Re: [tlug] genetic algorithm/optimization framework

[Claus Aranha <caranha@example.com>]

Have you checked ECJ?

https://cs.gmu.edu/~eclab/projects/ecj/

It has support for multi-objective evaluation, is highly customizable,
and has a very helpful/prompt mailing list.

It also has some modules for parsimony pressure (which I'm guessing
you are going to need if you are easily reaching depth 80 trees).

I do work on this area (although not with this particular piece of
software), so If you need someone to bounce ideas over, I'd be happy
to lend an ear.

Cheers


On Thu, Apr 28, 2016 at 5:39 AM, Lars Kotthoff <lists@example.com> wrote:
> Hi Attila,
>
>  can you give us some more background on what you need to optimise and what the
> application is? Are you totally set on GAs? Something else may be more suitable.
>
> Cheers,
>
> Lars
>
>
> On Wed, 27 Apr 2016 22:22:25 +0200
> Attila Kinali <attila@example.com> wrote:
>
>> Moin,
>>
>> While i'm writing here anyways... a small question:
>>
>> Does anyone have a recomendation for a genetic algorithm/optimization
>> framework? I have a problem that would be best solved with such a system
>> (intractable, highly non-linear problem, with no way of having anything
>> that can be fed into another optimization algorithm).
>>
>> I googled a bit but most of the stuff out there is optimized for
>> single objective optimization with a very simple model of a genome
>> (usually lists of fixed length). While i would ideally use an list
>> of variable length of structs, or a tree like structure.
>>
>> DEAP seems to be ok, but has the problem that it can support at least
>> a tree depth of 80-90, which is way too small. MOEA looks better,
>> but any usefull documentation is only available for $$$ which i'm not
>> willing to spend before i actually know whether it can do what i want.
>> The others are either non-open source or are lacking severly at
>> the documentation.
>>
>> Of course, i could write my own framework, but that would take
>> more time than i want to spend... and reinventing the wheel
>> sucks, even if all other wheels are square.
>>
>>
>>                       Attila Kinali
>
>
> --
> To unsubscribe from this mailing list,
> please see the instructions at http://lists.tlug.jp/list.html
>
> The TLUG mailing list is hosted by ASAHI Net, provider of mobile and
> fixed broadband Internet services to individuals and corporations.
> Visit ASAHI Net's English-language Web page: http://asahi-net.jp/en/



-- 
Claus Aranha
http://claus.castelodelego.org