
fsolve().


From fredmfp at gmail.com  Mon Jun 16 07:24:09 2008
From: fredmfp at gmail.com (fred)
Date: Mon, 16 Jun 2008 13:24:09 +0200
Subject: [SciPy-user] reading binary files written by a gfortran code...
In-Reply-To: <1213605538.21575.11.camel@iagpc71.iag.uni-stuttgart.de>
References: <485502EE.6030708@gmail.com>	<1213599451.20630.7.camel@iagpc71.iag.uni-stuttgart.de>	<4856204A.4040503@gmail.com>
	<1213605538.21575.11.camel@iagpc71.iag.uni-stuttgart.de>
Message-ID: <48564D59.70701@gmail.com>

Benedikt Koenig a ?crit :

> Unfortunately, I am not really familiar with fortran. The code I am
> using was not written by myself, I am just using the source code to
> compile on different platforms I am working on. So I am not the right
> one to ask about ways of Fortran.
Ok.

Maybe a fortran guru ?
I can't believe there is no fortran guru here ;-)

> However, IIRC Fortran writes a marker before and after every record. If
> you write one real (4 byte) and you end up with 20 byte output this
> seems to me like you still have the markers of 8 byte length rather than
> the 4 bytes as should be the case using  -frecord-marker=4. Maybe you
> want to check the hex code, that is acutally written? The record markers
> should be 4 bytes length and contain the length of the record (in your
> case 1 real)
Well, hexdump gives me
0000000 0004 0000 0000 0000 0000 0000 0004 0000
0000010 0000 0000
0000014

On a binary file created with the intel version (so a 4 bytes file),
I get
0000000 0000 0000
0000004

The trick I don't understand is that if my file is 4 bytes length, there 
is no marker, no, as in the case of a binary file written by say a C code ?
Or I'm completely stupid ?

> BTW, which version of gfortran are you using? 
4.1.1 on debian etch box.


Cheers,

-- 
Fred


From hasslerjc at comcast.net  Mon Jun 16 08:13:38 2008
From: hasslerjc at comcast.net (John Hassler)
Date: Mon, 16 Jun 2008 08:13:38 -0400
Subject: [SciPy-user] reading binary files written by a gfortran code...
In-Reply-To: <48564D59.70701@gmail.com>
References: <485502EE.6030708@gmail.com>	<1213599451.20630.7.camel@iagpc71.iag.uni-stuttgart.de>	<4856204A.4040503@gmail.com>	<1213605538.21575.11.camel@iagpc71.iag.uni-stuttgart.de>
	<48564D59.70701@gmail.com>
Message-ID: <485658F2.1060808@comcast.net>

An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080616/d0087022/attachment.html>

From fredmfp at gmail.com  Mon Jun 16 08:55:38 2008
From: fredmfp at gmail.com (fred)
Date: Mon, 16 Jun 2008 14:55:38 +0200
Subject: [SciPy-user] reading binary files written by a gfortran code...
In-Reply-To: <485658F2.1060808@comcast.net>
References: <485502EE.6030708@gmail.com>	<1213599451.20630.7.camel@iagpc71.iag.uni-stuttgart.de>	<4856204A.4040503@gmail.com>	<1213605538.21575.11.camel@iagpc71.iag.uni-stuttgart.de>	<48564D59.70701@gmail.com>
	<485658F2.1060808@comcast.net>
Message-ID: <485662CA.20805@gmail.com>

John Hassler a ?crit :

> Reading FORTRAN unformatted binary files in C/C++
> Or....FORTRAN Weirdness, what were they thinking?
> Written by Paul Bourke
> April 2003
> http://local.wasp.uwa.edu.au/~pbourke/dataformats/fortran/
> 
> ... which is basically what you've already figured out.
Yes, I have already read this.
But this does not give me a solution to my issue.

Cheers,

-- 
Fred


From w.henney at astrosmo.unam.mx  Mon Jun 16 09:44:43 2008
From: w.henney at astrosmo.unam.mx (William Henney)
Date: Mon, 16 Jun 2008 13:44:43 +0000 (UTC)
Subject: [SciPy-user] reading binary files written by a gfortran code...
References: <485502EE.6030708@gmail.com>	<1213599451.20630.7.camel@iagpc71.iag.uni-stuttgart.de>	<4856204A.4040503@gmail.com>
	<1213605538.21575.11.camel@iagpc71.iag.uni-stuttgart.de>
	<48564D59.70701@gmail.com>
Message-ID: <loom.20080616T133621-538@post.gmane.org>

fred <fredmfp <at> gmail.com> writes:

> The trick I don't understand is that if my file is 4 bytes length,
> there is no marker, no, as in the case of a binary file written by
> say a C code ? Or I'm completely stupid ?

I think you had the right answer to start with: you must use
access='stream'

This is a new feature of Fortran 2003, but it is already supported by
most compilers.

> > BTW, which version of gfortran are you using? 
> 4.1.1 on debian etch box.

And here is your trouble - this is a very old version of gfortran. I
have 4.3.0 and the following code works fine:

----------------- streamio.f90 --------------------------
program streamio
  implicit none
  real :: a = 0.0
  open(unit=20, file='foo.dat', form='unformatted', access='stream')
  write(20) a
  close(20)
end program streamio
------------------------------------------------------

$ gfortran -o streamio streamio.f90 && ./streamio && hexdump foo.dat
0000000 00 00 00 00                                    
0000004

You might also want to read this page:

http://www.star.le.ac.uk/~cgp/streamIO.html

Cheers

Will









From fredmfp at gmail.com  Mon Jun 16 10:07:10 2008
From: fredmfp at gmail.com (fred)
Date: Mon, 16 Jun 2008 16:07:10 +0200
Subject: [SciPy-user] extracting z, v from a 3D array
Message-ID: <4856738E.8050408@gmail.com>

Hi,

In a 3D array, each cell [i, j, k] has a value v which is the value at 
the point x=i*dx, y=j*dy, z=k*dz.

How can I extract, "efficiently" (fast and memory resource) the 2D array 
(v, z), ie z vs. v ?

"fast and memory resource" means "in acceptable time": for instance, I 
compute an histogram using scipy.histogram for a 800x850x900 cells float 
array in 3 mn.


TIA.

Cheers,

-- 
Fred


From david.huard at gmail.com  Mon Jun 16 10:17:32 2008
From: david.huard at gmail.com (David Huard)
Date: Mon, 16 Jun 2008 10:17:32 -0400
Subject: [SciPy-user] reading binary files written by a gfortran code...
In-Reply-To: <48555398.70100@gmail.com>
References: <485502EE.6030708@gmail.com> <48553EF6.1060001@gmail.com>
	<48555398.70100@gmail.com>
Message-ID: <91cf711d0806160717gd7cb9c8ga460f23621bfcefd@mail.gmail.com>

According to the Fortran95 handbook (1997) ACCESS arguments takes either
DIRECT or SEQUENTIAL as values.

Also, if the ACCESS specifier is DIRECT, a RECL specifier must be present.

RECL is an integer specifying the length of each record in default
characters if access is direct, or the maximum length of a record if the
access method is sequential.

HTH,

David

2008/6/15 fred <fredmfp at gmail.com>:

> Phil Austin a ?crit :
> >
> > You want to write/read direct access files, which are just the bytes
> > without the compiler-specific recordlength information.  Here is
> > an example of a some tutorial code I wrote for myself exercising
> > memmap and a fortran direct-access read -- just replace read with
> > write in the fortran code to get output
> Thanks Phil.
>
> But I would rather prefer the "magic line" fortran code (if it exists)
> to write directly the binary file with the right format.
>
> For instance, as seen on the web,
>
> open(unit=20, file='foo.dat', form='unformatted', access='stream')
> does not work:
>
>  Fortran runtime error: Bad ACCESS parameter in OPEN statement
>
> In fact, I would like to know all access parameters for the open
> statement. Where could I get them ?
>
> I did not find relevant info, in gfortran manpage, gfortran info, etc.
>
>
> Cheers,
>
> --
> Fred
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080616/2b99577b/attachment.html>

From fredmfp at gmail.com  Mon Jun 16 10:17:29 2008
From: fredmfp at gmail.com (fred)
Date: Mon, 16 Jun 2008 16:17:29 +0200
Subject: [SciPy-user] reading binary files written by a gfortran code...
In-Reply-To: <loom.20080616T133621-538@post.gmane.org>
References: <485502EE.6030708@gmail.com>	<1213599451.20630.7.camel@iagpc71.iag.uni-stuttgart.de>	<4856204A.4040503@gmail.com>	<1213605538.21575.11.camel@iagpc71.iag.uni-stuttgart.de>	<48564D59.70701@gmail.com>
	<loom.20080616T133621-538@post.gmane.org>
Message-ID: <485675F9.70303@gmail.com>

William Henney a ?crit :


> And here is your trouble - this is a very old version of gfortran. I
> have 4.3.0 and the following code works fine:
Hurray !! :-)))

I just did not think my gfortran version was too old...

Tons of thanks, William !

> You might also want to read this page:
> 
> http://www.star.le.ac.uk/~cgp/streamIO.html
Ok.

Cheers,

-- 
Fred


From ivo.maljevic at gmail.com  Mon Jun 16 10:32:00 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Mon, 16 Jun 2008 10:32:00 -0400
Subject: [SciPy-user] More on speed comparisons
Message-ID: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>

I was planing to refrain from this, but I believe this might be an
interesting comparison for people who seriously think about switching to
SciPy/NumPy.

Few days ago I wrote about speed comparison between the scalar and
vectorized versions of function calls. Based on several comments, I
concluded that the same story that applies to Matlab and Octave applies
here: vectorize thy code, and speed gain will come.

Before I show the results of vectorized vs. non-vectorized results, just
want to go on the record and say that I am by no means sayhing that
SciPy/NumPy is not good. I still like what has been done here. There is a
particular scenario that I use at my work where SciPy, combined with
matplotlib, is extremely useful. That scenario is the following.

In my wireless lab, I have basestations, mobile stations, whole bunch of
instruments and PCs either connected via network or GPIB cables (for
instruments).I use Python here to automate test cases and data collection
and the ability to do SSH and GPIB communication is very useful. Once I
collect data, I use SciPy for some simple postprocessing and I generate PNG
plots, and finally, I generate HTML pages with results shown as tables and
plots. So, it is all done in a single language/script instead of having to
break the processing into several languages/scripts.

However, I wanted to see if SciPy would be good enough speedwise to
completely replace Matlab. An, at least for the type of processing I do, it
comes nowhere near it. I wrote a small toy program that does some simple
random variable manipulation in several languages. The python code consists
of two versions, one uses for loop and basic pathon libraries and the other
uses nympy's vectorized form (there was no difference between numpy and
scipy).

Here are the relative results after running the code on two machines:

64-bit Ubunty 8.04:
=================

Fortran    C       Octave    SciPy   Pure Python
=================================================
1          1.2      2.2      16      20


32-Bit openSUSE 10.3:
==================

Fortran    C       Octave    SciPy   Pure Python
=================================================
1          1.2      2.4      15      19.4


The numbers are rounded a little bit, but they are in that range. I see two
problems here:

1. SciPy is very slow, even when compared to Octave 3.0
2. It is only sligtly faster than Python with a for loop.

Below is the source code for the two python versions. While this processing
is not from any real application,
it is not very different from the processing I normally do.

Now, it is very likely that for different type of processing people will
find SciPy fast enough (matrix inversions, eigenvalues, etc), but for the
type of
processing I need it is not fast enough.

Ivo

##########################################################################
# rand_test_1.py
from random import random
from math import sqrt, sin

N = 1000000

mean = 0
var  = 0

for i in range(N):
  x = random()
  x = 3.14*sqrt(x)
  x = sin(x)
  mean += x
  var += x**2

mean = mean/N
var  = var/N - mean**2
print 'Mean=%g, var=%g' % (mean, var)

# rand_test_2.py
from numpy import random, sin, sqrt

N = 1000000

x = random.rand(N)
x = 3.14*sqrt(x)
x = sin(x)

mean = sum(x)/N
var  = sum(x**2)/N - mean**2

print 'Mean=%g, var=%g' % (mean, var)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080616/f2273631/attachment.html>

From matthieu.brucher at gmail.com  Mon Jun 16 10:43:33 2008
From: matthieu.brucher at gmail.com (Matthieu Brucher)
Date: Mon, 16 Jun 2008 16:43:33 +0200
Subject: [SciPy-user] More on speed comparisons
In-Reply-To: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>
References: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>
Message-ID: <e76aa17f0806160743s477b7c79v33b633ceaf8c2141@mail.gmail.com>

Hi,

Try with complete Numpy support :

# rand_test_2.py
from numpy import random, sin, sqrt, mean, var

N = 1000000
x = random.rand(N)
x = 3.14*sqrt(x)
x = sin(x)

mean_ = mean(x)
var_  = var(x)

print 'Mean=%g, var=%g' % (mean_, var_)

You can use numpy.sum as well.

Matthieu

2008/6/16 Ivo Maljevic <ivo.maljevic at gmail.com>:
> I was planing to refrain from this, but I believe this might be an
> interesting comparison for people who seriously think about switching to
> SciPy/NumPy.
>
> Few days ago I wrote about speed comparison between the scalar and
> vectorized versions of function calls. Based on several comments, I
> concluded that the same story that applies to Matlab and Octave applies
> here: vectorize thy code, and speed gain will come.
>
> Before I show the results of vectorized vs. non-vectorized results, just
> want to go on the record and say that I am by no means sayhing that
> SciPy/NumPy is not good. I still like what has been done here. There is a
> particular scenario that I use at my work where SciPy, combined with
> matplotlib, is extremely useful. That scenario is the following.
>
> In my wireless lab, I have basestations, mobile stations, whole bunch of
> instruments and PCs either connected via network or GPIB cables (for
> instruments).I use Python here to automate test cases and data collection
> and the ability to do SSH and GPIB communication is very useful. Once I
> collect data, I use SciPy for some simple postprocessing and I generate PNG
> plots, and finally, I generate HTML pages with results shown as tables and
> plots. So, it is all done in a single language/script instead of having to
> break the processing into several languages/scripts.
>
> However, I wanted to see if SciPy would be good enough speedwise to
> completely replace Matlab. An, at least for the type of processing I do, it
> comes nowhere near it. I wrote a small toy program that does some simple
> random variable manipulation in several languages. The python code consists
> of two versions, one uses for loop and basic pathon libraries and the other
> uses nympy's vectorized form (there was no difference between numpy and
> scipy).
>
> Here are the relative results after running the code on two machines:
>
> 64-bit Ubunty 8.04:
> =================
>
> Fortran    C       Octave    SciPy   Pure Python
> =================================================
> 1          1.2      2.2      16      20
>
>
> 32-Bit openSUSE 10.3:
> ==================
>
> Fortran    C       Octave    SciPy   Pure Python
> =================================================
> 1          1.2      2.4      15      19.4
>
>
> The numbers are rounded a little bit, but they are in that range. I see two
> problems here:
>
> 1. SciPy is very slow, even when compared to Octave 3.0
> 2. It is only sligtly faster than Python with a for loop.
>
> Below is the source code for the two python versions. While this processing
> is not from any real application,
> it is not very different from the processing I normally do.
>
> Now, it is very likely that for different type of processing people will
> find SciPy fast enough (matrix inversions, eigenvalues, etc), but for the
> type of
> processing I need it is not fast enough.
>
> Ivo
>
> ##########################################################################
> # rand_test_1.py
> from random import random
> from math import sqrt, sin
>
> N = 1000000
>
> mean = 0
> var  = 0
>
> for i in range(N):
>   x = random()
>   x = 3.14*sqrt(x)
>   x = sin(x)
>   mean += x
>   var += x**2
>
> mean = mean/N
> var  = var/N - mean**2
> print 'Mean=%g, var=%g' % (mean, var)
>
> # rand_test_2.py
> from numpy import random, sin, sqrt
>
> N = 1000000
>
> x = random.rand(N)
> x = 3.14*sqrt(x)
> x = sin(x)
>
> mean = sum(x)/N
> var  = sum(x**2)/N - mean**2
>
> print 'Mean=%g, var=%g' % (mean, var)
>
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
>



-- 
French PhD student
Website : http://matthieu-brucher.developpez.com/
Blogs : http://matt.eifelle.com and http://blog.developpez.com/?blog=92
LinkedIn : http://www.linkedin.com/in/matthieubrucher


From ivo.maljevic at gmail.com  Mon Jun 16 10:45:10 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Mon, 16 Jun 2008 10:45:10 -0400
Subject: [SciPy-user] More on speed comparisons
In-Reply-To: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>
References: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>
Message-ID: <826c64da0806160745mf57f37bqc5f1e87adddb590a@mail.gmail.com>

I knew I should have refrained from sending my "findings". Especially
because I was quick to
jump to conclusions :(

It turns out SciPy is as fast as octave when written in a vector form. The
correct version of
the array form of the script is (it was not using proper version of the sum
function):

from numpy import random, sin, sqrt, sum

N = 1000000

x = random.rand(N)
x = 3.14*sqrt(x)
x = sin(x)

mean = sum(x)/N
var  = sum(x**2)/N - mean**2

print 'Mean=%g, var=%g' % (mean, var)


2008/6/16 Ivo Maljevic <ivo.maljevic at gmail.com>:

> I was planing to refrain from this, but I believe this might be an
> interesting comparison for people who seriously think about switching to
> SciPy/NumPy.
>
> Few days ago I wrote about speed comparison between the scalar and
> vectorized versions of function calls. Based on several comments, I
> concluded that the same story that applies to Matlab and Octave applies
> here: vectorize thy code, and speed gain will come.
>
> Before I show the results of vectorized vs. non-vectorized results, just
> want to go on the record and say that I am by no means sayhing that
> SciPy/NumPy is not good. I still like what has been done here. There is a
> particular scenario that I use at my work where SciPy, combined with
> matplotlib, is extremely useful. That scenario is the following.
>
> In my wireless lab, I have basestations, mobile stations, whole bunch of
> instruments and PCs either connected via network or GPIB cables (for
> instruments).I use Python here to automate test cases and data collection
> and the ability to do SSH and GPIB communication is very useful. Once I
> collect data, I use SciPy for some simple postprocessing and I generate PNG
> plots, and finally, I generate HTML pages with results shown as tables and
> plots. So, it is all done in a single language/script instead of having to
> break the processing into several languages/scripts.
>
> However, I wanted to see if SciPy would be good enough speedwise to
> completely replace Matlab. An, at least for the type of processing I do, it
> comes nowhere near it. I wrote a small toy program that does some simple
> random variable manipulation in several languages. The python code consists
> of two versions, one uses for loop and basic pathon libraries and the other
> uses nympy's vectorized form (there was no difference between numpy and
> scipy).
>
> Here are the relative results after running the code on two machines:
>
> 64-bit Ubunty 8.04:
> =================
>
> Fortran    C       Octave    SciPy   Pure Python
> =================================================
> 1          1.2      2.2      16      20
>
>
> 32-Bit openSUSE 10.3:
> ==================
>
> Fortran    C       Octave    SciPy   Pure Python
> =================================================
> 1          1.2      2.4      15      19.4
>
>
> The numbers are rounded a little bit, but they are in that range. I see two
> problems here:
>
> 1. SciPy is very slow, even when compared to Octave 3.0
> 2. It is only sligtly faster than Python with a for loop.
>
> Below is the source code for the two python versions. While this processing
> is not from any real application,
> it is not very different from the processing I normally do.
>
> Now, it is very likely that for different type of processing people will
> find SciPy fast enough (matrix inversions, eigenvalues, etc), but for the
> type of
> processing I need it is not fast enough.
>
> Ivo
>
> ##########################################################################
> # rand_test_1.py
> from random import random
> from math import sqrt, sin
>
> N = 1000000
>
> mean = 0
> var  = 0
>
> for i in range(N):
>   x = random()
>   x = 3.14*sqrt(x)
>   x = sin(x)
>   mean += x
>   var += x**2
>
> mean = mean/N
> var  = var/N - mean**2
> print 'Mean=%g, var=%g' % (mean, var)
>
> # rand_test_2.py
> from numpy import random, sin, sqrt
>
> N = 1000000
>
> x = random.rand(N)
> x = 3.14*sqrt(x)
> x = sin(x)
>
> mean = sum(x)/N
> var  = sum(x**2)/N - mean**2
>
> print 'Mean=%g, var=%g' % (mean, var)
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080616/f3f38430/attachment.html>

From ivo.maljevic at gmail.com  Mon Jun 16 10:55:06 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Mon, 16 Jun 2008 10:55:06 -0400
Subject: [SciPy-user] More on speed comparisons
In-Reply-To: <e76aa17f0806160743s477b7c79v33b633ceaf8c2141@mail.gmail.com>
References: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>
	<e76aa17f0806160743s477b7c79v33b633ceaf8c2141@mail.gmail.com>
Message-ID: <826c64da0806160755u314913adg5ec7a2dcd801d037@mail.gmail.com>

I know, the wrong sum call was the problem. The built in mean and var don't
make any difference, as long as I include sum (which I didn't the first
time).

Thanks,
Ivo

2008/6/16 Matthieu Brucher <matthieu.brucher at gmail.com>:

> Hi,
>
> Try with complete Numpy support :
>
> # rand_test_2.py
> from numpy import random, sin, sqrt, mean, var
>
> N = 1000000
> x = random.rand(N)
> x = 3.14*sqrt(x)
> x = sin(x)
>
> mean_ = mean(x)
> var_  = var(x)
>
> print 'Mean=%g, var=%g' % (mean_, var_)
>
> You can use numpy.sum as well.
>
> Matthieu
>
> 2008/6/16 Ivo Maljevic <ivo.maljevic at gmail.com>:
> > I was planing to refrain from this, but I believe this might be an
> > interesting comparison for people who seriously think about switching to
> > SciPy/NumPy.
> >
> > Few days ago I wrote about speed comparison between the scalar and
> > vectorized versions of function calls. Based on several comments, I
> > concluded that the same story that applies to Matlab and Octave applies
> > here: vectorize thy code, and speed gain will come.
> >
> > Before I show the results of vectorized vs. non-vectorized results, just
> > want to go on the record and say that I am by no means sayhing that
> > SciPy/NumPy is not good. I still like what has been done here. There is a
> > particular scenario that I use at my work where SciPy, combined with
> > matplotlib, is extremely useful. That scenario is the following.
> >
> > In my wireless lab, I have basestations, mobile stations, whole bunch of
> > instruments and PCs either connected via network or GPIB cables (for
> > instruments).I use Python here to automate test cases and data collection
> > and the ability to do SSH and GPIB communication is very useful. Once I
> > collect data, I use SciPy for some simple postprocessing and I generate
> PNG
> > plots, and finally, I generate HTML pages with results shown as tables
> and
> > plots. So, it is all done in a single language/script instead of having
> to
> > break the processing into several languages/scripts.
> >
> > However, I wanted to see if SciPy would be good enough speedwise to
> > completely replace Matlab. An, at least for the type of processing I do,
> it
> > comes nowhere near it. I wrote a small toy program that does some simple
> > random variable manipulation in several languages. The python code
> consists
> > of two versions, one uses for loop and basic pathon libraries and the
> other
> > uses nympy's vectorized form (there was no difference between numpy and
> > scipy).
> >
> > Here are the relative results after running the code on two machines:
> >
> > 64-bit Ubunty 8.04:
> > =================
> >
> > Fortran    C       Octave    SciPy   Pure Python
> > =================================================
> > 1          1.2      2.2      16      20
> >
> >
> > 32-Bit openSUSE 10.3:
> > ==================
> >
> > Fortran    C       Octave    SciPy   Pure Python
> > =================================================
> > 1          1.2      2.4      15      19.4
> >
> >
> > The numbers are rounded a little bit, but they are in that range. I see
> two
> > problems here:
> >
> > 1. SciPy is very slow, even when compared to Octave 3.0
> > 2. It is only sligtly faster than Python with a for loop.
> >
> > Below is the source code for the two python versions. While this
> processing
> > is not from any real application,
> > it is not very different from the processing I normally do.
> >
> > Now, it is very likely that for different type of processing people will
> > find SciPy fast enough (matrix inversions, eigenvalues, etc), but for the
> > type of
> > processing I need it is not fast enough.
> >
> > Ivo
> >
> >
> ##########################################################################
> > # rand_test_1.py
> > from random import random
> > from math import sqrt, sin
> >
> > N = 1000000
> >
> > mean = 0
> > var  = 0
> >
> > for i in range(N):
> >   x = random()
> >   x = 3.14*sqrt(x)
> >   x = sin(x)
> >   mean += x
> >   var += x**2
> >
> > mean = mean/N
> > var  = var/N - mean**2
> > print 'Mean=%g, var=%g' % (mean, var)
> >
> > # rand_test_2.py
> > from numpy import random, sin, sqrt
> >
> > N = 1000000
> >
> > x = random.rand(N)
> > x = 3.14*sqrt(x)
> > x = sin(x)
> >
> > mean = sum(x)/N
> > var  = sum(x**2)/N - mean**2
> >
> > print 'Mean=%g, var=%g' % (mean, var)
> >
> >
> > _______________________________________________
> > SciPy-user mailing list
> > SciPy-user at scipy.org
> > http://projects.scipy.org/mailman/listinfo/scipy-user
> >
> >
>
>
>
> --
> French PhD student
> Website : http://matthieu-brucher.developpez.com/
> Blogs : http://matt.eifelle.com and http://blog.developpez.com/?blog=92
> LinkedIn : http://www.linkedin.com/in/matthieubrucher
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080616/0bb8ee57/attachment.html>

From gael.varoquaux at normalesup.org  Mon Jun 16 10:59:16 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Mon, 16 Jun 2008 16:59:16 +0200
Subject: [SciPy-user] More on speed comparisons
In-Reply-To: <826c64da0806160745mf57f37bqc5f1e87adddb590a@mail.gmail.com>
References: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>
	<826c64da0806160745mf57f37bqc5f1e87adddb590a@mail.gmail.com>
Message-ID: <20080616145916.GC3938@phare.normalesup.org>

On Mon, Jun 16, 2008 at 10:45:10AM -0400, Ivo Maljevic wrote:
>    I knew I should have refrained from sending my "findings". Especially
>    because I was quick to
>    jump to conclusions :(

>    It turns out SciPy is as fast as octave when written in a vector form. The
>    correct version of
>    the array form of the script is (it was not using proper version of the
>    sum function):

Indeed. I am also curious to know how you measure timings. The proper way
of mesuring timings (ie measuring CPU time, and not wall time) is using
the timeit module. You can either use the timeit shell command, or the
timeit magic, in ipython. Here are the results I get:

##########################################################################
# rand_test_1.py
from random import random
from math import sqrt, sin

def do_python():
    N = 1000000

    mean = 0
    var  = 0

    for i in range(N):
        x = random()
        x = 3.14*sqrt(x)
        x = sin(x)
        mean += x
        var += x**2

    mean = mean/N
    var  = var/N - mean**2
    print 'Mean=%g, var=%g' % (mean, var)

# rand_test_2.py

def do_numpy():
    import numpy as np
    
    N = 1000000
    
    x = np.random.rand(N)
    x = 3.14*np.sqrt(x)
    x = np.sin(x)
    
    mean = x.mean()
    var  = np.sum(x**2)/N - mean**2
    
    print 'Mean=%g, var=%g' % (mean, var)
##########################################################################

And in ipython:

    In [1]: %run test.py

    In [2]: %timeit do_python()
    [... snip ]
    10 loops, best of 3: 1.18 s per loop

    In [3]: %timeit do_numpy()
    10 loops, best of 3: 147 ms per loop

This is significantly different from your timings. The numbers do not
mean the same thing, but I trust these one more, and would try to use CPU
time only for comparisons between different approachs.

My 2 cents.

Ga?l


From david at ar.media.kyoto-u.ac.jp  Mon Jun 16 10:46:48 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Mon, 16 Jun 2008 23:46:48 +0900
Subject: [SciPy-user] More on speed comparisons
In-Reply-To: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>
References: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>
Message-ID: <48567CD8.5070207@ar.media.kyoto-u.ac.jp>

Ivo Maljevic wrote:
> However, I wanted to see if SciPy would be good enough speedwise to 
> completely replace Matlab. An, at least for the type of processing I 
> do, it comes nowhere near it.

Note that random number generator greatly vary across languages and 
implementations.

> The numbers are rounded a little bit, but they are in that range. I 
> see two problems here:
>
> 1. SciPy is very slow, even when compared to Octave 3.0
> 2. It is only sligtly faster than Python with a for loop.

That's really surprising. Also, I quickly checked: random is as fast 
under scipy as under matlab. The problem is not random (which takes less 
than 10 % of the running time; as always, use profiling :) ).

And I found your problem: sum. You are not using numpy sum, but python 
sum, which is extremely slow (it has to convert the array to a sequence 
first I think, which means it may well be slower than looping :) ).

Here is my version:

from numpy import random, sin, sqrt, mean, var

def compute():
    N = 1000000
  
    x = random.rand(N)
    x = 3.14*sqrt(x)
    x = sin(x)
                                                                                         

    m = mean(x)
    v = var(x)
       
    print 'Mean=', m, ', var=', 
v                                                       
       
if __name__ == 
'__main__':                                                              
    compute()

This is roughly ten times faster than your scipy version on my computer. 
Which means we are pretty close to C performances :)

cheers,

David


From ivo.maljevic at gmail.com  Mon Jun 16 11:13:48 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Mon, 16 Jun 2008 11:13:48 -0400
Subject: [SciPy-user] More on speed comparisons
In-Reply-To: <20080616145916.GC3938@phare.normalesup.org>
References: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>
	<826c64da0806160745mf57f37bqc5f1e87adddb590a@mail.gmail.com>
	<20080616145916.GC3938@phare.normalesup.org>
Message-ID: <826c64da0806160813q6d2377d3pdad6811f10a7d8f7@mail.gmail.com>

Aside from good speed with SciPy (I feel embarrassed that I made that 'sum'
mistake), I am impressed with the speed of reaction of you guys (David,
Gael, Anne, Matthieu, Robert). These are the names I have picked up so far
to be the most frequent.

Geal, you are working the internals of the SciPy and therefore you might be
interested in timing for individual loop execution time, but for me the most
relevant time is how long does it take to execute simulation from the moment
I press return to the moment it is all done. Typically, I use:

time command (e.g., time python ./rand_test_2.py). While you may find this
methodology incorrect,
I do the same with fortran and C code (time ./rand_test_c or time
./rand_test_f).

Some of my matlab simulation run for more than one hour and I use tic/toc
there.

Ivo

2008/6/16 Gael Varoquaux <gael.varoquaux at normalesup.org>:

> On Mon, Jun 16, 2008 at 10:45:10AM -0400, Ivo Maljevic wrote:
> >    I knew I should have refrained from sending my "findings". Especially
> >    because I was quick to
> >    jump to conclusions :(
>
> >    It turns out SciPy is as fast as octave when written in a vector form.
> The
> >    correct version of
> >    the array form of the script is (it was not using proper version of
> the
> >    sum function):
>
> Indeed. I am also curious to know how you measure timings. The proper way
> of mesuring timings (ie measuring CPU time, and not wall time) is using
> the timeit module. You can either use the timeit shell command, or the
> timeit magic, in ipython. Here are the results I get:
>
> ##########################################################################
> # rand_test_1.py
> from random import random
> from math import sqrt, sin
>
> def do_python():
>     N = 1000000
>
>    mean = 0
>    var  = 0
>
>    for i in range(N):
>        x = random()
>        x = 3.14*sqrt(x)
>        x = sin(x)
>        mean += x
>        var += x**2
>
>    mean = mean/N
>    var  = var/N - mean**2
>    print 'Mean=%g, var=%g' % (mean, var)
>
> # rand_test_2.py
>
> def do_numpy():
>    import numpy as np
>
>    N = 1000000
>
>    x = np.random.rand(N)
>    x = 3.14*np.sqrt(x)
>    x = np.sin(x)
>
>    mean = x.mean()
>    var  = np.sum(x**2)/N - mean**2
>
>    print 'Mean=%g, var=%g' % (mean, var)
> ##########################################################################
>
> And in ipython:
>
>    In [1]: %run test.py
>
>    In [2]: %timeit do_python()
>    [... snip ]
>    10 loops, best of 3: 1.18 s per loop
>
>    In [3]: %timeit do_numpy()
>    10 loops, best of 3: 147 ms per loop
>
> This is significantly different from your timings. The numbers do not
> mean the same thing, but I trust these one more, and would try to use CPU
> time only for comparisons between different approachs.
>
> My 2 cents.
>
> Ga?l
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080616/882af886/attachment.html>

From ivo.maljevic at gmail.com  Mon Jun 16 11:22:36 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Mon, 16 Jun 2008 11:22:36 -0400
Subject: [SciPy-user] More on speed comparisons
In-Reply-To: <20080616145916.GC3938@phare.normalesup.org>
References: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>
	<826c64da0806160745mf57f37bqc5f1e87adddb590a@mail.gmail.com>
	<20080616145916.GC3938@phare.normalesup.org>
Message-ID: <826c64da0806160822p284fef40rbe5712270b9cbbd4@mail.gmail.com>

Ga?l,
If you look at the numbers you have obtained, you will notice that
your ratio is 1180/147 = 8

After I fixed the sum error, SciPy code became as fast as Octave, therefore:

16/2.2 = 7.3

=> 8 vs. 7.3 not significant difference

Ivo

##########################################################################
>
> And in ipython:
>
>    In [1]: %run test.py
>
>    In [2]: %timeit do_python()
>    [... snip ]
>    10 loops, best of 3: 1.18 s per loop
>
>    In [3]: %timeit do_numpy()
>    10 loops, best of 3: 147 ms per loop
>
> This is significantly different from your timings. The numbers do not
> mean the same thing, but I trust these one more, and would try to use CPU
> time only for comparisons between different approachs.
>
> My 2 cents.
>
> Ga?l
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080616/c210e54d/attachment.html>

From david at ar.media.kyoto-u.ac.jp  Mon Jun 16 11:12:43 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Tue, 17 Jun 2008 00:12:43 +0900
Subject: [SciPy-user] More on speed comparisons
In-Reply-To: <826c64da0806160813q6d2377d3pdad6811f10a7d8f7@mail.gmail.com>
References: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>	<826c64da0806160745mf57f37bqc5f1e87adddb590a@mail.gmail.com>	<20080616145916.GC3938@phare.normalesup.org>
	<826c64da0806160813q6d2377d3pdad6811f10a7d8f7@mail.gmail.com>
Message-ID: <485682EB.4060504@ar.media.kyoto-u.ac.jp>

Ivo Maljevic wrote:
> Aside from good speed with SciPy (I feel embarrassed that I made that 
> 'sum' mistake)

No need to be embarrassed, I think most of us went through the same.

>
>
> time command (e.g., time python ./rand_test_2.py). While you may find 
> this methodology incorrect,
> I do the same with fortran and C code (time ./rand_test_c or time 
> ./rand_test_f).

It may be correct depending on what you want to measure: here you are 
measuring start up times (which is negligeable for C and fortran on a 
decent OS if you are not running cold, that is the C /F runtime is 
already in memory; it is really hard not to have the C runtime already 
loaded on unix :) ).

Numpy startup time is significant for such short computations (a few 
seconds). If for some reason you need to call the script often, it may 
be the good way to measure things. If you just want to benchmark 
different methods, that's certainly the wrong approach, and Gael's one 
is the right one. What you are doing is similar to matlab -r, except 
that python is much more powerful for scripting here.

>
> Some of my matlab simulation run for more than one hour and I use 
> tic/toc there.

On matlab, I think you should use cputime, but I cannot find this 
recommendation in matlab's online help anymore.

cheers,

David


From ivo.maljevic at gmail.com  Mon Jun 16 11:34:25 2008
From: ivo.maljevic at gmail.com (Ivo Maljevic)
Date: Mon, 16 Jun 2008 11:34:25 -0400
Subject: [SciPy-user] More on speed comparisons
In-Reply-To: <485682EB.4060504@ar.media.kyoto-u.ac.jp>
References: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>
	<826c64da0806160745mf57f37bqc5f1e87adddb590a@mail.gmail.com>
	<20080616145916.GC3938@phare.normalesup.org>
	<826c64da0806160813q6d2377d3pdad6811f10a7d8f7@mail.gmail.com>
	<485682EB.4060504@ar.media.kyoto-u.ac.jp>
Message-ID: <826c64da0806160834o105a4cdfkcb437f1090db5bcf@mail.gmail.com>

>
> > Some of my matlab simulation run for more than one hour and I use
> > tic/toc there.
>
> On matlab, I think you should use cputime, but I cannot find this
> recommendation in matlab's online help anymore.
>
> cheers,
>

I must admit I was never consistent with matlab's execution time
measurements. I used all of them:
clock/etime, cputime, tic/toc. Don't know if there is a preferred method.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080616/16600626/attachment.html>

From bsouthey at gmail.com  Mon Jun 16 11:41:26 2008
From: bsouthey at gmail.com (Bruce Southey)
Date: Mon, 16 Jun 2008 10:41:26 -0500
Subject: [SciPy-user] More on speed comparisons
In-Reply-To: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>
References: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>
Message-ID: <485689A6.1040205@gmail.com>

Ivo Maljevic wrote:
> I was planing to refrain from this, but I believe this might be an 
> interesting comparison for people who seriously think about switching 
> to SciPy/NumPy.
>
> Few days ago I wrote about speed comparison between the scalar and 
> vectorized versions of function calls. Based on several comments, I
> concluded that the same story that applies to Matlab and Octave 
> applies here: vectorize thy code, and speed gain will come.
>
> Before I show the results of vectorized vs. non-vectorized results, 
> just want to go on the record and say that I am by no means sayhing 
> that SciPy/NumPy is not good. I still like what has been done here. 
> There is a particular scenario that I use at my work where SciPy, 
> combined with matplotlib, is extremely useful. That scenario is the 
> following.
>
> In my wireless lab, I have basestations, mobile stations, whole bunch 
> of instruments and PCs either connected via network or GPIB cables 
> (for instruments).I use Python here to automate test cases and data 
> collection and the ability to do SSH and GPIB communication is very 
> useful. Once I collect data, I use SciPy for some simple 
> postprocessing and I generate PNG plots, and finally, I generate HTML 
> pages with results shown as tables and plots. So, it is all done in a 
> single language/script instead of having to break the processing into 
> several languages/scripts.
>
> However, I wanted to see if SciPy would be good enough speedwise to 
> completely replace Matlab. An, at least for the type of processing I 
> do, it comes nowhere near it. I wrote a small toy program that does 
> some simple random variable manipulation in several languages. The 
> python code consists of two versions, one uses for loop and basic 
> pathon libraries and the other uses nympy's vectorized form (there was 
> no difference between numpy and scipy).
>
> Here are the relative results after running the code on two machines:
>
> 64-bit Ubunty 8.04:
> =================
>
> Fortran    C       Octave    SciPy   Pure Python
> =================================================
> 1          1.2      2.2      16      20
>
>
> 32-Bit openSUSE 10.3:
> ==================
>
> Fortran    C       Octave    SciPy   Pure Python
> =================================================
> 1          1.2      2.4      15      19.4
>
>
> The numbers are rounded a little bit, but they are in that range. I 
> see two problems here:
>
> 1. SciPy is very slow, even when compared to Octave 3.0
> 2. It is only sligtly faster than Python with a for loop.
>
> Below is the source code for the two python versions. While this 
> processing is not from any real application,
> it is not very different from the processing I normally do.
>
> Now, it is very likely that for different type of processing people 
> will find SciPy fast enough (matrix inversions, eigenvalues, etc), but 
> for the type of
> processing I need it is not fast enough.
>
> Ivo
>
> ##########################################################################
> # rand_test_1.py
> from random import random
> from math import sqrt, sin
>
> N = 1000000
>
> mean = 0
> var  = 0
>
> for i in range(N):
>   x = random()
>   x = 3.14*sqrt(x)
>   x = sin(x)
>   mean += x
>   var += x**2
>
> mean = mean/N
> var  = var/N - mean**2
> print 'Mean=%g, var=%g' % (mean, var)
>
> # rand_test_2.py
> from numpy import random, sin, sqrt
>
> N = 1000000
>
> x = random.rand(N)
> x = 3.14*sqrt(x)
> x = sin(x)
>
> mean = sum(x)/N
> var  = sum(x**2)/N - mean**2
>
> print 'Mean=%g, var=%g' % (mean, var)
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>   
Hi,
Actually this was addressed in Hans Petter Langtangen's book 'Python 
scripting for Computational Science' and probably elsewhere before that. 
While the book is somewhat dated (2004) it does contain considerable 
useful information on numerical python. For example the section 4.2.3 
shows this problem of using scalar arguments in Numeric and numarray 
compared using  Python's math module.

Bruce


From anand.prabhakar.patil at gmail.com  Mon Jun 16 11:58:04 2008
From: anand.prabhakar.patil at gmail.com (Anand Patil)
Date: Mon, 16 Jun 2008 16:58:04 +0100
Subject: [SciPy-user] Bus error on Intel mac pro w/Leopard, gcc 4.2,
	numpy from svn head
Message-ID: <FD058F3B-B911-4140-9EA4-9ADF1845D7C0@gmail.com>

Hi all,

I'm getting a bus error using the numpy head on an Intel Mac Pro  
running Leopard with gcc 4.2. It happens intermittently 1-2 hours into  
a long computation, so I haven't been able to boil the situation  
down... I'm hoping the gdb output and crash report below are enough to  
go on.

I built Python from source using gcc 4.2, so I had to get rid of the  
'no-cpp-precomp' and 'Wno-long-double' options.

The really long call stack in the crash report below makes me wonder  
whether the problem is in numpy itself, or other code is misusing  
numpy somehow. I'm using several f2py'ed Fortran modules from PyMC and  
also PyTables here.

Any suggestions are appreciated, including ideas on how to chase down  
the problem... I'll post if I get anywhere.

Thanks,
Anand


----------------GDB output:

(gdb) attach 28942
Attaching to process 28942.
0x9400f5e2 in select$DARWIN_EXTSN ()
(gdb) continue
Continuing.
Reading symbols for shared libraries + done
Reading symbols for shared libraries + done
Reading symbols for shared libraries + done
Reading symbols for shared libraries + done
Reading symbols for shared libraries + done
Reading symbols for shared libraries + done
Reading symbols for shared libraries + done
Reading symbols for shared libraries + done
Reading symbols for shared libraries + done


Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_PROTECTION_FAILURE at address: 0xb0082ff0
[Switching to process 28942 thread 0x2603]
0x0122c8a6 in array_dealloc (self=0x3fe5e80) at arrayobject.c:2079
2079	        Py_DECREF(self->base);
(gdb)
Continuing.

