Pyevolve with multi processing using more RAM with each GA generation

Question

I am using pyevolve to run a GA using multi processing. The code runs fine but uses more memory with every new generation of GA. The code is creating new jobs for each generation while the previous ones are idle while reserving RAM.It finally crashes after around 10 generations. How do I correct this?[Ram Usage][1][1]: https://i.stack.imgur.com/knCC0.jpg

def main_run():

    flowargs, rhoargs, zindex=get_params()
    genome = G1DList.G1DList(18)
    genome.initializator.set(myGA.myInitializator)
    genome.crossover.set(myGA.mycrossover)
    genome.mutator.set(pl.Mutators.G1DListMutatorSwap)
    genome.setParams(z=zindex,f=flowargs,r=rhoargs)
    genome.evaluator.set(get_rho)
    ga = GSimpleGA.GSimpleGA(genome)
    ga.minimax = Consts.minimaxType["minimize"]
    ga.setGenerations(100)
    ga.selector.set(Selectors.GRouletteWheel)
    ga.setInteractiveMode(False)
    ga.setMultiProcessing(True)
    ga.evolve(freq_stats=1)
    best = ga.bestIndividual()
    print(best)

if __name__ == "__main__":
    main_run()

score 0 · Accepted Answer · edited Aug 29 '18 at 12:29

I came across this problem a while back, and after much research I found the answer,

You have to modify the GPopulation.py in the library source code,

--Add the following two commands:

proc_pool.close()
proc_pool.join()

--before the last line of the function evaluate(self, **args)

--Rebuild and re-install the libraries.
i.e. C:\Pevolve Source code\Pyevolve-0.6rc1\pyevolve\GPopulation (mine was here)

Edit with IDLE

--Add proc_pool.close() & proc_pool.join() just before self.clearFlags()

--Save

Open Command Prompt

Go to C:\Pevolve Source code\Pyevolve-0.6rc1
setup.py build
setup.py install

Pyevolve with multi processing using more RAM with each GA generation

1 Answers1

--Save