def test_episode_runner_run(self):
crossing_state_params = CrossingStateDefaultParametersInt()
CrossingStateParametersInt.CHAIN_LENGTH = 21
runner = CrossingStateEpisodeRunnerInt(
{
1: AgentPolicyCrossingStateInt((5, 5), crossing_state_params),
2: AgentPolicyCrossingStateInt((5, 5), crossing_state_params)
}, [
AgentPolicyCrossingStateInt((4, 5), crossing_state_params),
AgentPolicyCrossingStateInt((5, 6), crossing_state_params)
], default_mcts_parameters(), crossing_state_params, 30, 200,
10000, None)
episode_result = runner.run(False)
print(episode_result)
def test_pickle_unpickle(self):
from mamcts import MctsParameters, HypothesisBeliefTracker
from mamcts import CrossingStateDefaultParametersFloat, CrossingStateDefaultParametersInt
params_mcts = MctsParameters()
params_mcts.DISCOUNT_FACTOR = 0.9
params_mcts.RANDOM_SEED = 1000
params_mcts.MAX_SEARCH_TIME = 1232423
params_mcts.MAX_NUMBER_OF_ITERATIONS = 2315677
params_mcts.random_heuristic.MAX_SEARCH_TIME = 10
params_mcts.random_heuristic.MAX_NUMBER_OF_ITERATIONS = 1000
params_mcts.uct_statistic.LOWER_BOUND = -1000
params_mcts.uct_statistic.UPPER_BOUND = 100
params_mcts.uct_statistic.EXPLORATION_CONSTANT = 0.7
params_mcts.hypothesis_statistic.COST_BASED_ACTION_SELECTION = True
params_mcts.hypothesis_statistic.PROGRESSIVE_WIDENING_HYPOTHESIS_BASED = True
params_mcts.hypothesis_statistic.LOWER_COST_BOUND = 0
params_mcts.hypothesis_statistic.UPPER_COST_BOUND = 1
params_mcts.hypothesis_statistic.PROGRESSIVE_WIDENING_ALPHA = 0.5
params_mcts.hypothesis_statistic.PROGRESSIVE_WIDENING_K = 1
params_mcts.hypothesis_statistic.EXPLORATION_CONSTANT = 0.7
params_mcts.hypothesis_belief_tracker.RANDOM_SEED_HYPOTHESIS_SAMPLING = 1000
params_mcts.hypothesis_belief_tracker.HISTORY_LENGTH = 4
params_mcts.hypothesis_belief_tracker.PROBABILITY_DISCOUNT = 1.0
params_mcts.hypothesis_belief_tracker.POSTERIOR_TYPE = HypothesisBeliefTracker.PosteriorType.PRODUCT
params_mcts.hypothesis_belief_tracker.FIXED_HYPOTHESIS_SET = {
1: 5,
10: 4,
3: 100
}
params_mcts_unpickle = pu(params_mcts)
self.assertTrue(is_equal_mcts_params(params_mcts,
params_mcts_unpickle))
params_crossing_state = CrossingStateDefaultParametersFloat()
params_crossing_state_unpickle = pu(params_crossing_state)
self.assertTrue(
is_equal_crossing_state_params(params_crossing_state,
params_crossing_state_unpickle))
params_crossing_state = CrossingStateDefaultParametersInt()
params_crossing_state_unpickle = pu(params_crossing_state)
self.assertTrue(
is_equal_crossing_state_params(params_crossing_state,
params_crossing_state_unpickle))
def test_episode_runner_step(self):
crossing_state_params = CrossingStateDefaultParametersInt()
CrossingStateParametersInt.CHAIN_LENGTH = 21
viewer = PyViewer()
runner = CrossingStateEpisodeRunnerInt(
{
1: AgentPolicyCrossingStateInt((5, 5), crossing_state_params),
2: AgentPolicyCrossingStateInt((5, 5), crossing_state_params)
}, [
AgentPolicyCrossingStateInt((4, 5), crossing_state_params),
AgentPolicyCrossingStateInt((5, 6), crossing_state_params)
], default_mcts_parameters(), crossing_state_params, 30, 200,
10000, viewer)
for _ in range(0, 20):
viewer.clear()
runner.step()
viewer.show()
def test_draw_state(self):
crossing_state_params = CrossingStateDefaultParametersInt()
viewer = PyViewer()
state = CrossingStateInt({}, crossing_state_params)
state.draw(viewer)
viewer.show(block=True)