One Max Problem: Short Version

The short One Max genetic algorithm example is very similar to the full one One Max Problem. The only difference is that it makes use of the algorithms module which implements some basic evolutionary algorithms. The initializations are almost the same. We only have to import some additional packages and modules.

import array
import numpy

from deap import algorithms

In order to use the evolution functions implemented in algorithms, we have to register some functions from the tools module: evaluate(), mate(), mutate(), and select().

toolbox.register("evaluate", evalOneMax)
toolbox.register("mate", tools.cxTwoPoint)
toolbox.register("mutate", tools.mutFlipBit, indpb=0.05)
toolbox.register("select", tools.selTournament, tournsize=3)

The toolbox is then passed to the algorithm and via stats it uses the registered functions.

def main():
    pop = toolbox.population(n=300)
    hof = tools.HallOfFame(1)
    stats = tools.Statistics(lambda ind: ind.fitness.values)
    stats.register("avg", numpy.mean)
    stats.register("std", numpy.std)
    stats.register("min", numpy.min)
    stats.register("max", numpy.max)

    pop, log = algorithms.eaSimple(pop, toolbox, cxpb=0.5, mutpb=0.2, ngen=40, 
                                   stats=stats, halloffame=hof, verbose=True)

The short GA One max example makes use of a HallOfFame in order to keep track of the best individual to appear in the evolution (it keeps it even in the case of extinction), and a Statistics object to compile the population statistics during the evolution.

Every algorithm in the algorithms module can handle these objects. Finally, the verbose keyword indicates whether we want the algorithm to output the results after each generation or not.

The complete source code: examples/%sga/onemax_short.