Program received signal EXC_BAD_ACCESS, Could not access memory.
Reason: KERN_PROTECTION_FAILURE at address: 0xb0082ff0
0x0122c8a6 in array_dealloc (self=0x3fe5e80) at arrayobject.c:2079
2079	        Py_DECREF(self->base);

----------------Crash report:

Process:         Python [24936]
Path:            /Library/Frameworks/Python.framework/Versions/2.5/ 
Resources/Python.app/Contents/MacOS/Python
Identifier:      Python
Version:         ??? (???)
Code Type:       X86 (Native)
Parent Process:  bash [18452]

Date/Time:       2008-06-16 13:36:45.992 +0100
OS Version:      Mac OS X 10.5.3 (9D34)
Report Version:  6

Exception Type:  EXC_BAD_ACCESS (SIGBUS)
Exception Codes: KERN_PROTECTION_FAILURE at 0x00000000b0082ff0
Crashed Thread:  2

Thread 0:
0   libSystem.B.dylib             	0x9400f5e2 select$DARWIN_EXTSN + 10
1   org.python.python             	0x00197cae PyOS_Readline + 254
2   org.python.python             	0x002223d5 builtin_raw_input + 597
3   org.python.python             	0x0022c6a8 PyEval_EvalFrameEx + 22856
4   org.python.python             	0x0022d445 PyEval_EvalFrameEx + 26341
5   org.python.python             	0x0022dba5 PyEval_EvalCodeEx + 1845
6   org.python.python             	0x0022c211 PyEval_EvalFrameEx + 21681
7   org.python.python             	0x0022dba5 PyEval_EvalCodeEx + 1845
8   org.python.python             	0x0022c5eb PyEval_EvalFrameEx + 22667
9   org.python.python             	0x0022dba5 PyEval_EvalCodeEx + 1845
10  org.python.python             	0x0022c211 PyEval_EvalFrameEx + 21681
11  org.python.python             	0x0022dba5 PyEval_EvalCodeEx + 1845
12  org.python.python             	0x0022c211 PyEval_EvalFrameEx + 21681
13  org.python.python             	0x0022d445 PyEval_EvalFrameEx + 26341
14  org.python.python             	0x0022dba5 PyEval_EvalCodeEx + 1845
15  org.python.python             	0x0022c211 PyEval_EvalFrameEx + 21681
16  org.python.python             	0x0022dba5 PyEval_EvalCodeEx + 1845
17  org.python.python             	0x0022c211 PyEval_EvalFrameEx + 21681
18  org.python.python             	0x0022dba5 PyEval_EvalCodeEx + 1845
19  org.python.python             	0x0022c211 PyEval_EvalFrameEx + 21681
20  org.python.python             	0x0022dba5 PyEval_EvalCodeEx + 1845
21  org.python.python             	0x0022dcd7 PyEval_EvalCode + 87
22  org.python.python             	0x002518c7 PyRun_FileExFlags + 263
23  org.python.python             	0x00251bd0 PyRun_SimpleFileExFlags  
+ 496
24  org.python.python             	0x002601ba Py_Main + 3194
25  org.python.python             	0x00001fb6 0x1000 + 4022

Thread 1:
0   libSystem.B.dylib             	0x9400f5e2 select$DARWIN_EXTSN + 10
1   libSystem.B.dylib             	0x93ff06f5 _pthread_start + 321
2   libSystem.B.dylib             	0x93ff05b2 thread_start + 34

Thread 2 Crashed:
0   multiarray.so                 	0x0122c8a6 array_dealloc + 150  
(arrayobject.c:2079)
1   multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
2   multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
3   multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
4   multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
5   multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
6   multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
7   multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
8   multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
9   multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
10  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
11  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
12  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
13  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
14  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
15  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
16  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
17  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
18  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
19  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
20  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
21  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
22  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
23  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
24  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
25  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
26  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
27  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
28  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
29  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
30  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
31  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
32  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
33  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
34  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
35  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
36  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
37  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
38  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
39  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
40  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
41  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
42  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
43  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
44  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
45  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
46  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
47  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
48  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
49  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
50  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
51  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
52  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
53  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
54  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
55  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
56  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
57  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
58  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
59  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
60  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
61  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
62  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
63  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
64  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
65  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
66  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
67  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
68  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
69  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
70  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
71  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
72  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
73  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
74  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
75  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
76  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
77  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
78  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
79  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
80  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
81  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
82  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
83  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
84  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
85  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
86  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
87  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
88  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
89  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
90  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
91  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
92  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
93  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
94  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
95  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
96  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
97  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
98  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
99  multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
100 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
101 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
102 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
103 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
104 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
105 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
106 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
107 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
108 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
109 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
110 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
111 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
112 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
113 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
114 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
115 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
116 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
117 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
118 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
119 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
120 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
121 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
122 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
123 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
124 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
125 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
126 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
127 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
128 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
129 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
130 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
131 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
132 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
133 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
134 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
135 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
136 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
137 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
138 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
139 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
140 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
141 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
142 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
143 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
144 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
145 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
146 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
147 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
148 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
149 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
150 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
151 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
152 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
153 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
154 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
155 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
156 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
157 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
158 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
159 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
160 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
161 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
162 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
163 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
164 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
165 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
166 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
167 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
168 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
169 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
170 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
171 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
172 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
173 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
174 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
175 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
176 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
177 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
178 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
179 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
180 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
181 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
182 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
183 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
184 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
185 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
186 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
187 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
188 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
189 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
190 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
191 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
192 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
193 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
194 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
195 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
196 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
197 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
198 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
199 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
200 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
201 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
202 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
203 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
204 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
205 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
206 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
207 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
208 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
209 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
210 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
211 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
212 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
213 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
214 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
215 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
216 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
217 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
218 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
219 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
220 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
221 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
222 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
223 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
224 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
225 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
226 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
227 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
228 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
229 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
230 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
231 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
232 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
233 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
234 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
235 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
236 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
237 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
238 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
239 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
240 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
241 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
242 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
243 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
244 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
245 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
246 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
247 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
248 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
249 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
250 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
251 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
252 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
253 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
254 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
255 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
256 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
257 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
258 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
259 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
260 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
261 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
262 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
263 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
264 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
265 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
266 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
267 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
268 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
269 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
270 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
271 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
272 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
273 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
274 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
275 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
276 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
277 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
278 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
279 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
280 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
281 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
282 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
283 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
284 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
285 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
286 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
287 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
288 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
289 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
290 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
291 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
292 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
293 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
294 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
295 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
296 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
297 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
298 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
299 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
300 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
301 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
302 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
303 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
304 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
305 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
306 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
307 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
308 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
309 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
310 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
311 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
312 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
313 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
314 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
315 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
316 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
317 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
318 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
319 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
320 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
321 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
322 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
323 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
324 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
325 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
326 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
327 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
328 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
329 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
330 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
331 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
332 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
333 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
334 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
335 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
336 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
337 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
338 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
339 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
340 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
341 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
342 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
343 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
344 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
345 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
346 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
347 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
348 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
349 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
350 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
351 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
352 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
353 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
354 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
355 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
356 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
357 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
358 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
359 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
360 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
361 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
362 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
363 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
364 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
365 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
366 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
367 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
368 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
369 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
370 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
371 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
372 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
373 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
374 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
375 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
376 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
377 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
378 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
379 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
380 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
381 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
382 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
383 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
384 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
385 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
386 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
387 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
388 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
389 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
390 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
391 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
392 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
393 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
394 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
395 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
396 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
397 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
398 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
399 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
400 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
401 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
402 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
403 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
404 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
405 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
406 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
407 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
408 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
409 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
410 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
411 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
412 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
413 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
414 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
415 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
416 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
417 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
418 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
419 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
420 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
421 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
422 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
423 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
424 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
425 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
426 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
427 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
428 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
429 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
430 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
431 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
432 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
433 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
434 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
435 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
436 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
437 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
438 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
439 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
440 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
441 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
442 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
443 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
444 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
445 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
446 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
447 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
448 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
449 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
450 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
451 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
452 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
453 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
454 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
455 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
456 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
457 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
458 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
459 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
460 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
461 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
462 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
463 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
464 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
465 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
466 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
467 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
468 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
469 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
470 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
471 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
472 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
473 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
474 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
475 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
476 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
477 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
478 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
479 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
480 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
481 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
482 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
483 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
484 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
485 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
486 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
487 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
488 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
489 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
490 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
491 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
492 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
493 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
494 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
495 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
496 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
497 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
498 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
499 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
500 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
501 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
502 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
503 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
504 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
505 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
506 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
507 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
508 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
509 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
510 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)
511 multiarray.so                 	0x0122c8ac array_dealloc + 156  
(arrayobject.c:2079)

Thread 2 crashed with X86 Thread State (32-bit):
   eax: 0x041cf330  ebx: 0x001c7df1  ecx: 0x00000000  edx: 0x01252380
   edi: 0x0486c210  esi: 0x043d2ea0  ebp: 0xb0083008  esp: 0xb0082ff0
    ss: 0x0000001f  efl: 0x00010246  eip: 0x0122c8a6   cs: 0x00000017
    ds: 0x0000001f   es: 0x0000001f   fs: 0x0000001f   gs: 0x00000037
   cr2: 0xb0082ff0

Binary Images:
     0x1000 -     0x1ff8 +org.python.python 2.5a0 (2.5alpha0)  
<11eb706d026ecbfb1f78c260c082c516> /Library/Frameworks/ 
Python.framework/Versions/2.5/Resources/Python.app/Contents/MacOS/Python
    0x48000 -    0x49ff7 +cStringIO.so ??? (???)  
<d6ad044f1a7eda056e5f635d871e95e5> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/cStringIO.so
    0xa5000 -    0xa5ff7 +resource.so ??? (???)  
<56321f762531e2ef9edcc11a639aecea> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/resource.so
    0xa9000 -    0xacffb +_struct.so ??? (???)  
<9c3c7e367cf5f6d04ebdae0bcf52bd33> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_struct.so
    0xb2000 -    0xb2fff +gestalt.so ??? (???)  
<852918ac90f2d71a9e8b790274430cf5> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/gestalt.so
    0xb6000 -    0xb6ff7 +_weakref.so ??? (???)  
<3413d47de76b9c955b5c5a3e41939a91> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_weakref.so
    0xba000 -    0xc1fed +libgcc_s.1.dylib ??? (???) /usr/local/lib/ 
libgcc_s.1.dylib
    0xe4000 -    0xe7ff1 +strop.so ??? (???)  
<4761fef70b2f0e950a0309edd2f32d72> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/strop.so
    0xed000 -    0xefff3 +operator.so ??? (???)  
<58d5206170fe1e6b0889d2216f8caeab> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/operator.so
    0xf5000 -    0xf6fff +time.so ??? (???)  
<f0302ca61a11e89116a95b3a0bc72795> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/time.so
    0xfd000 -    0xfdff2 +_bisect.so ??? (???)  
<f4a993401f2016000f33321a72cb5311> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_bisect.so
   0x192000 -   0x292fef +org.python.python 2.5a0 (2.5)  
<0d30889f65289c605f069dc75cf9e498> /Library/Frameworks/ 
Python.framework/Versions/2.5/Python
   0x3a4000 -   0x3a7ff7 +itertools.so ??? (???)  
<2750f2667689ff0fea4d50ab019d59fe> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/itertools.so
   0x3ae000 -   0x3afffa +_heapq.so ??? (???)  
<35b276d5575f2fea7c63f50f4ce56ed0> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_heapq.so
   0x3b4000 -   0x3b5fff +math.so ??? (???)  
<0e51601fe1bc8d2e8c2768c0ab34768b> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/math.so
   0x3fa000 -   0x3fbfff +_random.so ??? (???)  
<c61e6a9772aa5a2eaa8ee36e440e96a2> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_random.so
   0x500000 -   0x502fff +binascii.so ??? (???)  
<58c7b821ffde5d10d82c178fd039316c> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/binascii.so
   0x507000 -   0x508fff +fcntl.so ??? (???)  
<90ffe24832cf1b3bfbf24d47bb78e1fc> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/fcntl.so
   0x50c000 -   0x50effa +collections.so ??? (???)  
<7c1097a3c3cceaf582791f175c1efcfb> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/collections.so
   0x513000 -   0x51effd +_curses.so ??? (???)  
<7a23a7622ae20237a3be87d931a11cb0> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_curses.so
   0x527000 -   0x558fe7 +libncurses.5.dylib ??? (???) /Library/ 
Frameworks/Python.framework/Versions/2.5/lib/libncurses.5.dylib
   0x574000 -   0x575fff +termios.so ??? (???)  
<ba80a1e473365e00a8b601477bd751d8> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/termios.so
   0x57a000 -   0x57bff1 +_hashlib.so ??? (???)  
<e06f553eac0db56a16af1a7c837b2eca> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_hashlib.so
   0x580000 -   0x582ff2 +_sha256.so ??? (???)  
<76bf92d04dd4d40b2ef259a967188e5d> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_sha256.so
   0x586000 -   0x590ff9 +_sha512.so ??? (???)  
<ce5ac5d4821135f7827e445f4d7e815f> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_sha512.so
   0x5a6000 -   0x5a9ff7 +Container_values.so ??? (???)  
<bb82fffa3f930b75f74b16155a5301a4> /Users/anand/renearch/pymc/pymc/ 
Container_values.so
   0x5ae000 -   0x5b2ff8 +LazyFunction.so ??? (???)  
<f6a119cff37ad3ea8a3f699983a4c66d> /Users/anand/renearch/pymc/pymc/ 
LazyFunction.so
   0x5b7000 -   0x5bfff7 +_sqlite3.so ??? (???)  
<b8dd6a03d6fe5a9ced3f7316d905d5d9> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_sqlite3.so
   0x5c8000 -   0x5dcfff +utilsExtension.so ??? (???)  
<fce5cbbb6010e559ecfc7ebfa017cef0> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/tables/ 
utilsExtension.so
   0x5e9000 -   0x5e9ff6 +_comp_lzo.so ??? (???)  
<725f6f7ec76003a55ffb8ec4f54d50ec> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/tables/ 
_comp_lzo.so
   0x5ed000 -   0x5edff1 +_comp_bzip2.so ??? (???)  
<141be1279ee2a60d047401a6dd9362ab> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/tables/ 
_comp_bzip2.so
   0x5f4000 -   0x61bffc +readline.so ??? (???)  
<afa10cc4dcf1c206940d8cbe5bcf706f> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/ 
readline-2.5.1-py2.5-macosx-10.5-i386.egg/readline.so
   0x66e000 -   0x67dfff +_ctypes.so ??? (???)  
<1020669adf8ac5a768dd23ee08e9ca6f> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_ctypes.so
   0x690000 -   0x691ff1 +_locale.so ??? (???)  
<3f5534c89080d70ceddbde1c824347d5> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_locale.so
   0x69c000 -   0x69dfff +_lsprof.so ??? (???)  
<f9d245a7363abde86e517665ddabf6ae> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_lsprof.so
   0x6a2000 -   0x6a3fff +_ssl.so ??? (???)  
<1b0ed0d4c6530fcad161516ca8e9e9f9> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_ssl.so
   0x729000 -   0x730fd3 +libintl.3.dylib ??? (???) /usr/local/lib/ 
libintl.3.dylib
   0x748000 -   0x756ff7 +cPickle.so ??? (???)  
<008dcb580c3f6022fa51f5e571058fa1> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/cPickle.so
   0x75e000 -   0x764ff7 +_socket.so ??? (???)  
<1aab699650fe8b337274a41a7f71cf3e> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_socket.so
  0x1100000 -  0x1113ff6 +_sort.so ??? (???)  
<ccde6ae0427813cca1efa817af9828b9> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/numpy/ 
core/_sort.so
  0x111a000 -  0x111efff +_dotblas.so ??? (???)  
<46eba2855880ea38b649be728e6b1041> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/numpy/ 
core/_dotblas.so
  0x11e8000 -  0x1251ff7 +multiarray.so ??? (???)  
<fff63f3f899f3884c18c5e30d00678db> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/numpy/ 
core/multiarray.so
  0x1284000 -  0x12a8fff +umath.so ??? (???)  
<5ce0337c2db27a055c4d86c6c07778a3> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/numpy/ 
core/umath.so
  0x1343000 -  0x134ffff +parser.so ??? (???)  
<ec663f0ce3cf69c646d09f1c42f663d1> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/parser.so
  0x1355000 -  0x1357ff3 +mmap.so ??? (???)  
<6c16036bb75555e57d47f2a94cddab18> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/mmap.so
  0x135c000 -  0x1376ff7 +scalarmath.so ??? (???)  
<1ef6ab8b6f0286fd3cbb461202220020> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/numpy/ 
core/scalarmath.so
  0x13c9000 -  0x13cbfff +_compiled_base.so ??? (???)  
<40c2d83b32c143918fca6303fb206710> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/numpy/ 
lib/_compiled_base.so
  0x13cf000 -  0x13d3ff5 +bz2.so ??? (???)  
<161d3f6681ced0347e88db273869de45> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/bz2.so
  0x141a000 -  0x141cffe +zlib.so ??? (???)  
<9825af0113467adf80182223e14e2a0f> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/zlib.so
  0x1427000 -  0x142cff1 +lapack_lite.so ??? (???)  
<02d2a15d1b6c16e8f8fff8ae8fc3900a> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/numpy/ 
linalg/lapack_lite.so
  0x1431000 -  0x1439ff7 +fftpack_lite.so ??? (???)  
<02245b3339ac24c7b5ef46dab78eec9f> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/numpy/ 
fft/fftpack_lite.so
  0x143d000 -  0x146afff +mtrand.so ??? (???)  
<ab2364841a03abce9c491e7f9dbbe9b5> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/numpy/ 
random/mtrand.so
  0x147d000 -  0x1488fff +datetime.so ??? (???)  
<2d3e277728c9a6c36585840a6da8483b> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/datetime.so
  0x14ef000 -  0x14f0ff0 +nxutils.so ??? (???)  
<412c09eb1e056be48fb73854e2537329> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/ 
matplotlib/nxutils.so
  0x14f4000 -  0x14f6ffa +_csv.so ??? (???)  
<732bc826fbeac2c94ceb1d8b5dd40a06> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_csv.so
  0x1640000 -  0x166eff7 +_path.so ??? (???)  
<d731798a570dd67802d408b94722f84d> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/ 
matplotlib/_path.so
  0x1692000 -  0x16c8ff3 +ft2font.so ??? (???)  
<c657996f1f0122a0883731d92ee81885> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/ 
matplotlib/ft2font.so
  0x17b6000 -  0x17b9fff +_cntr.so ??? (???)  
<f1d88d83a137b7dc3b4701531b0aff11> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/ 
matplotlib/_cntr.so
  0x17bd000 -  0x17d5ff7 +_png.so ??? (???)  
<84815bdd92d088c0aa41a45ed3281ab4> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/ 
matplotlib/_png.so
  0x1836000 -  0x1895fd7 +libfreetype.6.dylib ??? (???) /usr/local/lib/ 
libfreetype.6.dylib
  0x1b2d000 -  0x1b60ff7 +_image.so ??? (???)  
<749429e16158897756345754aa6ea806> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/ 
matplotlib/_image.so
  0x1bc3000 -  0x1be2ffb +libpng12.0.dylib ??? (???)  
<944211fe5ff1afb93c3101b840c56d18> /usr/local/lib/libpng12.0.dylib
  0x1c35000 -  0x1c3bff3 +_tkinter.so ??? (???)  
<0fbfb437897f096637bab8e73e600c0f> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/_tkinter.so
  0x1c94000 -  0x1c95ff7 +MacOS.so ??? (???)  
<5884ea28d1b1f2f5f88579da4679c1b1> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/MacOS.so
  0x1ca8000 -  0x1cadfff +minpack2.so ??? (???)  
<8ec48bcce4d9119b595e24fec46a214c> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
optimize/minpack2.so
  0x1cb3000 -  0x1cb4fff +_zeros.so ??? (???)  
<fede13bd01b6f9069a8b578a1b6ab2fd> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
optimize/_zeros.so
  0x1cfe000 -  0x1d7cfef  com.tcltk.tcllibrary 8.4.7 b (8.4.7 b)  
<ba64bd6b47532dacf102eb072c7594f1> /System/Library/Frameworks/ 
Tcl.framework/Versions/8.4/Tcl
  0x1d9f000 -  0x1db2ff3 +_tkagg.so ??? (???)  
<5dd15d7cebcb67fbb2995e587bc673a1> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/ 
matplotlib/backends/_tkagg.so
  0x1f89000 -  0x1fecfe7 +unicodedata.so ??? (???)  
<636e2c9bc33d10b95d2c4dcc4dd46e14> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/unicodedata.so
  0x1fff000 -  0x2088ffb +_backend_agg.so ??? (???)  
<49d5b507e3bfb76fab8bce9c999e866f> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/ 
matplotlib/backends/_backend_agg.so
  0x210f000 -  0x23acffb +libhdf5.5.dylib ??? (???)  
<56a4e109e3068f137d6fb9888e9a50b5> /usr/local/hdf5/lib/libhdf5.5.dylib
  0x2422000 -  0x243efff +hdf5Extension.so ??? (???)  
<7324f0d980097db706ae8a45917483df> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/tables/ 
hdf5Extension.so
  0x2450000 -  0x246cfff +tableExtension.so ??? (???)  
<165670450d16ad850f10e1dd80975b9d> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/tables/ 
tableExtension.so
  0x24bc000 -  0x24ddff7 +interpreter.so ??? (???)  
<c2fcfcc0df4c5e4dcdbc3f7750c5b1d4> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/tables/ 
numexpr/interpreter.so
  0x24e2000 -  0x24f9fff +_minpack.so ??? (???)  
<5507b3760fc1e2a240b27f587488f058> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
optimize/_minpack.so
  0x2600000 -  0x2634fe7 +flib.so ??? (???)  
<f7b37aec69fe0e7118a2583956d74c28> /Users/anand/renearch/pymc/pymc/ 
flib.so
  0x2e4e000 -  0x2ed9fff +libgfortran.2.dylib ??? (???) /usr/local/lib/ 
libgfortran.2.dylib
  0x30ce000 -  0x30e6fff +_lbfgsb.so ??? (???)  
<3dafdd8ea4e006706522bff5490a7e1c> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
optimize/_lbfgsb.so
  0x30ec000 -  0x30f1fff +moduleTNC.so ??? (???)  
<e580f362ee508bc222fb028cdc3f575d> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
optimize/moduleTNC.so
  0x30f5000 -  0x310cfff +_cobyla.so ??? (???)  
<6c5be5a3125efff0b7f323e2715a1395> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
optimize/_cobyla.so
  0x3112000 -  0x3123ff3 +_slsqp.so ??? (???)  
<1a6f70b918125e7b6be6364e2c3942fb> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
optimize/_slsqp.so
  0x3129000 -  0x3130ffc +linalg_utils.so ??? (???)  
<6f4cd5472d2a89dfc7b113a5e0ab7df1> /Users/anand/renearch/pymc/pymc/gp/ 
linalg_utils.so
  0x3143000 -  0x314bff0 +incomplete_chol.so ??? (???)  
<6bd7b381bd9c1e077e7419f3ce4f4b75> /Users/anand/renearch/pymc/pymc/gp/ 
incomplete_chol.so
  0x31aa000 -  0x31b8fff +isotropic_cov_funs.so ??? (???)  
<d5f8f44c5dd6414eeb058862cc3272cd> /Users/anand/renearch/pymc/pymc/gp/ 
cov_funs/isotropic_cov_funs.so
  0x31c3000 -  0x31ccfff +distances.so ??? (???)  
<d9b5e185fefb608ddf141e73d69119f7> /Users/anand/renearch/pymc/pymc/gp/ 
cov_funs/distances.so
  0x31e1000 -  0x32e8fff +_cephes.so ??? (???)  
<366d7cb3d7f641382d6d79019412f756> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
special/_cephes.so
  0x3300000 -  0x33b7ff3 +specfun.so ??? (???)  
<363b09a95f4a86b444cf807aaf30ce36> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
special/specfun.so
  0x33cb000 -  0x33cdff3 +clapack.so ??? (???)  
<1382d21c5011ae18278cd6f8d3283029> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
linalg/clapack.so
  0x3413000 -  0x341cfff +_flinalg.so ??? (???)  
<468f18986a2ed6f1c5e4f45fb52acd46> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
linalg/_flinalg.so
  0x3424000 -  0x345cfff +flapack.so ??? (???)  
<b911123cbb10018d97f36242809cae15> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
linalg/flapack.so
  0x3481000 -  0x3488fff +calc_lwork.so ??? (???)  
<7ff90d5038a1d7771564385e638f0526> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
linalg/calc_lwork.so
  0x3490000 -  0x34b3fff +fblas.so ??? (???)  
<5ac17c542f994b528a15c9af434bd5f3> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
linalg/fblas.so
  0x34c9000 -  0x34cbff7 +cblas.so ??? (???)  
<d89fdb8bd3196790fddde4d8256a4539> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
linalg/cblas.so
  0x34d0000 -  0x3590ff7 +_csr.so ??? (???)  
<b5ad64c99159d224f7c38dc1d70e4fdc> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
sparse/sparsetools/_csr.so
  0x35ab000 -  0x3621fff +_csc.so ??? (???)  
<15cb9a9ac8f6482003220df9f75d4c60> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
sparse/sparsetools/_csc.so
  0x362e000 -  0x3651ff3 +_coo.so ??? (???)  
<40c7657d221d8ea576a431beff4bb8d8> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
sparse/sparsetools/_coo.so
  0x3658000 -  0x3667ff3 +_dia.so ??? (???)  
<04419611d1acf2d6049a3eb96c5cab0f> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
sparse/sparsetools/_dia.so
  0x366e000 -  0x373bfff +_bsr.so ??? (???)  
<8be3e55fdca1a8ac9b8a0736d1cfbfbb> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
sparse/sparsetools/_bsr.so
  0x3798000 -  0x37c1fff +_iterative.so ??? (???)  
<7f8e6eb898ca17c0cf33a83e5fb2255e> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
sparse/linalg/isolve/_iterative.so
  0x37d3000 -  0x3801fff +_zsuperlu.so ??? (???)  
<572ed07a437d8b30f44491568f2af1de> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
sparse/linalg/dsolve/_zsuperlu.so
  0x3810000 -  0x383efff +_ssuperlu.so ??? (???)  
<3f22e09d10027738c2594e29ed7faf49> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
sparse/linalg/dsolve/_ssuperlu.so
  0x384d000 -  0x387bfff +_dsuperlu.so ??? (???)  
<92416bc4ddd6eeed6c955d40ff953669> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
sparse/linalg/dsolve/_dsuperlu.so
  0x388a000 -  0x38b8fff +_csuperlu.so ??? (???)  
<6d7eb884ab3ff7beed867d9e087908e8> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
sparse/linalg/dsolve/_csuperlu.so
  0x38c7000 -  0x394ffff +_arpack.so ??? (???)  
<bf7490fc5c8c122a78202cef56b8dcf6> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
sparse/linalg/eigen/arpack/_arpack.so
  0x3961000 -  0x3968fff +statlib.so ??? (???)  
<6fd2e4a41357a7c9a6d145683f997d2f> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
stats/statlib.so
  0x396e000 -  0x3972ff3 +futil.so ??? (???)  
<e96f9e514b5e93599c70b54fc663ab65> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
stats/futil.so
  0x3978000 -  0x3979fff +select.so ??? (???)  
<8c792315a80303fb2d3279c516889c4e> /Library/Frameworks/ 
Python.framework/Versions/2.5/lib/python2.5/lib-dynload/select.so
  0x3a1a000 -  0x3a27fe7 +mvn.so ??? (???)  
<67f2b43156b77f173da8a22360e04f24> /Library/Frameworks/ 
Python.framework/Versions/Current/lib/python2.5/site-packages/scipy/ 
stats/mvn.so
  0xb000000 -  0xb0adfeb  com.tcltk.tklibrary 8.4.7 b (8.4.7 b)  
<bfcdb5182a3c9cf2733442efac5c92a5> /System/Library/Frameworks/ 
Tk.framework/Versions/8.4/Tk
0x8fe00000 - 0x8fe2da53  dyld 96.2 (???)  
<5013f43c4d2c33c9619011f103ec3238> /usr/lib/dyld
0x90128000 - 0x90136ffd  libz.1.dylib ??? (???)  
<545ca09467025f77131cfac09d8b9375> /usr/lib/libz.1.dylib
0x90137000 - 0x9013efe9  libgcc_s.1.dylib ??? (???)  
<28a7cbc3a5ca2982d124668306f422d9> /usr/lib/libgcc_s.1.dylib
0x9013f000 - 0x901effff  edu.mit.Kerberos 6.0.12 (6.0.12)  
<1dc515ebe407292db8e603938c72d4e8> /System/Library/Frameworks/ 
Kerberos.framework/Versions/A/Kerberos
0x901f0000 - 0x9046bfe7  com.apple.Foundation 6.5.5 (677.19)  
<bfd4ebea1a7739dd6b523f15dca01a37> /System/Library/Frameworks/ 
Foundation.framework/Versions/C/Foundation
0x9046c000 - 0x90746ff3  com.apple.CoreServices.CarbonCore 786.4  
(786.4) <059c4803a7a95e3c1a95a332baeb1edf> /System/Library/Frameworks/ 
CoreServices.framework/Versions/A/Frameworks/CarbonCore.framework/ 
Versions/A/CarbonCore
0x90747000 - 0x90767ff2  libGL.dylib ??? (???) /System/Library/ 
Frameworks/OpenGL.framework/Versions/A/Libraries/libGL.dylib
0x90819000 - 0x908a4fff  com.apple.framework.IOKit 1.5.1 (???)  
<60cfc4b175c4ef60bb8e9036716a29f4> /System/Library/Frameworks/ 
IOKit.framework/Versions/A/IOKit
0x908ad000 - 0x908adffa  com.apple.CoreServices 32 (32)  
<2760719f7a81e8c2bdfd15b0939abc29> /System/Library/Frameworks/ 
CoreServices.framework/Versions/A/CoreServices
0x90a2e000 - 0x90ab0ffb  com.apple.CFNetwork 330 (330)  
<6c5eda16e640b09334809ba4c1df985d> /System/Library/Frameworks/ 
CoreServices.framework/Versions/A/Frameworks/CFNetwork.framework/ 
Versions/A/CFNetwork
0x90b34000 - 0x90ecafff  com.apple.QuartzCore 1.5.3 (1.5.3)  
<1b65c05f89e81a499302fd63295b242d> /System/Library/Frameworks/ 
QuartzCore.framework/Versions/A/QuartzCore
0x90ecc000 - 0x90eeafff  libresolv.9.dylib ??? (???)  
<32ccbe19e89a3fdd09a0c88151ea508c> /usr/lib/libresolv.9.dylib
0x90eeb000 - 0x90efbffc  com.apple.LangAnalysis 1.6.4 (1.6.4)  
<fc8a73eb3d8b94467d46b81759f65d97> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/Frameworks/ 
LangAnalysis.framework/Versions/A/LangAnalysis
0x90efc000 - 0x90f20feb  libssl.0.9.7.dylib ??? (???)  
<0ee18f8589ed06aabdc1df5b37a801cd> /usr/lib/libssl.0.9.7.dylib
0x90f21000 - 0x91067ff7  com.apple.ImageIO.framework 2.0.2 (2.0.2)  
<77dfee73f4c0d230425a5151ee0bce05> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/Frameworks/ImageIO.framework/ 
Versions/A/ImageIO
0x91086000 - 0x91559ffe  libGLProgrammability.dylib ??? (???)  
<475db64244e011cd8811e076035b2632> /System/Library/Frameworks/ 
OpenGL.framework/Versions/A/Libraries/libGLProgrammability.dylib
0x9155a000 - 0x91582fff  libcups.2.dylib ??? (???)  
<ece20dff2a2c8ed3ae6ef735ef440c37> /usr/lib/libcups.2.dylib
0x915bd000 - 0x915d3fe7  com.apple.CoreVideo 1.5.1 (1.5.1)  
<ed7bb95fb94817ea3212090aac5c65f3> /System/Library/Frameworks/ 
CoreVideo.framework/Versions/A/CoreVideo
0x915d4000 - 0x915d4ffd  com.apple.Accelerate 1.4.2 (Accelerate  
1.4.2) /System/Library/Frameworks/Accelerate.framework/Versions/A/ 
Accelerate
0x91648000 - 0x9164aff5  libRadiance.dylib ??? (???)  
<c307054a6d8337c85ef501780b54704d> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/Frameworks/ImageIO.framework/ 
Versions/A/Resources/libRadiance.dylib
0x9164b000 - 0x9164ffff  libGIF.dylib ??? (???)  
<75b4fd9684d792add088205f987fb02e> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/Frameworks/ImageIO.framework/ 
Versions/A/Resources/libGIF.dylib
0x91650000 - 0x91731ff7  libxml2.2.dylib ??? (???)  
<1baef3d4972ee789d8fa6c1fa44da45c> /usr/lib/libxml2.2.dylib
0x91810000 - 0x91bcefea  libLAPACK.dylib ??? (???) /System/Library/ 
Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/ 
Versions/A/libLAPACK.dylib
0x91c99000 - 0x91cc4fe7  libauto.dylib ??? (???)  
<2072d673706bbe463ed2426af57a28d7> /usr/lib/libauto.dylib
0x91d68000 - 0x91d6bfff  com.apple.help 1.1 (36)  
<175489f8adf287b3ebd259362b0292c0> /System/Library/Frameworks/ 
Carbon.framework/Versions/A/Frameworks/Help.framework/Versions/A/Help
0x91d6c000 - 0x91d78fe7  com.apple.opengl 1.5.6 (1.5.6)  
<125de77ea2434a91364e79a0905a7771> /System/Library/Frameworks/ 
OpenGL.framework/Versions/A/OpenGL
0x91d79000 - 0x91e03fe3  com.apple.DesktopServices 1.4.6 (1.4.6)  
<94d1a28b351b7dff77becadab0967772> /System/Library/PrivateFrameworks/ 
DesktopServicesPriv.framework/Versions/A/DesktopServicesPriv
0x91e04000 - 0x91e04fff  com.apple.Carbon 136 (136)  
<ec1d4184925e652dbe1b9200a5a552ec> /System/Library/Frameworks/ 
Carbon.framework/Versions/A/Carbon
0x91e05000 - 0x91e5eff7  libGLU.dylib ??? (???) /System/Library/ 
Frameworks/OpenGL.framework/Versions/A/Libraries/libGLU.dylib
0x91e5f000 - 0x91e9dff7  libGLImage.dylib ??? (???)  
<093b1b698ca93a0380f5fa262459ea28> /System/Library/Frameworks/ 
OpenGL.framework/Versions/A/Libraries/libGLImage.dylib
0x91e9e000 - 0x91f18ff8  com.apple.print.framework.PrintCore 5.5.3  
(245.3) <222dade7b33b99708b8c09d1303f93fc> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/Frameworks/ 
PrintCore.framework/Versions/A/PrintCore
0x91f19000 - 0x91f46feb  libvDSP.dylib ??? (???)  
<2ee4eb005babc90eaa352b33eb09226e> /System/Library/Frameworks/ 
Accelerate.framework/Versions/A/Frameworks/vecLib.framework/Versions/A/ 
libvDSP.dylib
0x925b0000 - 0x925b0ff8  com.apple.ApplicationServices 34 (34)  
<e9cd7c823062c4382d89e3c9997f4739> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/ApplicationServices
0x925c4000 - 0x925c4ffd  com.apple.vecLib 3.4.2 (vecLib 3.4.2) /System/ 
Library/Frameworks/vecLib.framework/Versions/A/vecLib
0x925c5000 - 0x92658fff  com.apple.ink.framework 101.3 (86)  
<dfa9debcd7537849d228021d1d9c0f63> /System/Library/Frameworks/ 
Carbon.framework/Versions/A/Frameworks/Ink.framework/Versions/A/Ink
0x92659000 - 0x92698fef  libTIFF.dylib ??? (???)  
<4b7d3b3b9a9c8335c2538371cb39b60b> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/Frameworks/ImageIO.framework/ 
Versions/A/Resources/libTIFF.dylib
0x926e9000 - 0x926effff  com.apple.print.framework.Print 218.0.2  
(220.1) <2979f3be4e7e8adc875bf21658e9be94> /System/Library/Frameworks/ 
Carbon.framework/Versions/A/Frameworks/Print.framework/Versions/A/Print
0x92732000 - 0x9276cfff  com.apple.coreui 1.1 (61) /System/Library/ 
PrivateFrameworks/CoreUI.framework/Versions/A/CoreUI
0x9276d000 - 0x92783fff  com.apple.DictionaryServices 1.0.0 (1.0.0)  
<7e9ff586b5c9d02b09e2a5527d98524f> /System/Library/Frameworks/ 
CoreServices.framework/Versions/A/Frameworks/ 
DictionaryServices.framework/Versions/A/DictionaryServices
0x92784000 - 0x92803ff5  com.apple.SearchKit 1.2.0 (1.2.0)  
<5abfde5537969168b8a8743ccb9ec735> /System/Library/Frameworks/ 
CoreServices.framework/Versions/A/Frameworks/SearchKit.framework/ 
Versions/A/SearchKit
0x92804000 - 0x92890ff7  com.apple.LaunchServices 289.2 (289.2)  
<3577886e3a6d56ee3949850c4fde76c9> /System/Library/Frameworks/ 
CoreServices.framework/Versions/A/Frameworks/LaunchServices.framework/ 
Versions/A/LaunchServices
0x92891000 - 0x928d7fef  com.apple.Metadata 10.5.2 (398.18)  
<adbb3a14e8f7da444e16d2fd61862771> /System/Library/Frameworks/ 
CoreServices.framework/Versions/A/Frameworks/Metadata.framework/ 
Versions/A/Metadata
0x92b50000 - 0x92b54fff  libmathCommon.A.dylib ??? (???) /usr/lib/ 
system/libmathCommon.A.dylib
0x92b7c000 - 0x92c0fff3  com.apple.ApplicationServices.ATS 3.3 (???)  
<064eb6d96417afa38a80b1735c4113aa> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/Frameworks/ATS.framework/ 
Versions/A/ATS
0x92c10000 - 0x92f17ff7  com.apple.HIToolbox 1.5.3 (???)  
<e36f5c553e5a32f64b7eb458dadadc71> /System/Library/Frameworks/ 
Carbon.framework/Versions/A/Frameworks/HIToolbox.framework/Versions/A/ 
HIToolbox
0x92f18000 - 0x92f1fffe  libbsm.dylib ??? (???)  
<5582985a86ea36504cca31788bccf963> /usr/lib/libbsm.dylib
0x92f20000 - 0x92f7aff7  com.apple.CoreText 2.0.2 (???)  
<9fde11f84a72e890bbf2aa8b0b13b79a> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/Frameworks/CoreText.framework/ 
Versions/A/CoreText
0x92f7b000 - 0x92f7bffb  com.apple.installserver.framework 1.0 (8) / 
System/Library/PrivateFrameworks/InstallServer.framework/Versions/A/ 
InstallServer
0x92f7c000 - 0x92f7efff  com.apple.securityhi 3.0 (30817)  
<020419ad33b8638b174e1a472728a894> /System/Library/Frameworks/ 
Carbon.framework/Versions/A/Frameworks/SecurityHI.framework/Versions/A/ 
SecurityHI
0x9313b000 - 0x93156ff3  libPng.dylib ??? (???)  
<c0484bec6e2432b406755591924fe664> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/Frameworks/ImageIO.framework/ 
Versions/A/Resources/libPng.dylib
0x9315d000 - 0x93224ff2  com.apple.vImage 3.0 (3.0) /System/Library/ 
Frameworks/Accelerate.framework/Versions/A/Frameworks/vImage.framework/ 
Versions/A/vImage
0x93225000 - 0x932acff7  libsqlite3.0.dylib ??? (???)  
<11311084bc4be9d4555dfac74fe7218a> /usr/lib/libsqlite3.0.dylib
0x932ad000 - 0x932dcff7  libncurses.5.4.dylib ??? (???)  
<00632d5180ac31e2cd437a1ce9d08562> /usr/lib/libncurses.5.4.dylib
0x933ac000 - 0x933c0ff3  com.apple.ImageCapture 4.0 (5.0.0) /System/ 
Library/Frameworks/Carbon.framework/Versions/A/Frameworks/ 
ImageCapture.framework/Versions/A/ImageCapture
0x933c1000 - 0x937d1fef  libBLAS.dylib ??? (???) /System/Library/ 
Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/ 
Versions/A/libBLAS.dylib
0x937d2000 - 0x9399ffe7  com.apple.security 5.0.3 (33532)  
<3bef414f3c6f433e707ac5abee340e16> /System/Library/Frameworks/ 
Security.framework/Versions/A/Security
0x939a0000 - 0x93a1dfef  libvMisc.dylib ??? (???) /System/Library/ 
Frameworks/Accelerate.framework/Versions/A/Frameworks/vecLib.framework/ 
Versions/A/libvMisc.dylib
0x93c65000 - 0x93d1ffe3  com.apple.CoreServices.OSServices 226.3  
(226.3) <456bdd65b936baf1ef497b74b4f960a8> /System/Library/Frameworks/ 
CoreServices.framework/Versions/A/Frameworks/OSServices.framework/ 
Versions/A/OSServices
0x93d20000 - 0x93d21ffc  libffi.dylib ??? (???)  
<596e0dbf626b211741cecaa9698f271b> /usr/lib/libffi.dylib
0x93e7f000 - 0x93edcffb  libstdc++.6.dylib ??? (???)  
<6106b1f2b0b303b06ae476253dbb5f3f> /usr/lib/libstdc++.6.dylib
0x93fbe000 - 0x9411eff3  libSystem.B.dylib ??? (???)  
<a12f397abf2285077b89bd726bff5b18> /usr/lib/libSystem.B.dylib
0x9411f000 - 0x9413effa  libJPEG.dylib ??? (???)  
<c776945af9d960ccf5a80aa2d89fd918> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/Frameworks/ImageIO.framework/ 
Versions/A/Resources/libJPEG.dylib
0x95295000 - 0x952c4fe3  com.apple.AE 402.2 (402.2)  
<d0da935e781446faa4563e7a9c587cce> /System/Library/Frameworks/ 
CoreServices.framework/Versions/A/Frameworks/AE.framework/Versions/A/AE
0x952c5000 - 0x95315ff7  com.apple.HIServices 1.7.0 (???)  
<f83f0cb838a646c69f08fd38a31265f3> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/Frameworks/ 
HIServices.framework/Versions/A/HIServices
0x95316000 - 0x953bdfeb  com.apple.QD 3.11.52 (???)  
<d6efc4827b91e646965d15c409a6610f> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/Frameworks/QD.framework/ 
Versions/A/QD
0x95425000 - 0x9545cfff  com.apple.SystemConfiguration 1.9.2 (1.9.2)  
<8b26ebf26a009a098484f1ed01ec499c> /System/Library/Frameworks/ 
SystemConfiguration.framework/Versions/A/SystemConfiguration
0x9545d000 - 0x954b9ff7  com.apple.htmlrendering 68 (1.1.3)  
<a9f65fa1c4668dc7c49af5bf7d5287ad> /System/Library/Frameworks/ 
Carbon.framework/Versions/A/Frameworks/HTMLRendering.framework/ 
Versions/A/HTMLRendering
0x954c0000 - 0x955f8ff7  libicucore.A.dylib ??? (???)  
<5031226ea28b371d8dfdbb32acfb48b5> /usr/lib/libicucore.A.dylib
0x95605000 - 0x95611fff  libbz2.1.0.dylib ??? (???)  
<e266a580d35c4518eee864dc064e4b98> /usr/lib/libbz2.1.0.dylib
0x95738000 - 0x9586afff  com.apple.CoreFoundation 6.5.2 (476.13)  
<b633d15f2901d73670cb1475628df1b3> /System/Library/Frameworks/ 
CoreFoundation.framework/Versions/A/CoreFoundation
0x9586b000 - 0x95874fff  com.apple.speech.recognition.framework 3.7.24  
(3.7.24) <6a6518b392d3d41ace3dcea69d6809d9> /System/Library/Frameworks/ 
Carbon.framework/Versions/A/Frameworks/SpeechRecognition.framework/ 
Versions/A/SpeechRecognition
0x958b2000 - 0x958bcfeb  com.apple.audio.SoundManager 3.9.2 (3.9.2)  
<caa41909dcb5a18a94bc68cd13999bd5> /System/Library/Frameworks/ 
Carbon.framework/Versions/A/Frameworks/CarbonSound.framework/Versions/ 
A/CarbonSound
0x95a97000 - 0x95b49ffb  libcrypto.0.9.7.dylib ??? (???)  
<8f92cbdc8777bea2ec49b06ee79fabc0> /usr/lib/libcrypto.0.9.7.dylib
0x95b7c000 - 0x95bf8feb  com.apple.audio.CoreAudio 3.1.0 (3.1)  
<be7120e91df91425e904d5327b4338b7> /System/Library/Frameworks/ 
CoreAudio.framework/Versions/A/CoreAudio
0x95d81000 - 0x95e60fff  libobjc.A.dylib ??? (???)  
<99a9ad33ca07114848fdd7580968a572> /usr/lib/libobjc.A.dylib
0x95e61000 - 0x95e66fff  com.apple.CommonPanels 1.2.4 (85)  
<3b64ef0de184d09c6f99a1a7e77e42be> /System/Library/Frameworks/ 
Carbon.framework/Versions/A/Frameworks/CommonPanels.framework/Versions/ 
A/CommonPanels
0x95e67000 - 0x95f32fff  com.apple.ColorSync 4.5.0 (4.5.0) /System/ 
Library/Frameworks/ApplicationServices.framework/Versions/A/Frameworks/ 
ColorSync.framework/Versions/A/ColorSync
0x95f39000 - 0x95f7bfef  com.apple.NavigationServices 3.5.2 (163)  
<91844980804067b07a0b6124310d3f31> /System/Library/Frameworks/ 
Carbon.framework/Versions/A/Frameworks/NavigationServices.framework/ 
Versions/A/NavigationServices
0x95f7c000 - 0x95f84fff  com.apple.DiskArbitration 2.2.1 (2.2.1)  
<42908e7ecc17a83cec4afef2850ec79e> /System/Library/Frameworks/ 
DiskArbitration.framework/Versions/A/DiskArbitration
0x95f85000 - 0x95f9dfff  com.apple.openscripting 1.2.6 (???)  
<4e0b05f9f47c6f7e2b01b321b2eb1413> /System/Library/Frameworks/ 
Carbon.framework/Versions/A/Frameworks/OpenScripting.framework/ 
Versions/A/OpenScripting
0x95f9e000 - 0x9663afff  com.apple.CoreGraphics 1.351.31 (???)  
<c97a42498636b2596764e48669f98e00> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/Frameworks/ 
CoreGraphics.framework/Versions/A/CoreGraphics
0x9663b000 - 0x9672fff4  libiconv.2.dylib ??? (???)  
<3f183527811098bb7332f67a1f902bfd> /usr/lib/libiconv.2.dylib
0x96741000 - 0x96751fff  com.apple.speech.synthesis.framework 3.7.1  
(3.7.1) <06d8fc0307314f8ffc16f206ad3dbf44> /System/Library/Frameworks/ 
ApplicationServices.framework/Versions/A/Frameworks/ 
SpeechSynthesis.framework/Versions/A/SpeechSynthesis
0x96f55000 - 0x96f79fff  libxslt.1.dylib ??? (???)  
<59399cc446ed903fd9479526ee9f116b> /usr/lib/libxslt.1.dylib
0x972b9000 - 0x972b9ffd  com.apple.Accelerate.vecLib 3.4.2 (vecLib  
3.4.2) /System/Library/Frameworks/Accelerate.framework/Versions/A/ 
Frameworks/vecLib.framework/Versions/A/vecLib
0xfffe8000 - 0xfffebfff  libobjc.A.dylib ??? (???) /usr/lib/ 
libobjc.A.dylib
0xffff0000 - 0xffff1780  libSystem.B.dylib ??? (???) /usr/lib/ 
libSystem.B.dylib

  


From ndbecker2 at gmail.com  Mon Jun 16 12:03:04 2008
From: ndbecker2 at gmail.com (Neal Becker)
Date: Mon, 16 Jun 2008 12:03:04 -0400
Subject: [SciPy-user] [optimization] OpenOpt release v 0.18
References: <4855754F.8090207@scipy.org>
Message-ID: <g362rh$ki3$1@ger.gmane.org>

Easy_installable?

sudo easy_install -U openopt
Searching for openopt
Reading http://pypi.python.org/simple/openopt/
Couldn't find index page for 'openopt' (maybe misspelled?)
Scanning index of all packages (this may take a while)
Reading http://pypi.python.org/simple/
Reading http://pypi.python.org/simple/OpenOpt/
Reading http://scipy.org/scipy/scikits/wiki/OpenOpt
No local packages or download links found for openopt
error: Could not find suitable distribution for Requirement.parse('openopt')

sudo easy_install -U OpenOpt
Searching for OpenOpt
Reading http://pypi.python.org/simple/OpenOpt/
Reading http://scipy.org/scipy/scikits/wiki/OpenOpt
No local packages or download links found for OpenOpt
error: Could not find suitable distribution for Requirement.parse('OpenOpt')




From robert.kern at gmail.com  Mon Jun 16 12:49:12 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Mon, 16 Jun 2008 11:49:12 -0500
Subject: [SciPy-user] Bus error on Intel mac pro w/Leopard, gcc 4.2,
	numpy from svn head
In-Reply-To: <FD058F3B-B911-4140-9EA4-9ADF1845D7C0@gmail.com>
References: <FD058F3B-B911-4140-9EA4-9ADF1845D7C0@gmail.com>
Message-ID: <3d375d730806160949xa3e83eag5f419d4d6026e0ca@mail.gmail.com>

On Mon, Jun 16, 2008 at 10:58, Anand Patil
<anand.prabhakar.patil at gmail.com> wrote:
> Hi all,
>
> I'm getting a bus error using the numpy head on an Intel Mac Pro
> running Leopard with gcc 4.2. It happens intermittently 1-2 hours into
> a long computation, so I haven't been able to boil the situation
> down... I'm hoping the gdb output and crash report below are enough to
> go on.
>
> I built Python from source using gcc 4.2, so I had to get rid of the
> 'no-cpp-precomp' and 'Wno-long-double' options.
>
> The really long call stack in the crash report below makes me wonder
> whether the problem is in numpy itself, or other code is misusing
> numpy somehow. I'm using several f2py'ed Fortran modules from PyMC and
> also PyTables here.

