From kenneth.odoh at gmail.com Tue Nov 7 09:12:05 2017 From: kenneth.odoh at gmail.com (Kenneth Odoh) Date: Tue, 7 Nov 2017 06:12:05 -0800 Subject: [SciPy-Dev] pySmooth: An Object-oriented Library for Time Series Analysis In-Reply-To: References: Message-ID: Dear Sir, I was thinking of adding a time series library to scipy. The Python ecosystem seems to be lacking in time series tool in comparison to their R counterpart. I had written a library that implements a number of popular algorithms, alongside some exotic ones. Pysmooth ( https://github.com/kenluck2001/pySmooth ) is a new time series library that I had written that implements the following algorithms. - Time difference model - Online ARIMA Model - Discrete Kalman Filter - Extended Kalman Filter - Unscented Kalman Filter The design goals for the project consist of the following. - A real-time model for time series analysis. - A less computational intensive model thereby resulting in scalability. - A support for multivariate / univariate time series data. Kindly let me know of your opinion on this package. Kenneth Odoh Software Engineer ############################################################ ################### All communication through this channel is considered confidential, and should be deleted by the receiving party if it is a wrong recipient. Thanks. -------------- next part -------------- An HTML attachment was scrubbed... URL: From josef.pktd at gmail.com Tue Nov 7 09:23:52 2017 From: josef.pktd at gmail.com (josef.pktd at gmail.com) Date: Tue, 7 Nov 2017 09:23:52 -0500 Subject: [SciPy-Dev] pySmooth: An Object-oriented Library for Time Series Analysis In-Reply-To: References: Message-ID: On Tue, Nov 7, 2017 at 9:12 AM, Kenneth Odoh wrote: > > > Dear Sir, > > I was thinking of adding a time series library to scipy. The Python > ecosystem seems to be lacking in time series tool in comparison to their R > counterpart. I had written a library that implements a number of popular > algorithms, alongside some exotic ones. > > maybe not as much as R has, but there is quite a bit in Python http://www.statsmodels.org/devel/tsa.html http://www.statsmodels.org/devel/statespace.html https://github.com/RJT1990/pyflux https://github.com/bashtage/arch and a few more, e.g. https://github.com/MaxBenChrist/awesome_time_series_in_python Josef > > Pysmooth ( https://github.com/kenluck2001/pySmooth ) is a new time series > library that I had written that implements the following algorithms. > > > - Time difference model > - Online ARIMA Model > - Discrete Kalman Filter > - Extended Kalman Filter > - Unscented Kalman Filter > > The design goals for the project consist of the following. > > - A real-time model for time series analysis. > - A less computational intensive model thereby resulting in > scalability. > - A support for multivariate / univariate time series data. > > Kindly let me know of your opinion on this package. > > > Kenneth Odoh > Software Engineer > > > ############################################################ > ################### > All communication through this channel is considered confidential, and > should be deleted by the receiving party if it is a wrong recipient. Thanks. > > > > _______________________________________________ > SciPy-Dev mailing list > SciPy-Dev at python.org > https://mail.python.org/mailman/listinfo/scipy-dev > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From ralf.gommers at gmail.com Thu Nov 9 03:15:25 2017 From: ralf.gommers at gmail.com (Ralf Gommers) Date: Thu, 9 Nov 2017 21:15:25 +1300 Subject: [SciPy-Dev] welcome Antonio to the core team Message-ID: Hi all, On behalf of the SciPy developers I'd like to welcome Antonio Horta Ribeiro as a member of the core dev team. Antonio has been contributing to scipy.optimize and scipy.signal for a year now, and has recently finished his GSoC project on a large scale optimization algorithm ( https://antonior92.github.io/posts/2017/24/GSoC-submission/). See https://github.com/scipy/scipy/pulls/antonior92 for his contributions so far; I'm looking forward to more to come! Cheers, Ralf -------------- next