There are some frameworks available for running GAs…
like: SFeresV2 (uses TBB for m-core support), ParadisEO → Both from french universitie.
PUGACE is a framework for cellular automata - but relatively in infancy.
All of them were presented in WCCI 2010 held in Barcelona…
May be, some1 can add GPU support for them.
I’ve done the Genetic Algorithm in CUDA…
I simply parallelize the initiation of populatioon and evalutation of individus
I think there’s more that you can parallelize.
But there’s limitation that you can’t generate random values in the GPU code :(
Try this paper:
Mariusz Nowostawski, Riccardo Poli. 1999. Parallel Genetic Algorithm Taxonomy. (i forgot the link)
Bydal, Asbjron. 2008. Implementation of genetic algorithm on CUDA. [url=“http://fruit.grm.hia.no/ikt502/year2008/projects/reports/Implementation_of_genetic_algorithm_on_CUDA.pdf”]http://fruit.grm.hia.no/ikt502/year2008/proj...thm_on_CUDA.pdf[/url]
I’ve done the Genetic Algorithm in CUDA…
I simply parallelize the initiation of populatioon and evalutation of individus
I think there’s more that you can parallelize.
But there’s limitation that you can’t generate random values in the GPU code :(
Try this paper:
Mariusz Nowostawski, Riccardo Poli. 1999. Parallel Genetic Algorithm Taxonomy. (i forgot the link)
Bydal, Asbjron. 2008. Implementation of genetic algorithm on CUDA. [url=“http://fruit.grm.hia.no/ikt502/year2008/projects/reports/Implementation_of_genetic_algorithm_on_CUDA.pdf”]http://fruit.grm.hia.no/ikt502/year2008/proj...thm_on_CUDA.pdf[/url]
Not true. You just can’t use the rand() implementation from cstdlib.
There’s an example how to generate random numbers in the SDK (Marsenne Twister) and many threads on the forums.
Not true. You just can’t use the rand() implementation from cstdlib.
There’s an example how to generate random numbers in the SDK (Marsenne Twister) and many threads on the forums.
oh, really?
I’ll try that, thanks :D
oh, really?
I’ll try that, thanks :D
Anyone know where i can get a Simple Genetic Algorithm in Cuda, c++ or python?
check out the software and documentation at http://metabolism.munch.life - it’s a sort of digital metabolism that exposes a genetic programming environment - it’s modeled after a cell cycle and is primarily designed to build a cell atlas that maps and evolves bits of information signaling from fields in the database to the actual evolution of the DNA in the life cycling through the digital metabolism
it sequences both online and offline genes … genes can be thought of as web services and a sequence of binary data signaling between them represents an atomic string of “digital DNA”
I am the founder - if You have any questions I’d be happy to expand - You can also email Me at marc@munchmode.com