It could be either. The long chain of array_dealloc() calls means that
you have a view of a view of a view ... of and array. When the tail
end of the view chain gets decrefed to 0 and deallocated, it will
decref its reference to the array it is a view of (myarray.base, if
you want to check this out at the Python level). If that was the only
reference, it gets deallocated and decrefs its .base array.

It is possible, though unlikely, that a cycle got formed somehow.

Does your program crash using the www.python.org binary built with gcc 4.0.1?

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From gael.varoquaux at normalesup.org  Mon Jun 16 13:00:51 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Mon, 16 Jun 2008 19:00:51 +0200
Subject: [SciPy-user] More on speed comparisons
In-Reply-To: <20080616145916.GC3938@phare.normalesup.org>
References: <826c64da0806160732i1ec11089wc3400d754276e8b2@mail.gmail.com>
	<826c64da0806160745mf57f37bqc5f1e87adddb590a@mail.gmail.com>
	<20080616145916.GC3938@phare.normalesup.org>
Message-ID: <20080616170051.GD3938@phare.normalesup.org>

On Mon, Jun 16, 2008 at 04:59:16PM +0200, Gael Varoquaux wrote:
> Indeed. I am also curious to know how you measure timings. The proper way
> of mesuring timings (ie measuring CPU time, and not wall time) is using
> the timeit module.

That was bullshit: timeit measures Wall time, it seems.

For the numbers, I was just being heavy, and had not notice that you
where normalize everything to fortran = 1. Sorry

Ga?l 


From anand.prabhakar.patil at gmail.com  Mon Jun 16 13:06:03 2008
From: anand.prabhakar.patil at gmail.com (Anand Patil)
Date: Mon, 16 Jun 2008 18:06:03 +0100
Subject: [SciPy-user] Bus error on Intel mac pro w/Leopard, gcc 4.2,
	numpy from svn head
In-Reply-To: <3d375d730806160949xa3e83eag5f419d4d6026e0ca@mail.gmail.com>
References: <FD058F3B-B911-4140-9EA4-9ADF1845D7C0@gmail.com>
	<3d375d730806160949xa3e83eag5f419d4d6026e0ca@mail.gmail.com>
Message-ID: <5216ED6D-7899-445A-8224-3E138F7DE970@gmail.com>

Robert,

Thanks for the reply.

On 16 Jun 2008, at 17:49, Robert Kern wrote:

> On Mon, Jun 16, 2008 at 10:58, Anand Patil
>
> It could be either. The long chain of array_dealloc() calls means that
> you have a view of a view of a view ... of and array. When the tail
> end of the view chain gets decrefed to 0 and deallocated, it will
> decref its reference to the array it is a view of (myarray.base, if
> you want to check this out at the Python level). If that was the only
> reference, it gets deallocated and decrefs its .base array.
>
> It is possible, though unlikely, that a cycle got formed somehow.
>
> Does your program crash using the www.python.org binary built with  
> gcc 4.0.1?

Actually, I wasn't able to get my Python environment set up with the  
python.org binary . I need OpenMP, which means I need gcc 4.2 ... but  
the python.org binary has no-cpp-precomp and Wno-long-double baked in.  
In addition it sets MACOSX_DEPLOYMENT_TARGET=10.3 and it uses the 10.4  
SDK's library directory, both of which confuse some packages' setup on  
Leopard. I'd be happy to try the binary if I could get past those  
problems...

Anand


From nwagner at iam.uni-stuttgart.de  Mon Jun 16 14:14:19 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Mon, 16 Jun 2008 20:14:19 +0200
Subject: [SciPy-user] scikits.timeseries
Message-ID: <web-103386599@uni-stuttgart.de>

Hi all,

I cannot install timeseries from svn.
Here is the output

linux:/home/nwagner/svn/timeseries # /usr/bin/python 
setup.py install
Traceback (most recent call last):
   File "setup.py", line 6, in ?
     from scikits.timeseries.version import version
   File 
"/home/nwagner/svn/timeseries/scikits/timeseries/__init__.py", 
line 14, in ?
     import const
   File 
"/home/nwagner/svn/timeseries/scikits/timeseries/const.py", 
line 80, in ?
     from cseries import freq_constants
ImportError: No module named cseries

Any idea ?

Nils

  


From pgmdevlist at gmail.com  Mon Jun 16 14:26:25 2008
From: pgmdevlist at gmail.com (Pierre GM)
Date: Mon, 16 Jun 2008 14:26:25 -0400
Subject: [SciPy-user] scikits.timeseries
In-Reply-To: <web-103386599@uni-stuttgart.de>
References: <web-103386599@uni-stuttgart.de>
Message-ID: <200806161426.26277.pgmdevlist@gmail.com>

On Monday 16 June 2008 14:14:19 Nils Wagner wrote:
> Hi all,
>
> I cannot install timeseries from svn.
> Here is the output

Mmh, strange. Seems to work OK on my machine. Have you tried a clean install 
(viz, removing any build directory beforehand, and any previous installation 
of the package ?).
Note that it could very well be a problem recently introduced, so contact me 
offlist if this doesn't work, we need to sort that out.
Thanks.



From robert.kern at gmail.com  Mon Jun 16 15:10:59 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Mon, 16 Jun 2008 14:10:59 -0500
Subject: [SciPy-user] Bus error on Intel mac pro w/Leopard, gcc 4.2,
	numpy from svn head
In-Reply-To: <5216ED6D-7899-445A-8224-3E138F7DE970@gmail.com>
References: <FD058F3B-B911-4140-9EA4-9ADF1845D7C0@gmail.com>
	<3d375d730806160949xa3e83eag5f419d4d6026e0ca@mail.gmail.com>
	<5216ED6D-7899-445A-8224-3E138F7DE970@gmail.com>
Message-ID: <3d375d730806161210x7d5b2c46w83a817c1d58aae74@mail.gmail.com>

On Mon, Jun 16, 2008 at 12:06, Anand Patil
<anand.prabhakar.patil at gmail.com> wrote:
> Robert,
>
> Thanks for the reply.
>
> On 16 Jun 2008, at 17:49, Robert Kern wrote:
>
>> On Mon, Jun 16, 2008 at 10:58, Anand Patil
>>
>> It could be either. The long chain of array_dealloc() calls means that
>> you have a view of a view of a view ... of and array. When the tail
>> end of the view chain gets decrefed to 0 and deallocated, it will
>> decref its reference to the array it is a view of (myarray.base, if
>> you want to check this out at the Python level). If that was the only
>> reference, it gets deallocated and decrefs its .base array.
>>
>> It is possible, though unlikely, that a cycle got formed somehow.
>>
>> Does your program crash using the www.python.org binary built with
>> gcc 4.0.1?
>
> Actually, I wasn't able to get my Python environment set up with the
> python.org binary . I need OpenMP, which means I need gcc 4.2 ... but
> the python.org binary has no-cpp-precomp and Wno-long-double baked in.

Good enough reason.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From zachary.pincus at yale.edu  Mon Jun 16 15:13:12 2008
From: zachary.pincus at yale.edu (Zachary Pincus)
Date: Mon, 16 Jun 2008 15:13:12 -0400
Subject: [SciPy-user] The IO library and image file formats -- compare
	with with PIL
In-Reply-To: <32CCE7B7-F876-41E8-AD5B-0347BFC756DF@yale.edu>
References: <bc657ead0804180136x5bf54ebax50cef6c0b5092a84@mail.gmail.com>
	<10B72505-ACEB-4EAE-A1B9-39A5038224CD@yale.edu>
	<9457e7c80804210704n348d2129l7228e5fce3ab164d@mail.gmail.com>
	<53E48B1E-91A0-422A-8309-CA58D4BCE486@yale.edu>
	<9457e7c80804211013i27a32020l4d704bf75d318c5f@mail.gmail.com>
	<3d375d730804211020n22ba0d72k8ae86cf8724a4c5b@mail.gmail.com>
	<0D36E713-05E6-43FA-9634-AD8BEA2B8743@yale.edu>
	<3d375d730804211349p4e75ab5cs6603fc13869f519@mail.gmail.com>
	<bc657ead0806150000t11818686xfb5ddb4fbc061cab@mail.gmail.com>
	<53B94D38-4F29-4FB6-9AED-A9647FE3A905@yale.edu>
	<bc657ead0806150950q1dbecacfud799c96436792c8f@mail.gmail.com>
	<32CCE7B7-F876-41E8-AD5B-0347BFC756DF@yale.edu>
Message-ID: <8CE7BF5F-36A9-4F25-B555-EB3353DEC0AA@yale.edu>

Hi all,

I just heard back from Fredrik. He's supportive of the idea, and made  
some helpful suggestions for how to proceed. It might be simpler than  
I had thought, actually...

Here's his email:

> hi zachary,
>
>> Our general idea is to use the python standard library to handle most
>> compressed data, and use numpy's internal unpacking features to  
>> decode the
>> uncompressed and assembled data streams. Anyhow, it struck me that  
>> the
>> *ImagePlugin.py files from the PIL would be a pretty useful (either
>> unmodified, or more likely, slightly-modified) as the front end of  
>> this IO
>> system, much as they are used in the PIL today.
>
> that definitely makes sense.  to avoid fragmentation, I'd prefer if
> you use unmodified versions (and submit any bug fixes etc upstream).
> the ImagePlugin modules have very few dependencies, on purpose; you
> should be able to create a light-weight "pil emulator" simply by
> plugging in Image, ImageFile, and ImagePalette objects in sys.modules,
> and then use the modules right away.  e.g.
>
>    class ImageEmulator:
>         ... stuff that implements necessary portions of the Image  
> interface ...
>    class ImageFileEmulator:
>         ... etc
>    class ImagePaletteEmulator:
>         ...
>    sys.modules["Image"] = ImageEmulator()
>    sys.modules["ImageFile"] = ImageFileEmulator()
>    sys.modules["ImagePalette"] = ImagePaletteEmulator()
>
>    import PngImagePlugin
>
> see the "open" and "save" code in Image.py to get some ideas on how to
> use the plugins.
>
> (and feel free to mail me if you want further integration ideas)
>
> cheers /F

Best,
Zach


From robert.kern at gmail.com  Mon Jun 16 15:19:35 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Mon, 16 Jun 2008 14:19:35 -0500
Subject: [SciPy-user] The IO library and image file formats -- compare
	with with PIL
In-Reply-To: <8CE7BF5F-36A9-4F25-B555-EB3353DEC0AA@yale.edu>
References: <bc657ead0804180136x5bf54ebax50cef6c0b5092a84@mail.gmail.com>
	<9457e7c80804211013i27a32020l4d704bf75d318c5f@mail.gmail.com>
	<3d375d730804211020n22ba0d72k8ae86cf8724a4c5b@mail.gmail.com>
	<0D36E713-05E6-43FA-9634-AD8BEA2B8743@yale.edu>
	<3d375d730804211349p4e75ab5cs6603fc13869f519@mail.gmail.com>
	<bc657ead0806150000t11818686xfb5ddb4fbc061cab@mail.gmail.com>
	<53B94D38-4F29-4FB6-9AED-A9647FE3A905@yale.edu>
	<bc657ead0806150950q1dbecacfud799c96436792c8f@mail.gmail.com>
	<32CCE7B7-F876-41E8-AD5B-0347BFC756DF@yale.edu>
	<8CE7BF5F-36A9-4F25-B555-EB3353DEC0AA@yale.edu>
Message-ID: <3d375d730806161219lfd13002p33a8b9af14e8c225@mail.gmail.com>

On Mon, Jun 16, 2008 at 14:13, Zachary Pincus <zachary.pincus at yale.edu> wrote:
> Hi all,
>
> I just heard back from Fredrik. He's supportive of the idea, and made
> some helpful suggestions for how to proceed. It might be simpler than
> I had thought, actually...

Excellent. All of my concerns are addressed. Thank you.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From jdh2358 at gmail.com  Mon Jun 16 15:32:03 2008
From: jdh2358 at gmail.com (John Hunter)
Date: Mon, 16 Jun 2008 14:32:03 -0500
Subject: [SciPy-user] scipy.stats.lognorm.rvs signature
Message-ID: <88e473830806161232x10c4ebhf771d7f3d9d87518@mail.gmail.com>

Is there a bug in the rvs method of scipy.stats.lognorm?

In [241]: scipy.__version__
Out[241]: '0.7.0.dev4388'

In [242]: dist = scipy.stats.lognorm()

In [243]: dist.rvs(10)
------------------------------------------------------------
Traceback (most recent call last):
  File "<ipython console>", line 1, in ?
  File "/home/titan/johnh/dev/lib/python2.4/site-packages/scipy/stats/distributions.py",
line 117, in rvs
    return self.dist.rvs(*self.args,**kwds)
  File "/home/titan/johnh/dev/lib/python2.4/site-packages/scipy/stats/distributions.py",
line 446, in rvs
    vals = reshape(self._rvs(*args),size)
TypeError: _rvs() takes exactly 2 arguments (1 given)



I was trying to plot empirical histograms against the pdf with the
following script when I bunped into this:

import matplotlib.pyplot as plt
import scipy.stats

for name in 'uniform', 'norm', 'expon', 'lognorm':
    print 'making', name
    dist = getattr(scipy.stats, name)()
    samples = dist.rvs(1000)
    fig = plt.figure()
    ax = fig.add_subplot(111)
    n, bins, patches = ax.hist(samples, 30, normed=True,
facecolor='blue', alpha=0.5)
    ax.set_title(name)
    ax.grid(True)
    binc = 0.5*(bins[1:]+bins[:-1])
    p = dist.pdf(binc)
    ax.plot(binc, p, lw=2, color='black')

plt.show()


From timmichelsen at gmx-topmail.de  Mon Jun 16 18:16:35 2008
From: timmichelsen at gmx-topmail.de (Tim Michelsen)
Date: Tue, 17 Jun 2008 00:16:35 +0200
Subject: [SciPy-user] converting 01-24h logger data for timeseries.date_array
Message-ID: <g36oo3$52d$1@ger.gmane.org>

Hello,
I have an array with date and time and data values.
I would like to read it into a timeseries. Therefore I would like to 
read the time information in the data file into a list of dates in order 
to create a timeseries.date_array.

Unfortunately, many logging devices put the data out in hours 1-24 
format (see A below).

How can I reformat this into a format (B below) that's acceped by 
datetime.datetime?

Example:

* How to read in such a array:

### A: orignial data ###

DATE; VALUE (tab separated)
03.08.99	23:50:00	10.
03.08.99	24:00:00	11.
04.08.99	00:10:00	10.5

###B: needed data for datetime ###

DATE; VALUE (tab separated)
03.08.99	23:50:00	10.
04.08.99	20:00:00; 	11.;
04.08.99	00:10:00	10.5

###

### Code I use to read the data:
data_in = numpy.loadtxt(input_file, dtype=numpy.str_, skiprows=1)
dates_list = ["%s %s" % (d[0], d[1]) for d in data_in]
dates_dt = [(datetime.datetime.strptime(d, "%d.%m.%Y %H:%M:%S")) for d 
in dates_list]
date_arr = ts.date_array(dates_dt, freq='minute')
series_in = ts.time_series(data_in[:,3].astype(numpy.float_), date_arr, 
mask=(data_in[:,3]==nodata_string_input), freq='minute')

=> Now, if I read in the raw data A the row
"03.08.99	24:00:00	11."
gets pre-pended to all values of the day 03.08.99 because it get's 
parsed as hour 0 of that day.
In reality it's hour 0 of day 04.08.99. Therefore I have to reformat the 
raw data to the way represented by B.
My current workaround is to open the file in a spreadsheet application 
and save it as ascii again. but I would prefer a python only solution 
because there are data sets which even don't fit in spreadsheets due to 
their length.

So far, I could find a way to do this efficently. I would really 
appreciate if someone could point me into a direction on how to achieve 
this.

Kind regards,
Timmie



From elmico.filos at gmail.com  Mon Jun 16 19:10:28 2008
From: elmico.filos at gmail.com (=?ISO-8859-1?Q?Mico_Fil=F3s?=)
Date: Tue, 17 Jun 2008 01:10:28 +0200
Subject: [SciPy-user] Integration with precalculated values
Message-ID: <e700999f0806161610v45b105d9x3da8191cc2d317d5@mail.gmail.com>

Hi,

I need to evaluate this function for an array of ys

    f(y) = integral( exp(x*x)  * psi(x), A -y, B -y) # 2nd and 3rd
arguments: lower and upper limits

where

    psi(x) = integral( exp(t*t) * ( 1 + erf(t) )**2 , -Inf, x)

Apart from the fact that the integrand blows up easily, I would gain
some speed by
precalculating psi for the ranges I need for the computation of f(y).
Is there any proper way to do that?

I would really appreciate any hint or suggestion on how to evaluate
the integral :)

Best


From robert.kern at gmail.com  Mon Jun 16 19:18:54 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Mon, 16 Jun 2008 18:18:54 -0500
Subject: [SciPy-user] Integration with precalculated values
In-Reply-To: <e700999f0806161610v45b105d9x3da8191cc2d317d5@mail.gmail.com>
References: <e700999f0806161610v45b105d9x3da8191cc2d317d5@mail.gmail.com>
Message-ID: <3d375d730806161618q389c0aafi8f7b632f47a479bd@mail.gmail.com>

On Mon, Jun 16, 2008 at 18:10, Mico Fil?s <elmico.filos at gmail.com> wrote:
> Hi,
>
> I need to evaluate this function for an array of ys
>
>    f(y) = integral( exp(x*x)  * psi(x), A -y, B -y) # 2nd and 3rd
> arguments: lower and upper limits
>
> where
>
>    psi(x) = integral( exp(t*t) * ( 1 + erf(t) )**2 , -Inf, x)
>
> Apart from the fact that the integrand blows up easily, I would gain
> some speed by
> precalculating psi for the ranges I need for the computation of f(y).
> Is there any proper way to do that?

Only if you can afford to give up adaptive sampling. Then you can use
fixed_quad() for Gaussian quadrature over your function (it will do
just one vectorized evaluation), or trapz(), simps() or romb() for
uniform samples that you have computed yourself. If your function
blows up easily, though, you may not be able to afford to give up on
adaptive sampling.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco

From peridot.faceted at gmail.com  Mon Jun 16 21:29:31 2008
From: peridot.faceted at gmail.com (Anne Archibald)
Date: Mon, 16 Jun 2008 19:29:31 -0600
Subject: [SciPy-user] Integration with precalculated values
In-Reply-To: <e700999f0806161610v45b105d9x3da8191cc2d317d5@mail.gmail.com>
References: <e700999f0806161610v45b105d9x3da8191cc2d317d5@mail.gmail.com>
Message-ID: <ce557a360806161829n1b97cfd2ib138917f3fd1e287@mail.gmail.com>

2008/6/16 Mico Fil?s <elmico.filos at gmail.com>:
> Hi,
>
> I need to evaluate this function for an array of ys
>
>    f(y) = integral( exp(x*x)  * psi(x), A -y, B -y) # 2nd and 3rd
> arguments: lower and upper limits
>
> where
>
>    psi(x) = integral( exp(t*t) * ( 1 + erf(t) )**2 , -Inf, x)
>
> Apart from the fact that the integrand blows up easily, I would gain
> some speed by
> precalculating psi for the ranges I need for the computation of f(y).
> Is there any proper way to do that?
>
> I would really appreciate any hint or suggestion on how to evaluate
> the integral :)

Hmm.

I would start with an implementation you know to be reliable; it may
be fast enough, and if not, it can serve as a good test case. For this
I'd use scipy.integrate.quad, more or less the way you wrote it there.

If you want to accelerate things by precalculating psi, how you handle
it depends on the y values and the accuracy you need. The most direct
way assumes you have no control of the y values; then you can use a
fairly standard trick to replace psi(t) by a function that is faster
to calculate. (At least, I often use it.) Just compute the function at
a reasonable set of values (I'm afraid you'll have to figure out what
qualifies as "reasonable"), and use scipy.interpolate.splrep to fit a
spline through them. Then instead of evaluating psi(x) directly, you
can evaluate the spline (which will be very quick). Since your
function is always positive and probably varies a lot in magnitude,
I'd fit a spline to the log and then exponentiate on evaluation. Also,
you can drastically accelerate the evaluation of psi(t) at a sequence
of values by using an ODE integrator (say scipy.integrate.odeint)
once. (Since integration needs to walk across t values anyway, you
might as well use an integrator that can keep track of all the values.
Also, if it's psi that likes to blow up, odeint can cope with that.
See its docstring.) Given a fast way to evaluate psi(t), quad should
allow you to evaluate the function you actually want rather quickly.

If this is too slow, you can probably do something involving more
carefully discretizing your two functions, but it will probably
require you to restrict the range of y values.

Looking at your function, I strongly suspect things can be simplified
by rewriting it as a multidimensional integral and rethinking how you
do the integral. In particular, I'd look to see if anything can be
done with rewriting the erf as an integral, since its exp(-u**2) might
help keep the other exp(x**2) and exp(t**2) under control. But as
always, it's a tradeoff between work for you and work for the machine.

Anne

From anand.prabhakar.patil at gmail.com  Tue Jun 17 04:34:22 2008
From: anand.prabhakar.patil at gmail.com (Anand Patil)
Date: Tue, 17 Jun 2008 09:34:22 +0100
Subject: [SciPy-user] Bus error on Intel mac pro w/Leopard, gcc 4.2,
	numpy from svn head
In-Reply-To: <3d375d730806161210x7d5b2c46w83a817c1d58aae74@mail.gmail.com>
References: <FD058F3B-B911-4140-9EA4-9ADF1845D7C0@gmail.com>
	<3d375d730806160949xa3e83eag5f419d4d6026e0ca@mail.gmail.com>
	<5216ED6D-7899-445A-8224-3E138F7DE970@gmail.com>
	<3d375d730806161210x7d5b2c46w83a817c1d58aae74@mail.gmail.com>
Message-ID: <3F964DF4-5AAB-4378-BD03-4126ED55AB3D@gmail.com>

On 16 Jun 2008, at 20:10, Robert Kern wrote:

> On Mon, Jun 16, 2008 at 12:06, Anand Patil
> <anand.prabhakar.patil at gmail.com> wrote:
>>
>> On 16 Jun 2008, at 17:49, Robert Kern wrote:
>>
>>> On Mon, Jun 16, 2008 at 10:58, Anand Patil
>>>
>>> Does your program crash using the www.python.org binary built with
>>> gcc 4.0.1?
>>
>> Actually, I wasn't able to get my Python environment set up with the
>> python.org binary . I need OpenMP, which means I need gcc 4.2 ... but
>> the python.org binary has no-cpp-precomp and Wno-long-double baked  
>> in.
>
> Good enough reason.

Can I be pretty sure that the problem is caused by a reference cycle,  
and how can I:
- Reproduce the problem with a simple program, assuming it were in  
numpy itself;
- Look for problematic reference cycles in my program?

Thanks,
Anand


From anand.prabhakar.patil at gmail.com  Tue Jun 17 07:24:49 2008
From: anand.prabhakar.patil at gmail.com (Anand Patil)
Date: Tue, 17 Jun 2008 12:24:49 +0100
Subject: [SciPy-user] Bus error on Intel mac pro w/Leopard, gcc 4.2,
	numpy from svn head
In-Reply-To: <3F964DF4-5AAB-4378-BD03-4126ED55AB3D@gmail.com>
References: <FD058F3B-B911-4140-9EA4-9ADF1845D7C0@gmail.com>
	<3d375d730806160949xa3e83eag5f419d4d6026e0ca@mail.gmail.com>
	<5216ED6D-7899-445A-8224-3E138F7DE970@gmail.com>
	<3d375d730806161210x7d5b2c46w83a817c1d58aae74@mail.gmail.com>
	<3F964DF4-5AAB-4378-BD03-4126ED55AB3D@gmail.com>
Message-ID: <465CCB8C-B480-47F6-97F7-86DBF78792C7@gmail.com>

On 17 Jun 2008, at 09:34, Anand Patil wrote:

> On 16 Jun 2008, at 20:10, Robert Kern wrote:
>
>> On Mon, Jun 16, 2008 at 12:06, Anand Patil
>> <anand.prabhakar.patil at gmail.com> wrote:
>>>
>>> On 16 Jun 2008, at 17:49, Robert Kern wrote:
>>>
>>>> On Mon, Jun 16, 2008 at 10:58, Anand Patil
>>>>
>>>> Does your program crash using the www.python.org binary built with
>>>> gcc 4.0.1?
>>>
>>> Actually, I wasn't able to get my Python environment set up with the
>>> python.org binary . I need OpenMP, which means I need gcc 4.2 ...  
>>> but
>>> the python.org binary has no-cpp-precomp and Wno-long-double baked  
>>> in.
>>
>> Good enough reason.
>
> Can I be pretty sure that the problem is caused by a reference  
> cycle, and how can I:
> - Reproduce the problem with a simple program, assuming it were in  
> numpy itself;
> - Look for problematic reference cycles in my program?



I found the bit of PyMC that was making views of views & changed it,  
and the problem seems to be gone (now that I have announced that I'm  
sure it'll come back this afternoon)... so this isn't a personal  
emergency anymore. So thanks for the tip about views.


I tried reproducing the bug as follows:

In [1]: from numpy import *

In [2]: A = zeros(10)

In [3]: for i in xrange(100000000):
    ...:     B = A.view(ndarray)

but no errors happened. Any other tests I can try? Would it be  
productive to submit the crash report as a bug, or is it too vague?


Anand


From elmico.filos at gmail.com  Tue Jun 17 07:29:26 2008
From: elmico.filos at gmail.com (=?ISO-8859-1?Q?Mico_Fil=F3s?=)
Date: Tue, 17 Jun 2008 13:29:26 +0200
Subject: [SciPy-user] Integration with precalculated values
In-Reply-To: <ce557a360806161829n1b97cfd2ib138917f3fd1e287@mail.gmail.com>
References: <e700999f0806161610v45b105d9x3da8191cc2d317d5@mail.gmail.com>
	<ce557a360806161829n1b97cfd2ib138917f3fd1e287@mail.gmail.com>
Message-ID: <e700999f0806170429k6beec359ma35fc16bc9237c63@mail.gmail.com>

Thanks Robert and Anne,

I will use scipy.interpolate.splrep for my precalculation.

I finally managed to keep the factors exp(x**2) under control. The
error function can be expanded in series that contain a common factor
equal to exp(-x**2). I use in particular the approximation of the
complementary error function with rational functions, as described in
"Rational Chebyshev approximations of the error function"  by WJ Cody,
Math. Comp., 1969, pp 631--638. I have also had to split the
definition of the integrand into positive and negative values of the
argument, in order for the integral to go smoothly.

Thanks for your help.


From robert.kern at gmail.com  Tue Jun 17 14:05:23 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Tue, 17 Jun 2008 13:05:23 -0500
Subject: [SciPy-user] Bus error on Intel mac pro w/Leopard, gcc 4.2,
	numpy from svn head
In-Reply-To: <465CCB8C-B480-47F6-97F7-86DBF78792C7@gmail.com>
References: <FD058F3B-B911-4140-9EA4-9ADF1845D7C0@gmail.com>
	<3d375d730806160949xa3e83eag5f419d4d6026e0ca@mail.gmail.com>
	<5216ED6D-7899-445A-8224-3E138F7DE970@gmail.com>
	<3d375d730806161210x7d5b2c46w83a817c1d58aae74@mail.gmail.com>
	<3F964DF4-5AAB-4378-BD03-4126ED55AB3D@gmail.com>
	<465CCB8C-B480-47F6-97F7-86DBF78792C7@gmail.com>
Message-ID: <3d375d730806171105n1198dbe8t5fdedbe868a2247@mail.gmail.com>

On Tue, Jun 17, 2008 at 06:24, Anand Patil
<anand.prabhakar.patil at gmail.com> wrote:
> I found the bit of PyMC that was making views of views & changed it,
> and the problem seems to be gone (now that I have announced that I'm
> sure it'll come back this afternoon)... so this isn't a personal
> emergency anymore. So thanks for the tip about views.

Okay, good. Can you post your patch? That might give us some clues.

> I tried reproducing the bug as follows:
>
> In [1]: from numpy import *
>
> In [2]: A = zeros(10)
>
> In [3]: for i in xrange(100000000):
>    ...:     B = A.view(ndarray)
>
> but no errors happened.

In [1]: from numpy import *

In [2]: A = zeros(1, dtype=uint8)

In [3]: B = A[:]

In [4]: for i in xrange(1000000):
   ...:     B = B[:]
   ...:
   ...:

In [5]: del B
zsh: segmentation fault  ipython

Yay! Reproducibility!

http://scipy.org/scipy/numpy/ticket/822

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From pav at iki.fi  Tue Jun 17 15:56:04 2008
From: pav at iki.fi (Pauli Virtanen)
Date: Tue, 17 Jun 2008 19:56:04 +0000 (UTC)
Subject: [SciPy-user] Bus error on Intel mac pro w/Leopard, gcc 4.2,
	numpy from svn head
References: <FD058F3B-B911-4140-9EA4-9ADF1845D7C0@gmail.com>
	<3d375d730806160949xa3e83eag5f419d4d6026e0ca@mail.gmail.com>
	<5216ED6D-7899-445A-8224-3E138F7DE970@gmail.com>
	<3d375d730806161210x7d5b2c46w83a817c1d58aae74@mail.gmail.com>
	<3F964DF4-5AAB-4378-BD03-4126ED55AB3D@gmail.com>
	<465CCB8C-B480-47F6-97F7-86DBF78792C7@gmail.com>
	<3d375d730806171105n1198dbe8t5fdedbe868a2247@mail.gmail.com>
Message-ID: <g394sk$ioi$1@ger.gmane.org>

Tue, 17 Jun 2008 13:05:23 -0500, Robert Kern wrote:

[clip: very long chain of views -> crash]
> Yay! Reproducibility!
> 
> http://scipy.org/scipy/numpy/ticket/822

It's probably the same as this one:
http://scipy.org/scipy/numpy/ticket/466

-- 
Pauli Virtanen



From robert.kern at gmail.com  Tue Jun 17 16:54:39 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Tue, 17 Jun 2008 15:54:39 -0500
Subject: [SciPy-user] Bus error on Intel mac pro w/Leopard, gcc 4.2,
	numpy from svn head
In-Reply-To: <g394sk$ioi$1@ger.gmane.org>
References: <FD058F3B-B911-4140-9EA4-9ADF1845D7C0@gmail.com>
	<3d375d730806160949xa3e83eag5f419d4d6026e0ca@mail.gmail.com>
	<5216ED6D-7899-445A-8224-3E138F7DE970@gmail.com>
	<3d375d730806161210x7d5b2c46w83a817c1d58aae74@mail.gmail.com>
	<3F964DF4-5AAB-4378-BD03-4126ED55AB3D@gmail.com>
	<465CCB8C-B480-47F6-97F7-86DBF78792C7@gmail.com>
	<3d375d730806171105n1198dbe8t5fdedbe868a2247@mail.gmail.com>
	<g394sk$ioi$1@ger.gmane.org>
Message-ID: <3d375d730806171354o7efc360bs5a28deecfe935f2e@mail.gmail.com>

On Tue, Jun 17, 2008 at 14:56, Pauli Virtanen <pav at iki.fi> wrote:
> Tue, 17 Jun 2008 13:05:23 -0500, Robert Kern wrote:
>
> [clip: very long chain of views -> crash]
>> Yay! Reproducibility!
>>
>> http://scipy.org/scipy/numpy/ticket/822
>
> It's probably the same as this one:
> http://scipy.org/scipy/numpy/ticket/466

Yes.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From anand.prabhakar.patil at gmail.com  Tue Jun 17 17:32:38 2008
From: anand.prabhakar.patil at gmail.com (Anand Patil)
Date: Tue, 17 Jun 2008 22:32:38 +0100
Subject: [SciPy-user] Bus error on Intel mac pro w/Leopard, gcc 4.2,
	numpy from svn head
In-Reply-To: <3d375d730806171354o7efc360bs5a28deecfe935f2e@mail.gmail.com>
References: <FD058F3B-B911-4140-9EA4-9ADF1845D7C0@gmail.com>
	<3d375d730806160949xa3e83eag5f419d4d6026e0ca@mail.gmail.com>
	<5216ED6D-7899-445A-8224-3E138F7DE970@gmail.com>
	<3d375d730806161210x7d5b2c46w83a817c1d58aae74@mail.gmail.com>
	<3F964DF4-5AAB-4378-BD03-4126ED55AB3D@gmail.com>
	<465CCB8C-B480-47F6-97F7-86DBF78792C7@gmail.com>
	<3d375d730806171105n1198dbe8t5fdedbe868a2247@mail.gmail.com>
	<g394sk$ioi$1@ger.gmane.org>
	<3d375d730806171354o7efc360bs5a28deecfe935f2e@mail.gmail.com>
Message-ID: <2bc7a5a50806171432obc98a6cm28f963af22f68890@mail.gmail.com>

On Tue, Jun 17, 2008 at 9:54 PM, Robert Kern <robert.kern at gmail.com> wrote:

> On Tue, Jun 17, 2008 at 14:56, Pauli Virtanen <pav at iki.fi> wrote:
> > Tue, 17 Jun 2008 13:05:23 -0500, Robert Kern wrote:
> >
> > [clip: very long chain of views -> crash]
> >> Yay! Reproducibility!
> >>
> >> http://scipy.org/scipy/numpy/ticket/822
> >
> > It's probably the same as this one:
> > http://scipy.org/scipy/numpy/ticket/466
>
> Yes.
>

Great, glad to hear it's being addressed & thanks again for the help. MCMC
can be a royal hassle; if there's the slightest chance of something going
wrong it'll eventually find it, often after hours or days of normal
operation.

In case it can still be helpful, my patch was to replace many consecutive
manipulations of PyMC variable 'x' as follows:

x.value = new_value
x.value = x.last_value

with:

x.value = new_value
x._value = x.last_value.

That bypassed calls to x.set_value via the property x.value. set_value casts
input arguments to the declared type of x, if provided, and then stores them
as x._value:

    def set_value(self, value):
        # Record new value and increment counter

        if self.verbose > 0:
            print '\t' + self.__name__ + ': value set to ', value

        # Value can't be updated if isdata=True
        if self.isdata:
            raise AttributeError, 'Stochastic '+self.__name__+'\'s value
cannot be updated if isdata flag is set'

        # Save current value as last_value
        # Don't copy because caching depends on the object's reference.
        self.last_value = self._value

        if isinstance(value, ndarray):
            value.flags['W'] = False
            if self.dtype is not None:
                if not self.dtype is value.dtype:
                    self._value = asarray(value,
dtype=self.dtype).view(value.__class__)
                else:
                    self._value = value
            else:
                self._value = value
        elif self.dtype and self.dtype is not object:
            try:
                self._value = self.dtype(value)
            except TypeError:
                self._value = asarray(value, dtype=self.dtype)
        else:
            self._value = value


    value = property(fget=get_value, fset=set_value, doc="Self's current
value.")

Anand
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080617/f304287c/attachment.html>

From bryan.fodness at gmail.com  Tue Jun 17 19:10:52 2008
From: bryan.fodness at gmail.com (Bryan Fodness)
Date: Tue, 17 Jun 2008 19:10:52 -0400
Subject: [SciPy-user] three interpolation scenarios
Message-ID: <fbf64d2b0806171610l4491c80dsc9dcd35ade7ce669@mail.gmail.com>

i have three different sets of data that need interpolation.

the first with a single x and y was easy and works.

x    y
0.0 0.999
0.1 1.006
0.2 1.014
0.3 1.021
0.4 1.000
0.5 1.001
...

data = loadtxt('data.txt', skiprows=1)
x, y = data[:,0], data[:,1]
interp = interpolate.interp1d(x, y)

the second with multiple y values is a little more complicated.
x    y1     y2      y3     y4      y5
0.0 1.000 0.999 0.999 0.999 1.000
0.1 1.000 1.000 1.001 1.005 1.006
0.2 1.000 1.001 1.004 1.008 1.014
0.3 1.000 1.002 1.006 1.012 1.021
0.4 1.000 1.003 1.009 1.015 1.028
0.5 1.001 1.005 1.012 1.019 1.036
...

y = 'y3'
data = loadtxt('oaf.txt', skiprows=1)
label = line.split()
d = label[0]
if d == 'd':
    dlist = label
x, y = data[:,0], data[:,dlist.index(y)]
interp = interpolate.interp1d(x, y)

and the third, is the 2d case.

       10     20      30     40      50
0.0 1.000 0.999 0.999 0.999 1.000
0.1 1.000 1.000 1.001 1.005 1.006
0.2 1.000 1.001 1.004 1.008 1.014
0.3 1.000 1.002 1.006 1.012 1.021
0.4 1.000 1.003 1.009 1.015 1.028
0.5 1.001 1.005 1.012 1.019 1.036
...

not sure how to get the x, y and z's

is the 1st and 2nd way the best way to do this, and can someone help with
the third case.


-- 
"The game of science can accurately be described as a never-ending insult to
human intelligence." - Jo?o Magueijo
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080617/f0485f63/attachment.html>

From Bryan.Fodness at gmail.com  Tue Jun 17 19:19:45 2008
From: Bryan.Fodness at gmail.com (Bryan)
Date: Tue, 17 Jun 2008 16:19:45 -0700 (PDT)
Subject: [SciPy-user] three interpolation scenarios
In-Reply-To: <fbf64d2b0806171610l4491c80dsc9dcd35ade7ce669@mail.gmail.com>
References: <fbf64d2b0806171610l4491c80dsc9dcd35ade7ce669@mail.gmail.com>
Message-ID: <97ef1481-e4cc-449c-89d5-714e5cfa547f@z72g2000hsb.googlegroups.com>



On Jun 17, 7:10?pm, "Bryan Fodness" <bryan.fodn... at gmail.com> wrote:
> i have three different sets of data that need interpolation.
>
> the first with a single x and y was easy and works.
>
> x ? ?y
> 0.0 0.999
> 0.1 1.006
> 0.2 1.014
> 0.3 1.021
> 0.4 1.000
> 0.5 1.001
> ...
>
> data = loadtxt('data.txt', skiprows=1)
> x, y = data[:,0], data[:,1]
> interp = interpolate.interp1d(x, y)
>
> the second with multiple y values is a little more complicated.
> x ? ?y1 ? ? y2 ? ? ?y3 ? ? y4 ? ? ?y5
> 0.0 1.000 0.999 0.999 0.999 1.000
> 0.1 1.000 1.000 1.001 1.005 1.006
> 0.2 1.000 1.001 1.004 1.008 1.014
> 0.3 1.000 1.002 1.006 1.012 1.021
> 0.4 1.000 1.003 1.009 1.015 1.028
> 0.5 1.001 1.005 1.012 1.019 1.036
> ...
>
> y = 'y3'
> data = loadtxt('oaf.txt', skiprows=1)
> label = line.split()
> d = label[0]
this is supposed to be
   if d == 'x':
> if d == 'd':
> ? ? dlist = label
> x, y = data[:,0], data[:,dlist.index(y)]
> interp = interpolate.interp1d(x, y)
>
> and the third, is the 2d case.
>
> ? ? ? ?10 ? ? 20 ? ? ?30 ? ? 40 ? ? ?50
> 0.0 1.000 0.999 0.999 0.999 1.000
> 0.1 1.000 1.000 1.001 1.005 1.006
> 0.2 1.000 1.001 1.004 1.008 1.014
> 0.3 1.000 1.002 1.006 1.012 1.021
> 0.4 1.000 1.003 1.009 1.015 1.028
> 0.5 1.001 1.005 1.012 1.019 1.036
> ...
>
> not sure how to get the x, y and z's
>
> is the 1st and 2nd way the best way to do this, and can someone help with
> the third case.
>
> --
> "The game of science can accurately be described as a never-ending insult to
> human intelligence." - Jo?o Magueijo
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-u... at scipy.orghttp://projects.scipy.org/mailman/listinfo/scipy-user


From C.J.Lee at tnw.utwente.nl  Wed Jun 18 03:26:57 2008
From: C.J.Lee at tnw.utwente.nl (Chris Lee)
Date: Wed, 18 Jun 2008 09:26:57 +0200
Subject: [SciPy-user] netCDF files, ctypes, and numpy
Message-ID: <2ADCDFB4-9E69-4578-86E4-56C9F5DAE4BB@tnw.utwente.nl>

Hi All,

I am trying to develop a script to read and write netCDF files. I  
could use Scientific.IO to do this (and in fact I already do this on a  
linux box) but I would rather use a single script that talks directly  
to the library so that I avoid needing to ensure that Scientific,  
Numeric, and numpy are all installed on every system the code gets  
used on.

I have made a start, using ctypes, I can load the netCDF file, read  
the attributes of each variable, count the dimensions and the  
variables.... but there are a few things that baffle me...

The file I am working with has multiple variables that use a single  
dimension. When I try and discover which dimensions are used, or load  
values from the variable, the buffer remains null. It appears that  
ctypes uses create_string_buffer(some_size) for all buffers.

Does anyone have any experience with this?

Cheers
Chris


From robert.kern at gmail.com  Wed Jun 18 04:18:03 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Wed, 18 Jun 2008 03:18:03 -0500
Subject: [SciPy-user] netCDF files, ctypes, and numpy
In-Reply-To: <2ADCDFB4-9E69-4578-86E4-56C9F5DAE4BB@tnw.utwente.nl>
References: <2ADCDFB4-9E69-4578-86E4-56C9F5DAE4BB@tnw.utwente.nl>
Message-ID: <3d375d730806180118h6bc7f335hc4916f376d18be8a@mail.gmail.com>

On Wed, Jun 18, 2008 at 02:26, Chris Lee <C.J.Lee at tnw.utwente.nl> wrote:
> Hi All,
>
> I am trying to develop a script to read and write netCDF files. I
> could use Scientific.IO to do this (and in fact I already do this on a
> linux box) but I would rather use a single script that talks directly
> to the library so that I avoid needing to ensure that Scientific,
> Numeric, and numpy are all installed on every system the code gets
> used on.