part -------------- An HTML attachment was scrubbed... URL: From josh.craig.wilson at gmail.com Thu Nov 9 10:21:42 2017 From: josh.craig.wilson at gmail.com (Joshua Wilson) Date: Thu, 9 Nov 2017 09:21:42 -0600 Subject: [SciPy-Dev] welcome Antonio to the core team In-Reply-To: References: Message-ID: Keep up the good work Antonio! On Thu, Nov 9, 2017 at 2:15 AM, Ralf Gommers wrote: > Hi all, > > On behalf of the SciPy developers I'd like to welcome Antonio Horta Ribeiro > as a member of the core dev team. > > Antonio has been contributing to scipy.optimize and scipy.signal for a year > now, and has recently finished his GSoC project on a large scale > optimization algorithm > (https://antonior92.github.io/posts/2017/24/GSoC-submission/). See > https://github.com/scipy/scipy/pulls/antonior92 for his contributions so > far; I'm looking forward to more to come! > > Cheers, > Ralf > > > _______________________________________________ > SciPy-Dev mailing list > SciPy-Dev at python.org > https://mail.python.org/mailman/listinfo/scipy-dev > From josh.craig.wilson at gmail.com Thu Nov 9 10:28:11 2017 From: josh.craig.wilson at gmail.com (Joshua Wilson) Date: Thu, 9 Nov 2017 09:28:11 -0600 Subject: [SciPy-Dev] Old release still on SourceForge Message-ID: Hey everyone, It seems that SciPy 0.16 is still up on SourceForge: https://sourceforge.net/projects/scipy/ According to the site, people are still downloading it, which seems bad given that it's a fairly old version at this point. I doubt that we want to put releases on SourceForge, so is there a way to take it down, or at least indicate that it's an old release? - Josh From charlesr.harris at gmail.com Thu Nov 9 10:31:38 2017 From: charlesr.harris at gmail.com (Charles R Harris) Date: Thu, 9 Nov 2017 08:31:38 -0700 Subject: [SciPy-Dev] welcome Antonio to the core team In-Reply-To: References: Message-ID: On Thu, Nov 9, 2017 at 1:15 AM, Ralf Gommers wrote: > Hi all, > > On behalf of the SciPy developers I'd like to welcome Antonio Horta > Ribeiro as a member of the core dev team. > > Antonio has been contributing to scipy.optimize and scipy.signal for a > year now, and has recently finished his GSoC project on a large scale > optimization algorithm (https://antonior92.github.io/ > posts/2017/24/GSoC-submission/). See https://github.com/scipy/ > scipy/pulls/antonior92 for his contributions so far; I'm looking forward > to more to come! > Welcome to the team Antonio. Chuck -------------- next part -------------- An HTML attachment was scrubbed... URL: From warren.weckesser at gmail.com Thu Nov 9 11:11:03 2017 From: warren.weckesser at gmail.com (Warren Weckesser) Date: Thu, 9 Nov 2017 11:11:03 -0500 Subject: [SciPy-Dev] welcome Antonio to the core team In-Reply-To: References: Message-ID: On Thu, Nov 9, 2017 at 3:15 AM, Ralf Gommers wrote: > Hi all, > > On behalf of the SciPy developers I'd like to welcome Antonio Horta > Ribeiro as a member of the core dev team. > > Antonio has been contributing to scipy.optimize and scipy.signal for a > year now, and has recently finished his GSoC project on a large scale > optimization algorithm (https://antonior92.github.io/ > posts/2017/24/GSoC-submission/). See https://github.com/scipy/ > scipy/pulls/antonior92 for his contributions so far; I'm looking forward > to more to come! > > Cheers, > Ralf > > Welcome Antonio! Thanks for the great work you've done--keep it up! Warren > _______________________________________________ > SciPy-Dev mailing list > SciPy-Dev at python.org > https://mail.python.org/mailman/listinfo/scipy-dev > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From haberland at ucla.edu Thu Nov 9 14:02:44 2017 From: haberland at ucla.edu (Matt Haberland) Date: Thu, 9 Nov 2017 11:02:44 -0800 Subject: [SciPy-Dev] welcome Antonio to the core team In-Reply-To: References: Message-ID: Great to hear! Nice work, Antonio. On Thu, Nov 9, 2017 at 12:15 AM, Ralf Gommers wrote: > Hi all, > > On behalf of the SciPy developers I'd like to welcome Antonio Horta > Ribeiro as a member of the core dev team. > > Antonio has been contributing to scipy.optimize