def test_expected_utility():
num_players = 2
node_map = {i: {} for i in range(num_players)}
action_map = {i: {} for i in range(num_players)}
cards = [Card(14, 1), Card(13, 1), Card(12, 1)]
learn(10000, cards, 2, node_map, action_map)
exploit = exploitability(cards, 2, node_map, action_map)
print(exploit)
assert exploit < .001, f"Exploitability was : {exploit}"
util = expected_utility(cards, 2, 2, node_map, action_map)
print(util)
print(json.dumps(action_map, indent=4))
print(node_map)
assert util.sum() == 0
assert np.isclose(np.abs(util[0]), np.abs(util[1])), f"Util was {util}"
assert np.abs(util).sum() > 0, f"Util was {util}"
regret = util[action] - node_util[state.turn]
node.regret_sum[action] += regret * reach_prob
return node_util
if __name__ == '__main__':
num_players = 2
node_map = {i: {} for i in range(num_players)}
action_map = {i: {} for i in range(num_players)}
cards = [
Card(14, 1),
Card(13, 1),
Card(12, 1),
Card(14, 2),
Card(13, 2),
Card(12, 2)
]
learn(10000, cards, 3, node_map, action_map)
exploit = exploitability(cards, 3, node_map, action_map)
print(exploit)
for player in node_map:
print(f"Player {player}")
print('Number of info sets', len(node_map[player]))
for info_set, node in node_map[player].items():
avg_strat = node.avg_strategy()
print(f"{info_set}: {avg_strat}")
util = expected_utility(cards, 3, 2, node_map, action_map)
print(util)