Numeric is unnecessary with recent versions of Scientific.

pupynere is a pure-Python netcdf reader. You may want to look into
implementing writer capabilities for it without using the netcdf C
library.

http://pypi.python.org/pypi/pupynere/

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From c.j.lee at tnw.utwente.nl  Wed Jun 18 04:44:41 2008
From: c.j.lee at tnw.utwente.nl (Chris Lee)
Date: Wed, 18 Jun 2008 10:44:41 +0200
Subject: [SciPy-user] netCDF files, ctypes, and numpy
In-Reply-To: <3d375d730806180118h6bc7f335hc4916f376d18be8a@mail.gmail.com>
References: <2ADCDFB4-9E69-4578-86E4-56C9F5DAE4BB@tnw.utwente.nl>
	<3d375d730806180118h6bc7f335hc4916f376d18be8a@mail.gmail.com>
Message-ID: <C7F44985-8452-4907-9ABD-BA4239CA7D04@tnw.utwente.nl>


On Jun 18, 2008, at 10:18 AM, Robert Kern wrote:

> On Wed, Jun 18, 2008 at 02:26, Chris Lee <C.J.Lee at tnw.utwente.nl>  
> wrote:
>> Hi All,
>>
>> I am trying to develop a script to read and write netCDF files. I
>> could use Scientific.IO to do this (and in fact I already do this  
>> on a
>> linux box) but I would rather use a single script that talks directly
>> to the library so that I avoid needing to ensure that Scientific,
>> Numeric, and numpy are all installed on every system the code gets
>> used on.
>
> Numeric is unnecessary with recent versions of Scientific.

Oh, that is good to know--certainly the Scientific website recommends  
a particular version of Numeric.

>
>
> pupynere is a pure-Python netcdf reader. You may want to look into
> implementing writer capabilities for it without using the netcdf C
> library.
>
> http://pypi.python.org/pypi/pupynere/

A life-saver :) I had actually started making real progress.... but  
this is sooo much simpler :)
/me stops development on alpha version of the wheel
Thanks for your help
Chris


From lopmart at gmail.com  Wed Jun 18 05:30:21 2008
From: lopmart at gmail.com (Jose Lopez)
Date: Wed, 18 Jun 2008 02:30:21 -0700
Subject: [SciPy-user] error at optimize.fmin
Message-ID: <4eeef9d40806180230y72fd83fdu8bcce6b6ccdfa7ff@mail.gmail.com>

Hi, my code is the next and i  have a error, but i not know what i do:

from pylab import *
from scipy import *

def func(b,Hder):
    return (Hder[0]-(b[0]-b[1]))**2 + (Hder[1]-(b[1]-b[2]))**2+
(Hder[2]-(b[2]-b[3]))**2+ (Hder[3]-(b[3]-100.0))**2


b0=[0.0,0.0,0.0,0.0]
H0=[0.0,-50.0,20.0,-20.0]

xopt =optimize.fmin_l_bfgs_b(func,b0,args=(H0))



error is


Traceback (most recent call last):
  File "C:/Users/Valeria2/JL-MAESTRIA/programas 3
avance/resultado2/pruebafmin_l.py", line 13, in <module>
    xopt =optimize.fmin_l_bfgs_b(funcion,b0,args=(H0))
  File "C:\Python25\Lib\site-packages\scipy\optimize\lbfgsb.py", line 205,
in fmin_l_bfgs_b
    f, g = func_and_grad(x)
  File "C:\Python25\Lib\site-packages\scipy\optimize\lbfgsb.py", line 156,
in func_and_grad
    f, g = func(x, *args)
TypeError: funcion() takes exactly 2 arguments (5 given)


thanks
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080618/24930b82/attachment.html>

From lopmart at gmail.com  Wed Jun 18 05:29:02 2008
From: lopmart at gmail.com (Jose Lopez)
Date: Wed, 18 Jun 2008 02:29:02 -0700
Subject: [SciPy-user] (no subject)
Message-ID: <4eeef9d40806180229y6456cb59m75250b2eb007a75e@mail.gmail.com>

Hi, my code is the next and i  have a error, but i not know what i do:

from pylab import *
from scipy import *

def func(b,Hder):
    return (Hder[0]-(b[0]-b[1]))**2 + (Hder[1]-(b[1]-b[2]))**2+
(Hder[2]-(b[2]-b[3]))**2+ (Hder[3]-(b[3]-100.0))**2


b0=[0.0,0.0,0.0,0.0]
H0=[0.0,-50.0,20.0,-20.0]

xopt =optimize.fmin_l_bfgs_b(func,b0,args=(H0))



error is


Traceback (most recent call last):
  File "C:/Users/Valeria2/JL-MAESTRIA/programas 3
avance/resultado2/pruebafmin_l.py", line 13, in <module>
    xopt =optimize.fmin_l_bfgs_b(funcion,b0,args=(H0))
  File "C:\Python25\Lib\site-packages\scipy\optimize\lbfgsb.py", line 205,
in fmin_l_bfgs_b
    f, g = func_and_grad(x)
  File "C:\Python25\Lib\site-packages\scipy\optimize\lbfgsb.py", line 156,
in func_and_grad
    f, g = func(x, *args)
TypeError: funcion() takes exactly 2 arguments (5 given)


thanks
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080618/1e130899/attachment.html>

From alexander.borghgraef.rma at gmail.com  Wed Jun 18 08:07:08 2008
From: alexander.borghgraef.rma at gmail.com (Alexander Borghgraef)
Date: Wed, 18 Jun 2008 14:07:08 +0200
Subject: [SciPy-user] Mlab doesn't work
In-Reply-To: <20080613133323.GG3573@phare.normalesup.org>
References: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>
	<4852671C.9080402@gmail.com>
	<9e8c52a20806130534x12a6c087jf486daca9a28c6e1@mail.gmail.com>
	<20080613125150.GA3573@phare.normalesup.org>
	<9e8c52a20806130627ib41e905s3c92a7c10fbf4834@mail.gmail.com>
	<20080613133323.GG3573@phare.normalesup.org>
Message-ID: <9e8c52a20806180507u4e862eebt72d3eda59e42d9b2@mail.gmail.com>

I've been playing with mlab some more, and I ran into another problem.
I upgraded to enthought 2.7.1 to get rid of the mayavi.tools thing,
and tried plotting some things, which worked fine. Yay. Then I tried
to use mlab in my existing code, and I kept running into the following
error:

 python: Python/ceval.c:2624: PyEval_EvalCodeEx: Assertion `tstate !=
((void *)0)' failed.
 Abort

After commenting out nearly everyting, it seemed that there was some
incompatibility between mlab and pylab (which I use for 2D plotting).
For example, this works fine:

  from enthought.mayavi import mlab
  from scipy import lena
  mlab.surf(lena())

But this results in the above mentioned error:

  import pylab
  from enthought.mayavi import mlab
  from scipy import lena
  mlab.surf(lena())

All .py files were run with ipython -wthread. Calling an mlab function
when pylab is loaded results in this crash. Pylab functions OTOH are
fine when mayavi.mlab is loaded. Is this a known problem? Is this
being fixed in 3.0.1b? Is there a workaround?

-- 
Alex Borghgraef


From se.berg at stud.uni-goettingen.de  Wed Jun 18 08:08:09 2008
From: se.berg at stud.uni-goettingen.de (Sebastian Stephan Berg)
Date: Wed, 18 Jun 2008 14:08:09 +0200
Subject: [SciPy-user] error at optimize.fmin
In-Reply-To: <4eeef9d40806180230y72fd83fdu8bcce6b6ccdfa7ff@mail.gmail.com>
References: <4eeef9d40806180230y72fd83fdu8bcce6b6ccdfa7ff@mail.gmail.com>
Message-ID: <1213790889.5906.17.camel@sebook>

Hi,

very simple little error ;). Python will see (H0) as H0 and not a tuple
with the list as element. Using args=(H0,) should fix it.

Sebastian



From gael.varoquaux at normalesup.org  Wed Jun 18 09:01:37 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Wed, 18 Jun 2008 15:01:37 +0200
Subject: [SciPy-user] netCDF files, ctypes, and numpy
In-Reply-To: <C7F44985-8452-4907-9ABD-BA4239CA7D04@tnw.utwente.nl>
References: <2ADCDFB4-9E69-4578-86E4-56C9F5DAE4BB@tnw.utwente.nl>
	<3d375d730806180118h6bc7f335hc4916f376d18be8a@mail.gmail.com>
	<C7F44985-8452-4907-9ABD-BA4239CA7D04@tnw.utwente.nl>
Message-ID: <20080618130137.GB7214@phare.normalesup.org>

On Wed, Jun 18, 2008 at 10:44:41AM +0200, Chris Lee wrote:
> > Numeric is unnecessary with recent versions of Scientific.

> Oh, that is good to know--certainly the Scientific website recommends  
> a particular version of Numeric.

Beware that you explicitely need to pass a build switch (can't numpy be
default ?)

Ga?l


From gael.varoquaux at normalesup.org  Wed Jun 18 09:17:29 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Wed, 18 Jun 2008 15:17:29 +0200
Subject: [SciPy-user] Mlab doesn't work
In-Reply-To: <9e8c52a20806180507u4e862eebt72d3eda59e42d9b2@mail.gmail.com>
References: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>
	<4852671C.9080402@gmail.com>
	<9e8c52a20806130534x12a6c087jf486daca9a28c6e1@mail.gmail.com>
	<20080613125150.GA3573@phare.normalesup.org>
	<9e8c52a20806130627ib41e905s3c92a7c10fbf4834@mail.gmail.com>
	<20080613133323.GG3573@phare.normalesup.org>
	<9e8c52a20806180507u4e862eebt72d3eda59e42d9b2@mail.gmail.com>
Message-ID: <20080618131729.GB10375@phare.normalesup.org>

On Wed, Jun 18, 2008 at 02:07:08PM +0200, Alexander Borghgraef wrote:
> After commenting out nearly everyting, it seemed that there was some
> incompatibility between mlab and pylab (which I use for 2D plotting).
> For example, this works fine:

>   from enthought.mayavi import mlab
>   from scipy import lena
>   mlab.surf(lena())

> But this results in the above mentioned error:

>   import pylab
>   from enthought.mayavi import mlab
>   from scipy import lena
>   mlab.surf(lena())

Interesting. I was not aware of this problem, but it partly makes sense.
The problem is that pylab, as you are probably using it, uses the TK
toolkit, whereas mayavi uses the Wx. Running both event loops in the same
time results in a nice segfault due to race conditions.

The solution is to have pylab use the Wx event loop. You can do this by
doing (before importing pylab):
"""
import matplotlib
matplotlib.use('WxAdd')
"""
Alternatively, in the recent versions of ipython, Fernando has made sure
that if you use both the "-pylab" switch and the "-wthread" switch to
ipython, this is enforced, and mayavi works fine.

HTH,

Ga?l


From c.j.lee at tnw.utwente.nl  Wed Jun 18 09:27:52 2008
From: c.j.lee at tnw.utwente.nl (Chris Lee)
Date: Wed, 18 Jun 2008 15:27:52 +0200
Subject: [SciPy-user] netCDF files, ctypes, and numpy
In-Reply-To: <20080618130137.GB7214@phare.normalesup.org>
References: <2ADCDFB4-9E69-4578-86E4-56C9F5DAE4BB@tnw.utwente.nl>
	<3d375d730806180118h6bc7f335hc4916f376d18be8a@mail.gmail.com>
	<C7F44985-8452-4907-9ABD-BA4239CA7D04@tnw.utwente.nl>
	<20080618130137.GB7214@phare.normalesup.org>
Message-ID: <695AF992-A9A5-4F38-B274-203E40E0E8C5@tnw.utwente.nl>


On Jun 18, 2008, at 3:01 PM, Gael Varoquaux wrote:

> On Wed, Jun 18, 2008 at 10:44:41AM +0200, Chris Lee wrote:
>>> Numeric is unnecessary with recent versions of Scientific.
>
>> Oh, that is good to know--certainly the Scientific website recommends
>> a particular version of Numeric.
>
> Beware that you explicitely need to pass a build switch (can't numpy  
> be
> default ?)

Well since discovering pupynere, I have no need for Scientific at all.  
Though, it should be noted that the current version of pupynere will  
only load netCDF files with a small number of variables. I think I  
have patched the code to fix that now. If I find that it works I will  
send the patched code on to the author.

Cheers
Chris


From ggellner at uoguelph.ca  Wed Jun 18 09:12:55 2008
From: ggellner at uoguelph.ca (Gabriel Gellner)
Date: Wed, 18 Jun 2008 09:12:55 -0400
Subject: [SciPy-user] error at optimize.fmin
In-Reply-To: <4eeef9d40806180230y72fd83fdu8bcce6b6ccdfa7ff@mail.gmail.com>
References: <4eeef9d40806180230y72fd83fdu8bcce6b6ccdfa7ff@mail.gmail.com>
Message-ID: <20080618131255.GA7844@basestar>

On Wed, Jun 18, 2008 at 02:30:21AM -0700, Jose Lopez wrote:
>    Hi, my code is the next and i  have a error, but i not know what i do:
> 
>    from pylab import *
>    from scipy import *
> 
>    def func(b,Hder):
>        return (Hder[0]-(b[0]-b[1]))**2 + (Hder[1]-(b[1]-b[2]))**2+
>    (Hder[2]-(b[2]-b[3]))**2+ (Hder[3]-(b[3]-100.0))**2
> 
> 
>    b0=[0.0,0.0,0.0,0.0]
>    H0=[0.0,-50.0,20.0,-20.0]
> 
>    xopt =optimize.fmin_l_bfgs_b(func
>    ,b0,args=(H0))
> 
>    error is
> 
>    Traceback (most recent call last):
>      File "C:/Users/Valeria2/JL-MAESTRIA/programas 3
>    avance/resultado2/pruebafmin_l.py", line 13, in <module>
>        xopt =optimize.fmin_l_bfgs_b(funcion,b0,args=(H0))
>      File "C:\Python25\Lib\site-packages\scipy\optimize\lbfgsb.py", line 205,
>    in fmin_l_bfgs_b
>        f, g = func_and_grad(x)
>      File "C:\Python25\Lib\site-packages\scipy\optimize\lbfgsb.py", line 156,
>    in func_and_grad
>        f, g = func(x, *args)
>    TypeError: funcion() takes exactly 2 arguments (5 given)
> 
>    thanks
> 
You have a couple problems.

the error is because you are passing a single item, H0, in parenthesis, which
is not the same as a tuple (as parenthesis are also used as grouping
specifiers), instead you need to do args=(H0,). If you add this you will get a
new traceback:

Traceback (most recent call last):
  File "opt_error.py", line 15, in <module>
    xopt = optimize.fmin_l_bfgs_b(func, b0, args=(H0,))
  File "/usr/lib/python2.5/site-packages/scipy/optimize/lbfgsb.py", line 205,
in fmin_l_bfgs_b
    f, g = func_and_grad(x)
  File "/usr/lib/python2.5/site-packages/scipy/optimize/lbfgsb.py", line 156,
in func_and_grad
    f, g = func(x, *args)
TypeError: 'numpy.float64' object is not iterable

Which is a cryptic way of noting that you are using the range bounded version
of bfgs, but you haven't provided bounds. If you meant to have the problem be
unconstrained, then you should have

xopt = optimize.fmin_bfgs(func, b0, args=(H0,))

And all will work. Otherwise provide bounds.

Okay so the problem is fixed, but to give another way of passing arguments,
which I find nicer (and is also the preferred way in modern matlab, using
function handles) . . .

Instead of using the args keyword for fmin_bfgs, make another function that
calls func with Hder bound to H0, like so:

def obj(b):
    return func(b, H0)

now just pass obj to fmin_bfgs. So the code would be:

from pylab import *
from scipy import *

def func(b,Hder):
    return (Hder[0]-(b[0]-b[1]))**2 +
(Hder[1]-(b[1]-b[2]))**2+(Hder[2]-(b[2]-b[3]))**2+ (Hder[3]-(b[3]-100.0))**2


b0=[0.0,0.0,0.0,0.0]
H0=[0.0,-50.0,20.0,-20.0]

def obj(b):
    return func(b, H0)

xopt=optimize.fmin_bfgs(obj, b0)

I find this prettier, you can also do it more concisely using lambda functions
(which makes it even closer to the matlab way), if you want to know how, send
me a message.

Gabriel


From gael.varoquaux at normalesup.org  Wed Jun 18 09:54:08 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Wed, 18 Jun 2008 15:54:08 +0200
Subject: [SciPy-user] error at optimize.fmin
In-Reply-To: <1213790889.5906.17.camel@sebook>
References: <4eeef9d40806180230y72fd83fdu8bcce6b6ccdfa7ff@mail.gmail.com>
	<1213790889.5906.17.camel@sebook>
Message-ID: <20080618135408.GD10375@phare.normalesup.org>

On Wed, Jun 18, 2008 at 02:08:09PM +0200, Sebastian Stephan Berg wrote:
> very simple little error ;). Python will see (H0) as H0 and not a tuple
> with the list as element. Using args=(H0,) should fix it.

There is more to it. You'll find another puzzling error once you fix this
one. If you read carefully the docs to optimize.fmin_l_bfgs_b, you can
see that if you do not specify the gradient of the function, fprime, func
should return two arguments: the value of the function you want to
optimize, and the gradient. If you don't want this behavior, you should
set "approx_grad" to true.

HTH,

Ga?l


From lopmart at gmail.com  Wed Jun 18 11:39:01 2008
From: lopmart at gmail.com (Jose Lopez)
Date: Wed, 18 Jun 2008 08:39:01 -0700
Subject: [SciPy-user] error at optimize.fmin
In-Reply-To: <20080618131255.GA7844@basestar>
References: <4eeef9d40806180230y72fd83fdu8bcce6b6ccdfa7ff@mail.gmail.com>
	<20080618131255.GA7844@basestar>
Message-ID: <4eeef9d40806180839l621b74b1lb9495004046ef656@mail.gmail.com>

hi

thanks for the answer. I want to know how to use the lambda functions.
thanks

and by the way, how i to provide bounds to optimize.fmin_l_bfgs_b? in my
code, each element of b are [0,1].

thaks

atte

JL



On Wed, Jun 18, 2008 at 6:12 AM, Gabriel Gellner <ggellner at uoguelph.ca>
wrote:

> On Wed, Jun 18, 2008 at 02:30:21AM -0700, Jose Lopez wrote:
> >    Hi, my code is the next and i  have a error, but i not know what i do:
> >
> >    from pylab import *
> >    from scipy import *
> >
> >    def func(b,Hder):
> >        return (Hder[0]-(b[0]-b[1]))**2 + (Hder[1]-(b[1]-b[2]))**2+
> >    (Hder[2]-(b[2]-b[3]))**2+ (Hder[3]-(b[3]-100.0))**2
> >
> >
> >    b0=[0.0,0.0,0.0,0.0]
> >    H0=[0.0,-50.0,20.0,-20.0]
> >
> >    xopt =optimize.fmin_l_bfgs_b(func
> >    ,b0,args=(H0))
> >
> >    error is
> >
> >    Traceback (most recent call last):
> >      File "C:/Users/Valeria2/JL-MAESTRIA/programas 3
> >    avance/resultado2/pruebafmin_l.py", line 13, in <module>
> >        xopt =optimize.fmin_l_bfgs_b(funcion,b0,args=(H0))
> >      File "C:\Python25\Lib\site-packages\scipy\optimize\lbfgsb.py", line
> 205,
> >    in fmin_l_bfgs_b
> >        f, g = func_and_grad(x)
> >      File "C:\Python25\Lib\site-packages\scipy\optimize\lbfgsb.py", line
> 156,
> >    in func_and_grad
> >        f, g = func(x, *args)
> >    TypeError: funcion() takes exactly 2 arguments (5 given)
> >
> >    thanks
> >
> You have a couple problems.
>
> the error is because you are passing a single item, H0, in parenthesis,
> which
> is not the same as a tuple (as parenthesis are also used as grouping
> specifiers), instead you need to do args=(H0,). If you add this you will
> get a
> new traceback:
>
> Traceback (most recent call last):
>   File "opt_error.py", line 15, in <module>
>    xopt = optimize.fmin_l_bfgs_b(func, b0, args=(H0,))
>  File "/usr/lib/python2.5/site-packages/scipy/optimize/lbfgsb.py", line
> 205,
> in fmin_l_bfgs_b
>    f, g = func_and_grad(x)
>   File "/usr/lib/python2.5/site-packages/scipy/optimize/lbfgsb.py", line
> 156,
> in func_and_grad
>    f, g = func(x, *args)
> TypeError: 'numpy.float64' object is not iterable
>
> Which is a cryptic way of noting that you are using the range bounded
> version
> of bfgs, but you haven't provided bounds. If you meant to have the problem
> be
> unconstrained, then you should have
>
> xopt = optimize.fmin_bfgs(func, b0, args=(H0,))
>
> And all will work. Otherwise provide bounds.
>
> Okay so the problem is fixed, but to give another way of passing arguments,
> which I find nicer (and is also the preferred way in modern matlab, using
> function handles) . . .
>
> Instead of using the args keyword for fmin_bfgs, make another function that
> calls func with Hder bound to H0, like so:
>
> def obj(b):
>    return func(b, H0)
>
> now just pass obj to fmin_bfgs. So the code would be:
>
> from pylab import *
> from scipy import *
>
> def func(b,Hder):
>    return (Hder[0]-(b[0]-b[1]))**2 +
> (Hder[1]-(b[1]-b[2]))**2+(Hder[2]-(b[2]-b[3]))**2+
> (Hder[3]-(b[3]-100.0))**2
>
>
> b0=[0.0,0.0,0.0,0.0]
> H0=[0.0,-50.0,20.0,-20.0]
>
> def obj(b):
>    return func(b, H0)
>
> xopt=optimize.fmin_bfgs(obj, b0)
>
> I find this prettier, you can also do it more concisely using lambda
> functions
> (which makes it even closer to the matlab way), if you want to know how,
> send
> me a message.
>
> Gabriel
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080618/ad619583/attachment.html>

From peridot.faceted at gmail.com  Wed Jun 18 11:47:05 2008
From: peridot.faceted at gmail.com (Anne Archibald)
Date: Wed, 18 Jun 2008 09:47:05 -0600
Subject: [SciPy-user] (no subject)
In-Reply-To: <4eeef9d40806180229y6456cb59m75250b2eb007a75e@mail.gmail.com>
References: <4eeef9d40806180229y6456cb59m75250b2eb007a75e@mail.gmail.com>
Message-ID: <ce557a360806180847p2eae0f86gaf7051274d6f93d8@mail.gmail.com>

2008/6/18 Jose Lopez <lopmart at gmail.com>:
> Hi, my code is the next and i  have a error, but i not know what i do:
>
> from pylab import *
> from scipy import *
>
> def func(b,Hder):
>     return (Hder[0]-(b[0]-b[1]))**2 + (Hder[1]-(b[1]-b[2]))**2+
> (Hder[2]-(b[2]-b[3]))**2+ (Hder[3]-(b[3]-100.0))**2
>
>
> b0=[0.0,0.0,0.0,0.0]
> H0=[0.0,-50.0,20.0,-20.0]
>
> xopt =optimize.fmin_l_bfgs_b(func,b0,args=(H0))

args should be a list or tuple of extra arguments to pass to func. In
python syntax, (H0) is the same thing has H0, which is a list with
four elements, so fmin_l_bfgs_b treats it as four additional
arguments, rather than one argument that is a list. You should instead
provide args= with a list or tuple of only one element. The most
readable way to do this is to write args=[H0], but if you want to use
a tuple, you should write (H0,) (that is, with a trailing comma, to
indicate that it's a tuple).

Anne


From hetland at tamu.edu  Wed Jun 18 11:48:01 2008
From: hetland at tamu.edu (Rob Hetland)
Date: Wed, 18 Jun 2008 17:48:01 +0200
Subject: [SciPy-user] netCDF files, ctypes, and numpy
In-Reply-To: <695AF992-A9A5-4F38-B274-203E40E0E8C5@tnw.utwente.nl>
References: <2ADCDFB4-9E69-4578-86E4-56C9F5DAE4BB@tnw.utwente.nl>
	<3d375d730806180118h6bc7f335hc4916f376d18be8a@mail.gmail.com>
	<C7F44985-8452-4907-9ABD-BA4239CA7D04@tnw.utwente.nl>
	<20080618130137.GB7214@phare.normalesup.org>
	<695AF992-A9A5-4F38-B274-203E40E0E8C5@tnw.utwente.nl>
Message-ID: <205150A8-AB0F-4324-8D7C-6ECB4C136086@tamu.edu>


On Jun 18, 2008, at 3:27 PM, Chris Lee wrote:
> Well since discovering pupynere, I have no need for Scientific at all.
> Though, it should be noted that the current version of pupynere will
> only load netCDF files with a small number of variables. I think I
> have patched the code to fix that now. If I find that it works I will
> send the patched code on to the author.

For writing (and reading netCDF3 and netCDF4 files), check out

http://netcdf4-python.googlecode.com

This package can now be installed linking to either the netcdf3 or  
netcdf4 libraries (which need to be installed prior).

-Rob

----
Rob Hetland, Associate Professor
Dept. of Oceanography, Texas A&M University
http://pong.tamu.edu/~rob
phone: 979-458-0096, fax: 979-845-6331





From hetland at tamu.edu  Wed Jun 18 11:48:21 2008
From: hetland at tamu.edu (Rob Hetland)
Date: Wed, 18 Jun 2008 17:48:21 +0200
Subject: [SciPy-user] netCDF files, ctypes, and numpy
In-Reply-To: <695AF992-A9A5-4F38-B274-203E40E0E8C5@tnw.utwente.nl>
References: <2ADCDFB4-9E69-4578-86E4-56C9F5DAE4BB@tnw.utwente.nl>
	<3d375d730806180118h6bc7f335hc4916f376d18be8a@mail.gmail.com>
	<C7F44985-8452-4907-9ABD-BA4239CA7D04@tnw.utwente.nl>
	<20080618130137.GB7214@phare.normalesup.org>
	<695AF992-A9A5-4F38-B274-203E40E0E8C5@tnw.utwente.nl>
Message-ID: <8D7591FA-7573-4F56-8651-8DAE3E9CF4F2@tamu.edu>


On Jun 18, 2008, at 3:27 PM, Chris Lee wrote:
> Well since discovering pupynere, I have no need for Scientific at all.
> Though, it should be noted that the current version of pupynere will
> only load netCDF files with a small number of variables. I think I
> have patched the code to fix that now. If I find that it works I will
> send the patched code on to the author.

For writing (and reading netCDF3 and netCDF4 files), check out

http://netcdf4-python.googlecode.com

This package can now be installed linking to either the netcdf3 or  
netcdf4 libraries (which need to be installed prior).

-Rob

----
Rob Hetland, Associate Professor
Dept. of Oceanography, Texas A&M University
http://pong.tamu.edu/~rob
phone: 979-458-0096, fax: 979-845-6331





From rmay31 at gmail.com  Wed Jun 18 20:20:30 2008
From: rmay31 at gmail.com (Ryan May)
Date: Wed, 18 Jun 2008 20:20:30 -0400
Subject: [SciPy-user] Mlab doesn't work
In-Reply-To: <20080618131729.GB10375@phare.normalesup.org>
References: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>	<4852671C.9080402@gmail.com>	<9e8c52a20806130534x12a6c087jf486daca9a28c6e1@mail.gmail.com>	<20080613125150.GA3573@phare.normalesup.org>	<9e8c52a20806130627ib41e905s3c92a7c10fbf4834@mail.gmail.com>	<20080613133323.GG3573@phare.normalesup.org>	<9e8c52a20806180507u4e862eebt72d3eda59e42d9b2@mail.gmail.com>
	<20080618131729.GB10375@phare.normalesup.org>
Message-ID: <4859A64E.2070206@gmail.com>

Gael Varoquaux wrote:
> On Wed, Jun 18, 2008 at 02:07:08PM +0200, Alexander Borghgraef wrote:
>> After commenting out nearly everyting, it seemed that there was some
>> incompatibility between mlab and pylab (which I use for 2D plotting).
>> For example, this works fine:
> 
>>   from enthought.mayavi import mlab
>>   from scipy import lena
>>   mlab.surf(lena())
> 
>> But this results in the above mentioned error:
> 
>>   import pylab
>>   from enthought.mayavi import mlab
>>   from scipy import lena
>>   mlab.surf(lena())
> 
> Interesting. I was not aware of this problem, but it partly makes sense.
> The problem is that pylab, as you are probably using it, uses the TK
> toolkit, whereas mayavi uses the Wx. Running both event loops in the same
> time results in a nice segfault due to race conditions.
> 
> The solution is to have pylab use the Wx event loop. You can do this by
> doing (before importing pylab):
> """
> import matplotlib
> matplotlib.use('WxAdd')
		    ^^^

You mean:
matplotlib.use('WxAgg')

Ryan

-- 
Ryan May
Graduate Research Assistant
School of Meteorology
University of Oklahoma


From gael.varoquaux at normalesup.org  Wed Jun 18 20:26:11 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Thu, 19 Jun 2008 02:26:11 +0200
Subject: [SciPy-user] Mlab doesn't work
In-Reply-To: <4859A64E.2070206@gmail.com>
References: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>
	<4852671C.9080402@gmail.com>
	<9e8c52a20806130534x12a6c087jf486daca9a28c6e1@mail.gmail.com>
	<20080613125150.GA3573@phare.normalesup.org>
	<9e8c52a20806130627ib41e905s3c92a7c10fbf4834@mail.gmail.com>
	<20080613133323.GG3573@phare.normalesup.org>
	<9e8c52a20806180507u4e862eebt72d3eda59e42d9b2@mail.gmail.com>
	<20080618131729.GB10375@phare.normalesup.org>
	<4859A64E.2070206@gmail.com>
Message-ID: <20080619002611.GA1641@phare.normalesup.org>

On Wed, Jun 18, 2008 at 08:20:30PM -0400, Ryan May wrote:
> > The solution is to have pylab use the Wx event loop. You can do this by
> > doing (before importing pylab):
> > """
> > import matplotlib
> > matplotlib.use('WxAdd')
> 		    ^^^

> You mean:
> matplotlib.use('WxAgg')

Indeed,

Thank you Ryan.


From alexander.borghgraef.rma at gmail.com  Thu Jun 19 05:01:46 2008
From: alexander.borghgraef.rma at gmail.com (Alexander Borghgraef)
Date: Thu, 19 Jun 2008 11:01:46 +0200
Subject: [SciPy-user] Mlab doesn't work
In-Reply-To: <20080619002611.GA1641@phare.normalesup.org>
References: <9e8c52a20806130516q2cab4081uaa2cce2ccc0b540f@mail.gmail.com>
	<4852671C.9080402@gmail.com>
	<9e8c52a20806130534x12a6c087jf486daca9a28c6e1@mail.gmail.com>
	<20080613125150.GA3573@phare.normalesup.org>
	<9e8c52a20806130627ib41e905s3c92a7c10fbf4834@mail.gmail.com>
	<20080613133323.GG3573@phare.normalesup.org>
	<9e8c52a20806180507u4e862eebt72d3eda59e42d9b2@mail.gmail.com>
	<20080618131729.GB10375@phare.normalesup.org>
	<4859A64E.2070206@gmail.com>
	<20080619002611.GA1641@phare.normalesup.org>
Message-ID: <9e8c52a20806190201s3b72cbcbyc3bde4cab8628d9f@mail.gmail.com>

On Thu, Jun 19, 2008 at 2:26 AM, Gael Varoquaux
<gael.varoquaux at normalesup.org> wrote:
> On Wed, Jun 18, 2008 at 08:20:30PM -0400, Ryan May wrote:
>> You mean:
>> matplotlib.use('WxAgg')

Thanks, that does the trick.

-- 
Alex Borghgraef


From fredmfp at gmail.com  Fri Jun 20 09:30:29 2008
From: fredmfp at gmail.com (fred)
Date: Fri, 20 Jun 2008 15:30:29 +0200
Subject: [SciPy-user] msked array & histogram...
Message-ID: <485BB0F5.90204@gmail.com>

Hi,

It seems that numpy.histogram does not work with masked arrays.

Is there some workaround ?

TIA.

Cheers,

-- 
Fred


From fredmfp at gmail.com  Fri Jun 20 09:45:07 2008
From: fredmfp at gmail.com (fred)
Date: Fri, 20 Jun 2008 15:45:07 +0200
Subject: [SciPy-user] msked array & histogram...
In-Reply-To: <485BB0F5.90204@gmail.com>
References: <485BB0F5.90204@gmail.com>
Message-ID: <485BB463.1020600@gmail.com>

fred a ?crit :

> Is there some workaround ?
Yes.
Convert them to NaN.

Sorry.

Cheers,

-- 
Fred


From y.copin at ipnl.in2p3.fr  Fri Jun 20 12:44:32 2008
From: y.copin at ipnl.in2p3.fr (Yannick Copin)
Date: Fri, 20 Jun 2008 18:44:32 +0200
Subject: [SciPy-user] fmin_tnc changes since scipy 0.6.0
Message-ID: <485BDE70.4020308@ipnl.in2p3.fr>

Hi,

I'm a bit puzzled: I had a code which worked fine using 
scipy.optimize.fmin_tnc with scipy 0.5.2. Unfortunately, since I upgraded 
scipy to 0.6.0, I hit some convergence issues...

Does it ring a bell to somebody? I understand I'm a bit vague for the moment 
(I did not look for a simple test case yet), but maybe somebody had this 
experience too. Besides the reverted output order, was there some significant 
algorithmic changes in fmin_tnc between 0.5.2 and 0.6.0?

Cheers.
-- 
    .~.   Yannick COPIN  (o:>*  Doctus cum libro
    /V\   Institut de physique nucleaire de Lyon (IN2P3 - France)
   // \\  Tel: (33/0) 472 431 968     AIM: YnCopin ICQ: 236931013
  /(   )\ http://snovae.in2p3.fr/ycopin/
   ^`~'^


From nwagner at iam.uni-stuttgart.de  Fri Jun 20 12:49:00 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Fri, 20 Jun 2008 18:49:00 +0200
Subject: [SciPy-user] fmin_tnc changes since scipy 0.6.0
In-Reply-To: <485BDE70.4020308@ipnl.in2p3.fr>
References: <485BDE70.4020308@ipnl.in2p3.fr>
Message-ID: <web-103617468@uni-stuttgart.de>

On Fri, 20 Jun 2008 18:44:32 +0200
  Yannick Copin <y.copin at ipnl.in2p3.fr> wrote:
> Hi,
> 
> I'm a bit puzzled: I had a code which worked fine using 
> scipy.optimize.fmin_tnc with scipy 0.5.2. Unfortunately, 
>since I upgraded 
> scipy to 0.6.0, I hit some convergence issues...
> 
> Does it ring a bell to somebody? I understand I'm a bit 
>vague for the moment 
> (I did not look for a simple test case yet), but maybe 
>somebody had this 
> experience too. Besides the reverted output order, was 
>there some significant 
> algorithmic changes in fmin_tnc between 0.5.2 and 0.6.0?
> 
> Cheers.
> -- 
>    .~.   Yannick COPIN  (o:>*  Doctus cum libro
>    /V\   Institut de physique nucleaire de Lyon (IN2P3 - 
>France)
>   // \\  Tel: (33/0) 472 431 968     AIM: YnCopin ICQ: 
>236931013
>  /(   )\ http://snovae.in2p3.fr/ycopin/
>   ^`~'^
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user


There is a new version of tnc (1.3) in svn

r2530 | jarrod.millman | 2007-01-11 07:19:21 +0100 (Do, 11 
Jan 2007) | 2 lines

initial check-in of tnc version 1.3

r3187 | dmitrey.kroshko | 2007-07-24 16:13:43 +0200 (Di, 
24 Jul 2007) | 2 lines

correct order of tnc return values + bugfix for tnc tests

  
Nils


From nwagner at iam.uni-stuttgart.de  Fri Jun 20 12:59:55 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Fri, 20 Jun 2008 18:59:55 +0200
Subject: [SciPy-user] fmin_tnc changes since scipy 0.6.0
In-Reply-To: <485BDE70.4020308@ipnl.in2p3.fr>
References: <485BDE70.4020308@ipnl.in2p3.fr>
Message-ID: <web-103617739@uni-stuttgart.de>

On Fri, 20 Jun 2008 18:44:32 +0200
  Yannick Copin <y.copin at ipnl.in2p3.fr> wrote:
> Hi,
> 
> I'm a bit puzzled: I had a code which worked fine using 
> scipy.optimize.fmin_tnc with scipy 0.5.2. Unfortunately, 
>since I upgraded 
> scipy to 0.6.0, I hit some convergence issues...
> 
> Does it ring a bell to somebody? I understand I'm a bit 
>vague for the moment 
> (I did not look for a simple test case yet), but maybe 
>somebody had this 
> experience too. Besides the reverted output order, was 
>there some significant 
> algorithmic changes in fmin_tnc between 0.5.2 and 0.6.0?
> 
> Cheers.
> -- 
>    .~.   Yannick COPIN  (o:>*  Doctus cum libro
>    /V\   Institut de physique nucleaire de Lyon (IN2P3 - 
>France)
>   // \\  Tel: (33/0) 472 431 968     AIM: YnCopin ICQ: 
>236931013
>  /(   )\ http://snovae.in2p3.fr/ycopin/
>   ^`~'^
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user

You might be also interested in

http://www.jeannot.org/~js/code/index.en.html
  
BTW, Openopt provides an interface to tnc as well.

p.solve('scipy_tnc')

Nils


From nwagner at iam.uni-stuttgart.de  Fri Jun 20 13:01:28 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Fri, 20 Jun 2008 19:01:28 +0200
Subject: [SciPy-user] fmin_tnc changes since scipy 0.6.0
In-Reply-To: <485BDE70.4020308@ipnl.in2p3.fr>
References: <485BDE70.4020308@ipnl.in2p3.fr>
Message-ID: <web-103617769@uni-stuttgart.de>

On Fri, 20 Jun 2008 18:44:32 +0200
  Yannick Copin <y.copin at ipnl.in2p3.fr> wrote:
> Hi,
> 
> I'm a bit puzzled: I had a code which worked fine using 
> scipy.optimize.fmin_tnc with scipy 0.5.2. Unfortunately, 
>since I upgraded 
> scipy to 0.6.0, I hit some convergence issues...
> 
> Does it ring a bell to somebody? I understand I'm a bit 
>vague for the moment 
> (I did not look for a simple test case yet), but maybe 
>somebody had this 
> experience too. Besides the reverted output order, was 
>there some significant 
> algorithmic changes in fmin_tnc between 0.5.2 and 0.6.0?
> 
> Cheers.
> -- 
>    .~.   Yannick COPIN  (o:>*  Doctus cum libro
>    /V\   Institut de physique nucleaire de Lyon (IN2P3 - 
>France)
>   // \\  Tel: (33/0) 472 431 968     AIM: YnCopin ICQ: 
>236931013
>  /(   )\ http://snovae.in2p3.fr/ycopin/
>   ^`~'^
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user

  
Here is just another link

http://openopt.blogspot.com/2007/08/tnc-connected-to-openopt.html


Nils


From yannick.copin at laposte.net  Fri Jun 20 15:26:48 2008
From: yannick.copin at laposte.net (Yannick Copin)
Date: Fri, 20 Jun 2008 19:26:48 +0000 (UTC)
Subject: [SciPy-user]
	=?utf-8?q?fmin=5Ftnc_changes_since_scipy_0=2E6=2E0?=
References: <485BDE70.4020308@ipnl.in2p3.fr>
Message-ID: <loom.20080620T191324-171@post.gmane.org>

Yannick Copin <y.copin <at> ipnl.in2p3.fr> writes:
> I'm a bit puzzled: I had a code which worked fine using 
> scipy.optimize.fmin_tnc with scipy 0.5.2. Unfortunately, since I upgraded 
> scipy to 0.6.0, I hit some convergence issues...
> 
> Does it ring a bell to somebody? I understand I'm a bit vague for the moment 
> (I did not look for a simple test case yet), but maybe somebody had this 
> experience too. Besides the reverted output order, was there some significant 
> algorithmic changes in fmin_tnc between 0.5.2 and 0.6.0?

To complete my previous email, it seems my convergence problems is related to
the new 'offsets' in fmin_tnc, which allows for fine-tuning of parameter
scaling. If I force these offsets to be all null (instead of their actual
defaults), I find results very similar to scipy 0.5.2 ones. 

PS: I suspect the TNC algo within OpenOpt is actually the one from Scipy, right?
So this would not solve my problem. Furthermore, I try to limit the number of
external packages to be needed by my code, so I restrict myself to routines from
numpy/scipy.




From rrothkop at MIT.EDU  Fri Jun 20 16:43:12 2008
From: rrothkop at MIT.EDU (rrothkop at MIT.EDU)
Date: Fri, 20 Jun 2008 16:43:12 -0400
Subject: [SciPy-user] Image Morphology
Message-ID: <20080620164312.re0odtu9fl08cg08@webmail.mit.edu>

Hello,
I am looking for python equivalents for the matlab functions BWselect and
BWmorph(skel). Can the outcomes of these functions be accomplished using
scipy.ndimage?
Thanks,
Rebecca


From nwagner at iam.uni-stuttgart.de  Fri Jun 20 16:56:19 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Fri, 20 Jun 2008 22:56:19 +0200
Subject: [SciPy-user] fmin_tnc changes since scipy 0.6.0
In-Reply-To: <loom.20080620T191324-171@post.gmane.org>
References: <485BDE70.4020308@ipnl.in2p3.fr>
	<loom.20080620T191324-171@post.gmane.org>
Message-ID: <web-103622918@uni-stuttgart.de>

On Fri, 20 Jun 2008 19:26:48 +0000 (UTC)
  Yannick Copin <yannick.copin at laposte.net> wrote:
> Yannick Copin <y.copin <at> ipnl.in2p3.fr> writes:
>> I'm a bit puzzled: I had a code which worked fine using 
>> scipy.optimize.fmin_tnc with scipy 0.5.2. Unfortunately, 
>>since I upgraded 
>> scipy to 0.6.0, I hit some convergence issues...
>> 
>> Does it ring a bell to somebody? I understand I'm a bit 
>>vague for the moment 
>> (I did not look for a simple test case yet), but maybe 
>>somebody had this 
>> experience too. Besides the reverted output order, was 
>>there some significant 
>> algorithmic changes in fmin_tnc between 0.5.2 and 0.6.0?
> 
> To complete my previous email, it seems my convergence 
>problems is related to
> the new 'offsets' in fmin_tnc, which allows for 
>fine-tuning of parameter
> scaling. If I force these offsets to be all null 
>(instead of their actual
> defaults), I find results very similar to scipy 0.5.2 
>ones. 
> 
> PS: I suspect the TNC algo within OpenOpt is actually 
>the one from Scipy, right?

Dmitrey should reply here since he is the author of 
openopt.

