genetic_algorithm.py

import random
import copy
import matplotlib.pyplot as plt


class GeneticAlgorithm:
    def __init__(self):
        self.chromosome_length = 50

        self.population = []

        self.n = 20
        self.crossovers = 5
        self.mutations = 5

        self.iterator = 0
        self.max_iterator = 500

        self.selection_cmp = None

    def initial(self):
        for i in range(self.n):
            self.population.append([random.randrange(-5, 5) for _ in range(self.chromosome_length)])

    def stop_condition(self):
        return self.iterator > self.max_iterator

    def do_crossover(self):
        chromo_a = self.population[random.randrange(0, self.n / 2)]
        chromo_b = self.population[random.randrange(self.n / 2, self.n)]

        random_index = random.randrange(self.chromosome_length)

        new_chromo_a = chromo_a[:random_index] + chromo_b[random_index:]
        new_chromo_b = chromo_b[:random_index] + chromo_a[random_index:]
        self.population.append(new_chromo_a)
        self.population.append(new_chromo_b)

    def do_mutation(self):
        for _ in range(self.mutations):
            chromo = copy.copy(self.population[random.randrange(len(self.population))])
            index = random.randrange(len(chromo) - 1)
            chromo[index] += random.randrange(-5, 5)
            self.population.append(chromo)

    def do_selection(self):
        self.population.sort(cmp=self.selection_cmp)
        self.population = self.population[:self.n]

    @staticmethod
    def fitness(chromo):
        result = 0
        for chromosome in chromo:
            result += 3 * (chromosome ** 4) - 4 * (chromosome ** 2) + 3 * chromosome - 7
        return result

    def __iter__(self):
        return self

    def next(self):
        if self.iterator < self.max_iterator:
            i = self.iterator
            self.iterator += 1
            return i
        else:
            raise StopIteration()


if __name__ == "__main__":
    perfect_fitness_list = []

    ga = GeneticAlgorithm()
    ga.initial()
    ga.selection_cmp = lambda x, y: ga.fitness(x) - ga.fitness(y)

    for iterator in ga:
        ga.do_crossover()
        ga.do_mutation()
        ga.do_selection()

        fitness_list = [ga.fitness(a) for a in ga.population]
        fitness = max(fitness_list)
        perfect_fitness_list.append([ga.iterator, fitness])

    fig, ax = plt.subplots()
    ax.plot([a[0] for a in perfect_fitness_list], [a[1] for a in perfect_fitness_list])
    ax.set(xlabel='iterates', ylabel='perfect fitness', title='Genetic algorithm')
    fig.savefig("plot.png")
    plt.show()