def main():
max_iteration = 100
bin_mutation_par = {}
# bin_mutation_par['steps'] = array([0.1, 0.2, 0.3, 0.7])*max_iteration
# bin_mutation_par['steps'] = bin_mutation_par['steps'].astype('int')
# bin_mutation_par['probs'] = [0.5, 0.2, 0.01, 0.005, 0.001]
bin_mutation_par["initial_prob"] = 0.5
bin_mutation_par['exp_coef'] = 0.001
bin_mutation_par['min_prob'] = 0.001
# Instantiating genetic operators
binmut_op = binmut.PunctualExpBinaryMutation(bin_mutation_par)
#binmut_op = binmut.PunctualStepBinaryMutation(bin_mutation_par)
cross_op = bincross.OnePointCrossover()
selection_op = selec.ElitistTournamentSelection(0.6)
# Population size
N = 30
# Generating population
init_sol = random.rand(N, GRID_SIZE**2) > 0.5
population = concatenate([init_sol, zeros((init_sol.shape[0], 1))], axis=1)
new_population = zeros((population.shape[0] * 2, population.shape[1]))
# Creating fitness tracking
fitness_tracking = zeros((max_iteration, 3))
for i in arange(max_iteration):
# Population fitness evaluation
population[:, -1] = objective_fun(population)
# Population mutation
population = binmut_op.mutate(population, i)
# Population crossover
new_population, idx_sons = cross_op.cross(population)
# Fitness evaluation
new_population[:, -1] = objective_fun(new_population)
# Selecting individuals
population = selection_op.select(new_population)
fitness_tracking[i, 0] = max(population[:, -1])
fitness_tracking[i, 1] = min(population[:, -1])
fitness_tracking[i, 2] = mean(population[:, -1])
print 'Max %s, Min %s, Mean %s, It %s' % (max(
population[:, -1]), min(population[:, -1]), mean(
population[:, -1]), i)
idx_sort_fitness = argmax(population[:, -1])
best_one = population[idx_sort_fitness, :-1]
print float(
sum(A == reshape(best_one, (GRID_SIZE, GRID_SIZE)))) / GRID_SIZE**2
savetxt('fitness_evol', fitness_tracking, fmt='%1.4f')
savetxt('best_sol', population[argmax(population[:, -1])], fmt='%1.4f')
def main():
objects = loadtxt('knapsack.txt')
max_capacity = 6404180
max_iteration = 200
bin_mutation_par = {}
bin_mutation_par['steps'] = array([0.1, 0.2, 0.3, 0.4, 0.8])*max_iteration
bin_mutation_par['steps'] = bin_mutation_par['steps'].astype('int')
bin_mutation_par['probs'] = [0.5, 0.3, 0.2, 0.2, 0.15, 0.08]
# Instantiating genetic operators
binmut_op = binmut.ElitistPunctualStepBinaryMutation(bin_mutation_par)
cross_op = bincross.ElitistTwoPointCrossover()
selection_op = selec.ElitistTournamentSelection(0.45)
N = 50
# Generating population
init_sol = zeros((N, objects.shape[0]))
i = 0
while i < N:
init_sol[i] = random.rand(1, objects.shape[0]) > 0.5
if(sum(init_sol[i])/objects.shape[0] > 0.6 or sum(init_sol[i])/objects.shape[0] < 0.4):
continue
else:
i += 1
population = concatenate([init_sol, zeros((init_sol.shape[0], 1))], axis = 1)
new_population = zeros((population.shape[0]*2, population.shape[1]))
# Creating fitness tracking
fitness_tracking = zeros((max_iteration, 3))
for i in arange(max_iteration):
# Population fitness evaluation
population[:, -1] = objective_fun(objects, max_capacity, population)
# Double the population thought mutation in populatipopulation.shape[0]on
keyboard;
new_population[0:population.shape[0], :] = population
new_population[population.shape[0]:, :] = binmut_op.mutate(population, i)
# Fitness evaluation
new_population[:, -1] = objective_fun(objects, max_capacity, new_population)
# Selecting individuals
population = selection_op.select(new_population)
fitness_tracking[i, 0] = max(population[:, -1])
fitness_tracking[i, 1] = min(population[:, -1])
fitness_tracking[i, 2] = mean(population[:, -1])