> So this would not solve my problem. Furthermore, I try 
>to limit the number of
> external packages to be needed by my code, so I restrict 
>myself to routines from
> numpy/scipy.
> 

IMHO, one should compare the results obtained by different 
optimizations tools, wrt. efficiency, convergence 
behaviour, etc.

openopt provides solvers for different optimizations 
tasks.

and it's easy to install

svn co http://svn.scipy.org/svn/scikits/trunk/openopt 
openopt

cd openopt
python setup.py install


Nils
  


From dmitrey.kroshko at scipy.org  Fri Jun 20 17:42:23 2008
From: dmitrey.kroshko at scipy.org (dmitrey)
Date: Sat, 21 Jun 2008 00:42:23 +0300
Subject: [SciPy-user] fmin_tnc changes since scipy 0.6.0
In-Reply-To: <web-103622918@uni-stuttgart.de>
References: <485BDE70.4020308@ipnl.in2p3.fr>	<loom.20080620T191324-171@post.gmane.org>
	<web-103622918@uni-stuttgart.de>
Message-ID: <485C243F.5030102@scipy.org>

Nils Wagner wrote:
> On Fri, 20 Jun 2008 19:26:48 +0000 (UTC)
>   Yannick Copin <yannick.copin at laposte.net> wrote:
>   
>> Yannick Copin <y.copin <at> ipnl.in2p3.fr> writes:
>>     
>>> I'm a bit puzzled: I had a code which worked fine using 
>>> scipy.optimize.fmin_tnc with scipy 0.5.2. Unfortunately, 
>>> since I upgraded 
>>> scipy to 0.6.0, I hit some convergence issues...
>>>
>>> Does it ring a bell to somebody? I understand I'm a bit 
>>> vague for the moment 
>>> (I did not look for a simple test case yet), but maybe 
>>> somebody had this 
>>> experience too. Besides the reverted output order, was 
>>> there some significant 
>>> algorithmic changes in fmin_tnc between 0.5.2 and 0.6.0?
>>>       
>> To complete my previous email, it seems my convergence 
>> problems is related to
>> the new 'offsets' in fmin_tnc, which allows for 
>> fine-tuning of parameter
>> scaling. If I force these offsets to be all null 
>> (instead of their actual
>> defaults), I find results very similar to scipy 0.5.2 
>> ones. 
>>
>> PS: I suspect the TNC algo within OpenOpt is actually 
>> the one from Scipy, right?
>>     
>
> Dmitrey should reply here since he is the author of 
> openopt.
>   
yes, indeed, versions are same since tnc is called from scipy.
>   
>> So this would not solve my problem. Furthermore, I try 
>> to limit the number of
>> external packages to be needed by my code, so I restrict 
>> myself to routines from
>> numpy/scipy.
>>
>>     
>
> IMHO, one should compare the results obtained by different 
> optimizations tools, wrt. efficiency, convergence 
> behaviour, etc.
>   
Maybe Yannick's demands are completely covered by scipy solvers, so 
making additional dependencies really are not appreciated here. First of 
I write  openopt for some classes of problems beyond scipy could solve.
D.
> Nils
>   



From listservs at mac.com  Fri Jun 20 19:51:25 2008
From: listservs at mac.com (Chris)
Date: Fri, 20 Jun 2008 23:51:25 +0000 (UTC)
Subject: [SciPy-user] =?utf-8?q?weave=3A_ambiguous_overload_for_=E2=80=98o?=
	=?utf-8?b?cGVyYXRvcirigJk=?=
Message-ID: <loom.20080620T234907-384@post.gmane.org>

I have some pretty simple weave code that multiplies floats:

/* Calculate new Q-value */
qval_raw += alpha * delta * traceval_raw;

all 3 elements are floats -- I have checked this. Yet, I
get the following error:

ReinforcementLearning.py:461: error: ambiguous overload 
for ?operator*? in ?alpha * delta?

I dont see anything ambiguous here -- any ideas?

Thanks,
cf




From brian.lewis17 at gmail.com  Fri Jun 20 21:48:15 2008
From: brian.lewis17 at gmail.com (Brian Lewis)
Date: Fri, 20 Jun 2008 18:48:15 -0700
Subject: [SciPy-user] Naive Question about Data Representations
Message-ID: <b26744770806201848p417ceee6rcb1f5245c34432da@mail.gmail.com>

Sorry for my naivety.

Can someone explain how it is possible to store more than 1 32-bit integer
on a 32-bit system?  I hope this questions makes my confusion obvious.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080620/eb2d690d/attachment.html>

From robert.kern at gmail.com  Fri Jun 20 22:41:29 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Fri, 20 Jun 2008 21:41:29 -0500
Subject: [SciPy-user] Naive Question about Data Representations
In-Reply-To: <b26744770806201848p417ceee6rcb1f5245c34432da@mail.gmail.com>
References: <b26744770806201848p417ceee6rcb1f5245c34432da@mail.gmail.com>
Message-ID: <3d375d730806201941r6f148834ubfc10a3c3fc3710d@mail.gmail.com>

On Fri, Jun 20, 2008 at 20:48, Brian Lewis <brian.lewis17 at gmail.com> wrote:
> Sorry for my naivety.
>
> Can someone explain how it is possible to store more than 1 32-bit integer
> on a 32-bit system?  I hope this questions makes my confusion obvious.

*That* you're confused, yes. *How* you are confused, no.

Can you give us a little more context? What are you trying to do?

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From dwf at cs.toronto.edu  Fri Jun 20 23:10:39 2008
From: dwf at cs.toronto.edu (David Warde-Farley)
Date: Fri, 20 Jun 2008 23:10:39 -0400
Subject: [SciPy-user] Naive Question about Data Representations
In-Reply-To: <b26744770806201848p417ceee6rcb1f5245c34432da@mail.gmail.com>
References: <b26744770806201848p417ceee6rcb1f5245c34432da@mail.gmail.com>
Message-ID: <BB75F63F-B0BB-43F9-8547-976C3334CA5C@cs.toronto.edu>

On 20-Jun-08, at 9:48 PM, Brian Lewis wrote:

> Sorry for my naivety.
>
> Can someone explain how it is possible to store more than 1 32-bit  
> integer on a 32-bit system?  I hope this questions makes my  
> confusion obvious.

If by "32-bit system" you mean "32-bit processor/operating system",  
this refers to the number of bits in a memory address, and by  
extension that the system can deal with 2^32 distinct memory  
addresses / locations (4 GB on a 32-bit system, though you can't  
usually use a full 4GB of physical memory since lots of addresses are  
reserved for use by the system for things like I/O devices).

A 32-bit integer will occupy 4 bytes of system memory (8 bits per  
byte), and any valid memory address will have to be such a 32-bit  
integer, but you could store much, much bigger numbers by just  
occupying more bytes of memory. For example, the long long type in C  
occupies 8 bytes = 64 bits.

While any given programming language may not support it directly, you  
could store an integer as big as your system's memory would allow, for  
example you could allocate 4 megabytes and use it to store a 33554432  
bit number if you wanted, or 2 gigabytes and store a 17179869184-bit  
number. Working with it in your programs wouldn't be pleasant, but it  
would be possible.

David


From brian.lewis17 at gmail.com  Sat Jun 21 00:47:07 2008
From: brian.lewis17 at gmail.com (Brian Lewis)
Date: Fri, 20 Jun 2008 21:47:07 -0700
Subject: [SciPy-user] Naive Question about Data Representations
In-Reply-To: <BB75F63F-B0BB-43F9-8547-976C3334CA5C@cs.toronto.edu>
References: <b26744770806201848p417ceee6rcb1f5245c34432da@mail.gmail.com>
	<BB75F63F-B0BB-43F9-8547-976C3334CA5C@cs.toronto.edu>
Message-ID: <b26744770806202147n1622015jc7ad617cbdf9fa90@mail.gmail.com>

On Fri, Jun 20, 2008 at 8:10 PM, David Warde-Farley <dwf at cs.toronto.edu>
wrote:

> On 20-Jun-08, at 9:48 PM, Brian Lewis wrote:
>
> > Sorry for my naivety.
> >
> > Can someone explain how it is possible to store more than 1 32-bit
> > integer on a 32-bit system?  I hope this questions makes my
> > confusion obvious.
>
> If by "32-bit system" you mean "32-bit processor/operating system",
> this refers to the number of bits in a memory address, and by
> extension that the system can deal with 2^32 distinct memory
> addresses / locations (4 GB on a 32-bit system, though you can't
> usually use a full 4GB of physical memory since lots of addresses are
> reserved for use by the system for things like I/O devices).
>

How do we go from 2^32 addresses to 4 GiB?  To make this jump, it seems we
associate each address with 1 B (8 bits).  Then 2^32 = 4 * 2^30 = 4
Gibi-addresses = 4 GibiBytes.  I understand that now, but if we associated
each address with something larger than 1 B, then the upper limit would be
larger.  So why does 1 address == 1 B?  It seems that having a 32-bit
processor/operating system, alone, is not the only constraint on why there
is a 4 GiB upper bound.


>
> A 32-bit integer will occupy 4 bytes of system memory (8 bits per
> byte), and any valid memory address will have to be such a 32-bit
> integer, but you could store much, much bigger numbers by just
> occupying more bytes of memory. For example, the long long type in C
> occupies 8 bytes = 64 bits.
>

Now I understand that we can store 2^32 Bytes / 4 Bytes = 2^30 integers
(upper bound).  Previously, I would have said:  "We have 2^32 locations,
which is just 32 bits....each integer requires 32 bits....so we should only
be able to store one integer".  Obviously, I know this not to be the case,
and the missing link was that each location corresponded to 1 B.   But why?
If we could associate each address with 2 Bytes, shouldn't the upper bound
for a 32-bit system be 8 GiB instead?

Relevance: I'm trying to understand the largest array (upper bounds) I can
make on my system, and I am not so familiar with these things.

>>> import struct; print struct.calcsize('P')
4

This means each pointer will take up 4 Bytes. So it is the size of an
integer, and I should be able to store 2^30, 32-bit integers (on a sunny
day).  Approx: 4 GiB of RAM.

Thanks again for your patience.   Please correct my mistakes and if
possible, shed light on why each address represents 1 B.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080620/000cb532/attachment.html>

From peridot.faceted at gmail.com  Sat Jun 21 01:23:48 2008
From: peridot.faceted at gmail.com (Anne Archibald)
Date: Sat, 21 Jun 2008 01:23:48 -0400
Subject: [SciPy-user] Naive Question about Data Representations
In-Reply-To: <b26744770806202147n1622015jc7ad617cbdf9fa90@mail.gmail.com>
References: <b26744770806201848p417ceee6rcb1f5245c34432da@mail.gmail.com>
	<BB75F63F-B0BB-43F9-8547-976C3334CA5C@cs.toronto.edu>
	<b26744770806202147n1622015jc7ad617cbdf9fa90@mail.gmail.com>
Message-ID: <ce557a360806202223w7f95ef75nf163822d08c22992@mail.gmail.com>

2008/6/21 Brian Lewis <brian.lewis17 at gmail.com>:

> How do we go from 2^32 addresses to 4 GiB?  To make this jump, it seems we
> associate each address with 1 B (8 bits).  Then 2^32 = 4 * 2^30 = 4
> Gibi-addresses = 4 GibiBytes.  I understand that now, but if we associated
> each address with something larger than 1 B, then the upper limit would be
> larger.  So why does 1 address == 1 B?  It seems that having a 32-bit
> processor/operating system, alone, is not the only constraint on why there
> is a 4 GiB upper bound.

> Now I understand that we can store 2^32 Bytes / 4 Bytes = 2^30 integers
> (upper bound).  Previously, I would have said:  "We have 2^32 locations,
> which is just 32 bits....each integer requires 32 bits....so we should only
> be able to store one integer".  Obviously, I know this not to be the case,
> and the missing link was that each location corresponded to 1 B.   But why?
> If we could associate each address with 2 Bytes, shouldn't the upper bound
> for a 32-bit system be 8 GiB instead?
>
> Relevance: I'm trying to understand the largest array (upper bounds) I can
> make on my system, and I am not so familiar with these things.
>
>>>> import struct; print struct.calcsize('P')
> 4
>
> This means each pointer will take up 4 Bytes. So it is the size of an
> integer, and I should be able to store 2^30, 32-bit integers (on a sunny
> day).  Approx: 4 GiB of RAM.
>
> Thanks again for your patience.   Please correct my mistakes and if
> possible, shed light on why each address represents 1 B.

It's a design decision. It's cumbersome to address memory on a finer
granularity than your addresses allow (though it is possible), so many
years ago computer designers settled on 8-bit bytes being the basic
unit of memory (for PCs). To get at memory in units smaller than a
byte you have to do bit-fiddling, which is slow and painful. Since
bytes are a reasonably common unit - for strings, for example - having
to do bit-fiddling to get at them would be a nuisance. That said, I
think some specialized architectures, for example some DSPs, do
exactly this, for efficiency. But since the only benefit would have
been a factor of four in address space, it didn't seem worth it -
especially as the decision was being made when a whole gigabyte of RAM
was barely imaginable. Since any change would have meant breaking
backward compatibility, nobody bothered until 64-bit addresses became
available, at which point it seemed moot again.

Anne


From dwf at cs.toronto.edu  Sat Jun 21 03:58:16 2008
From: dwf at cs.toronto.edu (David Warde-Farley)
Date: Sat, 21 Jun 2008 03:58:16 -0400
Subject: [SciPy-user] Naive Question about Data Representations
In-Reply-To: <b26744770806202147n1622015jc7ad617cbdf9fa90@mail.gmail.com>
References: <b26744770806201848p417ceee6rcb1f5245c34432da@mail.gmail.com>
	<BB75F63F-B0BB-43F9-8547-976C3334CA5C@cs.toronto.edu>
	<b26744770806202147n1622015jc7ad617cbdf9fa90@mail.gmail.com>
Message-ID: <AFBF4FD1-7215-433E-8058-095C8BC62161@cs.toronto.edu>

On 21-Jun-08, at 12:47 AM, Brian Lewis wrote:

> Now I understand that we can store 2^32 Bytes / 4 Bytes = 2^30  
> integers (upper bound).  Previously, I would have said:  "We have  
> 2^32 locations, which is just 32 bits....each integer requires 32  
> bits....so we should only be able to store one integer".  Obviously,  
> I know this not to be the case, and the missing link was that each  
> location corresponded to 1 B.   But why?  If we could associate each  
> address with 2 Bytes, shouldn't the upper bound for a 32-bit system  
> be 8 GiB instead?

Anne's reply is far more astute than I could ever manage; suffice it  
to say, for a variety of reasons (ASCII being one of them, I think),  
it's just The Way It Is on all modern CPUs, and changing it would  
require fabricating new chips and updating everything from the OS  
upward to work with the new addressing scheme.

> Relevance: I'm trying to understand the largest array (upper bounds)  
> I can make on my system, and I am not so familiar with these things.
>
> >>> import struct; print struct.calcsize('P')
> 4
>
> This means each pointer will take up 4 Bytes. So it is the size of  
> an integer, and I should be able to store 2^30, 32-bit integers (on  
> a sunny day).  Approx: 4 GiB of RAM.

Yes, though keep in mind that on modern multi-tasking OSes, you've got  
some overhead from the kernel and the always-running services.

Also there's virtual memory to think about, where all programs are  
tricked into thinking there's more memory than actually available, and  
the deficit is just handled by swapping out unused portions to disk  
(all transparent to the user). The address space limitation still  
holds though, in terms of the upper limit of what you can _address_  
(even though the totality of the memory your program uses may not  
physically be in RAM all the time).

David


From contact at pythonxy.com  Sat Jun 21 04:24:44 2008
From: contact at pythonxy.com (Pierre Raybaut)
Date: Sat, 21 Jun 2008 10:24:44 +0200
Subject: [SciPy-user] Advanced Image Processing with Python
In-Reply-To: <mailman.59127.1214017846.27196.scipy-user@scipy.org>
References: <mailman.59127.1214017846.27196.scipy-user@scipy.org>
Message-ID: <485CBACC.6050500@pythonxy.com>

Hi all,

I'm about to start a project at work and I would really appreciate to 
have your opinion.

The purpose of this project is to develop a modular image and signal 
processing software (I would really like it to be open-source but it's 
far from easy at my work, so I'll see in time) with an "object-oriented 
GUI" (i.e. from the user point of view) and an interactive console. 
Image processing features will be quite advanced (not just filters and 
basic operations) and signal processing features will be basic (just 
what's needed to process image profiles: smoothing, curve fitting, and 
so on).

I am sure that some of you are using advanced image processing tools in 
Python.
So, according to your experience in this field, what would be the most 
complete, advanced and up-to-date Python module for this application?

Apparently, VTK, OpenCV, and PIL (too basic?) are generally used in this 
field.

Thanks
Pierre


From contact at pythonxy.com  Sat Jun 21 06:09:36 2008
From: contact at pythonxy.com (Pierre Raybaut)
Date: Sat, 21 Jun 2008 12:09:36 +0200
Subject: [SciPy-user] [ Python(x,y) ] New release : 1.2.7
Message-ID: <485CD360.8000707@pythonxy.com>

Hi all,

Python(x,y) 1.2.7 is now available on http://www.pythonxy.com.

 From now on, two Python(x,y) versions will be released:
 > Python(x,y).exe : standard Windows installer (~250Mb)
 > Python(x,y)-Lite.exe : exactly the same installer but without Eclipse 
(~150Mb) - but please note that Eclipse (as other packages) is still 
optional in the standard installer

Linux version:
Thanks to a lot of discussions with some of you on Python(x,y) Google 
group or in private, we will distribute the linux version as a metapackage.

Changes history
06 -21 -2008 - Version 1.2.7 :

     * Added:
           o VPython 4.beta26 : creating 3D interactive models of 
physical systems
     * Updated:
           o Pydev 1.3.18
           o py2exe 0.6.8 (see release notes)
           o pywin32 2.11 (see release notes)
           o PySerial 2.3 (see release notes)

Regards,
Pierre Raybaut









From zachary.pincus at yale.edu  Sat Jun 21 09:53:28 2008
From: zachary.pincus at yale.edu (Zachary Pincus)
Date: Sat, 21 Jun 2008 09:53:28 -0400
Subject: [SciPy-user] Advanced Image Processing with Python
In-Reply-To: <485CBACC.6050500@pythonxy.com>
References: <mailman.59127.1214017846.27196.scipy-user@scipy.org>
	<485CBACC.6050500@pythonxy.com>
Message-ID: <CED9D23A-AA56-47B4-ACEE-AF0728BB1722@yale.edu>

Hello,

> I am sure that some of you are using advanced image processing tools  
> in
> Python.
> So, according to your experience in this field, what would be the most
> complete, advanced and up-to-date Python module for this application?
>
> Apparently, VTK, OpenCV, and PIL (too basic?) are generally used in  
> this
> field.

VTK is, as the name suggests, a visualization toolkit. It has some  
image-processing features, but by-and-large, they're the kind of thing  
you'd need for 2/3D image display (isocontouring, etc.). Its sister  
project, ITK, has more of the hard-core image-processing algorithms in  
general, and in particular, lots of tools for segmentation and  
registration.

A while ago, I participated in getting some modern python wrappers for  
ITK working. I'm not sure what the state of this work is now, but  
hopefully there should be some decent wrappers. (It's a pain, because  
ITK is designed almost completely around C++ templates, which are  
instantiated at compile-time, so making a complete run-time library  
for a dynamic language is rather tricky.)

There are some recent ctypes wrappers for OpenCV, but I haven't worked  
with those. OpenCV is sort of tricky because a lot of it was designed  
to be really low-level (so you could put it on stripped down or even  
embedded systems), but it's got good stuff.

I wind up using scipy.ndimage a lot because it is a pretty convenient  
library and does a lot of the basics, and in particular, has nice (if  
slow-ish) resampling routines.

Good luck,
Zach


From mconroy at uga.edu  Sat Jun 21 12:54:17 2008
From: mconroy at uga.edu (Mike Conroy)
Date: Sat, 21 Jun 2008 12:54:17 -0400
Subject: [SciPy-user] problems with python and numpy
Message-ID: <485D3239.7050800@uga.edu>

All- I am having some baffling conflicts between python and numpy that  
seem to have arisen in recent installation. These seem to center around 
the tempfile in the Python Lib
directory.

What happens is that when I open python in an directory other than c:\  
or perhaps c:\python25, and then attempt to load numpy (or pymc which 
also loads numpy) I get:
the following error from tempfile.py
"from random import Random as _Random   ImportError: cannot import name 
Random"

Because things work fine in C:\ and some other directories, I am betting 
this is a path issue. Here is my path in case anyone sees the issue>

PATH=C:\Python25\;c:\Python25\Lib;c:\program 
files\miktex2.5\miktex\bin;C:\WINDOWS\system32;C:\WINDOWS;C:\WINDOWS\system32\wbem;c:\python25;c:\python25\bin;c:\python24\enthought\mingw\bin;c:\python24\enthought\graphviz\bin;c:\program 
files\jedit;C:\Program Files\QuickTime\QTSystem\;C:\Program 
Files\Subversion\bin;C:\MinGW\bin;C:\Program 
Files\Intel\MKL\10.0.1.015\ia64\bin;C:\Program Files\Common 
Files\Intuit\QBPOSSDKRuntime

I have installed the following version of python and packages:

ActivePython-2.5.2.2-win32-x86
numpy-1.1.0-win32-superpack-python2.5
matplotlib-0.98.0.win32-py2.5.
scipy-0.6.0.win32-p3-py2.5
and the latest svn-updated pymc2.0

Thanks for any help. Meantime things work but what a pain to be stuck in 
one director

M




-- 
Dr. Michael J. Conroy
Adjunct Professor and Assistant Unit Leader
Georgia Cooperative Fish and Wildlife Research Unit
Warnell School of Forestry and Natural Resources
University of Georgia
Athens, GA 30602 USA
Off. 3-427 Forestry Building
tel. +706-542-1167
fax +706-542-8356

Unit web page
http://coopunit.forestry.uga.edu/unit_homepage
My web page
http://coopunit.forestry.uga.edu/unit_homepage/Conroy



From stefan at sun.ac.za  Sat Jun 21 13:01:21 2008
From: stefan at sun.ac.za (=?ISO-8859-1?Q?St=E9fan_van_der_Walt?=)
Date: Sat, 21 Jun 2008 19:01:21 +0200
Subject: [SciPy-user] Advanced Image Processing with Python
In-Reply-To: <485CBACC.6050500@pythonxy.com>
References: <mailman.59127.1214017846.27196.scipy-user@scipy.org>
	<485CBACC.6050500@pythonxy.com>
Message-ID: <9457e7c80806211001q7e9db52fs93ecf8dac31b5d4a@mail.gmail.com>

2008/6/21 Pierre Raybaut <contact at pythonxy.com>:
> The purpose of this project is to develop a modular image and signal
> processing software (I would really like it to be open-source but it's
> far from easy at my work, so I'll see in time) with an "object-oriented
> GUI" (i.e. from the user point of view) and an interactive console.

Traits is a fantastic way to build an object-oriented GUI.

> Image processing features will be quite advanced (not just filters and
> basic operations) and signal processing features will be basic (just
> what's needed to process image profiles: smoothing, curve fitting, and
> so on).

Do stay in contact with the list, because many of us have implemented
some of these features on our own.  If you release your work under a
free license, I'm sure you'll have no trouble finding contributors.
Enthought has set a stellar example of how to open software whilst
running a business.

Regards
St?fan


From Scott.Daniels at Acm.Org  Sat Jun 21 15:54:06 2008
From: Scott.Daniels at Acm.Org (Scott David Daniels)
Date: Sat, 21 Jun 2008 12:54:06 -0700
Subject: [SciPy-user] problems with python and numpy
In-Reply-To: <485D3239.7050800@uga.edu>
References: <485D3239.7050800@uga.edu>
Message-ID: <g3jlmg$nem$1@ger.gmane.org>

Mike Conroy wrote:
> ... when I open python in an directory other than c:\ or perhaps c:\python25,
 > and then attempt to load numpy (or pymc which also loads numpy) I get:
> the following error from tempfile.py
> "from random import Random as _Random   ImportError: cannot import name 
> Random"
> ...
Here's how to debug this:
Run python -v  (You'll get a lot of output -- -v shows every import).
Then to the prompt type "import numpy".  Somewhere near the end you
will find enough information about where things started going wrong.

-Scott



From haase at msg.ucsf.edu  Sat Jun 21 16:38:40 2008
From: haase at msg.ucsf.edu (Sebastian Haase)
Date: Sat, 21 Jun 2008 22:38:40 +0200
Subject: [SciPy-user] Advanced Image Processing with Python
In-Reply-To: <485CBACC.6050500@pythonxy.com>
References: <mailman.59127.1214017846.27196.scipy-user@scipy.org>
	<485CBACC.6050500@pythonxy.com>
Message-ID: <bc657ead0806211338w375a3118ia0225f86769eb7b7@mail.gmail.com>

On Sat, Jun 21, 2008 at 10:24 AM, Pierre Raybaut <contact at pythonxy.com> wrote:
> Hi all,
>
> I'm about to start a project at work and I would really appreciate to
> have your opinion.
>
> The purpose of this project is to develop a modular image and signal
> processing software (I would really like it to be open-source but it's
> far from easy at my work, so I'll see in time) with an "object-oriented
> GUI" (i.e. from the user point of view) and an interactive console.
> Image processing features will be quite advanced (not just filters and
> basic operations) and signal processing features will be basic (just
> what's needed to process image profiles: smoothing, curve fitting, and
> so on).
>
> I am sure that some of you are using advanced image processing tools in
> Python.
> So, according to your experience in this field, what would be the most
> complete, advanced and up-to-date Python module for this application?
>
> Apparently, VTK, OpenCV, and PIL (too basic?) are generally used in this
> field.
>

Hi Pierre,
You failed to say what you mean by "image" -- 2D or 3D or maybe even
sequences of the above ...
Color or gray-scale or multi-wavelength !?

How would your software compare to existing ones -- like ImageJ ?

- Sebastian Haase


From robert.kern at gmail.com  Sat Jun 21 19:39:30 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Sat, 21 Jun 2008 18:39:30 -0500
Subject: [SciPy-user] problems with python and numpy
In-Reply-To: <485D3239.7050800@uga.edu>
References: <485D3239.7050800@uga.edu>
Message-ID: <3d375d730806211639v6ad576ecn8d3126e11b3a1df7@mail.gmail.com>

On Sat, Jun 21, 2008 at 11:54, Mike Conroy <mconroy at uga.edu> wrote:
> All- I am having some baffling conflicts between python and numpy that
> seem to have arisen in recent installation. These seem to center around
> the tempfile in the Python Lib
> directory.
>
> What happens is that when I open python in an directory other than c:\
> or perhaps c:\python25, and then attempt to load numpy (or pymc which
> also loads numpy) I get:
> the following error from tempfile.py
> "from random import Random as _Random   ImportError: cannot import name
> Random"
>
> Because things work fine in C:\ and some other directories, I am betting
> this is a path issue. Here is my path in case anyone sees the issue>

You almost certainly have a module named random.py in the directories
where you see the problem. Python looks in the current directory for
modules to import before standard directories like c:\Python25\Lib
where the desired random.py module is.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From zachary.pincus at yale.edu  Sat Jun 21 22:56:59 2008
From: zachary.pincus at yale.edu (Zachary Pincus)
Date: Sat, 21 Jun 2008 22:56:59 -0400
Subject: [SciPy-user] Image Morphology
In-Reply-To: <20080620164312.re0odtu9fl08cg08@webmail.mit.edu>
References: <20080620164312.re0odtu9fl08cg08@webmail.mit.edu>
Message-ID: <F657F88F-37B2-4D58-882D-F44DD7F97A28@yale.edu>

Hi Rebecca,

> I am looking for python equivalents for the matlab functions  
> BWselect and
> BWmorph(skel). Can the outcomes of these functions be accomplished  
> using
> scipy.ndimage?

Could you describe those operations in a bit more detail? ndimage  
probably doesn't have any direct equivalent, but since many of the  
more-complex morphological operations are built out of simpler ones  
(ndimage has the full complement of "building block" morphological  
operators, more or less), it might be possible to get the same  
functionality out of scipy pretty easily.

Zach


From zhangchipr at gmail.com  Sun Jun 22 00:58:09 2008
From: zhangchipr at gmail.com (zhang chi)
Date: Sun, 22 Jun 2008 12:58:09 +0800
Subject: [SciPy-user] why my fft2 can't work?
Message-ID: <90c482ab0806212158p480316f4ne9263c849b4a9c77@mail.gmail.com>

from numpy import *
from scipy import fftpack


a = ones((64,64))
b = fftpack.fft2(a)

the system display:

Process Python ??? (core dumped)

why?

thank you
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080622/db2709e6/attachment.html>

From zhangchipr at gmail.com  Sun Jun 22 03:56:58 2008
From: zhangchipr at gmail.com (zhang chi)
Date: Sun, 22 Jun 2008 15:56:58 +0800
Subject: [SciPy-user] How to realize inverse DCT transformation by using
	scipy
Message-ID: <90c482ab0806220056l66986450wac3f0ee183703395@mail.gmail.com>

hi

How to realize inverse DCT transformation by using scipy?

Is there a function to realize this transformation in scipy?

Thank you very much
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080622/2fae6705/attachment.html>

From contact at pythonxy.com  Sun Jun 22 16:40:03 2008
From: contact at pythonxy.com (Pierre Raybaut)
Date: Sun, 22 Jun 2008 22:40:03 +0200
Subject: [SciPy-user] Advanced Image Processing with Python
In-Reply-To: <mailman.9.1214154004.15447.scipy-user@scipy.org>
References: <mailman.9.1214154004.15447.scipy-user@scipy.org>
Message-ID: <485EB8A3.9040400@pythonxy.com>

>
>> > Hi all,
>> >
>> > I'm about to start a project at work and I would really appreciate to
>> > have your opinion.
>> >
>> > The purpose of this project is to develop a modular image and signal
>> > processing software (I would really like it to be open-source but it's
>> > far from easy at my work, so I'll see in time) with an "object-oriented
>> > GUI" (i.e. from the user point of view) and an interactive console.
>> > Image processing features will be quite advanced (not just filters and
>> > basic operations) and signal processing features will be basic (just
>> > what's needed to process image profiles: smoothing, curve fitting, and
>> > so on).
>> >
>> > I am sure that some of you are using advanced image processing tools in
>> > Python.
>> > So, according to your experience in this field, what would be the most
>> > complete, advanced and up-to-date Python module for this application?
>> >
>> > Apparently, VTK, OpenCV, and PIL (too basic?) are generally used in this
>> > field.
>> >
>>     
>
> Hi Pierre,
> You failed to say what you mean by "image" -- 2D or 3D or maybe even
> sequences of the above ...
> Color or gray-scale or multi-wavelength !?
>
> How would your software compare to existing ones -- like ImageJ ?
>
> - Sebastian Haase
You're right, I've been intentionally vague, but I guess I could be more 
specific anyway.
I used the word "advanced" to illustrate the fact that it would be a 
scientific image processing tool, and not just a simple software to 
apply basic operations (maybe it was an unnecessary precaution since 
this is the *Sci*Py user list...). In fact, the "advanced" image 
processing would be based on some of our own algorithms which I will not 
explain here (and could never be a part of the open-source project if 
there is one). The rest of what I'm looking for is basically what you 
can find in MATLAB image processing toolbox, and some shape recognition 
algorithms eventually.

The main idea would be to develop a first software with commonly used 
image processing tools (basic ones: filters, geometric transformations, 
basic operations, histograms, profile extraction, fft, ...), and 
optional modules to extend its features.

This software will be used to process mainly gray-scale (and less often 
color) images, that can be stored in 2D data arrays (as well as 1D 
signals, but that's not a problem with Numpy/Scipy). So what I'm looking 
for would be an image processing library (and one only if possible) to 
handle this, and of course I'd prefer it to be widely used, still 
developed and maintained, with a great future ahead... the perfect library!

I didn't know ImageJ, it seems interesting, I'll have to take a closer 
look at it.

Thanks for your help
Pierre


From rmay31 at gmail.com  Sun Jun 22 17:45:38 2008
From: rmay31 at gmail.com (Ryan May)
Date: Sun, 22 Jun 2008 17:45:38 -0400
Subject: [SciPy-user] Help with 2D interpolation
Message-ID: <485EC802.3040105@gmail.com>

Hi,

Can anyone explain why this won't work:

import numpy as np
import scipy.interpolate as interp
from scipy.interpolate.fitpack2 import SmoothBivariateSpline
import matplotlib.pyplot as plt

x = np.linspace(-10, 10, 25)
y = np.linspace(-15, 15, 50)
X,Y = np.meshgrid(x,y)
Z = 1.5 * X**2 + 0.5 * Y**2

#tck = interp.bisplrep(X, Y, Z)
#z = interp.bisplev(X.flatten(), Y.flatten(), tck)

lut = SmoothBivariateSpline(X.ravel(), Y.ravel(), Z.ravel())
z = lut(X.ravel(), Y.ravel())

fig = plt.figure()
ax1 = fig.add_subplot(1,2,1)
ax1.pcolor(X,Y,Z)
ax1.set_title('Original')
ax2 = fig.add_subplot(1,2,2)
ax2.pcolor(X,Y,z.reshape(Z.shape))
ax2.set_title('Interpolated')
plt.show()


I get this error with scipy 0.6:
---------------------------------------------------------------------------
AssertionError                            Traceback (most recent call last)

/home/rmay/pypkg/scattering/test/test_interp.py in <module>()
      14
      15 lut = SmoothBivariateSpline(X.ravel(), Y.ravel(), Z.ravel())
---> 16 z = lut(X.ravel(), Y.ravel())
      17
      18 fig = plt.figure()

/usr/lib64/python2.5/site-packages/scipy/interpolate/fitpack2.py in 
__call__(self, x, y, mth)
     350             kx,ky = self.degrees
     351             z,ier = dfitpack.bispev(tx,ty,c,kx,ky,x,y)
--> 352             assert ier==0,'Invalid input: ier='+`ier`
     353             return z
     354         raise NotImplementedError


Any takers?

Ryan

-- 
Ryan May
Graduate Research Assistant
School of Meteorology
University of Oklahoma


From zachary.pincus at yale.edu  Sun Jun 22 18:10:14 2008
From: zachary.pincus at yale.edu (Zachary Pincus)
Date: Sun, 22 Jun 2008 18:10:14 -0400
Subject: [SciPy-user] Advanced Image Processing with Python
In-Reply-To: <485EB8A3.9040400@pythonxy.com>
References: <mailman.9.1214154004.15447.scipy-user@scipy.org>
	<485EB8A3.9040400@pythonxy.com>
Message-ID: <66DC272E-5676-4C7A-9CC8-68CB6CC45B73@yale.edu>

> The rest of what I'm looking for is basically what you
> can find in MATLAB image processing toolbox, and some shape  
> recognition
> algorithms eventually.
>
> The main idea would be to develop a first software with commonly used
> image processing tools (basic ones: filters, geometric  
> transformations,
> basic operations, histograms, profile extraction, fft, ...), and
> optional modules to extend its features.
>
> This software will be used to process mainly gray-scale (and less  
> often
> color) images, that can be stored in 2D data arrays (as well as 1D
> signals, but that's not a problem with Numpy/Scipy). So what I'm  
> looking
> for would be an image processing library (and one only if possible) to
> handle this, and of course I'd prefer it to be widely used, still
> developed and maintained, with a great future ahead... the perfect  
> library!

I think that OpenCV (with the ctypes python bindings) and  
scipy.ndimage should probably be pretty good candidates here...

Zach


From jjh at 42quarks.com  Mon Jun 23 01:59:53 2008
From: jjh at 42quarks.com (Jonathan Hunt)
Date: Mon, 23 Jun 2008 15:59:53 +1000
Subject: [SciPy-user] Test failure on OS X Leopard
Message-ID: <e26778a60806222259v584d1bc4ube1cf4ffa9b3e4d7@mail.gmail.com>

Hi,

I installed NumPy (1.1) and SciPy (0.6) both from source on OS X
(Leopard 10.5.3) . I removed existing numpy/scipy libraries. Both
install with no errors. And numpy.test() passes all tests. However,
running:

>>> import scipy
>>> scipy.test();
fails with:
....................................................................................................................warning:
specified build_dir '_bad_path_' does not exist or is not writable.
Trying default locations
.....warning: specified build_dir '_bad_path_' does not exist or is
not writable. Trying default locations
............................................................................................................................
======================================================================
FAIL: check_dot (scipy.lib.blas.tests.test_blas.test_fblas1_simple)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Library/Python/2.5/site-packages/scipy/lib/blas/tests/test_blas.py",
line 76, in check_dot
    assert_almost_equal(f([3j,-4,3-4j],[2,3,1]),-9+2j)
  File "/Library/Python/2.5/site-packages/numpy/testing/utils.py",
line 158, in assert_almost_equal
    assert round(abs(desired - actual),decimal) == 0, msg
AssertionError:
Items are not equal:
 ACTUAL: 2.1796638438433325e-36j
 DESIRED: (-9+2j)

----------------------------------------------------------------------
Ran 1574 tests in 5.679s

FAILED (failures=1)
<unittest._TextTestResult run=1574 errors=0 failures=1>


Is this bad? A problem? It seems to be related to LAPACK/BLAS but on
install SciPy said it detected LAPACK/BLAST.

Any help appreciated.

Thanks,
Jonny





Full test output is appended below:
Failed importing scipy.linsolve.umfpack: 'module' object has no
attribute 'umfpack'
  Found 9/9 tests for scipy.cluster.tests.test_vq
  Found 18/18 tests for scipy.fftpack.tests.test_basic
  Found 4/4 tests for scipy.fftpack.tests.test_helper
  Found 20/20 tests for scipy.fftpack.tests.test_pseudo_diffs
  Found 1/1 tests for scipy.integrate.tests.test_integrate
  Found 10/10 tests for scipy.integrate.tests.test_quadpack
  Found 3/3 tests for scipy.integrate.tests.test_quadrature
  Found 6/6 tests for scipy.tests.test_fitpack
  Found 6/6 tests for scipy.tests.test_interpolate
  Found 4/4 tests for scipy.io.tests.test_array_import
  Found 28/28 tests for scipy.io.tests.test_mio
  Found 13/13 tests for scipy.io.tests.test_mmio
  Found 5/5 tests for scipy.io.tests.test_npfile
  Found 4/4 tests for scipy.io.tests.test_recaster
  Found 16/16 tests for scipy.lib.blas.tests.test_blas
  Found 128/128 tests for scipy.lib.blas.tests.test_fblas

****************************************************************
WARNING: clapack module is empty
-----------
See scipy/INSTALL.txt for troubleshooting.
Notes:
* If atlas library is not found by numpy/distutils/system_info.py,
  then scipy uses flapack instead of clapack.
****************************************************************

  Found 42/42 tests for scipy.lib.lapack.tests.test_lapack
  Found 41/41 tests for scipy.linalg.tests.test_basic
<module 'scipy.linalg.fblas' from
'/Library/Python/2.5/site-packages/scipy/linalg/fblas.so'>
  Found 16/16 tests for scipy.linalg.tests.test_blas
  Found 72/72 tests for scipy.linalg.tests.test_decomp
  Found 128/128 tests for scipy.linalg.tests.test_fblas
  Found 6/6 tests for scipy.linalg.tests.test_iterative
  Found 4/4 tests for scipy.linalg.tests.test_lapack
  Found 7/7 tests for scipy.linalg.tests.test_matfuncs
Failed importing
/Library/Python/2.5/site-packages/scipy/linsolve/umfpack/tests/test_umfpack.py:
'module' object has no attribute 'umfpack'
  Found 2/2 tests for scipy.maxentropy.tests.test_maxentropy
Failed importing
/Library/Python/2.5/site-packages/scipy/misc/tests/test_pilutil.py: No
module named PIL.Image
  Found 399/399 tests for scipy.ndimage.tests.test_ndimage
  Found 5/5 tests for scipy.odr.tests.test_odr
  Found 1/1 tests for scipy.optimize.tests.test_cobyla
  Found 10/10 tests for scipy.optimize.tests.test_nonlin
  Found 8/8 tests for scipy.optimize.tests.test_optimize
  Found 4/4 tests for scipy.optimize.tests.test_zeros
  Found 5/5 tests for scipy.signal.tests.test_signaltools
  Found 4/4 tests for scipy.signal.tests.test_wavelets
  Found 152/152 tests for scipy.sparse.tests.test_sparse
  Found 342/342 tests for scipy.special.tests.test_basic
  Found 3/3 tests for scipy.special.tests.test_spfun_stats
  Found 73/73 tests for scipy.stats.tests.test_distributions
  Found 10/10 tests for scipy.stats.tests.test_morestats
  Found 107/107 tests for scipy.stats.tests.test_stats
  Found 1/1 tests for scipy.weave.tests.test_ast_tools
  Found 2/2 tests for scipy.weave.tests.test_blitz_tools
  Found 9/9 tests for scipy.weave.tests.test_build_tools
  Found 0/0 tests for scipy.weave.tests.test_c_spec
  Found 26/26 tests for scipy.weave.tests.test_catalog
building extensions here: /Users/uqjhunt2/.python25_compiled/m5
  Found 1/1 tests for scipy.weave.tests.test_ext_tools
  Found 0/0 tests for scipy.weave.tests.test_inline_tools
  Found 0/0 tests for scipy.weave.tests.test_scxx_dict
  Found 0/0 tests for scipy.weave.tests.test_scxx_object
  Found 0/0 tests for scipy.weave.tests.test_scxx_sequence
  Found 74/74 tests for scipy.weave.tests.test_size_check
  Found 16/16 tests for scipy.weave.tests.test_slice_handler
  Found 3/3 tests for scipy.weave.tests.test_standard_array_spec
  Found 0/0 tests for scipy.weave.tests.test_wx_spec
.../Library/Python/2.5/site-packages/scipy/cluster/vq.py:477:
UserWarning: One of the clusters is empty. Re-run kmean with a
different initialization.
  warnings.warn("One of the clusters is empty. "
exception raised as expected: One of the clusters is empty. Re-run
kmean with a different initialization.
................................................Residual: 1.05006987366e-07
............../Library/Python/2.5/site-packages/scipy/interpolate/fitpack2.py:458:
UserWarning:
The coefficients of the spline returned have been computed as the
minimal norm least-squares solution of a (numerically) rank deficient
system (deficiency=7). If deficiency is large, the results may be
inaccurate. Deficiency may strongly depend on the value of eps.
  warnings.warn(message)
............
Don't worry about a warning regarding the number of bytes read.
Warning: 1000000 bytes requested, 20 bytes read.
..............................................................F..........caxpy:n=4
..caxpy:n=3
....ccopy:n=4
..ccopy:n=3
.............cscal:n=4
....cswap:n=4
..cswap:n=3
.....daxpy:n=4
..daxpy:n=3
....dcopy:n=4
..dcopy:n=3
.............dscal:n=4
....dswap:n=4
..dswap:n=3
.....saxpy:n=4
..saxpy:n=3
....scopy:n=4
..scopy:n=3
.............sscal:n=4
....sswap:n=4
..sswap:n=3
.....zaxpy:n=4
..zaxpy:n=3
....zcopy:n=4
..zcopy:n=3
.............zscal:n=4
....zswap:n=4
..zswap:n=3
.....................................................................................
****************************************************************
WARNING: cblas module is empty
-----------
See scipy/INSTALL.txt for troubleshooting.
Notes:
* If atlas library is not found by numpy/distutils/system_info.py,
  then scipy uses fblas instead of cblas.
****************************************************************

...........................................................................................caxpy:n=4
..caxpy:n=3
....ccopy:n=4
..ccopy:n=3
.............cscal:n=4
....cswap:n=4
..cswap:n=3
.....daxpy:n=4
..daxpy:n=3
....dcopy:n=4
..dcopy:n=3
.............dscal:n=4
....dswap:n=4
..dswap:n=3
.....saxpy:n=4
..saxpy:n=3
....scopy:n=4
..scopy:n=3
.............sscal:n=4
....sswap:n=4
..sswap:n=3
.....zaxpy:n=4
..zaxpy:n=3
....zcopy:n=4
..zcopy:n=3
.............zscal:n=4
....zswap:n=4
..zswap:n=3
..........
****************************************************************
WARNING: clapack module is empty
-----------
See scipy/INSTALL.txt for troubleshooting.
Notes:
* If atlas library is not found by numpy/distutils/system_info.py,
  then scipy uses flapack instead of clapack.
****************************************************************

...Result may be inaccurate, approximate err = 1.23518201169e-08
...Result may be inaccurate, approximate err = 7.27595761418e-12
............................................................................................................/Library/Python/2.5/site-packages/scipy/ndimage/interpolation.py:41:
UserWarning: Mode "reflect" may yield incorrect results on boundaries.
Please use "mirror" instead.
  warnings.warn('Mode "reflect" may yield incorrect results on '
.....................................................................................................................................................................................................................................................................................................................................................................Use
minimum degree ordering on A'+A.
.....................................Use minimum degree ordering on A'+A.
.....................................Use minimum degree ordering on A'+A.
................................Use minimum degree ordering on A'+A.
....................................................................................................................................................................................................................................................................................................................................................0.2
0.2
0.2
......0.2
..0.2
0.2
0.2
0.2
0.2
..............................................................................................Ties
preclude use of exact statistic.
..Ties preclude use of exact statistic.
......................./Library/Python/2.5/site-packages/numpy/lib/function_base.py:166:
FutureWarning:
        The semantics of histogram will be modified in
        release 1.2 to improve outlier handling. The new behavior can be
        obtained using new=True. Note that the new version accepts/returns
        the bin edges instead of the left bin edges.
        Please read the docstring for more information.
  Please read the docstring for more information.""", FutureWarning)
/Library/Python/2.5/site-packages/numpy/lib/function_base.py:181:
FutureWarning:
                Outliers handling will change in version 1.2.
                Please read the docstring for details.
  Please read the docstring for details.""", FutureWarning)
.............................................................................................warning:
specified build_dir '_bad_path_' does not exist or is not writable.
Trying default locations
.....warning: specified build_dir '_bad_path_' does not exist or is
not writable. Trying default locations
............................................................................................................................
======================================================================
FAIL: check_dot (scipy.lib.blas.tests.test_blas.test_fblas1_simple)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/Library/Python/2.5/site-packages/scipy/lib/blas/tests/test_blas.py",
line 76, in check_dot
    assert_almost_equal(f([3j,-4,3-4j],[2,3,1]),-9+2j)
  File "/Library/Python/2.5/site-packages/numpy/testing/utils.py",
line 158, in assert_almost_equal
    assert round(abs(desired - actual),decimal) == 0, msg
AssertionError:
Items are not equal:
 ACTUAL: 8.5864323499206074e-37j
 DESIRED: (-9+2j)

----------------------------------------------------------------------
Ran 1848 tests in 6.136s

FAILED (failures=1)
<unittest._TextTestResult run=1848 errors=0 failures=1>
>>>


-- 
Jonathan J Hunt <jjh at 42quarks.com>
Homepage: http://www.42quarks.net.nz/wiki/JJH
(Further contact details there)
"Physics isn't the most important thing. Love is." Richard Feynman


From david at ar.media.kyoto-u.ac.jp  Mon Jun 23 08:02:00 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Mon, 23 Jun 2008 21:02:00 +0900
Subject: [SciPy-user] Test failure on OS X Leopard
In-Reply-To: <e26778a60806222259v584d1bc4ube1cf4ffa9b3e4d7@mail.gmail.com>
References: <e26778a60806222259v584d1bc4ube1cf4ffa9b3e4d7@mail.gmail.com>
Message-ID: <485F90B8.7080701@ar.media.kyoto-u.ac.jp>

Jonathan Hunt wrote:
> Is this bad? A problem? It seems to be related to LAPACK/BLAS but on
> install SciPy said it detected LAPACK/BLAST.
>   

Yes, this is potentially bad. It is more than likely caused by a buggy
interaction between C and Fortran. Which version of Mac OS X are you
using, and which fortran compiler are you using ? A full build log (from
scratch, that is after having removed build directory) would be useful,

cheers,

David


From jjh at 42quarks.com  Mon Jun 23 08:58:13 2008
From: jjh at 42quarks.com (Jonathan Hunt)
Date: Mon, 23 Jun 2008 22:58:13 +1000
Subject: [SciPy-user] Test failure on OS X Leopard
In-Reply-To: <e26778a60806230551w5cb2a6dby86bc211b686f945a@mail.gmail.com>
References: <e26778a60806222259v584d1bc4ube1cf4ffa9b3e4d7@mail.gmail.com>
	<485F90B8.7080701@ar.media.kyoto-u.ac.jp>
	<e26778a60806230551w5cb2a6dby86bc211b686f945a@mail.gmail.com>
Message-ID: <e26778a60806230558r422dedb9xb1549e83ad3708a1@mail.gmail.com>

Hi,

Build log is attached. I am running OS X 10.5.3 with gfortran 4.2.1
and gcc 4.0.1 (and python 2.5).