and scipy.signal for a > year now, and has recently finished his GSoC project on a large scale > optimization algorithm (https://antonior92.github.io/ > posts/2017/24/GSoC-submission/). See https://github.com/scipy/ > scipy/pulls/antonior92 for his contributions so far; I'm looking forward > to more to come! > > Cheers, > Ralf > > > _______________________________________________ > SciPy-Dev mailing list > SciPy-Dev at python.org > https://mail.python.org/mailman/listinfo/scipy-dev > > -- Matt Haberland Assistant Adjunct Professor in the Program in Computing Department of Mathematics 7620E Math Sciences Building, UCLA -------------- next part -------------- An HTML attachment was scrubbed... URL: From ralf.gommers at gmail.com Thu Nov 9 14:16:25 2017 From: ralf.gommers at gmail.com (Ralf Gommers) Date: Fri, 10 Nov 2017 08:16:25 +1300 Subject: [SciPy-Dev] Old release still on SourceForge In-Reply-To: References: Message-ID: On Fri, Nov 10, 2017 at 4:28 AM, Joshua Wilson wrote: > Hey everyone, > > It seems that SciPy 0.16 is still up on SourceForge: > > https://sourceforge.net/projects/scipy/ > > According to the site, people are still downloading it, which seems > bad given that it's a fairly old version at this point. I doubt that > we want to put releases on SourceForge, so is there a way to take it > down, or at least indicate that it's an old release? > Removing the whole project can only be done via a request, not the UI. SF staff then determine whether to honor the request or not; in the past they've been very bad at listening to projects (e.g. https://webapps.stackexchange.com/questions/42490/can-you-delete-a-project-on-sourceforge). There may still be some users out there who need the .exe installers for some reason (but not 782 in the last week...). So indicating the version is old by adding a file THIS_IS_AN_OLD_RELEASE_GO_TO_PYPI_OR_CONDAFORGE.txt may be the way to go. Ralf -------------- next part -------------- An HTML attachment was scrubbed... URL: From willsheffler at gmail.com Thu Nov 9 14:53:42 2017 From: willsheffler at gmail.com (William Sheffler) Date: Thu, 9 Nov 2017 11:53:42 -0800 Subject: [SciPy-Dev] Sparse matrix manipulation in c++ Message-ID: Hello scipy-dev, First, forgive me if this is not the right forum, I am no scipy-dev! I have a mixed c++/python/pybind11 project in which I'd like to use a couple of sparse matrix types (bsr_matrix and csr_matrix, if that matters). I don't want to 'reinvent the wheel'... using Scipys implementation would be (a) better and (2) easier. But, I will need to access the data from c++ as well as python. The scipy implementation is C++, so it seems this should be possible. But what is the 'right' way to make my c++ understand scipy sparse arrays? I don't suppose there is some nice buffer-protocol-like abstraction? Pybind has a nice, 'magical' solution to this for numpy compatibility that avoids having to link against specific versions of numpy, but the details are a bit beyond me. Many thanks for any advice! Will Sheffler -------------- next part -------------- An HTML attachment was scrubbed... URL: From pav at iki.fi Thu Nov 9 14:56:38 2017 From: pav at iki.fi (Pauli Virtanen) Date: Thu, 09 Nov 2017 20:56:38 +0100 Subject: [SciPy-Dev] Old release still on SourceForge In-Reply-To: References: Message-ID: <1510257398.3403.2.camel@iki.fi> pe, 2017-11-10 kello 08:16 +1300, Ralf Gommers kirjoitti: [clip] > There may still be some users out there who need the .exe installers > for > some reason (but not 782 in the last week...). So indicating the > version is > old by adding a file > THIS_IS_AN_OLD_RELEASE_GO_TO_PYPI_OR_CONDAFORGE.txt > may be the way to go. I added a notice that the project has "moved" to the project description, and uploaded a README file to the project file root. There already was a README pointing to pypi/github under the scipy/Latest folder. Pauli From ralf.gommers at gmail.com Sun Nov 12 05:18:14 2017 From: ralf.gommers at gmail.com (Ralf Gommers) Date: Sun, 12 Nov 2017 23:18:14 +1300 Subject: [SciPy-Dev] Sparse matrix manipulation in c++ In-Reply-To: References: Message-ID: On Fri, Nov 10, 2017 at 8:53 AM, William Sheffler wrote: > Hello scipy-dev, > > First, forgive me if this is not the right forum, I am no scipy-dev! > No worries, you're in the right place. I have a mixed c++/python/pybind11 project in which I'd like to use a > couple of sparse matrix types (bsr_matrix and csr_matrix, if that matters). > I don't want to 'reinvent the wheel'... using Scipys implementation would > be (a) better and (2) easier. But, I will need to access the data from c++ > as well as python. The scipy implementation is C++, so it seems this should > be possible. But what is the 'right' way to make my c++ understand scipy > sparse arrays? > That C++ implementation does not have a defined API and is in practice an implementation detail behind the Python API that we could change in any release. So I would not recommend to attempt to reach into the scipy.sparse internals here. Ralf > I don't suppose there is some nice buffer-protocol-like abstraction? > Pybind has a nice, 'magical' solution to this for numpy compatibility that > avoids having to link against specific versions of numpy, but the details > are a bit beyond me. > > Many thanks for any advice! > Will Sheffler > > _______________________________________________ > SciPy-Dev mailing list > SciPy-Dev at python.org > https://mail.python.org/mailman/listinfo/scipy-dev > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From benny.malengier at gmail.com Sun Nov 12 06:08:59 2017 From: benny.malengier at gmail.com (Benny Malengier) Date: Sun, 12 Nov 2017 12:08:59 +0100 Subject: [SciPy-Dev] Sparse matrix manipulation in c++ In-Reply-To: References: Message-ID: As Ralph says. Now, you should not have many instances to pump over between your structure and python, so you can use cython and the normal API when this is needed. Example for the sundials nvector, 2 functions to convert: https://github.com/bmcage/odes/blob/master/scikits/odes/sundials/common_defs.pyx If in your implementation you want to use the python methods on sparse structure, then I assume a cython project makes more sense than C++ working on internal python structure. It should be just as fast and much more stable. A pointer to get started: http://cythoncython.readthedocs.io/en/latest/src/userguide/wrapping_CPlusPlus.html Benny 2017-11-12 11:18 GMT+01:00 Ralf Gommers : > > > On Fri, Nov 10, 2017 at 8:53 AM, William Sheffler > wrote: > >> Hello scipy-dev, >> >> First, forgive me if this is not the right forum, I am no scipy-dev! >> > > No worries, you're in the right place. > > I have a mixed c++/python/pybind11 project in which I'd like to use a >> couple of sparse matrix types (bsr_matrix and csr_matrix, if that matters). >> I don't want to 'reinvent the wheel'... using Scipys implementation would >> be (a) better and (2) easier. But, I will need to access the data from c++ >> as well as python. The scipy implementation is C++, so it seems this should >> be possible. But what is the 'right' way to make my c++ understand scipy >> sparse arrays? >> > > That C++ implementation does not have a defined API and is in practice an > implementation detail behind the Python API that we could change in any > release. So I would not recommend to attempt to reach into the scipy.sparse > internals here. > > Ralf > > >> I don't suppose there is some nice buffer-protocol-like abstraction? >> Pybind has a nice, 'magical' solution to this for numpy compatibility that >> avoids having to link against specific versions of numpy, but the details >> are a bit beyond me. >> >> Many thanks for any advice! >> Will Sheffler >> >> _______________________________________________ >> SciPy-Dev mailing list >> SciPy-Dev at python.org >> https://mail.python.org/mailman/listinfo/scipy-dev >> >> > > _______________________________________________ > SciPy-Dev mailing list > SciPy-Dev at python.org > https://mail.python.org/mailman/listinfo/scipy-dev > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From cyrille.rossant at gmail.com Mon Nov 13 09:11:54 2017 From: cyrille.rossant at gmail.com (Cyrille Rossant) Date: Mon, 13 Nov 2017 