Problems with a custom generator with NSGA-II

[Alberto Tonda]

Dear all,

I am experimenting with a custom generator for NSGA-II, but I have some weird error when I try to pass it some arguments with a dictionary. This is the basic example of NSGA-II, for which my generator makes no sense, but I am able to reproduce the error:

from random import Random

from time import time

import inspyred

def nsga2generate(random, args) :

        phenotype_len = args.get("phenotype_len", 3)

        # extract random number in 1,400

        individual_size = random.randint(1,400)

        individual = [0] * individual_size

        for i in range(0, individual_size) : individual[i] = random.randint(1, phenotype_len)

        return individual

def main(prng=None, display=False):

    if prng is None:

        prng = Random()

        prng.seed(time())

    args = dict()

    args["phenotype_len"] = 10

    problem = inspyred.benchmarks.Kursawe(3)

    ea = inspyred.ec.emo.NSGA2(prng)

    ea.variator = [inspyred.ec.variators.blend_crossover,

                   inspyred.ec.variators.gaussian_mutation]

    ea.terminator = inspyred.ec.terminators.generation_termination

    final_pop = ea.evolve(generator=nsga2generate(prng, args),

                          #generator=nsga2generate

                          evaluator=problem.evaluator,

                          pop_size=100,

                          maximize=problem.maximize,

                          bounder=problem.bounder,

                          max_generations=80)

    if display:

        final_arc = ea.archive

        print('Best Solutions: \n')

        for f in final_arc:

            print(f)

        import pylab

        x = []

        y = []

        for f in final_arc:

            x.append(f.fitness[0])

            y.append(f.fitness[1])

        pylab.scatter(x, y, color='b')

        pylab.savefig('{0} Example ({1}).pdf'.format(ea.__class__.__name__,

                                                     problem.__class__.__name__),

                      format='pdf')

        pylab.show()

    return ea

if __name__ == '__main__':

    main(display=True)

If I just pass the generator with 

generator=nsga2generate

everything works properly. However, with the current code, I get:

Traceback (most recent call last):

  File "sample-nsga-ii.py", line 53, in <module>

    main(display=True)

  File "sample-nsga-ii.py", line 32, in main

    max_generations=80)

  File "/usr/local/lib/python3.5/dist-packages/inspyred/ec/emo.py", line 147, in evolve

    return ec.EvolutionaryComputation.evolve(self, generator, evaluator, pop_size, seeds, maximize, bounder, **args)

  File "/usr/local/lib/python3.5/dist-packages/inspyred/ec/ec.py", line 430, in evolve

    cs = generator(random=self._random, args=self._kwargs)

TypeError: 'list' object is not callable

There is clearly something I am missing. Can you help me?

Thank you for your time.

[Aaron Garrett]

When you pass that generator into the args, you're calling it (by passing the prng and args parameters). That means you're setting generator to the RESULT of the function call, which is a list. You just need to pass the function name as the generator like you have commented out. 

If I've misunderstood your question, let me know.

[Alberto Tonda]

I see! But then, how do you pass the arguments to the generator? 

--
You received this message because you are subscribed to a topic in the Google Groups "Inspyred" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/inspyred/mKwNHVAyRfk/unsubscribe.
To unsubscribe from this group and all its topics, send an email to inspyred+unsubscribe@googlegroups.com.
To post to this group, send email to insp...@googlegroups.com.
Visit this group at https://groups.google.com/group/inspyred.
For more options, visit https://groups.google.com/d/optout.

--

Alberto Tonda

[Aaron Garrett]

After reading your message again, I think I understand what your real question is. You probably want to know how to pass the dictionary stuff to your custom generator. You do that by putting those values in the arguments to the evolve function call.

    final_pop = ea.evolve(generator=nsga2generate,

                          evaluator=problem.evaluator,

                          pop_size=100,

                          maximize=problem.maximize,

                          bounder=problem.bounder,

                          max_generations=80,

                          phenotype_len=10)

Those are automatically collected and put into the args dictionary that is passed to bounder. You can consult it exactly as you are doing right now.

On Thursday, November 10, 2016 at 9:01:46 AM UTC-6, Alberto Tonda wrote:

[Aaron Garrett]

Ha! You ninja'd me on that reply. Or am I a mind reader?...

On Thursday, November 10, 2016 at 9:01:46 AM UTC-6, Alberto Tonda wrote:

[Alberto Tonda]

Excellent! Thanks a lot, I completely missed that.

PS: both. I am a ninja, and you're a mind reader ;-)

--

You received this message because you are subscribed to a topic in the Google Groups "Inspyred" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/inspyred/mKwNHVAyRfk/unsubscribe.
To unsubscribe from this group and all its topics, send an email to inspyred+unsubscribe@googlegroups.com.
To post to this group, send email to insp...@googlegroups.com.
Visit this group at https://groups.google.com/group/inspyred.
For more options, visit https://groups.google.com/d/optout.

--

Alberto Tonda