Thanks for any help.
Jonny

> On Mon, Jun 23, 2008 at 10:02 PM, David Cournapeau
> <david at ar.media.kyoto-u.ac.jp> wrote:
>> Jonathan Hunt wrote:
>>> Is this bad? A problem? It seems to be related to LAPACK/BLAS but on
>>> install SciPy said it detected LAPACK/BLAST.
>>>
>>
>> Yes, this is potentially bad. It is more than likely caused by a buggy
>> interaction between C and Fortran. Which version of Mac OS X are you
>> using, and which fortran compiler are you using ? A full build log (from
>> scratch, that is after having removed build directory) would be useful,
>
> --
> Jonathan J Hunt <jjh at 42quarks.com>
> Homepage: http://www.42quarks.net.nz/wiki/JJH
> (Further contact details there)
> "Physics isn't the most important thing. Love is." Richard Feynman
>



-- 
Jonathan J Hunt <jjh at 42quarks.com>
Homepage: http://www.42quarks.net.nz/wiki/JJH
(Further contact details there)
"Physics isn't the most important thing. Love is." Richard Feynman
-------------- next part --------------
A non-text attachment was scrubbed...
Name: build.log.bz2
Type: application/x-bzip2
Size: 25562 bytes
Desc: not available
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080623/242a15a4/attachment.bin>

From piscinero at gmail.com  Mon Jun 23 09:29:16 2008
From: piscinero at gmail.com (AndresGM)
Date: Mon, 23 Jun 2008 06:29:16 -0700 (PDT)
Subject: [SciPy-user] Test failure on OS X Leopard
In-Reply-To: <e26778a60806230558r422dedb9xb1549e83ad3708a1@mail.gmail.com>
References: <e26778a60806222259v584d1bc4ube1cf4ffa9b3e4d7@mail.gmail.com> 
	<485F90B8.7080701@ar.media.kyoto-u.ac.jp>
	<e26778a60806230551w5cb2a6dby86bc211b686f945a@mail.gmail.com> 
	<e26778a60806230558r422dedb9xb1549e83ad3708a1@mail.gmail.com>
Message-ID: <5f89059f-2266-47d8-bb7b-bfc34259ae8e@w7g2000hsa.googlegroups.com>

I'm getting a different error when I try to run the tests. I installed
numpy 1.1.0 from source and tried Scipy both 0.6 and from the SVN
trunk. Numpy builds without any problems and the all tests pass. Scipy
builds but when trying to run the tests I get:

import scipy
>>> scipy.test(1,10)
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/Library/Frameworks/Python.framework/Versions/2.5/lib/
python2.5/site-packages/scipy/testing/nosetester.py", line 133, in
test
    nose = import_nose()
  File "/Library/Frameworks/Python.framework/Versions/2.5/lib/
python2.5/site-packages/scipy/testing/nosetester.py", line 25, in
import_nose
    raise ImportError('Need nose >=0.10 for tests - see '
ImportError: Need nose >=0.10 for tests - see http://somethingaboutorange.com/mrl/projects/nose
>>>

I'm running python 2.5.2 from python.org and gfortran 4.2.3.

Any help is appreciated.

Thanks,

Andr?s


On 23 jun, 07:58, "Jonathan Hunt" <j... at 42quarks.com> wrote:
> Hi,
>
> Build log is attached. I am running OS X 10.5.3 with gfortran 4.2.1
> and gcc 4.0.1 (and python 2.5).
>
> Thanks for any help.
> Jonny
>
>
>
> > On Mon, Jun 23, 2008 at 10:02 PM, David Cournapeau
> > <da... at ar.media.kyoto-u.ac.jp> wrote:
> >> Jonathan Hunt wrote:
> >>> Is this bad? A problem? It seems to be related to LAPACK/BLAS but on
> >>> install SciPy said it detected LAPACK/BLAST.
>
> >> Yes, this is potentially bad. It is more than likely caused by a buggy
> >> interaction between C and Fortran. Which version of Mac OS X are you
> >> using, and which fortran compiler are you using ? A full build log (from
> >> scratch, that is after having removed build directory) would be useful,
>
> > --
> > Jonathan J Hunt <j... at 42quarks.com>
> > Homepage:http://www.42quarks.net.nz/wiki/JJH
> > (Further contact details there)
> > "Physics isn't the most important thing. Love is." Richard Feynman
>
> --
> Jonathan J Hunt <j... at 42quarks.com>
> Homepage:http://www.42quarks.net.nz/wiki/JJH
> (Further contact details there)
> "Physics isn't the most important thing. Love is." Richard Feynman
>
> ?build.log.bz2
> 33 KDescargar
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-u... at scipy.orghttp://projects.scipy.org/mailman/listinfo/scipy-user


From david at ar.media.kyoto-u.ac.jp  Mon Jun 23 09:29:55 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Mon, 23 Jun 2008 22:29:55 +0900
Subject: [SciPy-user] Test failure on OS X Leopard
In-Reply-To: <5f89059f-2266-47d8-bb7b-bfc34259ae8e@w7g2000hsa.googlegroups.com>
References: <e26778a60806222259v584d1bc4ube1cf4ffa9b3e4d7@mail.gmail.com>
	<485F90B8.7080701@ar.media.kyoto-u.ac.jp>	<e26778a60806230551w5cb2a6dby86bc211b686f945a@mail.gmail.com>
	<e26778a60806230558r422dedb9xb1549e83ad3708a1@mail.gmail.com>
	<5f89059f-2266-47d8-bb7b-bfc34259ae8e@w7g2000hsa.googlegroups.com>
Message-ID: <485FA553.2040109@ar.media.kyoto-u.ac.jp>

AndresGM wrote:
>
> ImportError: Need nose >=0.10 for tests - see http://somethingaboutorange.com/mrl/projects/nose
>   

What about installing nose :) Starting from 0.7 for scipy and 1.2 for
numpy (meaning svn trunk as well for both packages), nose is a hard
dependency.

cheers,

David


From david at ar.media.kyoto-u.ac.jp  Mon Jun 23 10:00:58 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Mon, 23 Jun 2008 23:00:58 +0900
Subject: [SciPy-user] Test failure on OS X Leopard
In-Reply-To: <e26778a60806230558r422dedb9xb1549e83ad3708a1@mail.gmail.com>
References: <e26778a60806222259v584d1bc4ube1cf4ffa9b3e4d7@mail.gmail.com>	<485F90B8.7080701@ar.media.kyoto-u.ac.jp>	<e26778a60806230551w5cb2a6dby86bc211b686f945a@mail.gmail.com>
	<e26778a60806230558r422dedb9xb1549e83ad3708a1@mail.gmail.com>
Message-ID: <485FAC9A.9060508@ar.media.kyoto-u.ac.jp>

Jonathan Hunt wrote:
> Hi,
>
> Build log is attached. I am running OS X 10.5.3 with gfortran 4.2.1
> and gcc 4.0.1 (and python 2.5).
>   

Thanks, as I expected, a problem with fortran/C wrapping. There is a
problem with the Accelerate framework, for which we have a workaround.
For some reason, the workaround is not setup in your case; I have
Leopard 10.5.3 too, and I don't have this problem (the workaround is
used in my case).

Even stranger, the workaround is used once in your case; it only does
not work for scipy.lib.blas. Since this module is not used anywhere in
scipy yet, it is not so much problematic (on the contrary, having this
problem in scipy.linalg would have been problematic).

If you are willing to help us solving this log, could you tell use what
does info['extra_link_flags'] contain in the function
needs_cblas_wrapper, in scipy.lib.blas.setup.py (line 30). It should
contain -Wl,Accelerate, as in the build.log, but I don't quite see how
the regex could fail...

thanks,

David


From twaite at berkeley.edu  Mon Jun 23 10:41:02 2008
From: twaite at berkeley.edu (Tom Waite)
Date: Mon, 23 Jun 2008 07:41:02 -0700
Subject: [SciPy-user] Image Morphology
In-Reply-To: <20080620164312.re0odtu9fl08cg08@webmail.mit.edu>
References: <20080620164312.re0odtu9fl08cg08@webmail.mit.edu>
Message-ID: <fe702a7c0806230741k5eba823crc3b1965c54f32a44@mail.gmail.com>

I wrote a package called segmenter (under development in ndimage) that has a
thinning method. It will operate on either a single binary image or with my
connected components methods to build connected regions and then thin each
region of interest, using the automatically derived bounding box of each
ROI. I need to clean up the code to complete it. I typically pre-filter the
images to prevent spurs and I had planned to add more advanced methods like
Zhang-Suen that deal with this (which may be what Matlab does).


On Fri, Jun 20, 2008 at 1:43 PM, <rrothkop at mit.edu> wrote:

> Hello,
> I am looking for python equivalents for the matlab functions BWselect and
> BWmorph(skel). Can the outcomes of these functions be accomplished using
> scipy.ndimage?
> Thanks,
> Rebecca
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080623/47831243/attachment.html>

From rrothkop at MIT.EDU  Mon Jun 23 11:41:49 2008
From: rrothkop at MIT.EDU (rrothkop at MIT.EDU)
Date: Mon, 23 Jun 2008 11:41:49 -0400
Subject: [SciPy-user] Image Morphology
Message-ID: <20080623114149.h22ww9augvksoog4@webmail.mit.edu>

Hello,
Thank you for your responses. To explain the functions further, BWselect returns
a binary image containing objects that overlap a given pixel and BWmorph(skel)
thins the edges of an image. It looks like Tom's package contains an
equivalent. I understand the package is not complete, but is the thinning
function working?
Thanks,
Rebecca


From nmb at wartburg.edu  Mon Jun 23 13:32:26 2008
From: nmb at wartburg.edu (Neil Martinsen-Burrell)
Date: Mon, 23 Jun 2008 17:32:26 +0000 (UTC)
Subject: [SciPy-user] Test failure on OS X Leopard
References: <e26778a60806222259v584d1bc4ube1cf4ffa9b3e4d7@mail.gmail.com>
	<485F90B8.7080701@ar.media.kyoto-u.ac.jp>	<e26778a60806230551w5cb2a6dby86bc211b686f945a@mail.gmail.com>
	<e26778a60806230558r422dedb9xb1549e83ad3708a1@mail.gmail.com>
	<5f89059f-2266-47d8-bb7b-bfc34259ae8e@w7g2000hsa.googlegroups.com>
	<485FA553.2040109@ar.media.kyoto-u.ac.jp>
Message-ID: <loom.20080623T173033-242@post.gmane.org>

David Cournapeau <david <at> ar.media.kyoto-u.ac.jp> writes:

> 
> AndresGM wrote:
> >
> > ImportError: Need nose >=0.10 for tests - see
http://somethingaboutorange.com/mrl/projects/nose
> >   
> 
> What about installing nose :) Starting from 0.7 for scipy and 1.2 for
> numpy (meaning svn trunk as well for both packages), nose is a hard
> dependency.

To be crystal-clear, nose is *not* a dependency for running anything in scipy or
numpy, nor will this ever be the case.  It *is* a hard dependency for running
the numpy and scipy test suites as of the current SVN versions of both scipy and
numpy.  (Scipy made the change before numpy but they will eventually converge in
the next releases.)

-Neil 






From twaite at berkeley.edu  Mon Jun 23 16:04:13 2008
From: twaite at berkeley.edu (Tom Waite)
Date: Mon, 23 Jun 2008 13:04:13 -0700
Subject: [SciPy-user] Image Morphology
In-Reply-To: <20080623114149.h22ww9augvksoog4@webmail.mit.edu>
References: <20080623114149.h22ww9augvksoog4@webmail.mit.edu>
Message-ID: <fe702a7c0806231304r7e756addge4930406c74e1180@mail.gmail.com>

Yes it is. I have a test_segmenter.py script that shows its use. I build 4
simple discs and get the Sobel edge "band". The thinning filter (I refer to
the method as mat-filter for medial axis) returns the single-pixel wide edge
and is shown to be comparable to the Canny edge filter which I also provide.
This will use the ROI when generating the thin edges.

On Mon, Jun 23, 2008 at 8:41 AM, <rrothkop at mit.edu> wrote:

> Hello,
> Thank you for your responses. To explain the functions further, BWselect
> returns
> a binary image containing objects that overlap a given pixel and
> BWmorph(skel)
> thins the edges of an image. It looks like Tom's package contains an
> equivalent. I understand the package is not complete, but is the thinning
> function working?
> Thanks,
> Rebecca
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080623/605f64d8/attachment.html>

From anand.prabhakar.patil at gmail.com  Mon Jun 23 18:12:56 2008
From: anand.prabhakar.patil at gmail.com (Anand Patil)
Date: Mon, 23 Jun 2008 23:12:56 +0100
Subject: [SciPy-user] Are cf2py threadsafe and cf2py inplace incompatible?
Message-ID: <FF275235-24AF-4A7F-BB46-81E76017B9DD@gmail.com>

Hi all,

I'm trying to write a multithreaded function where each thread fills  
in a portion of an array. I'm on a MacBook Pro using Python 2.5 and  
the numpy subversion head. The following program produces intermittent  
seg faults:

---------------------


from f2py_test import test
from numpy import *
from threading import Thread, Lock
from copy import copy

D = empty((60,60))

n_threads = 4
threads = []

for i in xrange(n_threads):

     new_thread= Thread(target=test, args=(D,))
     threads.append(new_thread)

     new_thread.start()

[thread.join() for thread in threads]


-----------------

where f2py_test.f is just

-----------------


       SUBROUTINE test(D,nx)

cf2py intent(inplace) D
cf2py intent(hide) nx
cf2py threadsafe

       DOUBLE PRECISION D(nx,nx)
       INTEGER nx

! Writing to D would happen here.
! Each thread would get some of the columns.

       RETURN
       END


------------------


Is there any way I can make this safe without computing the pieces  
independently and copying them in in serial?

Thanks very much,
Anand


The relevant bits of the crash report are

Process:         Python [6796]
Path:            /Library/Frameworks/Python.framework/Versions/2.5/ 
Resources/Python.app/Contents/MacOS/Python
Identifier:      Python
Version:         ??? (???)
Code Type:       X86 (Native)
Parent Process:  bash [2950]

Date/Time:       2008-06-23 22:58:21.777 +0100
OS Version:      Mac OS X 10.5.3 (9D34)
Report Version:  6

Exception Type:  EXC_BAD_ACCESS (SIGSEGV)
Exception Codes: KERN_INVALID_ADDRESS at 0x00000000039cc148
Crashed Thread:  3

...

Thread 3 Crashed:
0   multiarray.so                 	0x011c135b _strided_byte_copy + 427  
(arrayobject.c:350)
1   multiarray.so                 	0x011f071d _copy_from_same_shape +  
525 (arrayobject.c:978)
2   multiarray.so                 	0x012086eb _array_copy_into + 1195  
(arrayobject.c:1134)
3   distances.so                  	0x031d0453 array_from_pyobj + 2179  
(fortranobject.c:660)
4   distances.so                  	0x031cd5aa f2py_rout_distances_test  
+ 170 (distancesmodule.c:233)
5   distances.so                  	0x031cee54 fortran_call + 68  
(fortranobject.c:323)
6   org.python.python             	0x0019b5d2 PyObject_Call + 50
7   org.python.python             	0x0022a9e4 PyEval_EvalFrameEx + 15492
8   org.python.python             	0x0022d445 PyEval_EvalFrameEx + 26341
9   org.python.python             	0x0022d445 PyEval_EvalFrameEx + 26341
10  org.python.python             	0x0022dba5 PyEval_EvalCodeEx + 1845
11  org.python.python             	0x001bf4ce function_call + 446
12  org.python.python             	0x0019b5d2 PyObject_Call + 50
13  org.python.python             	0x001a3732 instancemethod_call + 354
14  org.python.python             	0x0019b5d2 PyObject_Call + 50
15  org.python.python             	0x00225c26  
PyEval_CallObjectWithKeywords + 118
16  org.python.python             	0x0026277f t_bootstrap + 63
17  libSystem.B.dylib             	0x93ff06f5 _pthread_start + 321
18  libSystem.B.dylib             	0x93ff05b2 thread_start + 34

...

Thread 3 crashed with X86 Thread State (32-bit):
   eax: 0x00000000  ebx: 0x011c11bb  ecx: 0x039cc148  edx: 0x00000000
   edi: 0x00000021  esi: 0x03c39008  ebp: 0xb0184098  esp: 0xb0184080
    ss: 0x0000001f  efl: 0x00010293  eip: 0x011c135b   cs: 0x00000017
    ds: 0x0000001f   es: 0x0000001f   fs: 0x0000001f   gs: 0x00000037
   cr2: 0x039cc148



From anand.prabhakar.patil at gmail.com  Mon Jun 23 18:25:30 2008
From: anand.prabhakar.patil at gmail.com (Anand Patil)
Date: Mon, 23 Jun 2008 23:25:30 +0100
Subject: [SciPy-user] Are cf2py threadsafe and cf2py inplace
	incompatible?
In-Reply-To: <FF275235-24AF-4A7F-BB46-81E76017B9DD@gmail.com>
References: <FF275235-24AF-4A7F-BB46-81E76017B9DD@gmail.com>
Message-ID: <2bc7a5a50806231525t241042c9s3533b5e66607ae9f@mail.gmail.com>

Sorry about the five copies of the last message I apparently just sent. I'll
stop using Apple Mail.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080623/7600d89e/attachment.html>

From haase at msg.ucsf.edu  Mon Jun 23 18:51:19 2008
From: haase at msg.ucsf.edu (Sebastian Haase)
Date: Tue, 24 Jun 2008 00:51:19 +0200
Subject: [SciPy-user] Image Morphology
In-Reply-To: <fe702a7c0806231304r7e756addge4930406c74e1180@mail.gmail.com>
References: <20080623114149.h22ww9augvksoog4@webmail.mit.edu>
	<fe702a7c0806231304r7e756addge4930406c74e1180@mail.gmail.com>
Message-ID: <bc657ead0806231551r14ec3956xc22289dc98b679e7@mail.gmail.com>

Hey Tom,
when you say """a package called segmenter (under development in ndimage)"""
what do you mean by
"""under development in ndimage"""
?
Is your code available (for public download) already ? Under a free license ?

Thanks,
Sebastian Haase



On Mon, Jun 23, 2008 at 10:04 PM, Tom Waite <twaite at berkeley.edu> wrote:
> Yes it is. I have a test_segmenter.py script that shows its use. I build 4
> simple discs and get the Sobel edge "band". The thinning filter (I refer to
> the method as mat-filter for medial axis) returns the single-pixel wide edge
> and is shown to be comparable to the Canny edge filter which I also provide.
> This will use the ROI when generating the thin edges.
>
> On Mon, Jun 23, 2008 at 8:41 AM, <rrothkop at mit.edu> wrote:
>>
>> Hello,
>> Thank you for your responses. To explain the functions further, BWselect
>> returns
>> a binary image containing objects that overlap a given pixel and
>> BWmorph(skel)
>> thins the edges of an image. It looks like Tom's package contains an
>> equivalent. I understand the package is not complete, but is the thinning
>> function working?
>> Thanks,
>> Rebecca


From robert.kern at gmail.com  Mon Jun 23 18:57:51 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Mon, 23 Jun 2008 17:57:51 -0500
Subject: [SciPy-user] Image Morphology
In-Reply-To: <bc657ead0806231551r14ec3956xc22289dc98b679e7@mail.gmail.com>
References: <20080623114149.h22ww9augvksoog4@webmail.mit.edu>
	<fe702a7c0806231304r7e756addge4930406c74e1180@mail.gmail.com>
	<bc657ead0806231551r14ec3956xc22289dc98b679e7@mail.gmail.com>
Message-ID: <3d375d730806231557r3afd01ddse760e4a8fe3b9e6@mail.gmail.com>

On Mon, Jun 23, 2008 at 17:51, Sebastian Haase <haase at msg.ucsf.edu> wrote:
> Hey Tom,
> when you say """a package called segmenter (under development in ndimage)"""
> what do you mean by
> """under development in ndimage"""
> ?

from scipy.ndimage import _segmenter

> Is your code available (for public download) already ?

scipy SVN.

> Under a free license ?

scipy license.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From jjh at 42quarks.com  Mon Jun 23 19:59:55 2008
From: jjh at 42quarks.com (Jonathan Hunt)
Date: Tue, 24 Jun 2008 09:59:55 +1000
Subject: [SciPy-user] Test failure on OS X Leopard
In-Reply-To: <485FAC9A.9060508@ar.media.kyoto-u.ac.jp>
References: <e26778a60806222259v584d1bc4ube1cf4ffa9b3e4d7@mail.gmail.com>
	<485F90B8.7080701@ar.media.kyoto-u.ac.jp>
	<e26778a60806230551w5cb2a6dby86bc211b686f945a@mail.gmail.com>
	<e26778a60806230558r422dedb9xb1549e83ad3708a1@mail.gmail.com>
	<485FAC9A.9060508@ar.media.kyoto-u.ac.jp>
Message-ID: <e26778a60806231659r47189deav7f33e3d4906051e8@mail.gmail.com>

Hi David,

I'm happy to help SciPy/NumPy are excellent software that I'm pleased
to be able to help with (in a small way). Btw. this problem occured on
both my laptop (MacBook Pro, Core2 Due) and desktop (MacPro 4 core)
which are running 10.5.3 and 10.5.2 respectively.

I had tried to answer you question so I opened
/Library/Python/2.5/site-packages/scipy/lib/blas/setup.py but I
couldn't find the function you were refering to. I have attached the
file in case I'm missing something. Am I looking in the wrong place?
Would you like me to update to the SVN version of SciPy and see if
that fixes the problem?

Also, if I understand your message the errors in this test function
won't cause problems when I actually use SciPy? Is that correct?

Thanks,
Jonny

On Tue, Jun 24, 2008 at 12:00 AM, David Cournapeau
<david at ar.media.kyoto-u.ac.jp> wrote:
> Jonathan Hunt wrote:
>> Hi,
>>
>> Build log is attached. I am running OS X 10.5.3 with gfortran 4.2.1
>> and gcc 4.0.1 (and python 2.5).
>>
>
> Thanks, as I expected, a problem with fortran/C wrapping. There is a
> problem with the Accelerate framework, for which we have a workaround.
> For some reason, the workaround is not setup in your case; I have
> Leopard 10.5.3 too, and I don't have this problem (the workaround is
> used in my case).
>
> Even stranger, the workaround is used once in your case; it only does
> not work for scipy.lib.blas. Since this module is not used anywhere in
> scipy yet, it is not so much problematic (on the contrary, having this
> problem in scipy.linalg would have been problematic).
>
> If you are willing to help us solving this log, could you tell use what
> does info['extra_link_flags'] contain in the function
> needs_cblas_wrapper, in scipy.lib.blas.setup.py (line 30). It should
> contain -Wl,Accelerate, as in the build.log, but I don't quite see how
> the regex could fail...
>
> thanks,
>
> David
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>



-- 
Jonathan J Hunt <jjh at 42quarks.com>
Homepage: http://www.42quarks.net.nz/wiki/JJH
(Further contact details there)
"Physics isn't the most important thing. Love is." Richard Feynman
-------------- next part --------------
A non-text attachment was scrubbed...
Name: setup.py
Type: application/octet-stream
Size: 3383 bytes
Desc: not available
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080624/1590452f/attachment.obj>

From david at ar.media.kyoto-u.ac.jp  Mon Jun 23 22:28:18 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Tue, 24 Jun 2008 11:28:18 +0900
Subject: [SciPy-user] Test failure on OS X Leopard
In-Reply-To: <loom.20080623T173033-242@post.gmane.org>
References: <e26778a60806222259v584d1bc4ube1cf4ffa9b3e4d7@mail.gmail.com>	<485F90B8.7080701@ar.media.kyoto-u.ac.jp>	<e26778a60806230551w5cb2a6dby86bc211b686f945a@mail.gmail.com>	<e26778a60806230558r422dedb9xb1549e83ad3708a1@mail.gmail.com>	<5f89059f-2266-47d8-bb7b-bfc34259ae8e@w7g2000hsa.googlegroups.com>	<485FA553.2040109@ar.media.kyoto-u.ac.jp>
	<loom.20080623T173033-242@post.gmane.org>
Message-ID: <48605BC2.5030900@ar.media.kyoto-u.ac.jp>

Neil Martinsen-Burrell wrote:
>
> To be crystal-clear, nose is *not* a dependency for running anything in scipy or
> numpy, nor will this ever be the case.  It *is* a hard dependency for running
> the numpy and scipy test suites as of the current SVN versions of both scipy and
> numpy.  (Scipy made the change before numpy but they will eventually converge in
> the next releases.)
>   

Ah yes, sorry for the confusion, I should have said it was a hard
dependency for *tests* only.

cheers,

David


From david at ar.media.kyoto-u.ac.jp  Mon Jun 23 22:48:23 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Tue, 24 Jun 2008 11:48:23 +0900
Subject: [SciPy-user] Test failure on OS X Leopard
In-Reply-To: <e26778a60806231659r47189deav7f33e3d4906051e8@mail.gmail.com>
References: <e26778a60806222259v584d1bc4ube1cf4ffa9b3e4d7@mail.gmail.com>	<485F90B8.7080701@ar.media.kyoto-u.ac.jp>	<e26778a60806230551w5cb2a6dby86bc211b686f945a@mail.gmail.com>	<e26778a60806230558r422dedb9xb1549e83ad3708a1@mail.gmail.com>	<485FAC9A.9060508@ar.media.kyoto-u.ac.jp>
	<e26778a60806231659r47189deav7f33e3d4906051e8@mail.gmail.com>
Message-ID: <48606077.8040007@ar.media.kyoto-u.ac.jp>

Jonathan Hunt wrote:
> I had tried to answer you question so I opened
> /Library/Python/2.5/site-packages/scipy/lib/blas/setup.py but I
> couldn't find the function you were refering to. I have attached the
> file in case I'm missing something. Am I looking in the wrong place?
>   

Hi Johnathan,

    You solved the problem :) I forgot that you were using 0.6, in which
the fix for this problem was not applied everywhere. I should have
thought about this...
> Also, if I understand your message the errors in this test function
> won't cause problems when I actually use SciPy? Is that correct?
>   

Yes. Unless you use directly scipy.lib.blas, no code inside scipy uses
it; it will always use the code from scipy.linalg, which does have the
fix even in 0.6,

cheers,

David


From jjh at 42quarks.com  Mon Jun 23 23:07:03 2008
From: jjh at 42quarks.com (Jonathan Hunt)
Date: Tue, 24 Jun 2008 13:07:03 +1000
Subject: [SciPy-user] Test failure on OS X Leopard
In-Reply-To: <48606077.8040007@ar.media.kyoto-u.ac.jp>
References: <e26778a60806222259v584d1bc4ube1cf4ffa9b3e4d7@mail.gmail.com>
	<485F90B8.7080701@ar.media.kyoto-u.ac.jp>
	<e26778a60806230551w5cb2a6dby86bc211b686f945a@mail.gmail.com>
	<e26778a60806230558r422dedb9xb1549e83ad3708a1@mail.gmail.com>
	<485FAC9A.9060508@ar.media.kyoto-u.ac.jp>
	<e26778a60806231659r47189deav7f33e3d4906051e8@mail.gmail.com>
	<48606077.8040007@ar.media.kyoto-u.ac.jp>
Message-ID: <e26778a60806232007p7f72b8dfxc1f2a8361f94af4f@mail.gmail.com>

Hi,

Sorry for the hassle. Thanks for that.

Jonny

On Tue, Jun 24, 2008 at 12:48 PM, David Cournapeau
<david at ar.media.kyoto-u.ac.jp> wrote:
> Jonathan Hunt wrote:
>> I had tried to answer you question so I opened
>> /Library/Python/2.5/site-packages/scipy/lib/blas/setup.py but I
>> couldn't find the function you were refering to. I have attached the
>> file in case I'm missing something. Am I looking in the wrong place?
>>
>
> Hi Johnathan,
>
>    You solved the problem :) I forgot that you were using 0.6, in which
> the fix for this problem was not applied everywhere. I should have
> thought about this...
>> Also, if I understand your message the errors in this test function
>> won't cause problems when I actually use SciPy? Is that correct?
>>
>
> Yes. Unless you use directly scipy.lib.blas, no code inside scipy uses
> it; it will always use the code from scipy.linalg, which does have the
> fix even in 0.6,
>
> cheers,
>
> David
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>



-- 
Jonathan J Hunt <jjh at 42quarks.com>
Homepage: http://www.42quarks.net.nz/wiki/JJH
(Further contact details there)
"Physics isn't the most important thing. Love is." Richard Feynman


From jjh at 42quarks.com  Mon Jun 23 23:56:42 2008
From: jjh at 42quarks.com (Jonathan Hunt)
Date: Tue, 24 Jun 2008 13:56:42 +1000
Subject: [SciPy-user] Reading MATLAB 7.4 (HDF5 based) .mat files
Message-ID: <e26778a60806232056q78d1eb0axc90d9ebc8dc6a57e@mail.gmail.com>

Hi,

The current SciPy libraries for reading MATLAB .mat files only support
v6 and v7.1 format. Just wondering if anyone is currently working on
adding support for 7.4 (HDF5 based) .mat files.

Thanks,
Jonny

-- 
Jonathan J Hunt <jjh at 42quarks.com>
Homepage: http://www.42quarks.net.nz/wiki/JJH
(Further contact details there)
"Physics isn't the most important thing. Love is." Richard Feynman


From peridot.faceted at gmail.com  Tue Jun 24 00:01:56 2008
From: peridot.faceted at gmail.com (Anne Archibald)
Date: Tue, 24 Jun 2008 00:01:56 -0400
Subject: [SciPy-user] Reading MATLAB 7.4 (HDF5 based) .mat files
In-Reply-To: <e26778a60806232056q78d1eb0axc90d9ebc8dc6a57e@mail.gmail.com>
References: <e26778a60806232056q78d1eb0axc90d9ebc8dc6a57e@mail.gmail.com>
Message-ID: <ce557a360806232101m5e0d05e8w6e2ee9f0c5617dfd@mail.gmail.com>

2008/6/23 Jonathan Hunt <jjh at 42quarks.com>:

> The current SciPy libraries for reading MATLAB .mat files only support
> v6 and v7.1 format. Just wondering if anyone is currently working on
> adding support for 7.4 (HDF5 based) .mat files.

My last use of MATLAB was some ten years ago so I haven't tried it,
but "pytables" is a general-purpose HDF5 reader, so presumably it
should be a fairly convenient way to get at MATLAB files.

Anne


From david at ar.media.kyoto-u.ac.jp  Mon Jun 23 23:47:38 2008
From: david at ar.media.kyoto-u.ac.jp (David Cournapeau)
Date: Tue, 24 Jun 2008 12:47:38 +0900
Subject: [SciPy-user] Reading MATLAB 7.4 (HDF5 based) .mat files
In-Reply-To: <e26778a60806232056q78d1eb0axc90d9ebc8dc6a57e@mail.gmail.com>
References: <e26778a60806232056q78d1eb0axc90d9ebc8dc6a57e@mail.gmail.com>
Message-ID: <48606E5A.1070307@ar.media.kyoto-u.ac.jp>

Jonathan Hunt wrote:
> Hi,
>
> The current SciPy libraries for reading MATLAB .mat files only support
> v6 and v7.1 format. Just wondering if anyone is currently working on
> adding support for 7.4 (HDF5 based) .mat files.
>   

For hdf5 files, pytables is your best bet I think: it uses hdf5 as its
implementation file format, and I used it to exchange data between
matlab and python (at a time where matlab support for hdf5 was really
poor; I would guess that since they base their new format on it, it is
much better now)

http://www.pytables.org/moin

cheers,

David


From grs2103 at columbia.edu  Tue Jun 24 00:36:40 2008
From: grs2103 at columbia.edu (Gideon Simpson)
Date: Tue, 24 Jun 2008 00:36:40 -0400
Subject: [SciPy-user] sinc function
Message-ID: <98DBB071-77BF-44E6-A50A-CE375D4F0C40@columbia.edu>

Why is sinc not in scipy.special?  Is it implemented in another python  
package?

-gideon



From robert.kern at gmail.com  Tue Jun 24 00:42:55 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Mon, 23 Jun 2008 23:42:55 -0500
Subject: [SciPy-user] sinc function
In-Reply-To: <98DBB071-77BF-44E6-A50A-CE375D4F0C40@columbia.edu>
References: <98DBB071-77BF-44E6-A50A-CE375D4F0C40@columbia.edu>
Message-ID: <3d375d730806232142i3ad51a0nf17c9cd51c2fe5c0@mail.gmail.com>

On Mon, Jun 23, 2008 at 23:36, Gideon Simpson <grs2103 at columbia.edu> wrote:
> Why is sinc not in scipy.special?  Is it implemented in another python
> package?

numpy.sinc()

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From anand.prabhakar.patil at gmail.com  Tue Jun 24 04:33:07 2008
From: anand.prabhakar.patil at gmail.com (Anand Patil)
Date: Tue, 24 Jun 2008 09:33:07 +0100
Subject: [SciPy-user] Are cf2py threadsafe and cf2py inplace
	incompatible?
In-Reply-To: <FF275235-24AF-4A7F-BB46-81E76017B9DD@gmail.com>
References: <FF275235-24AF-4A7F-BB46-81E76017B9DD@gmail.com>
Message-ID: <2bc7a5a50806240133j699f2606k659e4470f2985191@mail.gmail.com>

On Mon, Jun 23, 2008 at 11:12 PM, Anand Patil <
anand.prabhakar.patil at gmail.com> wrote:

> Hi all,
>
> I'm trying to write a multithreaded function where each thread fills in a
> portion of an array. I'm on a MacBook Pro using Python 2.5 and the numpy
> subversion head. The following program produces intermittent seg faults:
>

Well good, it looks like only one copy got through.

Also I realized what the problem is: the new empty array is row-major, so
f2py has to make a column-major copy of it to pass into Fortran, then copy
the results into the original array to make it seem like Fortran function is
changing the array in-place.

Cheers,
Anand
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080624/79297e7b/attachment.html>

From lorenzo.isella at gmail.com  Tue Jun 24 09:12:34 2008
From: lorenzo.isella at gmail.com (Lorenzo Isella)
Date: Tue, 24 Jun 2008 15:12:34 +0200
Subject: [SciPy-user] Again on array manipulation
Message-ID: <a2b3004b0806240612r78408478lb8877f5d1bcd9e14@mail.gmail.com>

Dear All,
This should be quick: say that you have e.g. the following two scipy
arrays:

a=[15 23 44 78|11 77| 33 45 89| 56 99| 12 654 81]

b=[10 10 10 10|40 40| 60 60 60| 22 22| 17 17 17]

(the vertical bar is just a guide for the eye).
No entry appears twice in a and no entry appears only once in b. I
would like to chop off all the
elements appearing more than twice in b and get rid of the
corresponding elements in a, thus ending up with:



a=[|11 77| 56 99|]

b=[|40 40| 22 22|]

what is the easiest way to achieve this?
Cheers

Lorenzo


From contact at pythonxy.com  Tue Jun 24 09:51:07 2008
From: contact at pythonxy.com (Pierre Raybaut)
Date: Tue, 24 Jun 2008 15:51:07 +0200
Subject: [SciPy-user] Advanced Image Processing with Python
Message-ID: <629b08a40806240651n49f6d208v3b7bbc4024d4dfa5@mail.gmail.com>

>
> Date: Sun, 22 Jun 2008 18:10:14 -0400
> From: Zachary Pincus <zachary.pincus at yale.edu>
> Subject: Re: [SciPy-user] Advanced Image Processing with Python
> To: SciPy Users List <scipy-user at scipy.org>
> Message-ID: <66DC272E-5676-4C7A-9CC8-68CB6CC45B73 at yale.edu>
> Content-Type: text/plain; charset=US-ASCII; format=flowed; delsp=yes
>
> > The rest of what I'm looking for is basically what you
> > can find in MATLAB image processing toolbox, and some shape
> > recognition
> > algorithms eventually.
> >
> > The main idea would be to develop a first software with commonly used
> > image processing tools (basic ones: filters, geometric
> > transformations,
> > basic operations, histograms, profile extraction, fft, ...), and
> > optional modules to extend its features.
> >
> > This software will be used to process mainly gray-scale (and less
> > often
> > color) images, that can be stored in 2D data arrays (as well as 1D
> > signals, but that's not a problem with Numpy/Scipy). So what I'm
> > looking
> > for would be an image processing library (and one only if possible) to
> > handle this, and of course I'd prefer it to be widely used, still
> > developed and maintained, with a great future ahead... the perfect
> > library!
>
> I think that OpenCV (with the ctypes python bindings) and
> scipy.ndimage should probably be pretty good candidates here...
>
> Zach


Hi Zach,

Among all Python image processing libraries, ITK appears to be the most
promising (thanks to WrapITK - as you know, Zach, of course:
http://voxel.jouy.inra.fr/darcs/contrib-itk/WrapITK/WrapITK_-_Enhanced_languages_support_for_the_Insight_Toolkit.pdf).
But for Windows users (and I am one, unfortunately), ITK seems difficult to
build compared to VTK for instance (and it's a very long process - 3 hours
on my machine which is quite powerful). I've already tried and failed with
MinGW and VSC2003.
That is really weird for such an interesting library to be almost closed to
Windows users: there is no binaries available for Python 2.5 (actually there
is none for VTK either but at least it's easy to build). And when I see what
it can do (at least on paper), it surprises me that no one tried to build it
on Windows before (and shared the binaries with the community), except for
Python 2.4 (
http://cpbotha.net/2007/08/02/python-enabled-vtk-51-and-itk-32-windows-binaries/).
Is there any reason for that? (is this library not so good as it seems?)

Anyway, if I succeed, I will be glad to share the result with the community.

Pierre
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080624/8e6ab216/attachment.html>

From wnbell at gmail.com  Tue Jun 24 10:06:41 2008
From: wnbell at gmail.com (Nathan Bell)
Date: Tue, 24 Jun 2008 09:06:41 -0500
Subject: [SciPy-user] Again on array manipulation
In-Reply-To: <a2b3004b0806240612r78408478lb8877f5d1bcd9e14@mail.gmail.com>
References: <a2b3004b0806240612r78408478lb8877f5d1bcd9e14@mail.gmail.com>
Message-ID: <d05265cb0806240706l65cd4d32s6acb40b73604bf01@mail.gmail.com>

On Tue, Jun 24, 2008 at 8:12 AM, Lorenzo Isella
<lorenzo.isella at gmail.com> wrote:
>
> a=[15 23 44 78|11 77| 33 45 89| 56 99| 12 654 81]
>
> b=[10 10 10 10|40 40| 60 60 60| 22 22| 17 17 17]
>
> (the vertical bar is just a guide for the eye).
> No entry appears twice in a and no entry appears only once in b. I
> would like to chop off all the
> elements appearing more than twice in b and get rid of the
> corresponding elements in a, thus ending up with:

If b consists of only small integers, then I'd used bincount()

In [15]: from scipy import *

In [16]: a = array([15, 23, 44, 78, 11, 77, 33, 45, 89, 56, 99, 12, 654, 81])

In [17]: b = array([10, 10, 10, 10, 40, 40, 60, 60, 60, 22, 22, 17, 17, 17])

In [18]: mask = (bincount(b) <= 2)[b]

In [19]: print a[mask]
[11 77 56 99]

In [20]: print b[mask]
[40 40 22 22]

Otherwise, it depends on whether the repeated elements of b are
contiguous or not.

-- 
Nathan Bell wnbell at gmail.com
http://graphics.cs.uiuc.edu/~wnbell/


From zachary.pincus at yale.edu  Tue Jun 24 10:11:16 2008
From: zachary.pincus at yale.edu (Zachary Pincus)
Date: Tue, 24 Jun 2008 10:11:16 -0400
Subject: [SciPy-user] Advanced Image Processing with Python
In-Reply-To: <629b08a40806240651n49f6d208v3b7bbc4024d4dfa5@mail.gmail.com>
References: <629b08a40806240651n49f6d208v3b7bbc4024d4dfa5@mail.gmail.com>
Message-ID: <E76AD7D8-8493-4EDF-B1CA-0D29D4C68F19@yale.edu>

> Hi Zach,
>
> Among all Python image processing libraries, ITK appears to be the  
> most promising (thanks to WrapITK - as you know, Zach, of course: http://voxel.jouy.inra.fr/darcs/contrib-itk/WrapITK/WrapITK_-_Enhanced_languages_support_for_the_Insight_Toolkit.pdf) 
> . But for Windows users (and I am one, unfortunately), ITK seems  
> difficult to build compared to VTK for instance (and it's a very  
> long process - 3 hours on my machine which is quite powerful). I've  
> already tried and failed with MinGW and VSC2003.
> That is really weird for such an interesting library to be almost  
> closed to Windows users: there is no binaries available for Python  
> 2.5 (actually there is none for VTK either but at least it's easy to  
> build). And when I see what it can do (at least on paper), it  
> surprises me that no one tried to build it on Windows before (and  
> shared the binaries with the community), except for Python 2.4 (http://cpbotha.net/2007/08/02/python-enabled-vtk-51-and-itk-32-windows-binaries/ 
> ). Is there any reason for that? (is this library not so good as it  
> seems?)
>
> Anyway, if I succeed, I will be glad to share the result with the  
> community.

Hi Pierre,

I'd advise asking for help on the ITK list -- they take quite  
seriously their commitment to cross-platform build-ability (to the  
extent of having built that CMake tool precisely to that end).

As for build time, that's another issue. Basically, ITK is (to a  
frightening degree) driven by C++ template programming, which really  
strains compilers. You can speed the build time for WrapITK (when you  
get it working) by turning off the template instantiations for all but  
a few pixel types (say, uint8, uint16, and float32) and image  
dimensions (only 2D, say). Build time will be nearly linear in the  
product of pixel types and image dimension...

Now, it's been a while since I was involved with WrapITK, but last I  
checked, the ITK people were pretty enthusiastic about getting it  
integrated into ITK and fully supported. Hopefully this is still the  
case! (Though coaxing the CMake build system into performing the magic  
required to run everything was non-trivial, due to limitations in the  
CMake language, so perhaps it has been difficult to maintain.)

