def run_simulation(self):
if not self.pause:
if self.current_timestep < self.max_timestep:
for i in range(len(self.environments)):
self.environments[i].drop_object_one_level()
agent_sensor_output = self.environments[i].agent.get_sensor_array(self.environments[i])
ann_inputs = agent_sensor_output
prediction = self.ann.predict(inputs=ann_inputs)
self.environments[i].prediction_to_maneuver(prediction=prediction)
self.draw_agents()
self.draw_beer_objects()
self.update_text()
self.current_timestep += 1
else:
print "Simulation over"
print self.ann.weights
PhenotypeBeerTracker.environments_for_fitness = self.environments
for l in range(len(self.environments)):
PhenotypeBeerTracker.environments_for_fitness[l].reset()
fitness = PhenotypeBeerTracker.fitness_evaluation(self.phenotype)
self.canvas.itemconfig(self.fitness_texts[l], text="Fitness:" + str(round(fitness[0], 3)) + " "+ str(fitness[1][l]))
#self.plot_data(self.fitness_log_average, self.fitness_log_best, self.standard_deviation_log)
return
self.after(self.delay, lambda: self.run_simulation())
def run_simulation(self):
if not self.pause:
if self.current_timestep < self.max_timestep:
for i in range(len(self.environments)):
self.environments[i].drop_object_one_level()
agent_sensor_output = self.environments[
i].agent.get_sensor_array(self.environments[i])
ann_inputs = agent_sensor_output
prediction = self.ann.predict(inputs=ann_inputs)
self.environments[i].prediction_to_maneuver(
prediction=prediction)
self.draw_agents()
self.draw_beer_objects()
self.update_text()
self.current_timestep += 1
else:
print "Simulation over"
print self.ann.weights
PhenotypeBeerTracker.environments_for_fitness = self.environments
for l in range(len(self.environments)):
PhenotypeBeerTracker.environments_for_fitness[l].reset()
fitness = PhenotypeBeerTracker.fitness_evaluation(
self.phenotype)
self.canvas.itemconfig(self.fitness_texts[l],
text="Fitness:" +
str(round(fitness[0], 3)) + " " +
str(fitness[1][l]))
#self.plot_data(self.fitness_log_average, self.fitness_log_best, self.standard_deviation_log)
return
self.after(self.delay, lambda: self.run_simulation())
def do_fitness_testing(self, environments):
PhenotypeBeerTracker.environments_for_fitness = environments
res = []
if MULTIPROCESSING:
res = parmap(PhenotypeBeerTracker.fitness_evaluation, self.phenotype_children_pool)
else:
for phenotype in self.phenotype_children_pool:
res.append(PhenotypeBeerTracker.fitness_evaluation(phenotype))
for i in range(len(res)):
self.phenotype_children_pool[i].fitness_value = res[i][0]
self.phenotype_children_pool[i].fitness_components = res[i][1]
# When using mixing in adult selection, the previous generation have to be reevaluated on the new scenarios
if self.adult_selection_protocol == 3:
res = parmap(PhenotypeBeerTracker.fitness_evaluation, self.phenotype_adult_pool)
for i in range(len(res)):
self.phenotype_adult_pool[i].fitness_value = res[i][0]
self.phenotype_adult_pool[i].fitness_components = res[i][1]
def init_phenotype_type(self, genotype):
return PhenotypeBeerTracker(genotype,
symbol_set_size=self.symbol_set_size,
hidden_layers=self.hidden_layers,
activation_functions=self.activation_functions)