Re: OpenOpt v 0.25 - free optimization toolbox

Hi,
This looks pretty interesting. I'll have a look at it when I get some
time, which unfortunately I don't have now. Did you know that there is
a simulated annealing solver in octave-forge (samin)? It is written as
an .oct function so that the loops are done in C. There is a
samin_example program that shows how to use it. Also, there is a bfgs
(and l-bfgs) solver, also written as an .oct function, with an .m file
frontend (bfgsmin.m). At some point, someone should try to do some
benchmarking for the available solvers. It would be nice to select the
ones that work best and that complement eachother, do some
standardization of interfaces, and weed out the ones that are
dominated by other alternatives. If you like what you see in samin and
bfgsmin, please feel free to chop out  and use anything that's useful.
Regards, Michael

On 12/15/06, Dmitrey Kroshko <[email protected]> wrote:
> Hi all,
> So OpenOpt v 0.25 (free optimization toolbox, GNU GPL2) is reliesed
> Link for download:
> http://www.box.net/public/6bsuq765t4#main
> Installation: just add reqursively to path.
> Now 6 solvers are available:
> 4 global (1 of them uses inner local 1st-order solver);
> 2 local non-smooth:
> ShorElipsoid() for small nVars ~ 1...10
> ralg() for nVars up to ~ some thouzands, fminsearch analog, that can use
> (sub)gradient provided by user, can handle lb-ub but using hand-turned
> penalty (functions for all type of constraints) currently is preferrable.
>
> it can be running in both MATLAB & Octave but
> 1) I have problems with plot()
> even plot([1 2], [3 4]) don't work - some problems with connection to
> GNUplot
> May be I use something wrong?
> So currently graphic output is unavailable for Octave.
> things I use along with plot(): legend(), xlabel(), ylabel(), drawnow(),
> title(), specifiers and colors as plot() parameters
> I have Mandriva Linux 2007 & Octave 2.9.9 from octave.org mirror
> (octave.org was unavailable for some days)
> 2) I intend to connect rsqp toolbox v 1.0 by Institute of Industrial
> Control, Zhejiang University
> It handles some tasks much better than fmincon (those that I construct &
> tested - in factor of 2-5), but not all constraints are available for now.
> however it requires at least colamd() analog - this one for some reasons
> says he don't want to work in Octave.
> BTW in the homepage colamd authors says version for 64-bit is not ready yet.
> Also, some problems with license exist - rsqp authors call Matworks QP
> routine.
> 3)currently OpenOpt abilities are very limited, its' further development
> will strongly depends on how much money we will found.
> 4)some solvers run very slow in Octave vs MATLAB. I guess this is due to
> JIT accelerator, that is currently absent in Octave. In my solvers I
> used vectorization where possible, and some more I updated, but some
> cycles still present in, for example, buscarnd() & anneal().
> Best regards,
> Dmitrey (last-year postgraduate),
> Instityte of cybernetics, Ukraine national sciences academy.
> www.icyb.kiev.ua
> _______________________________________________
> Octave-sources mailing list
> [email protected]
> https://www.cae.wisc.edu/mailman/listinfo/octave-sources
>