15:11:54 +0100 Subject: [SciPy-Dev] Looking for technical reviewers Message-ID: Hello, I'm looking for technical reviewers for 11 recipes of the upcoming IPython Cookbook, 2nd edition (most of the contents of the book will be freely available on GitHub) If you're familiar with IPython/Jupyter, NumPy/SciPy, matplotlib, etc. please feel free to go through these recipes and let me know before Nov 24th of any comments on GitHub : https://github.com/ipython-b ooks/recipes-for-review Here are the recipe names: 2.1. Using the latest features of Python 3 3.3. Mastering widgets in the Jupyter notebook 3.4. Creating custom Jupyter notebook widgets in Python, HTML, and JavaScript 3.5. Configuring the Jupyter notebook 3.6. Introducing JupyterLab 5.1. Knowing Python to write faster code 5.11. Performing out-of-core computations on large arrays with Dask 6.1. Using matplotlib styles 6.2. Creating beautiful statistical plots easily with seaborn 6.5. Discovering interactive visualization libraries in the notebook 6.6. Creating plots with Altair and the Vega-Lite specification Thanks, Cyrille -------------- next part -------------- An HTML attachment was scrubbed... URL: From Former at physicist.net Mon Nov 13 16:07:34 2017 From: Former at physicist.net (Adam) Date: Mon, 13 Nov 2017 13:07:34 -0800 Subject: [SciPy-Dev] Travis CI fails for test_ndimage which is unrelated to my PR Message-ID: <1510607254.5863.7.camel@physicist.net> Hello everybody, I've submitted PR?8151?which fixes some issues with the hypergeometric function hyp2f1 that I discussed in an earlier thread. Right now, it's failing the Travis CI checks. ?From I can tell, it's claiming that a test script called 'test_ndimage.py' is failing. ?From the Travis output: FAIL scipy/ndimage/tests/test_ndimage.py::TestNdimage::()::test_map_coordina tes_endianness_with_output_parameter (link:?https://travis-ci.org/scipy/scipy/jobs/301127108#L946-L9 49) My PR has nothing to do with test_ndimage. (Although I did add a new test script called?test_hyp2f1.py, but as far as I can tell all the test cases with this file passed). Is this another bug with the Travis CI checks? Or is there something wrong with my PR? Thanks! --Adam K. ? ?? -------------- next part -------------- An HTML attachment was scrubbed... URL: From pav at iki.fi Mon Nov 13 17:02:06 2017 From: pav at iki.fi (Pauli Virtanen) Date: Mon, 13 Nov 2017 23:02:06 +0100 Subject: [SciPy-Dev] Travis CI fails for test_ndimage which is unrelated to my PR In-Reply-To: <1510607254.5863.7.camel@physicist.net> References: <1510607254.5863.7.camel@physicist.net> Message-ID: <1510610526.8184.3.camel@iki.fi> Hi, ma, 2017-11-13 kello 13:07 -0800, Adam kirjoitti: > Right now, it's failing the Travis CI checks. From I can tell, it's > claiming that a test script called 'test_ndimage.py' is failing. > From > the Travis output: > FAIL > scipy/ndimage/tests/test_ndimage.py::TestNdimage::()::test_map_coordi > na > tes_endianness_with_output_parameter > (link: https://travis-ci.org/scipy/scipy/jobs/301127108#L946-L9 > 49) > My PR has nothing to do with test_ndimage. (Although I did add a new > test script called test_hyp2f1.py, but as far as I can tell all the > test cases with this file passed). > Is this another bug with the Travis CI checks? Or is there something > wrong with my PR? If it's clear the failure is unrelated, you can simply ignore it, and maybe note that it's unrelated in a comment on the PR. In this case it's unrelated --- one of the CI checks runs using current numpy master branch, and it breaks when Numpy adds deprecation warnings or makes some backward incompatible changes. This is also the cause for the failure you are seeing. Generally we try to address these as they turn up, but it can take some days. Pauli From Former at physicist.net Mon Nov 13 17:29:57 2017 From: Former at physicist.net (Adam) Date: Mon, 13 Nov 2017 14:29:57 -0800 Subject: [SciPy-Dev] Travis CI fails for test_ndimage which is unrelated to my PR In-Reply-To: <1510610526.8184.3.camel@iki.fi> References: <1510607254.5863.7.camel@physicist.net> <1510610526.8184.3.camel@iki.fi> Message-ID: <1510612197.5863.8.camel@physicist.net> Alright, thanks! ?I'll add a comment to my PR. Adam On Mon, 2017-11-13 at 23:02 +0100, Pauli Virtanen wrote: > Hi, > > ma, 2017-11-13 kello 13:07 -0800, Adam kirjoitti: > > > > Right now, it's failing the Travis CI checks.??From I can tell, > > it's > > claiming that a test script called 'test_ndimage.py' is failing. > > ?From > > the Travis output: > > FAIL > > scipy/ndimage/tests/test_ndimage.py::TestNdimage::()::test_map_coor > > di > > na > > tes_endianness_with_output_parameter > > (link: https://travis-ci.org/scipy/scipy/jobs/301127108#L946-L9 > > 49) > > My PR has nothing to do with test_ndimage. (Although I did add a > > new > > test script called test_hyp2f1.py, but as far as I can tell all the > > test cases with this file passed). > > Is this another bug with the Travis CI checks? Or is there > > something > > wrong with my PR? > If it's clear the failure is unrelated, you can simply ignore it, and > maybe note that it's unrelated in a comment on the PR. > > In this case it's unrelated --- one of the CI checks runs using > current > numpy master branch, and it breaks when Numpy adds deprecation > warnings > or makes some backward incompatible changes. This is also the cause > for > the failure you are seeing. Generally we try to address these as they > turn up, but it can take some days. > > Pauli > > _______________________________________________ > SciPy-Dev mailing list > SciPy-Dev at python.org > https://mail.python.org/mailman/listinfo/scipy-dev From njs at pobox.com Tue Nov 14 22:37:59 2017 From: njs at pobox.com (Nathaniel Smith) Date: Tue, 14 Nov 2017 19:37:59 -0800 Subject: [SciPy-Dev] Upcoming revision of the BLAS standard Message-ID: Hi NumPy and SciPy developers, Apparently there is some work afoot to update the BLAS standard, with a working document here: https://docs.google.com/document/d/1DY4ImZT1coqri2382GusXgBTTTVdBDvtD5I14QHp9OE/edit This seems like something where we might want to get involved in, so that the new standard works for us, and James Demmel (the first author on that proposal and a professor here at Berkeley) suggested they'd be interested to hear our thoughts. I'm not sure exactly what the process is here -- apparently there have been some workshops, and there was going to be a BoF today at Supercomputing, but I don't know what the schedule is or how they'll be making decisions. It's possible for anyone interested to click on that google doc above and make "suggestions", but it seems like maybe it would be useful for the NumPy/SciPy teams to come up with some sort of shared document on what we want? I'm really, really not the biggest linear algebra expert on these lists, so I'm hoping those with more experience will jump in, but to get started here are some initial ideas for things we might want to ask for: - Support for arbitrary strided memory layout - Replacing xerbla with proper error codes (already in that proposal) - There's some discussion about NaN handling where I think we might have opinions. (Am I remember right that currently we have to check for NaNs ourselves all the time because there are libraries that blow up if we don't, and we don't know which ones those are?) - Where the spec ends up giving implementors flexibility, some way to detect at compile time what options they chose. -n -- Nathaniel J. Smith -- https://vorpus.org From andyfaff at gmail.com Tue Nov 14 23:09:03 2017 From: andyfaff at gmail.com (Andrew Nelson) Date: Wed, 15 Nov 2017 15:09:03 +1100 Subject: [SciPy-Dev] Upcoming revision of the BLAS standard In-Reply-To: References: Message-ID: Linalg noob here. can we have it in a c library so we don't need to bother with getting Fortran compilers on windows? On 15 Nov 2017 2:38 pm, "Nathaniel Smith" wrote: Hi NumPy and SciPy developers, Apparently there is some work afoot to update the BLAS standard, with a working document here: https://docs.google.com/document/d/1DY4ImZT1coqri2382GusXgBTTTVdB DvtD5I14QHp9OE/edit This seems like something where we might want to get involved in, so that the new standard works for us, and James Demmel (the first author on that proposal and a professor here at Berkeley) suggested they'd be interested to hear our thoughts. I'm not sure exactly what the process is here -- apparently there have been some workshops, and there was going to be a BoF today at Supercomputing, but I don't know what the schedule is or how they'll be making decisions. It's possible for anyone interested to click on that google doc above and make "suggestions", but it seems like maybe it would be useful for the NumPy/SciPy teams to come up with some sort of shared document on what we want? I'm really, really not the biggest linear algebra expert on these lists, so I'm hoping those with more experience will jump in, but to get started here are some initial ideas for things we might want to ask for: - Support for arbitrary strided memory layout - Replacing xerbla with proper error codes (already in that proposal) - There's some discussion about NaN handling where I think we might have opinions. (Am I remember right that currently we have to check for NaNs ourselves all the time because there are libraries that blow up if we don't, and we don't know which ones those are?) - Where the spec ends up giving implementors flexibility, some way to detect at compile time what options they chose. -n -- Nathaniel J. Smith -- https://vorpus.org _______________________________________________ SciPy-Dev mailing list SciPy-Dev at python.org https://mail.python.org/mailman/listinfo/scipy-dev -------------- next part -------------- An HTML attachment was scrubbed... URL: From njs at pobox.com Tue Nov 14 23:18:39 2017 From: njs at pobox.com (Nathaniel Smith) Date: Tue, 14 Nov 2017 20:18:39 -0800 Subject: [SciPy-Dev] Upcoming revision of the BLAS standard In-Reply-To: References: Message-ID: On Tue, Nov 14, 2017 at 8:09 PM, Andrew Nelson wrote: > Linalg noob here. > > can we have it in a c library so we don't need to bother with getting > Fortran compilers on windows? I guess they could define the API as a C API -- in fact I think they might be doing that already -- but particular implementations may still use Fortran internally, and SciPy will still need Fortran for a bunch of other stuff (starting with LAPACK), so I don't think this will help much with the windows compiler problem. -n -- Nathaniel J. Smith -- https://vorpus.org From toddrjen at gmail.com Wed Nov 15 00:02:55 2017 From: toddrjen at gmail.com (Todd) Date: Wed, 15 Nov 2017 00:02:55 -0500 Subject: [SciPy-Dev] Upcoming revision of the BLAS standard In-Reply-To: References: Message-ID: Is getting decimal versions of the functions a possibility? On Nov 14, 2017 22:38, "Nathaniel Smith" wrote: > Hi NumPy and SciPy developers, > > Apparently there is some work afoot to update the BLAS standard, with > a working document here: > > https://docs.google.com/document/d/1DY4ImZT1coqri2382GusXgBTTTVdB > DvtD5I14QHp9OE/edit > > This seems like something where we might want to get involved in, so > that the new standard works for us, and James Demmel (the first author > on that proposal and a professor here at Berkeley) suggested they'd be > interested to hear our thoughts. > > I'm not sure exactly what the process is here -- apparently there have > been some workshops, and there was going to be a BoF today at > Supercomputing, but I don't know what the schedule is or how they'll > be making decisions. It's possible for anyone interested to click on > that google doc above and make "suggestions", but it seems like maybe > it would be useful for the NumPy/SciPy teams to come up with some sort > of shared document on what we want? > > I'm really, really not the biggest linear algebra expert on these > lists, so I'm hoping those with more experience will jump in, but to > get started here are some initial ideas for things we might want to > ask for: > > - Support for arbitrary strided memory layout > - Replacing xerbla with proper error codes (already in that proposal) > - There's some discussion about NaN handling where I think we might > have opinions. (Am I remember right that currently we have to check > for NaNs ourselves all the time