# print 'Max %s, Min %s, Mean %s, It %s' % (max(population[:, -1]), min(population[:, -1]), mean(population[:, -1]), i)
savetxt('fitness_evol', fitness_tracking, fmt='%1.4f')
savetxt('best_sol', population[argmax(population[:, -1])], fmt='%1.4f')
def main():
dist_matrix = loadtxt('dist.txt')
max_iteration = 2000
# Instantiating genetic operators
binmut_op = permut.ElitistPercentReverseMutation(1)
cross_op = permcross.ElitistPMXCrossover()
selection_op = selec.ElitistTournamentSelection(0.5)
# Population size
N = 50
# Generating population
init_sol = zeros((N, dist_matrix.shape[1]))
for i in arange(0, N):
init_sol[i, :] = random.permutation(dist_matrix.shape[1])
population = concatenate([init_sol, zeros((init_sol.shape[0], 1))], axis=1)
new_population = zeros((population.shape[0] * 2, population.shape[1]))
# Creating fitness tracking
fitness_tracking = zeros((max_iteration, 3))
for i in arange(max_iteration):
# Population fitness evaluation
population[:, -1] = objective_fun(dist_matrix, population)
# Population mutation
population = binmut_op.mutate(population, i)
# Population crossover
new_population, idx_sons = cross_op.cross(population)
# Fitness evaluation
new_population[:, -1] = objective_fun(dist_matrix, new_population)
# Selecting individuals
population = selection_op.select(new_population)
fitness_tracking[i, 0] = max(population[:, -1])
fitness_tracking[i, 1] = min(population[:, -1])
fitness_tracking[i, 2] = mean(population[:, -1])
print 'Max %s, Min %s, Mean %s, It %s' % (max(
population[:, -1]), min(population[:, -1]), mean(
population[:, -1]), i)
savetxt('fitness_evol', fitness_tracking, fmt='%1.4f')
savetxt('best_sol', population[argmax(population[:, -1])], fmt='%1.4f')
def main():
objects = loadtxt('sum_diff_order.txt')
max_iteration = 400
bin_mutation_par = {}
bin_mutation_par['steps'] = array([0.1, 0.2, 0.3, 0.4, 0.8
]) * max_iteration
bin_mutation_par['steps'] = bin_mutation_par['steps'].astype('int')
bin_mutation_par['probs'] = [0.3, 0.2, 0.08, 0.025, 0.015, 0.01]
# Instantiating genetic operators
binmut_op = binmut.PunctualStepBinaryMutation(bin_mutation_par)
cross_op = bincross.OnePointCrossover()
selection_op = selec.ElitistTournamentSelection(0.6)
# Population size
N = 200
# Population option
op = 1
# Generating population
if op == 1:
init_sol = random.rand(N, objects.shape[0]) > 0.5
else:
init_sol = zeros((N, objects.shape[0]))
i = 0
while i < N:
init_sol[i] = random.rand(1, objects.shape[0]) > 0.5
if (sum(init_sol[i]) / objects.shape[0] > 0.6
or sum(init_sol[i]) / objects.shape[0] < 0.4):
continue
else:
i += 1
population = concatenate([init_sol, zeros((init_sol.shape[0], 1))], axis=1)
new_population = zeros((population.shape[0] * 2, population.shape[1]))
# Creating fitness tracking
fitness_tracking = zeros((max_iteration, 3))
for i in arange(max_iteration):
# Population fitness evaluation
population[:, -1] = objective_fun(objects, population)
# Population mutation
population = binmut_op.mutate(population, i)
# Population crossover
new_population, idx_sons = cross_op.cross(population)
# Fitness evaluation
new_population[:, -1] = objective_fun(objects, new_population)
# Selecting individuals
population = selection_op.select(new_population)
fitness_tracking[i, 0] = max(population[:, -1])
fitness_tracking[i, 1] = min(population[:, -1])
fitness_tracking[i, 2] = mean(population[:, -1])
print 'Max %s, Min %s, Mean %s, It %s' % (max(
population[:, -1]), min(population[:, -1]), mean(
population[:, -1]), i)
savetxt('fitness_evol', fitness_tracking, fmt='%1.4f')
savetxt('best_sol', population[argmax(population[:, -1])], fmt='%1.4f')