def get_hv_for_algo(algo, max_fevals, pop_size, seed, problem, run):
max_gen = math.ceil(max_fevals / pop_size)
# same (random) starting population for both algos
pop = pg.population(problem, pop_size, seed)
f_list, x_list = pop.get_f(), pop.get_x()
for i in range(max_gen):
pop = algo.evolve(pop)
# Get all the f and x values for this generation
f_list = numpy.concatenate((f_list, pop.get_f()))
x_list = numpy.concatenate((x_list, pop.get_x()))
save_values(
'storedvalues/' + algo.get_name().split(':')[0] + '_x_' +
problem.get_name() + '_run' + str(run) + '.txt', x_list.tolist())
save_values(
'storedvalues/' + algo.get_name().split(':')[0] + '_f_' +
problem.get_name() + '_run' + str(run) + '.txt', f_list.tolist())
pop_empty = pg.population(prob=problem, seed=seed)
return reconstruct_hv_per_feval(max_fevals, x_list, f_list, pop_empty)
def calculate_mean_rbf(n,
max_fevals,
working_fevals,
seed,
problem,
cycle,
output_stream=None):
return_array = []
for i in range(n):
# Create dictionary for algo settings
dict_settings = {
'max_evaluations': max_fevals,
'rand_seed': seed,
}
var_types = ['R'] * problem.get_nx()
algo_rbfmopt = rbfmopt.RbfmoptWrapper(dict_settings, problem,
var_types, output_stream,
'tchebycheff', cycle)
# RBFMopt hypervolume calculations
algo_rbfmopt.evolve()
empty_pop = pg.population(prob=problem, seed=seed)
x_list = np.array(algo_rbfmopt.get_x_list())
f_list = np.array(algo_rbfmopt.get_f_list())
save_values(
'storedvalues/rbfmopt_x_cycle' + str(cycle) + '_' +
problem.get_name() + '_run' + str(i + 1) + '.txt', x_list.tolist())
save_values(
'storedvalues/rbfmopt_f_cycle' + str(cycle) + '_' +
problem.get_name() + '_run' + str(i + 1) + '.txt', f_list.tolist())
return_array.append(
reconstruct_hv_per_feval(working_fevals, algo_rbfmopt.get_x_list(),
algo_rbfmopt.get_f_list(), empty_pop))
# Make sure we change the seed each time the algo is being run
seed += (i + 1)
return numpy.mean(return_array, axis=0)
seed = 33
# For the each problem in the problem suite
for i in range(problem_number):
problem_function = getattr(pg.problems, problem_name)
if (problem_name == "dtlz"):
problem = pg.problem(problem_function(i + 1, dim=dim, fdim=fdim))
else:
problem = pg.problem(problem_function(i + 1, param=dim))
algo_moead = pg.algorithm(pg.moead(gen=1))
algo_nsga2 = pg.algorithm(pg.nsga2(gen=1))
# Hypervolume calculations, mean taken over n number of times
hv_rbfmopt_plot = calculate_mean_rbf(n, max_fevals, working_fevals, seed,
problem, cycle)
hv_moead_plot = calculate_mean_pyg(n, algo_moead, working_fevals, pop_size,
seed, problem)
hv_nsga2_plot = calculate_mean_pyg(n, algo_nsga2, working_fevals, pop_size,
seed, problem)
fevals_plot = range(0, max_fevals)
save_values(
'storedvalues/rbfmopt_hv_' + problem.get_name() + '_fevals' +
str(max_fevals) + '.txt', hv_rbfmopt_plot.tolist())
save_values(
'storedvalues/moead_hv_' + problem.get_name() + '_fevals' +
str(max_fevals) + '.txt', hv_moead_plot.tolist())
save_values(
'storedvalues/nsga2_hv_' + problem.get_name() + '_fevals' +
str(max_fevals) + '.txt', hv_nsga2_plot.tolist())
def calculate_mean_rbf(n, max_fevals, working_fevals, seed, problem, cycle,
max_filter):
return_array = []
for i in range(n):
# Create dictionary for algo settings
dict_settings = {
'max_evaluations': max_fevals,
'rand_seed': seed,
}
var_types = ['R'] * problem.get_nx()
file_string = 'log/rbfmopt_log_ncycle' + str(cycle) + '_filter' + str(
max_filter) + '_fevals' + str(
max_fevals) + problem.get_name() + '_run' + str(i + 1) + '.txt'
stream = open(file_string, 'a')
algo_rbfmopt = rbfmopt.RbfmoptWrapper(dict_settings, problem,
var_types, stream, 'tchebycheff',
cycle, max_filter)
# RBFMopt hypervolume calculations
start = time.time()
# RBFMopt hypervolume calculations
algo_rbfmopt.evolve()
end = time.time()
save_values(
'timer/rbfmopt_timer_ncycle' + str(cycle) + '_filter' +
str(max_filter) + '_fevals' + str(max_fevals) + '_' +
problem.get_name() + '_run' + str(i + 1) + '.txt', (end - start))
empty_pop = pg.population(prob=problem, seed=seed)
x_list = np.array(algo_rbfmopt.get_x_list())
f_list = np.array(algo_rbfmopt.get_f_list())
save_values(
'store_x/rbfmopt_x_ncycle' + str(cycle) + '_filter' +
str(max_filter) + '_fevals' + str(max_fevals) + '_' +
problem.get_name() + '_run' + str(i + 1) + '.txt', x_list.tolist())
save_values(
'store_f/rbfmopt_f_ncycle' + str(cycle) + '_filter' +
str(max_filter) + '_fevals' + str(max_fevals) + '_' +
problem.get_name() + '_run' + str(i + 1) + '.txt', f_list.tolist())
hv_for_run = reconstruct_hv_per_feval(working_fevals,
algo_rbfmopt.get_x_list(),
algo_rbfmopt.get_f_list(),
empty_pop)
save_values(
'store_hv/rbfmopt_hv_ncycle' + str(cycle) + '_filter' +
str(max_filter) + '_fevals' + str(max_fevals) +
problem.get_name() + '_run' + str(i + 1) + '.txt', hv_for_run)
return_array.append(hv_for_run)
# Make sure we change the seed each time the algo is being run
seed += (i + 1)
return numpy.mean(return_array, axis=0)
problem_number = 6
# max_fevals = (dim+1) * 50
max_fevals = 30
# To account for the fact that the zero index array in util functions
# are actually the 1st feval
working_fevals = max_fevals-1
pop_size = 24
seed = 33
cycle = 3
max_filter = 3
# For the each problem in the problem suite
for i in range(problem_number):
if i == 4:
continue
problem_function = getattr(pg.problems, problem_name)
problem = pg.problem(problem_function(i+1, param=dim))
hv_rbfmopt_plot = calculate_mean_rbf(n, max_fevals, working_fevals, seed, problem, cycle, None)
save_values('store_hv/rbfmopt_hv_ncycle' + str(cycle) + '_filter' + 'None' + '_fevals' + str(max_fevals) + problem.get_name() + '.txt', hv_rbfmopt_plot.tolist())
for j in range(3):
hv_rbfmopt_plot = calculate_mean_rbf(n, max_fevals, working_fevals, seed, problem, cycle, (j+1)*max_filter*dim)
save_values('store_hv/rbfmopt_hv_ncycle' + str(cycle) + '_filter' + str((j+1)*max_filter*dim) + '_fevals' + str(max_fevals) + problem.get_name() + '.txt', hv_rbfmopt_plot.tolist())
problem_name = 'zdt'
problem_number = 6
max_fevals = (dim + 1) * 2
# To account for the fact that the zero index array in util functions
# are actually the 1st feval
working_fevals = max_fevals - 1
pop_size = 24
seed = 33
default_rf = 3
# For the each problem in the problem suite
for i in range(problem_number):
problem_function = getattr(pg.problems, problem_name)
problem = pg.problem(problem_function(i + 1, param=dim))
for cycle_mult in range(3):
cycle = (cycle_mult + 1) * default_rf
hv_rbfmopt_plot = calculate_mean_rbf(n, max_fevals, working_fevals,
seed, problem, cycle)
save_values(
'storedvalues/rbfmopt_hv_cycle' + str(cycle) + '_' +
problem.get_name() + '_fevals' + str(max_fevals) + '.txt',
hv_rbfmopt_plot.tolist())