[galib] PSO and galib
frederic.ratle at polymtl.ca
Thu Feb 3 11:59:01 EST 2005
I don't know much about particle swarm, but it does not really use variation
operators. Applying operators such as mutation and crossover does not
really make sense since PSO isn't based on evolution. Course and velocity
are in fact deterministically ajusted at each step, weighed by a random
factor (an acceleration).
In fact it looks quite simple to code, so I would suggest implementing
Selon "Dickinson, Bill" <dickinsw at CMG.MFLDCLIN.EDU>:
> I know nothing about PSO, but, from your description it seems reasonable
> that the genetic operators could do their thing on particle vectors (alter
> their course and velocity), measure the affects of the changes against an
> objective function or functions, and then evolve.
More information about the galib