def expand(self, state: State):
if not self.already_expanded:
if not state.is_terminal():
if not state in self.tree:
self.tree[state] = [state.find_random_successor()]
else:
child = state.find_random_successor()
if not child in self.tree[state]:
self.tree[state] += [child]
def simulate(self, state: State) -> float:
"""
A simulation is rolled out using uniform random choices.
Return the simulation's reward (i.e., reward of the terminal state).
"""
while not state.is_terminal():
state = state.find_random_successor()
z = state.reward()
if state not in self.states_evaluated:
self.states_evaluated[state] = z
self.n_evaluations += 1
if z > 0:
self.n_positive_evaluations += 1
self.nof_reward_function_calls += 1
self.terminal_nodes_visits += 1
return z