Re: de_min.m: global optimisation (differential evolution)

[Christian Fischer-8]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA256

Hi Michael,

Thanks for the test script.

It has been a while. My first version I sent performed very well on
the simple rosenbrock function, outperforming SA by using much
less function evaluations.

I made a few adjustments which should make it easier to use.
I added a test, a demo and did a bit cleanup.

I attached the new version along with another very nasty test
function (rastrigin.m). I also made a cpp version if someone is
interested.

I think de_min.m is ready to be included in the optim package.

If there are people saying that they would benefit from a c++
implementation, I might do it. But for many real problems, the
function evaluation itself is the problem.

In order to make updates to this function, it might be a good idea
if have an svn account.

Comments and criticism welcome.

Best wishes
Christian

On Tue, 2009-07-21 at 10:00 +0200, Michael Creel wrote:

> Hi Christian,
> This function looks like it could find a home in the optim package
> (http://octave.sourceforge.net/optim/index.html). I have written
> simulated annealing code that is already part of the package.
> Following is a script that will use your code and the simulated
> annealing code to minimize a higher dimensional Rosenbrock function
> (also in optim).
> Best, Michael
>
> dim = 10;
> ctl.XVmin = -2*ones(1, dim);
> ctl.XVmax = 2*ones(1, dim);
> tic;
> [x, obj_value, nfeval] = de_min (@rosenbrock, ctl);
> toc
> printf("solution: \n");
> disp(x);
>
> # SA controls
> ub = 2*ones(dim,1);
> lb = -ub;
> nt = 20;
> ns = 5;
> rt = 0.75; # careful - this is too low for many problems
> maxevals = 1e10;
> neps = 5;
> functol = 1e-7;
> paramtol = 1e-5;
> verbosity = 1; # only final results. Inc
> minarg = 1;
> control = { lb, ub, nt, ns, rt, maxevals, neps, functol, paramtol,
> verbosity, 1};
> # do sa
> x = zeros(dim,1);
> tic;
> [theta, obj_value, convergence] = samin("rosenbrock", {x}, control);
> toc

-----BEGIN PGP SIGNATURE-----

Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Use GnuPG with Firefox : http://getfiregpg.org (Version: 0.7.6)
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=scNm
-----END PGP SIGNATURE-----

______________________________________________________
GRATIS für alle WEB.DE-Nutzer: Die maxdome Movie-FLAT!
Jetzt freischalten unter http://movieflat.web.de





------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Octave-dev mailing list
Octave-dev@...
https://lists.sourceforge.net/lists/listinfo/octave-dev

[Christian Fischer-8]

Hi

I need svn access to add my previously discussed function to the
repository.

My sourceforge username is: sffisch


I plan to look after and improve this function - I appreciate feedback
and criticism! :)

Best wishes
Christian



On Fri, 2009-09-04 at 11:24 +0200, Michael Creel wrote:

> Hi Christian,
> I have checked de_min.m, trying it with the test function you
> attached. I find that neighther de_min nor samin can successfully
> minimize this function when the dimension is high. Perhaps better
> tuning would help, but I haven't spent too much time on it.
>
> For the Rosenbrock function, I agree that the method works better than
> samin. I'll probably be trying it out on some of my research problems,
> where up to now I have been using samin quite extensively.
>
> To add this to optim, just follow the instructions at
> http://octave.sourceforge.net/developers.html
>
> The optim package has no single person responsible for its contents,
> it's just a collection of code by various authors. Don't expect anyone
> else to maintain the code you put there - it won't happen :-)   For
> that reason, I think that it would be better if you put the code in
> yourself, that way your name will be associated with it, and people
> will know to contact you about it.
>
> Thanks a lot for making this algorithm available, it looks quite
> promising to me. I'll let you know how it goes with my research
> problems.
>
> Cheers, Michael
>

.....

------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
Octave-dev mailing list
Octave-dev@...
https://lists.sourceforge.net/lists/listinfo/octave-dev

[Søren Hauberg]

fre, 18 09 2009 kl. 01:45 +0200, skrev Christian Fischer:
> I need svn access to add my previously discussed function to the
> repository.
>
> My sourceforge username is: sffisch
>
>
> I plan to look after and improve this function - I appreciate feedback
> and criticism! :)

I've given you access to SVN. Let us know if you have problems.

Søren


------------------------------------------------------------------------------
Come build with us! The BlackBerry® Developer Conference in SF, CA
is the only developer event you need to attend this year. Jumpstart your
developing skills, take BlackBerry mobile applications to market and stay
ahead of the curve. Join us from November 9-12, 2009. Register now!
http://p.sf.net/sfu/devconf
_______________________________________________
Octave-dev mailing list
Octave-dev@...
https://lists.sourceforge.net/lists/listinfo/octave-dev