def run_optimal():
vis = Visualiser(env, 80)
numActions = env.n_actions
playerA = QLearn(actions=list(range(numActions)))
playerA.load_Qtable("saved_players/QR")
playerB = QLearn(actions=list(range(numActions)))
playerB.load_Qtable("saved_players/QR_base")
for episode in range(500):
observation = env.reset()
vis.update_canvas(env)
while(True):
actionA = playerA.choose_action(str(observation))
actionB = playerB.choose_action(str(observation))
observation_, reward, done = env.step(actionA, actionB)
observation = observation_
vis.update_canvas(env)
if done:
vis.reset()
break
print("Games won: " + str(env.win_count))
vis.destroy()
def run_optimalB():
numActions = env.n_actions
drawProbability = 0.1
decay = 10**(-2. / N_EPISODES * 0.05)
vis = Visualiser(env, 80)
numActions = env.n_actions
playerB = MinimaxQPlayer(numActions, numActions, decay=decay, expl=0.00, gamma=1-drawProbability)
playerB.load_Qtable("MR")
playerA = QLearn(actions=list(range(numActions)))
playerA.load_Qtable("saved_players/MR_base")
for episode in range(20):
observation = env.reset()
vis.update_canvas(env)
while(True):
actionA = playerA.choose_action(str(observation))
actionB = playerB.choose_action(str(observation))
observation_, reward, done = env.step(actionA, actionB)
observation = observation_
vis.update_canvas(env)
if done:
vis.reset()
break
print("Games won: " + str(env.win_count))
vis.destroy()