Good luck,

Zach


From contact at pythonxy.com  Tue Jun 24 14:26:18 2008
From: contact at pythonxy.com (Pierre Raybaut)
Date: Tue, 24 Jun 2008 20:26:18 +0200
Subject: [SciPy-user] Advanced Image Processing with Python
In-Reply-To: <mailman.9.1214326803.31080.scipy-user@scipy.org>
References: <mailman.9.1214326803.31080.scipy-user@scipy.org>
Message-ID: <48613C4A.7070905@pythonxy.com>

>
>> > Hi Zach,
>> >
>> > Among all Python image processing libraries, ITK appears to be the  
>> > most promising (thanks to WrapITK - as you know, Zach, of course: http://voxel.jouy.inra.fr/darcs/contrib-itk/WrapITK/WrapITK_-_Enhanced_languages_support_for_the_Insight_Toolkit.pdf) 
>> > . But for Windows users (and I am one, unfortunately), ITK seems  
>> > difficult to build compared to VTK for instance (and it's a very  
>> > long process - 3 hours on my machine which is quite powerful). I've  
>> > already tried and failed with MinGW and VSC2003.
>> > That is really weird for such an interesting library to be almost  
>> > closed to Windows users: there is no binaries available for Python  
>> > 2.5 (actually there is none for VTK either but at least it's easy to  
>> > build). And when I see what it can do (at least on paper), it  
>> > surprises me that no one tried to build it on Windows before (and  
>> > shared the binaries with the community), except for Python 2.4 (http://cpbotha.net/2007/08/02/python-enabled-vtk-51-and-itk-32-windows-binaries/ 
>> > ). Is there any reason for that? (is this library not so good as it  
>> > seems?)
>> >
>> > Anyway, if I succeed, I will be glad to share the result with the  
>> > community.
>>     
>
> Hi Pierre,
>
> I'd advise asking for help on the ITK list -- they take quite  
> seriously their commitment to cross-platform build-ability (to the  
> extent of having built that CMake tool precisely to that end).
>   

I'm giving it a try with VSC2005 Express, and if it goes wrong I'll 
switch to ITK list for support indeed -- good suggestion.

> Now, it's been a while since I was involved with WrapITK, but last I  
> checked, the ITK people were pretty enthusiastic about getting it  
> integrated into ITK and fully supported. Hopefully this is still the  
> case! (Though coaxing the CMake build system into performing the magic  
> required to run everything was non-trivial, due to limitations in the  
> CMake language, so perhaps it has been difficult to maintain.)
>
>   
Apparently, WrapITK is now a part of ITK package but there is still a 
warning in CMake when you enable it (something like "WrapITK is still 
experimental [...] some problems have been reported when building on 
Windows platforms [...]").
I had problems which were in fact limitations of MinGW and VSC2003, so 
at this stage I can't really say that the CMake configuration does not 
work. I'll see tomorrow morning if it worked with VSC2005 when I get 
back to my work...
> Good luck,
Hope I won't need it ;)
Thanks
Pierre



From zachary.pincus at yale.edu  Tue Jun 24 15:33:41 2008
From: zachary.pincus at yale.edu (Zachary Pincus)
Date: Tue, 24 Jun 2008 15:33:41 -0400
Subject: [SciPy-user] simple PDE tools?
Message-ID: <2E41AD1B-5EF8-426A-99F3-B99ECCAB4EA3@yale.edu>

Hello all,

I'm trying to solve an optimization problem that might be best  
approached by formulating it as a "physical" system of interacting  
particles (certain of which have attractions to and repulsions from  
certain others). I could then write an energy term for the system at a  
given state, and a term for the forces on each particle in that state.

Are there any simple PDE tools available that I could use to time-step  
something like this until convergence? I've looked into FiPy, which  
looks like it's more set up for continuum modeling (like level sets,  
heat flow, etc.) than just time-integrating the forces on a set of  
particles. I could also write an integrator myself, but there are of  
course sufficiently many gotchas there that I'd rather not.

Thanks,

Zach


From robert.kern at gmail.com  Tue Jun 24 15:38:50 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Tue, 24 Jun 2008 14:38:50 -0500
Subject: [SciPy-user] simple PDE tools?
In-Reply-To: <2E41AD1B-5EF8-426A-99F3-B99ECCAB4EA3@yale.edu>
References: <2E41AD1B-5EF8-426A-99F3-B99ECCAB4EA3@yale.edu>
Message-ID: <3d375d730806241238k69136397gb875bf7038eb8c06@mail.gmail.com>

On Tue, Jun 24, 2008 at 14:33, Zachary Pincus <zachary.pincus at yale.edu> wrote:
> Hello all,
>
> I'm trying to solve an optimization problem that might be best
> approached by formulating it as a "physical" system of interacting
> particles (certain of which have attractions to and repulsions from
> certain others). I could then write an energy term for the system at a
> given state, and a term for the forces on each particle in that state.
>
> Are there any simple PDE tools available that I could use to time-step
> something like this until convergence? I've looked into FiPy, which
> looks like it's more set up for continuum modeling (like level sets,
> heat flow, etc.) than just time-integrating the forces on a set of
> particles.

Wouldn't that be an ODE, then?

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From zachary.pincus at yale.edu  Tue Jun 24 16:13:46 2008
From: zachary.pincus at yale.edu (Zachary Pincus)
Date: Tue, 24 Jun 2008 16:13:46 -0400
Subject: [SciPy-user] simple PDE tools?
In-Reply-To: <3d375d730806241238k69136397gb875bf7038eb8c06@mail.gmail.com>
References: <2E41AD1B-5EF8-426A-99F3-B99ECCAB4EA3@yale.edu>
	<3d375d730806241238k69136397gb875bf7038eb8c06@mail.gmail.com>
Message-ID: <80CABC80-D50B-4B7D-A631-C126531852EA@yale.edu>

Robert Kern:
> Wouldn't that be an ODE, then?

Hmm, it would appear that you are correct. (As usual!) I had somehow  
gotten myself confused into thinking that the problem wasn't an ODE  
one...

Sorry for the noise, and thanks,

Zach


From zachary.pincus at yale.edu  Tue Jun 24 19:45:10 2008
From: zachary.pincus at yale.edu (Zachary Pincus)
Date: Tue, 24 Jun 2008 19:45:10 -0400
Subject: [SciPy-user] integrate ODE to steady-state?
Message-ID: <042BF501-C3C2-451D-A0D3-0A3B762E3930@yale.edu>

Hi all,

So, after a brief bout of the stupids (thanks Robert), I have  
formulated my optimization problem as a physical system governed by an  
ODE, and I wish to learn the equilibrium configuration of the system.

Any thoughts on what the easiest way to do this with scipy.integrate  
is? Ideally, I'd just want the solver to take as large steps as  
possible until things converge, and so I don't really care about the  
"time" values. One option would be to use odeint and just tell it to  
integrate to a distant time-point when I'm sure things will be in  
equilibrium, but that seems dorky, wasteful, and potentially incorrect.

Alternately, I could use the ode class and keep asking it to integrate  
small time-steps until the RMS change drops below a threshold. There,  
still, I'd need to choose a reasonable time-step, and also the inner  
loop would be in python instead of fortran.

Any recommendations? (Or a I again being daft? I never really took a  
class in numerical methods, so sorry for dim-bulb questions!)

Zach


From wnbell at gmail.com  Tue Jun 24 20:51:57 2008
From: wnbell at gmail.com (Nathan Bell)
Date: Tue, 24 Jun 2008 19:51:57 -0500
Subject: [SciPy-user] integrate ODE to steady-state?
In-Reply-To: <042BF501-C3C2-451D-A0D3-0A3B762E3930@yale.edu>
References: <042BF501-C3C2-451D-A0D3-0A3B762E3930@yale.edu>
Message-ID: <d05265cb0806241751j56d413d9h93b64dc57fae31d0@mail.gmail.com>

On Tue, Jun 24, 2008 at 6:45 PM, Zachary Pincus <zachary.pincus at yale.edu> wrote:
>
> Any thoughts on what the easiest way to do this with scipy.integrate
> is? Ideally, I'd just want the solver to take as large steps as
> possible until things converge, and so I don't really care about the
> "time" values. One option would be to use odeint and just tell it to
> integrate to a distant time-point when I'm sure things will be in
> equilibrium, but that seems dorky, wasteful, and potentially incorrect.
>

Regarding numerical methods, you should read about the "Forward Euler"
integrator, which is very simple and intuitive.  For applications
where accuracy is unimportant and a reasonable time step is known a
priori, Forward Euler is sufficient.

Also, explicit ODE integrators (like Forward Euler and Runge-Kutta)
are generally cheap, so the dominant cost of the integration is
usually computing the time derivatives for each variable.

You are right that you'll want to take as-large-as-possible timesteps
when integrating through fictitious time.  However, even though the
timescale is arbitrary, the allowable timestep is limited by both the
accuracy of the trajectory (probably not a concern in your case) and
the stability of the system.  In short, there's no free lunch :)

I don't know what interface scipy's ode solvers support, but I would
suggest setting t_final to be a large value and instructing the solver
to use at most N iterations.  I would use an "adaptive" method like
RK45 (or whatever scipy supports) which automatically finds a timestep
that achieves numerical stability and the desired accuracy.

Alternatively, you could use your second proposal, and terminate
relaxation when the functional is sufficiently small.



BTW, are you using a Lennard-Jones like potential to distribute points
?  I did something similar to distribute points evenly over a mesh
surface using a simple integrator:
http://graphics.cs.uiuc.edu/~wnbell/publications/2005-05-SCA-Granular/BeYiMu2005.pdf

An interesting alternative is the CVT:
http://www.math.psu.edu/qdu/Res/Pic/gallery3.html

-- 
Nathan Bell wnbell at gmail.com
http://graphics.cs.uiuc.edu/~wnbell/


From rob.clewley at gmail.com  Tue Jun 24 20:55:42 2008
From: rob.clewley at gmail.com (Rob Clewley)
Date: Tue, 24 Jun 2008 20:55:42 -0400
Subject: [SciPy-user] integrate ODE to steady-state?
In-Reply-To: <042BF501-C3C2-451D-A0D3-0A3B762E3930@yale.edu>
References: <042BF501-C3C2-451D-A0D3-0A3B762E3930@yale.edu>
Message-ID: <a749952d0806241755tf68720es2488466dea0ddec7@mail.gmail.com>

Zach,

It somewhat depends on your equations, but in principle you don't need
to actually integrate the ODEs at all. An equilibrium corresponds to
no motion, i.e. when the right-hand sides of all the ODEs are
zero-valued. So for a system dx/dt = f(x), where x is a vector, you
just need to solve f(x) = 0. This may have multiple solutions, of
course, if f is nonlinear. There's no scipy code that will do all this
for you but fsolve will do in most cases. It's best to have a good
initial guess as a starting condition for the search. So, integration
can help you find that. If f is strongly nonlinear there could be many
equilibria and your task will be to identify which initial conditions
lead to which equilibria. This is not a trivial task - you may need an
exhaustive search of initial conditions (e.g. based on sampling your
phase space of x) to get started. I have some naive code that
pre-samples the phase space and then uses fsolve, and works OK on some
nonlinear problems. It's the find_fixedpoints function in PyDSTool,
but it's extremely easy to remove the PyDSTool dependence :)

If this approach turns out to be too numerically problematic, you'll
just have to go back to integrating for long times...

Rob

On Tue, Jun 24, 2008 at 7:45 PM, Zachary Pincus <zachary.pincus at yale.edu> wrote:
> Hi all,
>
> So, after a brief bout of the stupids (thanks Robert), I have
> formulated my optimization problem as a physical system governed by an
> ODE, and I wish to learn the equilibrium configuration of the system.
>
> Any thoughts on what the easiest way to do this with scipy.integrate
> is? Ideally, I'd just want the solver to take as large steps as
> possible until things converge, and so I don't really care about the
> "time" values. One option would be to use odeint and just tell it to
> integrate to a distant time-point when I'm sure things will be in
> equilibrium, but that seems dorky, wasteful, and potentially incorrect.
>
> Alternately, I could use the ode class and keep asking it to integrate
> small time-steps until the RMS change drops below a threshold. There,
> still, I'd need to choose a reasonable time-step, and also the inner
> loop would be in python instead of fortran.
>
> Any recommendations? (Or a I again being daft? I never really took a
> class in numerical methods, so sorry for dim-bulb questions!)
>
> Zach
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>


From lou_boog2000 at yahoo.com  Tue Jun 24 21:05:23 2008
From: lou_boog2000 at yahoo.com (Lou Pecora)
Date: Tue, 24 Jun 2008 18:05:23 -0700 (PDT)
Subject: [SciPy-user] integrate ODE to steady-state?
In-Reply-To: <a749952d0806241755tf68720es2488466dea0ddec7@mail.gmail.com>
Message-ID: <714288.82822.qm@web34408.mail.mud.yahoo.com>

I agree with Rob Clewley's approach.  I suspect that finding the zeros of the vector field (f(x)) is the best way to go.  Integration is harder and still leaves the problem of multiple equilibria.  This is often expressed as finding multiple basins of attraction (each initial condition of which goes to a different equilibrium point).  You may have an additional problem in that you want to find the stability of the equilibrium points.  If you want a point that a system will stay at even under the influence of some local noise or perturbations, then once you find the equilibrium points, you want to determine their stability.  This is done quite easily by evaluating the Jacobian of the vector field (the matrix which is the "gradient" of the vector field:  d f(x)/ dx, where f and x are vectors of the same dimension).  You then find the eigenvalues of the Jacobian.  If all are negative, you have a stable equilibrium.  If one if positive, you have an unstable
 equilibrium and the actual system will probably *not* ever end up there.  I hope that's clear and that helps.  Your problem is only as hard as the f(x) is complex and nonlinear.  It is not an ODE integration problem, really. 

-- Lou Pecora,   my views are my own.




      


From wnbell at gmail.com  Tue Jun 24 21:21:59 2008
From: wnbell at gmail.com (Nathan Bell)
Date: Tue, 24 Jun 2008 20:21:59 -0500
Subject: [SciPy-user] integrate ODE to steady-state?
In-Reply-To: <714288.82822.qm@web34408.mail.mud.yahoo.com>
References: <a749952d0806241755tf68720es2488466dea0ddec7@mail.gmail.com>
	<714288.82822.qm@web34408.mail.mud.yahoo.com>
Message-ID: <d05265cb0806241821s2066123fh366a5e473deeef05@mail.gmail.com>

On Tue, Jun 24, 2008 at 8:05 PM, Lou Pecora <lou_boog2000 at yahoo.com> wrote:
>  It is not an ODE integration problem, really.

Casting the minimization of a functional as a system of ODEs is a
completely reasonable thing to do.

-- 
Nathan Bell wnbell at gmail.com
http://graphics.cs.uiuc.edu/~wnbell/


From zachary.pincus at yale.edu  Tue Jun 24 22:50:20 2008
From: zachary.pincus at yale.edu (Zachary Pincus)
Date: Tue, 24 Jun 2008 22:50:20 -0400
Subject: [SciPy-user] integrate ODE to steady-state?
In-Reply-To: <d05265cb0806241821s2066123fh366a5e473deeef05@mail.gmail.com>
References: <a749952d0806241755tf68720es2488466dea0ddec7@mail.gmail.com>
	<714288.82822.qm@web34408.mail.mud.yahoo.com>
	<d05265cb0806241821s2066123fh366a5e473deeef05@mail.gmail.com>
Message-ID: <1CF00348-252B-4F68-A3DE-5AEA8A8ABBD1@yale.edu>

Hi all,

Thanks for the help!

Nathan:
> I don't know what interface scipy's ode solvers support, but I would
> suggest setting t_final to be a large value and instructing the solver
> to use at most N iterations.  I would use an "adaptive" method like
> RK45 (or whatever scipy supports) which automatically finds a timestep
> that achieves numerical stability and the desired accuracy.
>
> Alternatively, you could use your second proposal, and terminate
> relaxation when the functional is sufficiently small.

I'll try both of these -- I hadn't thought to use the solver's limit  
on iterations, which is a good idea...

Nathan:
> BTW, are you using a Lennard-Jones like potential to distribute points
> ?  I did something similar to distribute points evenly over a mesh
> surface using a simple integrator:
> http://graphics.cs.uiuc.edu/~wnbell/publications/2005-05-SCA-Granular/BeYiMu2005.pdf
>
> An interesting alternative is the CVT:
> http://www.math.psu.edu/qdu/Res/Pic/gallery3.html

My problem is similar, actually, but in 2D. I have an outline of a  
bacteria as a 2D polygon and want to lay down a coordinate system on  
it, with a midline running from pole to pole ("x-axis"), and well- 
defined lines normal to the midline ("y-axis"). The classic solution  
would be to use the medial axis transform, which I had been (using  
Martin Held's excellent if quirky VRONI code). However, the medial  
axis is only a partial midline, it may branch, and there's no  
guarantee that it is possible to find reasonable straight-line normals  
to the axis (you get a clearance radius at each point instead). So I  
formulated the problem as trying to find pairs of positions along the  
outline of the shape where the distance between the pairs is  
minimized, but adjacent points don't "clump" too much, which naturally  
led to this sort of spring-system picture. (This is a very brief  
explanation -- sorry if unclear.) It doesn't seem particularly elegant  
to solve it out like this, but I wanted to see if it worked at all...

Also, thanks for the references -- interesting indeed.

Rob and Lou:
> [try solving for dy/dt=0]

Good idea too! I'll try that one as well, and see which seems to work  
better...


Thanks again, everyone,

Zach





From peridot.faceted at gmail.com  Tue Jun 24 23:57:56 2008
From: peridot.faceted at gmail.com (Anne Archibald)
Date: Tue, 24 Jun 2008 23:57:56 -0400
Subject: [SciPy-user] integrate ODE to steady-state?
In-Reply-To: <1CF00348-252B-4F68-A3DE-5AEA8A8ABBD1@yale.edu>
References: <a749952d0806241755tf68720es2488466dea0ddec7@mail.gmail.com>
	<714288.82822.qm@web34408.mail.mud.yahoo.com>
	<d05265cb0806241821s2066123fh366a5e473deeef05@mail.gmail.com>
	<1CF00348-252B-4F68-A3DE-5AEA8A8ABBD1@yale.edu>
Message-ID: <ce557a360806242057t79c29a2cg2d6d69a74a987c00@mail.gmail.com>

2008/6/24 Zachary Pincus <zachary.pincus at yale.edu>:

> My problem is similar, actually, but in 2D. I have an outline of a
> bacteria as a 2D polygon and want to lay down a coordinate system on
> it, with a midline running from pole to pole ("x-axis"), and well-
> defined lines normal to the midline ("y-axis"). The classic solution
> would be to use the medial axis transform, which I had been (using
> Martin Held's excellent if quirky VRONI code). However, the medial
> axis is only a partial midline, it may branch, and there's no
> guarantee that it is possible to find reasonable straight-line normals
> to the axis (you get a clearance radius at each point instead). So I
> formulated the problem as trying to find pairs of positions along the
> outline of the shape where the distance between the pairs is
> minimized, but adjacent points don't "clump" too much, which naturally
> led to this sort of spring-system picture. (This is a very brief
> explanation -- sorry if unclear.) It doesn't seem particularly elegant
> to solve it out like this, but I wanted to see if it worked at all...
>
> Also, thanks for the references -- interesting indeed.
>
> Rob and Lou:
>> [try solving for dy/dt=0]
>
> Good idea too! I'll try that one as well, and see which seems to work
> better...

I'd set it up as a numerical minimization procedure - minimizing the
"potential energy" of all those springs. If you give it a good initial
guess and are lucky, the minimizer will follow a path down to the
optimum not too different from the one your ODEs would have followed,
but since the optimizer knows you don't care how it gets there it can
be much more efficient. The danger with both approaches is that there
may be some rather bad local minimum they can get trapped in. The best
approach is to choose a decent initial guess - maybe using the medial
axis transform, since you already have the code working - so that the
nearest, most obvious solution is the real optimum. Failing that, you
can look at some of the global optimizers in OpenOpt.

Better, of course, is to find an algorithm specifically adapted to
solving your particular problem, but I gather you've already done some
considerable searching.

Anne


From fredmfp at gmail.com  Wed Jun 25 13:28:31 2008
From: fredmfp at gmail.com (fred)
Date: Wed, 25 Jun 2008 19:28:31 +0200
Subject: [SciPy-user] msked array & histogram...
In-Reply-To: <485BB463.1020600@gmail.com>
References: <485BB0F5.90204@gmail.com> <485BB463.1020600@gmail.com>
Message-ID: <4862803F.2020604@gmail.com>

fred a ?crit :
> fred a ?crit :
> 
>> Is there some workaround ?
> Yes.
> Convert them to NaN.
No.

It does not work at all (say for 10 % of nan)
for masked arrays or arrays with NaN.

Any suggestion ?

I really need some help for this issue.

TIA.


Cheers,

-- 
Fred


From pgmdevlist at gmail.com  Wed Jun 25 13:49:59 2008
From: pgmdevlist at gmail.com (Pierre GM)
Date: Wed, 25 Jun 2008 13:49:59 -0400
Subject: [SciPy-user] msked array & histogram...
In-Reply-To: <4862803F.2020604@gmail.com>
References: <485BB0F5.90204@gmail.com> <485BB463.1020600@gmail.com>
	<4862803F.2020604@gmail.com>
Message-ID: <200806251350.00021.pgmdevlist@gmail.com>

Fred,
If you don't need to process the missing data are using a 1D array or an array 
that can be flattened, just use the `compressed()` method/function on your 
MaskedArray, and send the output to histogram, e.g.
numpy.histogram(your_masked_array.compressed())

If you need to keep track of the number of missing data, or qualify them as 
one particular category, just fill your masked array with a convenient value.



From phaustin at gmail.com  Wed Jun 25 13:54:39 2008
From: phaustin at gmail.com (Phil Austin)
Date: Wed, 25 Jun 2008 10:54:39 -0700
Subject: [SciPy-user] msked array & histogram...
In-Reply-To: <4862803F.2020604@gmail.com>
References: <485BB0F5.90204@gmail.com> <485BB463.1020600@gmail.com>
	<4862803F.2020604@gmail.com>
Message-ID: <4862865F.7050903@gmail.com>

fred wrote:
> fred a ?crit :
>> fred a ?crit :
>>
>>> Is there some workaround ?
>> Yes.
>> Convert them to NaN.
> No.
> 
> It does not work at all (say for 10 % of nan)
> for masked arrays or arrays with NaN.
> 
> Any suggestion ?

ma.compressed?

import numpy.ma as ma
In [8]:help ma.compressed
------>help(ma.compressed)
Help on function compressed in module numpy.ma.core:

compressed(x)
     Return a 1-D array of all the non-masked data.



From lists at vrbka.net  Wed Jun 25 14:20:07 2008
From: lists at vrbka.net (Lubos Vrbka)
Date: Wed, 25 Jun 2008 20:20:07 +0200
Subject: [SciPy-user] How to realize inverse DCT transformation by using
 scipy
In-Reply-To: <90c482ab0806220056l66986450wac3f0ee183703395@mail.gmail.com>
References: <90c482ab0806220056l66986450wac3f0ee183703395@mail.gmail.com>
Message-ID: <48628C57.9080507@vrbka.net>

hi

> How to realize inverse DCT transformation by using scipy?
if by iDCT you mean inverse dicrete cosine transform, you can use this code
http://projects.scipy.org/scipy/scipy/browser/trunk/scipy/sandbox/image/transforms.py

i asked similar question (but regarding DST) few days ago and the 
following travis' reply might be the right thing for you as well.
http://article.gmane.org/gmane.comp.python.scientific.user/16607/match=fourier+sine+transform+scipy

best,

-- 
Lubos _ at _"
http://www.lubos.vrbka.net


From fredmfp at gmail.com  Wed Jun 25 14:46:33 2008
From: fredmfp at gmail.com (fred)
Date: Wed, 25 Jun 2008 20:46:33 +0200
Subject: [SciPy-user] msked array & histogram...
In-Reply-To: <4862865F.7050903@gmail.com>
References: <485BB0F5.90204@gmail.com>
	<485BB463.1020600@gmail.com>	<4862803F.2020604@gmail.com>
	<4862865F.7050903@gmail.com>
Message-ID: <48629289.20900@gmail.com>

Phil Austin a ?crit :

> ma.compressed?
Great !! :-)))

Thank to you two, Pierre & Phil.


Cheers,

-- 
Fred


From williamhpurcell at gmail.com  Thu Jun 26 08:28:59 2008
From: williamhpurcell at gmail.com (William Purcell)
Date: Thu, 26 Jun 2008 07:28:59 -0500
Subject: [SciPy-user] import scipy - py2exe
Message-ID: <d9186c3b0806260528i303b8b85qbc3cd3efc3cf5747@mail.gmail.com>

Sorry for posting in two places for those of you on the py2exe list. The
py2exe list doesn't seem to be as active as this one.

After compiling a small wx gui with py2exe, the following is the traceback I
get when running the resulting .exe.

Traceback (most recent call last):
  File "simple.py", line 2, in <module>
  File "zipextimporter.pyo", line 82, in load_module
  File "scipy\__init__.pyo", line 48, in <module>
TypeError: unsupported operand type(s) for +=: 'NoneType' and 'str'

operating sys.           --> XP
scipy.__version__      --> 0.6.0
wx.__version__         --> 2.8.7.1
py2exe.__version__   --> 0.6.8

The gui script is simple....

import wx, scipy
app = wx.App()
frame = wx.Frame(None, -1, 'simple.py')
frame.Show()
app.MainLoop()

... and runs fine from the command line. I was having this problem with a
larger gui and I think I have narrowed the problem down to this.

Thanks in advance,
Bill
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080626/c11d6b8c/attachment.html>

From williamhpurcell at gmail.com  Thu Jun 26 09:36:40 2008
From: williamhpurcell at gmail.com (William Purcell)
Date: Thu, 26 Jun 2008 08:36:40 -0500
Subject: [SciPy-user] import scipy - py2exe
In-Reply-To: <d9186c3b0806260528i303b8b85qbc3cd3efc3cf5747@mail.gmail.com>
References: <d9186c3b0806260528i303b8b85qbc3cd3efc3cf5747@mail.gmail.com>
Message-ID: <d9186c3b0806260636k4e715d2aj67d906a783a24159@mail.gmail.com>

FYI: If i remove 'import scipy' the gui will compile with py2exe and run
fine.

On Thu, Jun 26, 2008 at 7:28 AM, William Purcell <williamhpurcell at gmail.com>
wrote:

> Sorry for posting in two places for those of you on the py2exe list. The
> py2exe list doesn't seem to be as active as this one.
>
> After compiling a small wx gui with py2exe, the following is the traceback
> I get when running the resulting .exe.
>
> Traceback (most recent call last):
>   File "simple.py", line 2, in <module>
>   File "zipextimporter.pyo", line 82, in load_module
>   File "scipy\__init__.pyo", line 48, in <module>
> TypeError: unsupported operand type(s) for +=: 'NoneType' and 'str'
>
> operating sys.           --> XP
> scipy.__version__      --> 0.6.0
> wx.__version__         --> 2.8.7.1
> py2exe.__version__   --> 0.6.8
>
> The gui script is simple....
>
> import wx, scipy
> app = wx.App()
> frame = wx.Frame(None, -1, 'simple.py')
> frame.Show()
> app.MainLoop()
>
> ... and runs fine from the command line. I was having this problem with a
> larger gui and I think I have narrowed the problem down to this.
>
> Thanks in advance,
> Bill
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080626/c4855b34/attachment.html>

From grs2103 at columbia.edu  Thu Jun 26 11:14:46 2008
From: grs2103 at columbia.edu (Gideon Simpson)
Date: Thu, 26 Jun 2008 11:14:46 -0400
Subject: [SciPy-user] optimize.fsolve output
Message-ID: <89F274AC-C1BB-4FB4-AA04-C52668995987@columbia.edu>

Is there a flag to give fsolve to see the convergence properties?

-gideon



From nwagner at iam.uni-stuttgart.de  Thu Jun 26 12:28:56 2008
From: nwagner at iam.uni-stuttgart.de (Nils Wagner)
Date: Thu, 26 Jun 2008 18:28:56 +0200
Subject: [SciPy-user] optimize.fsolve output
In-Reply-To: <89F274AC-C1BB-4FB4-AA04-C52668995987@columbia.edu>
References: <89F274AC-C1BB-4FB4-AA04-C52668995987@columbia.edu>
Message-ID: <web-103881185@uni-stuttgart.de>

On Thu, 26 Jun 2008 11:14:46 -0400
  Gideon Simpson <grs2103 at columbia.edu> wrote:
> Is there a flag to give fsolve to see the convergence 
>properties?
> 
> -gideon
  
AFAIK fsolve has no callback function like

cg(A, b, x0=None, tol=1.0000000000000001e-05, 
maxiter=None, xtype=None, M=None, callback=None)

So I guess you cannot monitor the convergence behavior
of fsolve.

Please correct me if I am missing something.


Nils


From gael.varoquaux at normalesup.org  Fri Jun 27 01:18:47 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Fri, 27 Jun 2008 07:18:47 +0200
Subject: [SciPy-user] Student sponsorship for the SciPy08 conference
Message-ID: <20080627051847.GM11323@phare.normalesup.org>

We are delighted to announce that the Python Software Foundation has
answered our call and is providing sponsoring to the SciPy08 conference.

We will use this money to sponsor the registration fees and travel for up
to 10 college or graduate students to attend the conference. The PSF did
not provide all the founds required for all 10 students and once again
Enthought Inc. (http://www.enthought.com) is stepping up to fill in.

To apply, please send a short description of what you are studying and
why you?d like to attend to info at enthought.com. Please include telephone
contact information.

Thanks a lot to Travis Vaught from Enthought for bringing this project to
a success.

Please don't hesitate to forward this announcement to anybody who might
be interested.

Ga?l, on behalf of the Scipy08 organisation committee

SciPy coneference site:   http://conference.scipy.org


From soren.skou.nielsen at gmail.com  Fri Jun 27 09:49:45 2008
From: soren.skou.nielsen at gmail.com (=?ISO-8859-1?Q?S=F8ren_Nielsen?=)
Date: Fri, 27 Jun 2008 15:49:45 +0200
Subject: [SciPy-user] weave problems, weave_imp.o no such file or directory
Message-ID: <de1aa14a0806270649v3b041f0p59ef06cb54b4e30a@mail.gmail.com>

Hi,

I've done a fresh install of MinGw 5.14, Python 2.4.4, Scipy 0.6.0, Numpy
1.1.0 and numarray 1.5.2

When I try to use weave i get this:

C:\>test_weave.py
<weave: compiling>
running build_ext
running build_src
building extension "sc_552cccf5dbf4f6eadd273cdcbd5860523" sources
customize Mingw32CCompiler
customize Mingw32CCompiler using build_ext
customize Mingw32CCompiler
customize Mingw32CCompiler using build_ext
building 'sc_552cccf5dbf4f6eadd273cdcbd5860523' extension
compiling C++ sources
C compiler: g++ -mno-cygwin -O2 -Wall

compile options: '-IC:\Python24\lib\site-packages\scipy\weave
-IC:\Python24\lib\
site-packages\scipy\weave\scxx
-IC:\Python24\lib\site-packages\numpy\core\includ
e -IC:\Python24\include -IC:\Python24\PC -c'
g++ -mno-cygwin -O2 -Wall -IC:\Python24\lib\site-packages\scipy\weave
-IC:\Pytho
n24\lib\site-packages\scipy\weave\scxx
-IC:\Python24\lib\site-packages\numpy\cor
e\include -IC:\Python24\include -IC:\Python24\PC -c
C:\Python24\lib\site-package
s\scipy\weave\scxx\weave_imp.cpp -o c:\docume~1\lisear~1\lokale~1\temp\Lise
Arle
th\python24_intermediate\compiler_08edc7e348e1c33f63a33ab500aef08e\Release\pytho
n24\lib\site-packages\scipy\weave\scxx\weave_imp.o
Found executable C:\MinGw\bin\g++.exe
g++.exe:
Arleth\python24_intermediate\compiler_08edc7e348e1c33f63a33ab500aef08e\
Release\python24\lib\site-packages\scipy\weave\scxx\weave_imp.o: No such
file or
 directory
Traceback (most recent call last):
  File "C:\test_weave.py", line 341, in ?
    main()
  File "C:\test_weave.py", line 224, in main
    weave.inline('printf("%d\\n",a);',['a'], verbose=2,
type_converters=converte
rs.blitz)  #, compiler = 'msvc', verbpse=2,
type_converters=converters.blitz, au
to_downcast=0) #'msvc' or 'gcc' or 'mingw32'
  File "C:\Python24\Lib\site-packages\scipy\weave\inline_tools.py", line
338, in
 inline
    auto_downcast = auto_downcast,
  File "C:\Python24\Lib\site-packages\scipy\weave\inline_tools.py", line
447, in
 compile_function
    verbose=verbose, **kw)
  File "C:\Python24\Lib\site-packages\scipy\weave\ext_tools.py", line 365,
in co
mpile
    verbose = verbose, **kw)
  File "C:\Python24\Lib\site-packages\scipy\weave\build_tools.py", line 269,
in
build_extension
    setup(name = module_name, ext_modules = [ext],verbose=verb)
  File "C:\Python24\Lib\site-packages\numpy\distutils\core.py", line 184, in
set
up
    return old_setup(**new_attr)
  File "C:\Python24\lib\distutils\core.py", line 166, in setup
    raise SystemExit, "error: " + str(msg)
distutils.errors.CompileError: error: Command "g++ -mno-cygwin -O2 -Wall
-IC:\Py
thon24\lib\site-packages\scipy\weave
-IC:\Python24\lib\site-packages\scipy\weave
\scxx -IC:\Python24\lib\site-packages\numpy\core\include
-IC:\Python24\include -
IC:\Python24\PC -c
C:\Python24\lib\site-packages\scipy\weave\scxx\weave_imp.cpp
-o c:\docume~1\lisear~1\lokale~1\temp\Lise
Arleth\python24_intermediate\compiler
_08edc7e348e1c33f63a33ab500aef08e\Release\python24\lib\site-packages\scipy\weave
\scxx\weave_imp.o" failed with exit status 1

the test_weave file was something i found on the scipy dev wiki... I also
tried some of my older files that also uses weave and they all give the same
error...

Can anyone help me with this?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080627/305627b4/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: test_weave.py
Type: text/x-python
Size: 11334 bytes
Desc: not available
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080627/305627b4/attachment.py>

From lbolla at gmail.com  Fri Jun 27 10:54:13 2008
From: lbolla at gmail.com (lorenzo bolla)
Date: Fri, 27 Jun 2008 16:54:13 +0200
Subject: [SciPy-user] Radial Basis Function Interpolation
Message-ID: <80c99e790806270754r6b7d8680ndf052020f23bb9ba@mail.gmail.com>

Using scipy.interpolate.Rbf, I found two bugs.
One is a typo in one of the possible values of the 'function' keyword:
'gausian' instead of 'gaussian'.
Another one is the lack of import of numpy.exp, required by 'gaussian'.

The following diff fixes these bugs.

=============================
$ diff rbf_orig.py rbf.py
45c45
< from numpy import sqrt, log, asarray, newaxis, all, dot, float64, eye
---
> from numpy import sqrt, exp, log, asarray, newaxis, all, dot, float64, eye
61c61
<         elif self.function.lower() == 'gausian':
---
>         elif self.function.lower() == 'gaussian':
87c87
<                 'gausian': exp(-(self.epsilon*r)**2)
---
>                 'gaussian': exp(-(self.epsilon*r)**2)
=============================

Lorenzo.

-- 
"Whereof one cannot speak, thereof one must be silent." -- Ludwig
Wittgenstein
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080627/1897bf68/attachment.html>

From rmay31 at gmail.com  Fri Jun 27 14:13:22 2008
From: rmay31 at gmail.com (Ryan May)
Date: Fri, 27 Jun 2008 14:13:22 -0400
Subject: [SciPy-user] 2D Interpolation
Message-ID: <48652DC2.2070807@gmail.com>

Hi,

Can anyone help me use scipy.interpolate correctly.  Here's my problem: 
I'm trying to make a 2D lookup table to save some calculations.  The two 
parameters over which the lookup table is generated are independent and 
I have complete control over how I divide up the domain.  Using this 
lookup table, I'd like to then calculate values over an unstructured set 
of parameter values (ie. a list of pairs of parameter values).  Is there 
a function in scipy.interpolate that can help here? What I'd really like 
to be able to do is generate an interpolator object from my 2D array, 
and then pass a pair of 1D arrays to the object and have it return 1D 
array of values.

Thanks in advance,

Ryan

-- 
Ryan May
Graduate Research Assistant
School of Meteorology
University of Oklahoma


From gael.varoquaux at normalesup.org  Fri Jun 27 14:24:29 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Fri, 27 Jun 2008 20:24:29 +0200
Subject: [SciPy-user] Scipy08 Paper submission deadline extention to Monday
	30th
Message-ID: <20080627182429.GH5103@phare.normalesup.org>

The deadline for submitting abstracts to the Scipy conference was tonight.
In order to give you more time to submit excellent abstracts, the review
committee is extending the deadline to Monday (June 30th), and will work
hastily to get all of them reviewed in time for the program announcement,
on Thursday July 3rd.

----

The SciPy 2008 Conference will be held 21-22 August 2008 at the
California Institute of Technology, Pasadena, California. SciPy is a
scientific computing package, written in the Python language. It is
widely used in research, the industry and academia.

The program features tutorials, contributed papers, lightning talks, and
bird-of-a-feather sessions. We are soliciting talks and accompanying
papers (either formal academic or magazine-style articles) that discuss
topics which center around scientific computing using Python. These
include applications, teaching, future development directions and
research. A collection of peer-reviewed articles will be published as
part of the proceedings.

Proposals for talks are submitted as extended abstracts. There are two
categories of talks:
Lightning talks

These talks are 10 minutes in duration. An abstract of between 300 and
700 words should describe the topic and motivate its relevance to
scientific computing. Lightning talks do not require an accompanying
article (although, if submitted, these will still be published).
Paper presentations

These talks are 35 minutes in duration (including questions). A one page
abstract of no less than 500 words (excluding figures and references)
should give an outline of the final paper. Papers are due two weeks
before the conference, and may be in a formal academic style, or in a
more relaxed magazine-style format.

If you wish to present a talk at the conference, please create an account
on the website http://conference.scipy.org. You may then submit an
abstract by logging in, clicking on your profile and following the "
Submit an abstract " link.

Ga?l, on behalf on the SciPy08 organizing committee.


From jtravs at gmail.com  Fri Jun 27 14:37:47 2008
From: jtravs at gmail.com (John Travers)
Date: Fri, 27 Jun 2008 19:37:47 +0100
Subject: [SciPy-user] 2D Interpolation
In-Reply-To: <48652DC2.2070807@gmail.com>
References: <48652DC2.2070807@gmail.com>
Message-ID: <3a1077e70806271137k4776e229ifd2850e9b460e79a@mail.gmail.com>

On Fri, Jun 27, 2008 at 7:13 PM, Ryan May <rmay31 at gmail.com> wrote:
> Hi,
>
> Can anyone help me use scipy.interpolate correctly.  Here's my problem:
> I'm trying to make a 2D lookup table to save some calculations.  The two
> parameters over which the lookup table is generated are independent and
> I have complete control over how I divide up the domain.  Using this
> lookup table, I'd like to then calculate values over an unstructured set
> of parameter values (ie. a list of pairs of parameter values).  Is there
> a function in scipy.interpolate that can help here? What I'd really like
> to be able to do is generate an interpolator object from my 2D array,
> and then pass a pair of 1D arrays to the object and have it return 1D
> array of values.

This should get you started:

import scipy
import scipy.interpolate

# the two axes
x = scipy.linspace(-1.0,1.0,10)
y = x

# make some pretend data
gridy, gridx = scipy.meshgrid(x,y)
z = scipy.sin(gridx)*scipy.sin(gridy)

# create a spline interpolator
spl = scipy.interpolate.RectBivariateSpline(x,y,z)

# make some new axes to interpolate to
nx = scipy.linspace(-1.0,1.0,100)
ny = nx

# evaluate
nz = spl(nx, ny)

# with matplotlib, compare:
import pylab
pylab.matshow(z)
pylab.matshow(nz)

Cheers,
John


From jtravs at gmail.com  Fri Jun 27 14:40:25 2008
From: jtravs at gmail.com (John Travers)
Date: Fri, 27 Jun 2008 19:40:25 +0100
Subject: [SciPy-user] 2D Interpolation
In-Reply-To: <3a1077e70806271137k4776e229ifd2850e9b460e79a@mail.gmail.com>
References: <48652DC2.2070807@gmail.com>
	<3a1077e70806271137k4776e229ifd2850e9b460e79a@mail.gmail.com>
Message-ID: <3a1077e70806271140y6dc8f6d4qf1f565a8b41f7e93@mail.gmail.com>

On Fri, Jun 27, 2008 at 7:37 PM, John Travers <jtravs at gmail.com> wrote:
> On Fri, Jun 27, 2008 at 7:13 PM, Ryan May <rmay31 at gmail.com> wrote:
>> Hi,
>>
>> Can anyone help me use scipy.interpolate correctly.  Here's my problem:
>> I'm trying to make a 2D lookup table to save some calculations.  The two
>> parameters over which the lookup table is generated are independent and
>> I have complete control over how I divide up the domain.  Using this
>> lookup table, I'd like to then calculate values over an unstructured set
>> of parameter values (ie. a list of pairs of parameter values).  Is there
>> a function in scipy.interpolate that can help here? What I'd really like
>> to be able to do is generate an interpolator object from my 2D array,
>> and then pass a pair of 1D arrays to the object and have it return 1D
>> array of values.
>
> This should get you started:
>

Ahhh, I miss-read your question and answered something else! Not sure
how to answer
your question...
Sorry.
John


From robert.kern at gmail.com  Fri Jun 27 15:38:11 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Fri, 27 Jun 2008 14:38:11 -0500
Subject: [SciPy-user] Radial Basis Function Interpolation
In-Reply-To: <80c99e790806270754r6b7d8680ndf052020f23bb9ba@mail.gmail.com>
References: <80c99e790806270754r6b7d8680ndf052020f23bb9ba@mail.gmail.com>
Message-ID: <3d375d730806271238r17f61699r199d67e9002b696a@mail.gmail.com>

On Fri, Jun 27, 2008 at 09:54, lorenzo bolla <lbolla at gmail.com> wrote:
> Using scipy.interpolate.Rbf, I found two bugs.
> One is a typo in one of the possible values of the 'function' keyword:
> 'gausian' instead of 'gaussian'.
> Another one is the lack of import of numpy.exp, required by 'gaussian'.

Fixed in r4486. Thank you!

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From pav at iki.fi  Fri Jun 27 16:13:21 2008
From: pav at iki.fi (Pauli Virtanen)
Date: Fri, 27 Jun 2008 20:13:21 +0000 (UTC)
Subject: [SciPy-user] 2D Interpolation
References: <48652DC2.2070807@gmail.com>
Message-ID: <g43hl1$6ta$1@ger.gmane.org>

Hi,

Fri, 27 Jun 2008 14:13:22 -0400, Ryan May wrote:
> Can anyone help me use scipy.interpolate correctly.  Here's my problem:
> I'm trying to make a 2D lookup table to save some calculations.  The two
> parameters over which the lookup table is generated are independent and
> I have complete control over how I divide up the domain.  Using this
> lookup table, I'd like to then calculate values over an unstructured set
> of parameter values (ie. a list of pairs of parameter values).  Is there
> a function in scipy.interpolate that can help here? What I'd really like
> to be able to do is generate an interpolator object from my 2D array,
> and then pass a pair of 1D arrays to the object and have it return 1D
> array of values.

I don't think there are currently any functions that do that, but 
certainly we'd like to have them.

I created a Scipy enhancement ticket for this feature:

	http://scipy.org/scipy/scipy/ticket/693

Attached to it is a quick patch that implements the necessary loop on the 
Fortran side. I suspect the patch needs further work, as there possibly 
are faster ways to vectorise this piece of computation than simply 
calling fpbisp at each point separately. (If someone more familiar with 
the spline code wants to bless it, I can commit it, though...)

Currently, you can use

	spl = scipy.interpolate.RectBivariateSpline(xi,yi,zi)
	z = scipy.array([spl(xp, yp)[0,0] for xp, yp in zip(x, y)])

The Python overhead isn't as bad as it looks like; moving the loop to
the Fortran side gains only a factor of 5 improvement in speed.

