def compute_avg(trial_path, overall=True):
num_trials = 0
cum_fitness = list()
for trial in parser.get_trial_list(trial_path):
# Only initialize the array on the first iteration
if not num_trials:
cum_fitness = np.zeros((2, trial.num_generations))
num_trials += 1
if overall:
for generation in trial:
cum_fitness[0][generation.number] += generation.max_fitness
cum_fitness[1][generation.number] += generation.ave_fitness
else:
for generation in trial:
selected_fitness = [
i.fitness for i in generation if i.selected
]
if selected_fitness:
cum_fitness[0][generation.number] += np.amax(
selected_fitness)
cum_fitness[1][generation.number] += np.mean(
selected_fitness)
return np.divide(cum_fitness, num_trials, dtype=np.float)
def compute_avg(trial_path, overall=True):
num_trials = 0
cum_fitness = list()
for trial in parser.get_trial_list(trial_path):
# Only initialize the array on the first iteration
if not num_trials:
cum_fitness = np.zeros((2, trial.num_generations))
num_trials += 1
if overall:
for generation in trial:
cum_fitness[0][generation.number] += generation.max_fitness
cum_fitness[1][generation.number] += generation.ave_fitness
else:
for generation in trial:
selected_fitness = [i.fitness for i in generation if i.selected]
if selected_fitness:
cum_fitness[0][generation.number] += np.amax(selected_fitness)
cum_fitness[1][generation.number] += np.mean(selected_fitness)
return np.divide(cum_fitness, num_trials, dtype=np.float)
def count(trial_path):
num_trials = 0
cum_nurturing = list()
for trial in parser.get_trial_list(trial_path):
if not num_trials:
cum_nurturing = np.zeros((3, trial.num_generations))
num_trials += 1
pop_size = trial.config['populationSize']
for generation in trial:
for i in xrange(pop_size):
a, b = generation[i], generation[i + pop_size]
if a.light_first or b.light_first:
# A nurtured B
if b.light_first and a.switch and a <= b:
cum_nurturing[0][generation.number] += 1
# B nurtured A
elif a.light_first and b.switch and a >= b:
cum_nurturing[0][generation.number] += 1
# Self care
else:
cum_nurturing[1][generation.number] += 1
# Self care
elif a.switch or b.switch:
if (a.fitness != 0) or (b.fitness != 0):
cum_nurturing[1][generation.number] += 1
else:
cum_nurturing[2][generation.number] += 1
elif (a.fitness == 0) and (b.fitness == 0):
cum_nurturing[2][generation.number] += 1
else:
cum_nurturing[2][generation.number] += 1
print 'Warning Case else: {} {}'.format(a, b)
return cum_nurturing, num_trials