because there are libraries that blow > up if we don't, and we don't know which ones those are?) > - Where the spec ends up giving implementors flexibility, some way to > detect at compile time what options they chose. > > -n > > -- > Nathaniel J. Smith -- https://vorpus.org > _______________________________________________ > SciPy-Dev mailing list > SciPy-Dev at python.org > https://mail.python.org/mailman/listinfo/scipy-dev > -------------- next part -------------- An HTML attachment was scrubbed... URL: From charlesr.harris at gmail.com Wed Nov 15 10:25:31 2017 From: charlesr.harris at gmail.com (Charles R Harris) Date: Wed, 15 Nov 2017 08:25:31 -0700 Subject: [SciPy-Dev] Upcoming revision of the BLAS standard In-Reply-To: References: Message-ID: On Tue, Nov 14, 2017 at 8:37 PM, Nathaniel Smith wrote: > Hi NumPy and SciPy developers, > > Apparently there is some work afoot to update the BLAS standard, with > a working document here: > > https://docs.google.com/document/d/1DY4ImZT1coqri2382GusXgBTTTVdB > DvtD5I14QHp9OE/edit > > This seems like something where we might want to get involved in, so > that the new standard works for us, and James Demmel (the first author > on that proposal and a professor here at Berkeley) suggested they'd be > interested to hear our thoughts. > > I'm not sure exactly what the process is here -- apparently there have > been some workshops, and there was going to be a BoF today at > Supercomputing, but I don't know what the schedule is or how they'll > be making decisions. It's possible for anyone interested to click on > that google doc above and make "suggestions", but it seems like maybe > it would be useful for the NumPy/SciPy teams to come up with some sort > of shared document on what we want? > > I'm really, really not the biggest linear algebra expert on these > lists, so I'm hoping those with more experience will jump in, but to > get started here are some initial ideas for things we might want to > ask for: > > - Support for arbitrary strided memory layout > - Replacing xerbla with proper error codes (already in that proposal) > - There's some discussion about NaN handling where I think we might > have opinions. (Am I remember right that currently we have to check > for NaNs ourselves all the time because there are libraries that blow > up if we don't, and we don't know which ones those are?) > - Where the spec ends up giving implementors flexibility, some way to > detect at compile time what options they chose. > Somewhat unrelated, but it would be nice to have 64 bit integers. That is already possible with compiler flags, but it would help if there was an easy way to tell what the compiled library was using. Chuck -------------- next part -------------- An HTML attachment was scrubbed... URL: From christoph.baumgarten at gmail.com Mon Nov 20 02:41:16 2017 From: christoph.baumgarten at gmail.com (Christoph Baumgarten) Date: Mon, 20 Nov 2017 08:41:16 +0100 Subject: [SciPy-Dev] added normal inverse gaussian distribution to scipy.stats Message-ID: Hi, I added the normal inverse gaussian distribution that is popular in financial mathematics: https://github.com/scipy/scipy/pull/8171 it is based on an old pull request #2648, main improvement is the addition of the rvs method that allows the generation of random samples without relying on the slow ppf method. Happy to get some feedback. Christoph P.S. Travis CI build still fails, I hope it is something easy to fix... -------------- next part -------------- An HTML attachment was scrubbed... URL: From Former at physicist.net Sun Nov 26 23:54:16 2017 From: Former at physicist.net (Adam) Date: Sun, 26 Nov 2017 20:54:16 -0800 Subject: [SciPy-Dev] Need review for hyp2f1 PR's Message-ID: <0MMBdz-1eDO8Y1WJB-0082Mg@mail.gmx.com> Hello everybody, I?ve responded to all the outstanding issues on my two PR?s relating to the hypergeometric function. Per the developer guide, I?m asking for further review so they can get merged into the master branch.? My PR?s are: 8110 8151 Thank you! --Adam -------------- next part -------------- An HTML attachment was scrubbed... URL: