genetic_algorithm.py

import csv
import random
import time

from poulation import whole_population, evolve_next_gen, population_member


if __name__ == '__main__':
    # set up dictionairy to contain longitude and latitude
    town_positions = {}

    # read in town names and positions
    with open("/home/kothgasser-paul/Downloads/european-no-duplicates.csv") as townfile:
        reader = csv.reader(townfile)
        # skip first line
        reader.__next__()
        for line in reader:
            town_positions[line[0]] = [float(i) for i in line[1:]]

    town_names = list(town_positions.keys())

    population_size = 1500
    parenthood_choice = int(0.25*population_size)

    first_generation = [population_member(random.sample(town_names, len(town_names)), town_positions) for i in range(population_size)]
    generation = 0
    elitism = int(0.1 * population_size)
    mutation_probability = 0.2

    current_population = whole_population(generation, elitism, mutation_probability, first_generation)
    current_fittest = current_population.order_by_fitness()[0].fitness_value()


    startime = time.time()
    while generation < 2500:

        current_fittest = current_population.order_by_fitness()[0].fitness_value()
        current_population = evolve_next_gen(current_population, parenthood_choice)
        generation = current_population.generation

        if generation in [1, 10, 20, 50, 80, 100, 150, 200, 400, 500, 700, 1000, 1200, 1400, 1600, 1800, 2000]:
            print(f'generation: {generation} || fitness : {current_population.order_by_fitness()[0].fitness_value()} || time {time.time() - startime}')
    print(f'generation: {generation} || fitness : {current_population.order_by_fitness()[0].fitness_value()} || time {time.time() - startime}')
    print(current_population.order_by_fitness()[0].ordered_towns)
    print(time.time() - startime)