-- 
Pauli Virtanen



From rmay31 at gmail.com  Fri Jun 27 16:46:30 2008
From: rmay31 at gmail.com (Ryan May)
Date: Fri, 27 Jun 2008 16:46:30 -0400
Subject: [SciPy-user] 2D Interpolation
In-Reply-To: <g43hl1$6ta$1@ger.gmane.org>
References: <48652DC2.2070807@gmail.com> <g43hl1$6ta$1@ger.gmane.org>
Message-ID: <486551A6.60203@gmail.com>

Pauli Virtanen wrote:
> Hi,
> 
> Fri, 27 Jun 2008 14:13:22 -0400, Ryan May wrote:
>> Can anyone help me use scipy.interpolate correctly.  Here's my problem:
>> I'm trying to make a 2D lookup table to save some calculations.  The two
>> parameters over which the lookup table is generated are independent and
>> I have complete control over how I divide up the domain.  Using this
>> lookup table, I'd like to then calculate values over an unstructured set
>> of parameter values (ie. a list of pairs of parameter values).  Is there
>> a function in scipy.interpolate that can help here? What I'd really like
>> to be able to do is generate an interpolator object from my 2D array,
>> and then pass a pair of 1D arrays to the object and have it return 1D
>> array of values.
> 
> I don't think there are currently any functions that do that, but 
> certainly we'd like to have them.
> 
> I created a Scipy enhancement ticket for this feature:
> 
> 	http://scipy.org/scipy/scipy/ticket/693
> 
> Attached to it is a quick patch that implements the necessary loop on the 
> Fortran side. I suspect the patch needs further work, as there possibly 
> are faster ways to vectorise this piece of computation than simply 
> calling fpbisp at each point separately. (If someone more familiar with 
> the spline code wants to bless it, I can commit it, though...)
> 
> Currently, you can use
> 
> 	spl = scipy.interpolate.RectBivariateSpline(xi,yi,zi)
> 	z = scipy.array([spl(xp, yp)[0,0] for xp, yp in zip(x, y)])
> 
> The Python overhead isn't as bad as it looks like; moving the loop to
> the Fortran side gains only a factor of 5 improvement in speed.

Thanks for that.  I'm surprised that the Python looping overhead isn't 
worse. I just automatically assumed looping of Python == death of my 
code.  I'll see what I get then just manually looping.

Thanks

-- 
Ryan May
Graduate Research Assistant
School of Meteorology
University of Oklahoma


From pav at iki.fi  Fri Jun 27 17:38:15 2008
From: pav at iki.fi (Pauli Virtanen)
Date: Fri, 27 Jun 2008 21:38:15 +0000 (UTC)
Subject: [SciPy-user] 2D Interpolation
References: <48652DC2.2070807@gmail.com> <g43hl1$6ta$1@ger.gmane.org>
	<486551A6.60203@gmail.com>
Message-ID: <g43mk7$6ta$2@ger.gmane.org>

Fri, 27 Jun 2008 16:46:30 -0400, Ryan May wrote:

> Pauli Virtanen wrote:
>> Hi,
>> 
>> Fri, 27 Jun 2008 14:13:22 -0400, Ryan May wrote:
>>> Can anyone help me use scipy.interpolate correctly.  Here's my
>>> problem: I'm trying to make a 2D lookup table to save some
>>> calculations.  The two parameters over which the lookup table is
>>> generated are independent and I have complete control over how I
>>> divide up the domain.  Using this lookup table, I'd like to then
>>> calculate values over an unstructured set of parameter values (ie. a
>>> list of pairs of parameter values).  Is there a function in
>>> scipy.interpolate that can help here? What I'd really like to be able
>>> to do is generate an interpolator object from my 2D array, and then
>>> pass a pair of 1D arrays to the object and have it return 1D array of
>>> values.
[clip]

Another hint: looking at

	scipy.ndimage.map_coordinates

may turn out to be useful: it seems to be able to interpolate from a 
regular grid to a vector of coordinates.

-- 
Pauli Virtanen



From travis at enthought.com  Fri Jun 27 19:04:01 2008
From: travis at enthought.com (Travis Vaught)
Date: Fri, 27 Jun 2008 18:04:01 -0500
Subject: [SciPy-user] SciPy Conference Updates
Message-ID: <66889DEE-AFE8-484C-9C15-582457DFD3DB@enthought.com>

Greetings,

The SciPy Conference is not too far away.  I thought I'd summarize  
some recent news about the conference in case some of you missed it:


  - Accommodations (news!): We've negotiated a group rate with a  
nearby Marriott hotel, for those that would like to take advantage of  
it.  The hotel has set up a web site for our event here:

http://cwp.marriott.com/laxot/scipyworkshop/

- Student Sponsorship: As you may have seen, the Python Software  
Foundation has agreed to partner with Enthought to sponsor 10  
students' travel, registration, and accommodation for the tutorials,  
conference and (most importantly) sprints.  If you're in college or a  
graduate program, please check out the details here:

http://conference.scipy.org/sponsoring

  - Abstracts Submission Deadline Extended:  The review committee is  
extending the deadline to Monday, June 30th.  Please see the Call for  
Papers for instructions on abstract submission here:

http://conference.scipy.org/call_for_papers


Please drop me an email if you have any questions or comments.

Best,

Travis


From emanuele at relativita.com  Sat Jun 28 09:46:50 2008
From: emanuele at relativita.com (Emanuele Olivetti)
Date: Sat, 28 Jun 2008 15:46:50 +0200
Subject: [SciPy-user] [OpenOpt] lb issue
Message-ID: <486640CA.8000303@relativita.com>

Dear all and dear Dmitrey,

I experience problems when setting lower bounds (lb) to
a non-linear problem. I have a simple bound for x: being
bigger than zero.

Here follows a simple example that shows the issue: even though
I set "lb=N.zeros(dimensions)", the "ralg" solver tries to compute
f() when x<0. Why?
----
import numpy as N
from scikits.openopt import NLP

size = 100
dimensions = 2
data = N.random.rand(size,dimensions)-0.5

def f(x):
    global data
    if (x<0).sum()>0:
        print "WARNING! Lower bound exceeded, x =",x
        pass
    return N.dot(data**2,x.T)

x0 = N.ones(dimensions)
p = NLP(f,x0,lb=N.zeros(dimensions),ftol=1.0e-3)
p.solve("ralg")
print p.ff,p.xf
----

I'm wondering if I've understood correctly how to use
p.lb. Any explanation will be very appreciated.

Emanuele


P.S.: OpenOpt updated from SVN, NumPy v1.0.3 and SciPy v0.5.2
provided by Ubuntu Gutsy 7.10.



From dmitrey.kroshko at scipy.org  Sat Jun 28 11:11:43 2008
From: dmitrey.kroshko at scipy.org (dmitrey)
Date: Sat, 28 Jun 2008 18:11:43 +0300
Subject: [SciPy-user] [OpenOpt] lb issue
In-Reply-To: <486640CA.8000303@relativita.com>
References: <486640CA.8000303@relativita.com>
Message-ID: <486654AF.8060905@scipy.org>

Hi Emanuele,
some solvers (ALGENCAN, scipy_lbfgsb) don't have to obtain objFunc(x) 
out of the lb-ub region (for box-bound problems), but some other, 
including current ralg implementation, do. Mb I could try to enhance 
ralg handling of the problems of the type during this gsoc, but it 
requires sufficient amount of time and I'm currently busy with other 
chapters from my gsoc shedule.
Regards, D.

Emanuele Olivetti wrote:
> Dear all and dear Dmitrey,
>
> I experience problems when setting lower bounds (lb) to
> a non-linear problem. I have a simple bound for x: being
> bigger than zero.
>
> Here follows a simple example that shows the issue: even though
> I set "lb=N.zeros(dimensions)", the "ralg" solver tries to compute
> f() when x<0. Why?
> ----
> import numpy as N
> from scikits.openopt import NLP
>
> size = 100
> dimensions = 2
> data = N.random.rand(size,dimensions)-0.5
>
> def f(x):
>     global data
>     if (x<0).sum()>0:
>         print "WARNING! Lower bound exceeded, x =",x
>         pass
>     return N.dot(data**2,x.T)
>
> x0 = N.ones(dimensions)
> p = NLP(f,x0,lb=N.zeros(dimensions),ftol=1.0e-3)
> p.solve("ralg")
> print p.ff,p.xf
> ----
>
> I'm wondering if I've understood correctly how to use
> p.lb. Any explanation will be very appreciated.
>
> Emanuele
>
>
> P.S.: OpenOpt updated from SVN, NumPy v1.0.3 and SciPy v0.5.2
> provided by Ubuntu Gutsy 7.10.
>
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>
>
>
>   



From dmitrey.kroshko at scipy.org  Sat Jun 28 14:49:10 2008
From: dmitrey.kroshko at scipy.org (dmitrey)
Date: Sat, 28 Jun 2008 21:49:10 +0300
Subject: [SciPy-user] [OpenOpt] lb issue
In-Reply-To: <486654AF.8060905@scipy.org>
References: <486640CA.8000303@relativita.com> <486654AF.8060905@scipy.org>
Message-ID: <486687A6.5020703@scipy.org>

Hi Emanuele,

I could propose you temporary solution (see below), this one doesn't 
require updating oo from svn. However, usually ALGENCAN, ipopt and 
scipy_slsqp work much better for box-bound constrained problems (w/o 
other constraints) than current ralg implementation.
D.

import numpy as N
from scikits.openopt import NLP
from numpy import any, inf
size = 100
dimensions = 2
data = N.random.rand(size,dimensions)-0.5

contol = 1e-6
lb=N.zeros(dimensions) + contol

def f(x):
    global data
    if any(x<0):
        #objective function is not defined here, let's use inf instead
        #however, some iters will show objFunVa= inf in text output
        # and graphic output is currently unavailable for the case
        return inf
    return N.dot(data**2,x.T)

x0 = N.ones(dimensions)
p = NLP(f,x0,lb=lb, contol = contol)
p.solve('ralg')
print p.ff,p.xf


From tomo.bbe at gmail.com  Sat Jun 28 17:09:55 2008
From: tomo.bbe at gmail.com (James)
Date: Sat, 28 Jun 2008 22:09:55 +0100
Subject: [SciPy-user] The other solvers in ODEPACK
Message-ID: <5a757d050806281409y6db27abck7d223f17c89480df@mail.gmail.com>

Hi all,

Recently I have been using ODEPACK (direct from Netlib) in a pure Fortran
project and have noticed dramatically different execution times between two
of the solvers (LSODA and LSODES).


I have also noticed the source for LSODES is in scipy's repository - so is
it possible to access the other solvers through scipy at the moment ?


Cheers
James
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080628/27d8c1a5/attachment.html>

From robert.kern at gmail.com  Sat Jun 28 17:13:02 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Sat, 28 Jun 2008 16:13:02 -0500
Subject: [SciPy-user] The other solvers in ODEPACK
In-Reply-To: <5a757d050806281409y6db27abck7d223f17c89480df@mail.gmail.com>
References: <5a757d050806281409y6db27abck7d223f17c89480df@mail.gmail.com>
Message-ID: <3d375d730806281413o702f6c85k432d41c7228fcddc@mail.gmail.com>

On Sat, Jun 28, 2008 at 16:09, James <tomo.bbe at gmail.com> wrote:
> Hi all,
>
> Recently I have been using ODEPACK (direct from Netlib) in a pure Fortran
> project and have noticed dramatically different execution times between two
> of the solvers (LSODA and LSODES).
>
>
> I have also noticed the source for LSODES is in scipy's repository - so is
> it possible to access the other solvers through scipy at the moment ?

Someone needs to write the f2py wrappers for them.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From emanuele at relativita.com  Sat Jun 28 18:12:24 2008
From: emanuele at relativita.com (Emanuele Olivetti)
Date: Sun, 29 Jun 2008 00:12:24 +0200
Subject: [SciPy-user] [OpenOpt] lb issue
In-Reply-To: <486654AF.8060905@scipy.org>
References: <486640CA.8000303@relativita.com> <486654AF.8060905@scipy.org>
Message-ID: <4866B748.3060805@relativita.com>

I'm trying scipy_lbfgsb because ALGENCAN requires extra libraries,
which is not desirable in my case. I hope not to incur in noise
instabilities which I know are handled well only by "ralg".

If you can improve ralg in near future I'll be very happy :)

Emanuele

dmitrey wrote:
> Hi Emanuele,
> some solvers (ALGENCAN, scipy_lbfgsb) don't have to obtain objFunc(x) 
> out of the lb-ub region (for box-bound problems), but some other, 
> including current ralg implementation, do. Mb I could try to enhance 
> ralg handling of the problems of the type during this gsoc, but it 
> requires sufficient amount of time and I'm currently busy with other 
> chapters from my gsoc shedule.
> Regards, D.
>
> Emanuele Olivetti wrote:
>> Dear all and dear Dmitrey,
>>
>> I experience problems when setting lower bounds (lb) to
>> a non-linear problem. I have a simple bound for x: being
>> bigger than zero.
>>
>> ...


From emanuele at relativita.com  Sat Jun 28 18:17:42 2008
From: emanuele at relativita.com (Emanuele Olivetti)
Date: Sun, 29 Jun 2008 00:17:42 +0200
Subject: [SciPy-user] [OpenOpt] lb issue
In-Reply-To: <486687A6.5020703@scipy.org>
References: <486640CA.8000303@relativita.com> <486654AF.8060905@scipy.org>
	<486687A6.5020703@scipy.org>
Message-ID: <4866B886.2050205@relativita.com>

Unfortunately it is not so simple to map this advice to my
real situation, which is more complex than the
proof-of-concept example of the initial message. Returning
a big positive value when x is outside the bounds is an option
I considered some time ago but then discarded. But I'll think
more about it now.

Ciao,

Emanuele

dmitrey wrote:
> Hi Emanuele,
>
> I could propose you temporary solution (see below), this one doesn't
> require updating oo from svn. However, usually ALGENCAN, ipopt and
> scipy_slsqp work much better for box-bound constrained problems (w/o
> other constraints) than current ralg implementation.
> D.
>
> import numpy as N
> from scikits.openopt import NLP
> from numpy import any, inf
> size = 100
> dimensions = 2
> data = N.random.rand(size,dimensions)-0.5
>
> contol = 1e-6
> lb=N.zeros(dimensions) + contol
>
> def f(x):
>    global data
>    if any(x<0):
>        #objective function is not defined here, let's use inf instead
>        #however, some iters will show objFunVa= inf in text output
>        # and graphic output is currently unavailable for the case
>        return inf
>    return N.dot(data**2,x.T)
>
> x0 = N.ones(dimensions)
> p = NLP(f,x0,lb=lb, contol = contol)
> p.solve('ralg')
> print p.ff,p.xf
>



From kdsudac at yahoo.com  Sat Jun 28 18:39:10 2008
From: kdsudac at yahoo.com (Keith Suda-Cederquist)
Date: Sat, 28 Jun 2008 15:39:10 -0700 (PDT)
Subject: [SciPy-user] Fastest Way to element-by-element operations in
	SciPy/NumPy
Message-ID: <886240.18505.qm@web54302.mail.re2.yahoo.com>

Hi All,

I'm a relatively new Python/SciPy/NumPy user who migrated from Matlab.? I've managed to put together some code that does some image processing on 2000x2000 sized images.? It all works well, but one of the image processing steps takes a long time and I'd like to try to speed it up.? I can think of a few ways that *might* speed it up, but I figured it'd? be good to ask the experts on how they would recommend to do it.

As is I import the image into an NumPy 2d array using PIL.? For each row, I do some signal processing to locate the zero crossing in-between a local maxima and a local minima (an edge detection algorithm).? So roughly my code is structured like this:


--Start Code
imarray=im2array(filename)? #reads file into array
steparray=scipy.zeros(shape(imarray))? #initialize array that will contain information of edge
? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? # locations

for row in xrange(0,shape(imarray)[0]):
???? imrow=imarray[row,:]
???? #some basic code that identifies the local minima and maxima then identifies a
???? columns that are first-guess zero crossing
???? for col in first_guesses:
????????? window=imrow[col-3:col+4]? #window of data around first guess zero crossing
????????? #code that does a scipy.polyfit operation to identify more precisely the????? 
????????? #zero-crossing.?? The zero-crossing from the fit is fit_zcross
????????? steparray[row,col]=fit_zcross
--End Code

As I said I'm new so my code is definitely not very 'pythonic', but I'm trying to learn how to do things better.

My two guesses for how to speed things up are:
1)? initializing the step array and then assigniging different values to different rows and columns of that array is probably a slow way of doing things?  
2)? write a function that takes as input a single row of the imarray and outputs an array giving the edge crossings.? Then use list comprehension to build the 2d array with my function

Am I on the right track, or would you suggest a different approach to speeding things up?

As is, this part of my code takes about 150 seconds to run, so for the 2000 rows that means 75ms per row.? So maybe my array is just too big and will take awhile to process.

Thanks in advance for your help.

Sincerely,
Keith



      
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080628/45d77e33/attachment.html>

From ted.sandler at gmail.com  Sat Jun 28 22:18:00 2008
From: ted.sandler at gmail.com (Ted Sandler)
Date: Sat, 28 Jun 2008 22:18:00 -0400
Subject: [SciPy-user] accessing only the nonzero elements of a sparse matrix
Message-ID: <fbd71dab0806281918r46daeb39rc34820080d898629@mail.gmail.com>

It doesn't seem like anything like the numpy arrays method ``nonzero''
is implemented for scipy's sparse matrices.  What I am looking for is
something like:

  A = sparse.spidentity(3);
  A.nonzero()
  ==>
    (array([0, 1, 2]), array([0, 1, 2]))

I guess I could convert to coordinate form 'COO' and then access the
"row" and "col" fields but this seems a little evil as it relies on
properties of the 'COO' matrix which could change in the future.  It
would be nicer to have a uniform method of all sparse matrix types.

Thanks for any info on this,
-Ted


From peridot.faceted at gmail.com  Sun Jun 29 02:01:31 2008
From: peridot.faceted at gmail.com (Anne Archibald)
Date: Sun, 29 Jun 2008 02:01:31 -0400
Subject: [SciPy-user] The other solvers in ODEPACK
In-Reply-To: <5a757d050806281409y6db27abck7d223f17c89480df@mail.gmail.com>
References: <5a757d050806281409y6db27abck7d223f17c89480df@mail.gmail.com>
Message-ID: <ce557a360806282301q3495f655n93ff2890213dc8b9@mail.gmail.com>

2008/6/28 James <tomo.bbe at gmail.com>:

> Recently I have been using ODEPACK (direct from Netlib) in a pure Fortran
> project and have noticed dramatically different execution times between two
> of the solvers (LSODA and LSODES).
>
>
> I have also noticed the source for LSODES is in scipy's repository - so is
> it possible to access the other solvers through scipy at the moment ?

No, it's not, as Robert said. I have written a (partial) wrapper for
one of them (LSODAR) before deciding it wasn't very useful. I do have
a few comments:

* The FORTRAN interface is a colossal pain to use in a python program,
so you'd want to come up with some sort of python wrapper. There are
two very different examples in scipy already, odeint() and the ode
class. the latter of which was intended to support multiple backends
(but only one was implemented). Neither interface did what I wanted,
though, which was part of my motivation for trying to wrap things
myself.

* If your concern is straight runtime, you should be aware that for
ODE solvers implemented in an ODEPACK-like way, where the right-hand
side is an arbitrary python function, there is a very substantial
overhead associated with calling back into python for each RHS
evaluation. I realize that different algorithms may conceivably make a
bigger difference than this, but do be aware that solveing ODEs in
python in this way is quite a slow process. For this reason I would be
much more interesting in wrapping an ODE solver that was more powerful
- that is, allowed solution of more problems - than one that was more
efficient.

* If you want a more full-featured ODE solver, you should look at
PyDSTool. As its name suggests, it is oriented towards the study of
dynamical systems, but it provides many useful tools for simply
solving ODEs. High on the list is a system for providing symbolic RHSs
which can be compiled to C and (I think) automatically differentiated.
It also includes some more modern ODE solvers (though I seem to recall
none of them support automatic stiff/non-stiff switching). There was
some discussion of integrating some of this into scipy.

Anne


From dmitrey.kroshko at scipy.org  Sun Jun 29 02:20:03 2008
From: dmitrey.kroshko at scipy.org (dmitrey)
Date: Sun, 29 Jun 2008 09:20:03 +0300
Subject: [SciPy-user] [OpenOpt] lb issue
Message-ID: <48672993.1040001@scipy.org>

I don't see any difficulties to map my advice to more difficult cases 
(just don't forget to increase your ineq constraints by contol, for to 
compare those ones with zero, not contol, and use inf, not the huge 
value you have mentioned). Current ralg implementation doesn't need 
objfunc value when point is outside of feasible region (i.e. if *any* 
constraint is bigger than p.contol, not only lb-ub constraint). OO calls 
objFunc outside feasible region just for to check some stop criteria, 
yield iter output point text and possible graphics output.

Regards, D.

Emanuele Olivetti wrote:
> Unfortunately it is not so simple to map this advice to my
> real situation, which is more complex than the
> proof-of-concept example of the initial message. Returning
> a big positive value when x is outside the bounds is an option
> I considered some time ago but then discarded. But I'll think
> more about it now.
>
> Ciao,
>
> Emanuele
>
> dmitrey wrote:
>   
>> Hi Emanuele,
>>
>> I could propose you temporary solution (see below), this one doesn't
>> require updating oo from svn. However, usually ALGENCAN, ipopt and
>> scipy_slsqp work much better for box-bound constrained problems (w/o
>> other constraints) than current ralg implementation.
>> D.
>>
>> import numpy as N
>> from scikits.openopt import NLP
>> from numpy import any, inf
>> size = 100
>> dimensions = 2
>> data = N.random.rand(size,dimensions)-0.5
>>
>> contol = 1e-6
>> lb=N.zeros(dimensions) + contol
>>
>> def f(x):
>>    global data
>>    if any(x<0):
>>        #objective function is not defined here, let's use inf instead
>>        #however, some iters will show objFunVa= inf in text output
>>        # and graphic output is currently unavailable for the case
>>        return inf
>>    return N.dot(data**2,x.T)
>>
>> x0 = N.ones(dimensions)
>> p = NLP(f,x0,lb=lb, contol = contol)
>> p.solve('ralg')
>> print p.ff,p.xf
>>
>>     
>
>
>
>
>   



From rmay31 at gmail.com  Sun Jun 29 14:00:06 2008
From: rmay31 at gmail.com (Ryan May)
Date: Sun, 29 Jun 2008 14:00:06 -0400
Subject: [SciPy-user] 2D Interpolation
In-Reply-To: <g43mk7$6ta$2@ger.gmane.org>
References: <48652DC2.2070807@gmail.com>
	<g43hl1$6ta$1@ger.gmane.org>	<486551A6.60203@gmail.com>
	<g43mk7$6ta$2@ger.gmane.org>
Message-ID: <4867CDA6.8070108@gmail.com>

Pauli Virtanen wrote:
> Fri, 27 Jun 2008 16:46:30 -0400, Ryan May wrote:
> 
>> Pauli Virtanen wrote:
>>> Hi,
>>>
>>> Fri, 27 Jun 2008 14:13:22 -0400, Ryan May wrote:
>>>> Can anyone help me use scipy.interpolate correctly.  Here's my
>>>> problem: I'm trying to make a 2D lookup table to save some
>>>> calculations.  The two parameters over which the lookup table is
>>>> generated are independent and I have complete control over how I
>>>> divide up the domain.  Using this lookup table, I'd like to then
>>>> calculate values over an unstructured set of parameter values (ie. a
>>>> list of pairs of parameter values).  Is there a function in
>>>> scipy.interpolate that can help here? What I'd really like to be able
>>>> to do is generate an interpolator object from my 2D array, and then
>>>> pass a pair of 1D arrays to the object and have it return 1D array of
>>>> values.
> [clip]
> 
> Another hint: looking at
> 
> 	scipy.ndimage.map_coordinates
> 
> may turn out to be useful: it seems to be able to interpolate from a 
> regular grid to a vector of coordinates.
> 

Jackpot!  This works really well.  There's a good example on using it here:

http://www.scipy.org/Cookbook/Interpolation

Which is important, because I had to read the docstring a dozen times to 
understand what was going on.  It should also be noted that while the 
example interpolates to a regular grid, there's nothing precluding 
interpolating to an irregular collection of points.  What's weird is 
that you need to manually scale the points to which you're interpolating 
  to be floating point indices within the original grid.  This is 
probably due to the ndimage-focused nature of map_coordinates.

Moving some of the functionality of map_coordinates, in a more generic 
fashion, into scipy.interpolate wouldn't be the worst idea in the world. 
  Then again, I don't know if anyone else is planning on improving 
scipy.interpolate to gain this functionality (interpolation to array of 
irregular set of points) in another way.  (I also can't volunteer to 
step up and do it at this time.)

Ryan

-- 
Ryan May
Graduate Research Assistant
School of Meteorology
University of Oklahoma


From travis at enthought.com  Sun Jun 29 14:18:20 2008
From: travis at enthought.com (Travis Vaught)
Date: Sun, 29 Jun 2008 13:18:20 -0500
Subject: [SciPy-user] 2D Interpolation
In-Reply-To: <4867CDA6.8070108@gmail.com>
References: <48652DC2.2070807@gmail.com>
	<g43hl1$6ta$1@ger.gmane.org>	<486551A6.60203@gmail.com>
	<g43mk7$6ta$2@ger.gmane.org> <4867CDA6.8070108@gmail.com>
Message-ID: <C0F1CE4B-4EA2-4826-A0E8-2F75E8ED5BF4@enthought.com>

On Jun 29, 2008, at 1:00 PM, Ryan May wrote:

> Pauli Virtanen wrote:
>> Fri, 27 Jun 2008 16:46:30 -0400, Ryan May wrote:
>>
>>> Pauli Virtanen wrote:
>>>> Hi,
>>>>
>>>> Fri, 27 Jun 2008 14:13:22 -0400, Ryan May wrote:
>>>>> Can anyone help me use scipy.interpolate correctly.  Here's my
>>>>> problem: I'm trying to make a 2D lookup table to save some
>>>>> calculations.  The two parameters over which the lookup table is
>>>>> generated are independent and I have complete control over how I
>>>>> divide up the domain.  Using this lookup table, I'd like to then
>>>>> calculate values over an unstructured set of parameter values  
>>>>> (ie. a
>>>>> list of pairs of parameter values).  Is there a function in
>>>>> scipy.interpolate that can help here? What I'd really like to be  
>>>>> able
>>>>> to do is generate an interpolator object from my 2D array, and  
>>>>> then
>>>>> pass a pair of 1D arrays to the object and have it return 1D  
>>>>> array of
>>>>> values.
>> [clip]
>>
>> Another hint: looking at
>>
>> 	scipy.ndimage.map_coordinates
>>
>> may turn out to be useful: it seems to be able to interpolate from a
>> regular grid to a vector of coordinates.
>>
>
> Jackpot!  This works really well.  There's a good example on using  
> it here:
>
> http://www.scipy.org/Cookbook/Interpolation
>
> Which is important, because I had to read the docstring a dozen  
> times to
> understand what was going on.  It should also be noted that while the
> example interpolates to a regular grid, there's nothing precluding
> interpolating to an irregular collection of points.  What's weird is
> that you need to manually scale the points to which you're  
> interpolating
>  to be floating point indices within the original grid.  This is
> probably due to the ndimage-focused nature of map_coordinates.
>
> Moving some of the functionality of map_coordinates, in a more generic
> fashion, into scipy.interpolate wouldn't be the worst idea in the  
> world.
>  Then again, I don't know if anyone else is planning on improving
> scipy.interpolate to gain this functionality (interpolation to array  
> of
> irregular set of points) in another way.  (I also can't volunteer to
> step up and do it at this time.)
>
> Ryan
>
> -- 
> Ryan May
> Graduate Research Assistant
> School of Meteorology
> University of Oklahoma

Ryan,

It looks like you were bitten by the same organizational problem that  
Shane mentions here (4th paragraph about where to find interpolation):

http://www.vetta.org/2008/05/scipy-some-more-thoughts/

Seems like we should address this both in the documentation and in the  
actual organization/location of the routines.  Thoughts?

Travis




From gael.varoquaux at normalesup.org  Sun Jun 29 14:29:48 2008
From: gael.varoquaux at normalesup.org (Gael Varoquaux)
Date: Sun, 29 Jun 2008 20:29:48 +0200
Subject: [SciPy-user] 2D Interpolation
In-Reply-To: <C0F1CE4B-4EA2-4826-A0E8-2F75E8ED5BF4@enthought.com>
References: <48652DC2.2070807@gmail.com> <g43hl1$6ta$1@ger.gmane.org>
	<486551A6.60203@gmail.com> <g43mk7$6ta$2@ger.gmane.org>
	<4867CDA6.8070108@gmail.com>
	<C0F1CE4B-4EA2-4826-A0E8-2F75E8ED5BF4@enthought.com>
Message-ID: <20080629182948.GA4828@phare.normalesup.org>

On Sun, Jun 29, 2008 at 01:18:20PM -0500, Travis Vaught wrote:
> Seems like we should address this both in the documentation and in the  
> actual organization/location of the routines.  Thoughts?

We need to get the numpy doc server[1] running also for scipy. But before
that we need to move it to scipy.org, elsewhere I'll be in trouble when I
get back home.

Ga?l

[1] http://sd-2116.dedibox.fr/pydocweb/wiki/Front%20Page/


From J.Anderson at hull.ac.uk  Sun Jun 29 14:45:48 2008
From: J.Anderson at hull.ac.uk (Joseph Anderson)
Date: Sun, 29 Jun 2008 19:45:48 +0100
Subject: [SciPy-user] advice on inner product, element by element
References: <c7009a550805291511l4b43d01ai9ab20eaa29b2955a@mail.gmail.com>
Message-ID: <D0C13D29BDA2C84A9F9DC7BD978B67E7079486@dougal.scar.hull.ac.uk>

Hello All,

I'm looking for advice on speeding up what looks to be a simple problem to me, but is illusive due to my numpy-naiveness. I'm attempting to evaluate the inner product of n-frames of a time varying signal with n-frames a time varying matrix. The code below does this using two approaches,  list comprehension and python's map function. Neither approach is particularly fast, which is problematic, as I'm usually needing to do this many times at once--bringing things to a grinding halt.

I'm expecting / hoping there is a much faster, fancy numpy way (clever indexing?) to do the trick. I'm considering having a go at the problem with weave.blitz or weave.inline, but this seems somewhat fussy for solving something that I expect is actually easy, but I'm just missing.

Thanks in advance for the help!



Example code:

# parms
cs = 4                          # number of channels
fs = 60 * 44100         # number of frames (a minute at sr = 44100)

# sigs
a = reshape(arange(fs * cs), (fs, cs))
b = reshape(arange(fs * cs * cs), (fs, cs, cs))

print "a = ", a                 # the 'signal'
print "b = ", b                 # the 'matrices'

print "list comp = ", array([inner(val_a, val_b) for val_a, val_b in zip(a, b)])

print "mapping = ", asarray(map(inner, a, b))






~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr Joseph Anderson
Lecturer in Music

School of Arts and New Media
University of Hull, Scarborough Campus,
Scarborough, North Yorkshire, YO11 3AZ, UK

T: +44.(0)1723.357341 T: +44.(0)1723.357370 F: +44.(0)1723.350815
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

-------------- next part --------------
A non-text attachment was scrubbed...
Name: winmail.dat
Type: application/ms-tnef
Size: 3348 bytes
Desc: not available
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080629/f5c05be3/attachment.bin>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: not available
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080629/f5c05be3/attachment.ksh>

From pav at iki.fi  Sun Jun 29 15:27:35 2008
From: pav at iki.fi (Pauli Virtanen)
Date: Sun, 29 Jun 2008 19:27:35 +0000 (UTC)
Subject: [SciPy-user] advice on inner product, element by element
References: <c7009a550805291511l4b43d01ai9ab20eaa29b2955a@mail.gmail.com>
	<D0C13D29BDA2C84A9F9DC7BD978B67E7079486@dougal.scar.hull.ac.uk>
Message-ID: <g48nn7$b45$1@ger.gmane.org>

Sun, 29 Jun 2008 19:45:48 +0100, Joseph Anderson wrote:
[clip]
> Example code:
> 
> # parms
> cs = 4
> fs = 60 * 44100    
> 
> a = reshape(arange(fs * cs), (fs, cs)) 
> b = reshape(arange(fs * cs * cs), (fs, cs, cs))
> 
> c1 = array([inner(val_a, val_b) for val_a, val_b in zip(a, b)])
> c2 = asarray(map(inner, a, b))

Something like this:

	c3 = (a[:,None,:]*b).sum(axis=2)

Note that "None" is the same as "newaxis". (Also note that inner
does an inner product along the last dimension by default, but you
probably knew that.)

Timings:

In [49]: %time c1 = array([inner(val_a, val_b) for val_a, val_b in zip(a, b)])
CPU times: user 1.07 s, sys: 0.04 s, total: 1.10 s
Wall time: 1.34 s

In [51]: %time c2 = asarray(map(inner, a, b))
CPU times: user 0.65 s, sys: 0.01 s, total: 0.66 s
Wall time: 0.82 s

In [53]: %time c3 = (a[:,None,:]*b).sum(axis=2)
CPU times: user 0.07 s, sys: 0.00 s, total: 0.07 s
Wall time: 0.13 s

In [60]: a[:,None,:].shape
Out[60]: (88200, 1, 4)

In [61]: b.shape
Out[61]: (88200, 4, 4)

In [55]: allclose(c1, c2)
Out[55]: True

In [56]: allclose(c1, c3)
Out[56]: True



From rmay31 at gmail.com  Sun Jun 29 15:30:51 2008
From: rmay31 at gmail.com (Ryan May)
Date: Sun, 29 Jun 2008 15:30:51 -0400
Subject: [SciPy-user] 2D Interpolation
In-Reply-To: <C0F1CE4B-4EA2-4826-A0E8-2F75E8ED5BF4@enthought.com>
References: <48652DC2.2070807@gmail.com>	<g43hl1$6ta$1@ger.gmane.org>	<486551A6.60203@gmail.com>	<g43mk7$6ta$2@ger.gmane.org>
	<4867CDA6.8070108@gmail.com>
	<C0F1CE4B-4EA2-4826-A0E8-2F75E8ED5BF4@enthought.com>
Message-ID: <4867E2EB.40404@gmail.com>

>> Jackpot!  This works really well.  There's a good example on using  
>> it here:
>>
>> http://www.scipy.org/Cookbook/Interpolation
>>
>> Which is important, because I had to read the docstring a dozen  
>> times to
>> understand what was going on.  It should also be noted that while the
>> example interpolates to a regular grid, there's nothing precluding
>> interpolating to an irregular collection of points.  What's weird is
>> that you need to manually scale the points to which you're  
>> interpolating
>>  to be floating point indices within the original grid.  This is
>> probably due to the ndimage-focused nature of map_coordinates.
>>
>> Moving some of the functionality of map_coordinates, in a more generic
>> fashion, into scipy.interpolate wouldn't be the worst idea in the  
>> world.
>>  Then again, I don't know if anyone else is planning on improving
>> scipy.interpolate to gain this functionality (interpolation to array  
>> of
>> irregular set of points) in another way.  (I also can't volunteer to
>> step up and do it at this time.)
>>
> Ryan,
> 
> It looks like you were bitten by the same organizational problem that  
> Shane mentions here (4th paragraph about where to find interpolation):
> 
> http://www.vetta.org/2008/05/scipy-some-more-thoughts/
> 
> Seems like we should address this both in the documentation and in the  
> actual organization/location of the routines.  Thoughts?
> 
> Travis

I think docs would be a good first step.  Ndimage definitely wasn't an 
area that I thought of, and even if I did look there, "map_coordinates" 
is not a name I associate with interpolation.  As far as a 
reorganization is concerned, I think a refactoring is more likely what 
is needed.  Even though it manages to do what I need, it still feels 
like I'm abusing the function a bit rather than it being a smooth 
solution to my problem.  I think what would be nice would be taking the 
functionality available in map_coordinates and wrapping it with the 
existing API's in scipy.interpolate (either OO-based or procedural). 
Obviously, this is a bit of work, and since I don't have the time ATM, I 
can't push too hard here. :)

I think a good interim solution would to at least add a see also link to 
map_coordinates from some of the relevant scipy.interpolate docstrings. 
  I think cleaning up the map_coordinates docstring and maybe adding 
some better examples would also help.

My 0.02.

Ryan

-- 
Ryan May
Graduate Research Assistant
School of Meteorology
University of Oklahoma


From wnbell at gmail.com  Mon Jun 30 04:28:11 2008
From: wnbell at gmail.com (Nathan Bell)
Date: Mon, 30 Jun 2008 03:28:11 -0500
Subject: [SciPy-user] accessing only the nonzero elements of a sparse
	matrix
In-Reply-To: <fbd71dab0806281918r46daeb39rc34820080d898629@mail.gmail.com>
References: <fbd71dab0806281918r46daeb39rc34820080d898629@mail.gmail.com>
Message-ID: <d05265cb0806300128m765c188andd6ef316f5f1e83a@mail.gmail.com>

On Sat, Jun 28, 2008 at 9:18 PM, Ted Sandler <ted.sandler at gmail.com> wrote:
>
> I guess I could convert to coordinate form 'COO' and then access the
> "row" and "col" fields but this seems a little evil as it relies on
> properties of the 'COO' matrix which could change in the future.  It
> would be nicer to have a uniform method of all sparse matrix types.
>
> Thanks for any info on this,

Ted, you are right that we should have a nonzero() member function.
Currently the best you can do is, as you say, converting to COO and
taking the .row,.col, and .data members.

I don't see the names of these variables changing, but we'll certainly
add .nonzero() in the future anyway.


-- 
Nathan Bell wnbell at gmail.com
http://graphics.cs.uiuc.edu/~wnbell/


From J.Anderson at hull.ac.uk  Mon Jun 30 11:38:56 2008
From: J.Anderson at hull.ac.uk (Joseph Anderson)
Date: Mon, 30 Jun 2008 16:38:56 +0100
Subject: [SciPy-user] advice on inner product, element by element
References: <c7009a550805291511l4b43d01ai9ab20eaa29b2955a@mail.gmail.com><D0C13D29BDA2C84A9F9DC7BD978B67E7079486@dougal.scar.hull.ac.uk>
	<g48nn7$b45$1@ger.gmane.org>
Message-ID: <D0C13D29BDA2C84A9F9DC7BD978B67E7079487@dougal.scar.hull.ac.uk>

Ah, ha. . . thanks Pauli, this was just the sort of thing I was hoping to see. . . the trouble being I'm still tending to think in terms of loops.

Thanks for the help!!


My best,

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Dr Joseph Anderson
Lecturer in Music

School of Arts and New Media
University of Hull, Scarborough Campus,
Scarborough, North Yorkshire, YO11 3AZ, UK

T: +44.(0)1723.357341 T: +44.(0)1723.357370 F: +44.(0)1723.350815
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~





-----Original Message-----
From: scipy-user-bounces at scipy.org on behalf of Pauli Virtanen
Sent: Sun 06/29/2008 8:27 PM
To: scipy-user at scipy.org
Subject: Re: [SciPy-user] advice on inner product, element by element
 
Sun, 29 Jun 2008 19:45:48 +0100, Joseph Anderson wrote:
[clip]
> Example code:
> 
> # parms
> cs = 4
> fs = 60 * 44100    
> 
> a = reshape(arange(fs * cs), (fs, cs)) 
> b = reshape(arange(fs * cs * cs), (fs, cs, cs))
> 
> c1 = array([inner(val_a, val_b) for val_a, val_b in zip(a, b)])
> c2 = asarray(map(inner, a, b))

Something like this:

	c3 = (a[:,None,:]*b).sum(axis=2)

Note that "None" is the same as "newaxis". (Also note that inner
does an inner product along the last dimension by default, but you
probably knew that.)

Timings:

In [49]: %time c1 = array([inner(val_a, val_b) for val_a, val_b in zip(a, b)])
CPU times: user 1.07 s, sys: 0.04 s, total: 1.10 s
Wall time: 1.34 s

In [51]: %time c2 = asarray(map(inner, a, b))
CPU times: user 0.65 s, sys: 0.01 s, total: 0.66 s
Wall time: 0.82 s

In [53]: %time c3 = (a[:,None,:]*b).sum(axis=2)
CPU times: user 0.07 s, sys: 0.00 s, total: 0.07 s
Wall time: 0.13 s

In [60]: a[:,None,:].shape
Out[60]: (88200, 1, 4)

In [61]: b.shape
Out[61]: (88200, 4, 4)

In [55]: allclose(c1, c2)
Out[55]: True

In [56]: allclose(c1, c3)
Out[56]: True

_______________________________________________
SciPy-user mailing list
SciPy-user at scipy.org
http://projects.scipy.org/mailman/listinfo/scipy-user

-------------- next part --------------
A non-text attachment was scrubbed...
Name: winmail.dat
Type: application/ms-tnef
Size: 3669 bytes
Desc: not available
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080630/6d8c3be2/attachment.bin>
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: not available
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080630/6d8c3be2/attachment.ksh>

From grs2103 at columbia.edu  Mon Jun 30 13:45:29 2008
From: grs2103 at columbia.edu (Gideon Simpson)
Date: Mon, 30 Jun 2008 13:45:29 -0400
Subject: [SciPy-user] single elements and arrays
Message-ID: <C24B3A1D-74FD-4531-BC93-34364B2F2670@columbia.edu>

Suppose I have a function foo

def foo(x):
	blah

and i have written foo in a vectorized fashion such that x can be a  
numpy array of #'s.  But suppose I also want to be able to do

foo(5.5)

where i am interested in the value at a single point.

Is there an easy way to handle both cases?  Should I just not permit  
this sort of ambiguity (force everything to be an array)?


-gideon



From robert.kern at gmail.com  Mon Jun 30 13:52:20 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Mon, 30 Jun 2008 12:52:20 -0500
Subject: [SciPy-user] single elements and arrays
In-Reply-To: <C24B3A1D-74FD-4531-BC93-34364B2F2670@columbia.edu>
References: <C24B3A1D-74FD-4531-BC93-34364B2F2670@columbia.edu>
Message-ID: <3d375d730806301052k703db955iac318c6bcfa7f200@mail.gmail.com>

On Mon, Jun 30, 2008 at 12:45, Gideon Simpson <grs2103 at columbia.edu> wrote:
> Suppose I have a function foo
>
> def foo(x):
>        blah
>
> and i have written foo in a vectorized fashion such that x can be a
> numpy array of #'s.  But suppose I also want to be able to do
>
> foo(5.5)
>
> where i am interested in the value at a single point.
>
> Is there an easy way to handle both cases?  Should I just not permit
> this sort of ambiguity (force everything to be an array)?

It depends. If you are just using ufuncs inside, you don't have to do
anything. Your function will work just fine with arrays and scalars.
If you're doing something else, like looking at .shape or using .sum()
or something, then you might need to actually do something. Exactly
what depends on the details, though. Can you provide more details
about what your function does?

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco


From grs2103 at columbia.edu  Mon Jun 30 14:14:06 2008
From: grs2103 at columbia.edu (Gideon Simpson)
Date: Mon, 30 Jun 2008 14:14:06 -0400
Subject: [SciPy-user] single elements and arrays
In-Reply-To: <3d375d730806301052k703db955iac318c6bcfa7f200@mail.gmail.com>
References: <C24B3A1D-74FD-4531-BC93-34364B2F2670@columbia.edu>
	<3d375d730806301052k703db955iac318c6bcfa7f200@mail.gmail.com>
Message-ID: <02AA0899-7A57-4CD3-8DF3-388135BD5D0A@columbia.edu>

foo finds a root where x is a parameter in the equation to be solved.   
If x is an array, I iterate through the elements of the array.



-gideon

On Jun 30, 2008, at 1:52 PM, Robert Kern wrote:

> On Mon, Jun 30, 2008 at 12:45, Gideon Simpson <grs2103 at columbia.edu>  
> wrote:
>> Suppose I have a function foo
>>
>> def foo(x):
>>       blah
>>
>> and i have written foo in a vectorized fashion such that x can be a
>> numpy array of #'s.  But suppose I also want to be able to do
>>
>> foo(5.5)
>>
>> where i am interested in the value at a single point.
>>
>> Is there an easy way to handle both cases?  Should I just not permit
>> this sort of ambiguity (force everything to be an array)?
>
> It depends. If you are just using ufuncs inside, you don't have to do
> anything. Your function will work just fine with arrays and scalars.
> If you're doing something else, like looking at .shape or using .sum()
> or something, then you might need to actually do something. Exactly
> what depends on the details, though. Can you provide more details
> about what your function does?
>
> -- 
> Robert Kern
>
> "I have come to believe that the whole world is an enigma, a harmless
> enigma that is made terrible by our own mad attempt to interpret it as
> though it had an underlying truth."
> -- Umberto Eco
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user



From robert.kern at gmail.com  Mon Jun 30 14:57:18 2008
From: robert.kern at gmail.com (Robert Kern)
Date: Mon, 30 Jun 2008 13:57:18 -0500
Subject: [SciPy-user] single elements and arrays
In-Reply-To: <02AA0899-7A57-4CD3-8DF3-388135BD5D0A@columbia.edu>
References: <C24B3A1D-74FD-4531-BC93-34364B2F2670@columbia.edu>
	<3d375d730806301052k703db955iac318c6bcfa7f200@mail.gmail.com>
	<02AA0899-7A57-4CD3-8DF3-388135BD5D0A@columbia.edu>
Message-ID: <3d375d730806301157t13d8d3eejc62b1f9ced005ed7@mail.gmail.com>

On Mon, Jun 30, 2008 at 13:14, Gideon Simpson <grs2103 at columbia.edu> wrote:
> foo finds a root where x is a parameter in the equation to be solved.
> If x is an array, I iterate through the elements of the array.

In that case, just special case it. Use numpy.isscalar() to do the test.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco