def _value_iteration_slow(self):
old_values = dict(self.mdp.values)
for i in range(100):
values = value_iteration(self.mdp.values, self.mdp, num_iter=1)
policy = policy_extraction(values, self.mdp)
self.gridworldwindow.update_grid(values, policy)
self.mdp.update_values(values)
self.mdp.update_policies(policy)
self.gridworldwindow.window.update()
time.sleep(0.25)
self.gridworldwindow.window.update()
new_values = dict(values)
if values_converged(new_values, old_values):
break
old_values = new_values
self.gridworldwindow.show_dialog('Value Iteration has converged in {} steps!'.format(i+1))
def _value_iteration_100_steps(self):
values = value_iteration(self.mdp.values, self.mdp, num_iter=100)
policy = policy_extraction(values, self.mdp)
self.gridworld.update_grid(values, policy)
self.mdp.update_values(values)
self.mdp.update_policy(policy)
def _value_iteration_1_step(self):
values = value_iteration(self.mdp.values, self.mdp, num_iter=1)
policy = policy_extraction(values, self.mdp)
self.gridworldwindow.update_grid(values, policy)
self.mdp.update_values(values)
self.mdp.update_policies(policy)
