def generation(env, old_population, new_population, p_mutation, p_crossover):
for i in range(0, len(old_population) - 1, 2):
print('generating',i)
# Selection
# parent1 = roulette_wheel_selection(old_population)
# parent2 = roulette_wheel_selection(old_population)
parent1, parent2 = ranking_selection(old_population)
# Crossover
child1 = copy.deepcopy(parent1)
child2 = copy.deepcopy(parent2)
child1.weights_biases, child2.weights_biases = crossover(parent1.weights_biases,
parent2.weights_biases,
p_crossover)
# Mutation
child1.weights_biases = mutation(child1.weights_biases, p_mutation)
child2.weights_biases = mutation(child2.weights_biases, p_mutation)
# Update model weights and biases
child1.update_model()
child2.update_model()
child1.calculate_fitness(env)
child2.calculate_fitness(env)
# If children fitness is greater thant parents update population
if child1.fitness + child2.fitness > parent1.fitness + parent2.fitness:
new_population[i] = child1
new_population[i + 1] = child2
else:
new_population[i] = parent1
new_population[i + 1] = parent2
def generation(env,
old_population: list,
new_population: list,
p_mutation: float,
p_crossover: float,
p_inversion: float = 0.0):
for i in range(0, len(old_population) - 1, 2):
# Selection
parent1, parent2 = ranking_selection(old_population)
# Crossover
child1 = copy.deepcopy(parent1)
child2 = copy.deepcopy(parent2)
child1.weights_biases, child2.weights_biases = crossover(
parent1.weights_biases, parent2.weights_biases, p_crossover)
# Mutation
child1.weights_biases = mutation(child1.weights_biases, p_mutation)
child2.weights_biases = mutation(child2.weights_biases, p_mutation)
# Update model weights and biases
child1.update_model()
child2.update_model()
child1.calculate_fitness(env)
child2.calculate_fitness(env)
# If children fitness is greater thant parents update population
if child1.fitness + child2.fitness > parent1.fitness + parent2.fitness:
new_population[i] = child1
new_population[i + 1] = child2
else:
new_population[i] = parent1
new_population[i + 1] = parent2