[SciPy-User] criteria's to get the exit mode '0' in slsqp

[jagan prabhu]

Hi all,

Thank you, it was a good answer. one step ahead and specific...

What are the criteria's to get the exit mode '0'  in case of 'fmin_slsqp'?

------
Jagan


--- On Sat, 19/9/09, Robert Kern <robert.kern at gmail.com> wrote:

From: Robert Kern <robert.kern at gmail.com>
Subject: Re: [SciPy-User] criteria's to get the exit mode '0
To: "SciPy Users List" <scipy-user at scipy.org>
Date: Saturday, 19 September, 2009, 12:17 AM

On Fri, Sep 18, 2009 at 08:40, jagan prabhu <jagan_cbe2003 at yahoo.co.in> wrote:
>
> Hi,
>
> I am using the scipy optimization routine 'fmin_slsqp' and  'fmin_l_bfgs_b' in both the case exit mode '0' represents optimization terminated successfully / convergence is achieved.
>
> What are the criteria's to get the exit mode '0' ?
>
> Because if i change my initial parameters by very small increment or decrement, i am getting huge difference in my optimized functional value & optimized parameter values.
>
> so i like to know,
> How do the optimization routine determines this the optimized parameters and optimized functional value?

It's slightly different for each routine, but basically, it stops when
the derivatives at the test point are close enough to zero and the
derivatives nearby show that you are at a minimum rather than a maximum
or a saddle point.  These are all local minimizers, meaning that they
can get trapped in so-called "local minima" where there are little
"valleys" in the function which are not the deepest. You want the
deepest valley of them all, or the global minimum, but the fmin routines
cannot guarantee that you will find it. They basically require that you
start with an initial guess that is close enough to the global minimum
that it manages to avoid all of the local minima.

--
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://mail.scipy.org/mailman/listinfo/scipy-user



      From cricket scores to your friends. Try the Yahoo! India Homepage! http://in.yahoo.com/trynew
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20090925/975f7a28/attachment.html>