def __init__(
self,
# alpha=0.01, gamma=0.9, epsilon=0.9,
verbose=False):
self.states = [0, 1, 2, 3, 4, 5, 6, 7]
self.actions = [-4, -1, 0, -1, 4]
self.S = SoccerEnv().S
self.A = SoccerEnv().A
self.q_table_A = pd.DataFrame(
1,
# np.random.rand(len(self.S), len(self.A)),
index=pd.MultiIndex.from_tuples(self.S),
columns=pd.MultiIndex.from_tuples(self.A),
dtype=np.float64)
self.q_table_B = pd.DataFrame(
1,
# np.random.rand(len(self.S), len(self.A)),
index=pd.MultiIndex.from_tuples(self.S),
columns=pd.MultiIndex.from_tuples(self.A),
dtype=np.float64)
self.verbose = verbose
self.vi_A = pd.DataFrame(1,
index=pd.MultiIndex.from_tuples(self.S),
columns=['value'],
dtype=np.float64)
self.vi_B = pd.DataFrame(1,
index=pd.MultiIndex.from_tuples(self.S),
columns=['value'],
dtype=np.float64)
def __init__(
self,
# alpha=0.01, gamma=0.9, epsilon=0.9,
verbose=False):
self.states = [0, 1, 2, 3, 4, 5, 6, 7]
self.actions = [-4, -1, 0, -1, 4]
self.S = SoccerEnv().S
self.A = SoccerEnv().A
# self.q_table_A = pd.DataFrame(np.random.randint(-10,10, size=(len(self.S),len(self.A))),
# index=pd.MultiIndex.from_tuples(self.S),
# columns=pd.MultiIndex.from_tuples(self.A), dtype=np.float64) /1000
# self.q_table_B = pd.DataFrame(np.random.randint(-10,10, size=(len(self.S),len(self.A))),
# index=pd.MultiIndex.from_tuples(self.S),
# columns=pd.MultiIndex.from_tuples(self.A), dtype=np.float64) /1000
self.q_table_A = pd.DataFrame(0,
index=pd.MultiIndex.from_tuples(self.S),
columns=pd.MultiIndex.from_tuples(
self.A),
dtype=np.float64)
self.q_table_B = pd.DataFrame(0,
index=pd.MultiIndex.from_tuples(self.S),
columns=pd.MultiIndex.from_tuples(
self.A),
dtype=np.float64)
self.verbose = verbose
def __init__(self,
# alpha=0.01, gamma=0.9, epsilon=0.9,
verbose=False):
self.states = [0, 1, 2, 3, 4, 5, 6, 7]
self.actions = [-4,-1,0,1,4]
self.S = SoccerEnv().S
self.A = SoccerEnv().A
self.q_table_A = pd.DataFrame(0, index=self.states, columns= self.actions, dtype=np.float64)
self.q_table_B = pd.DataFrame(0, index=self.states, columns=self.actions, dtype=np.float64)
self.verbose = verbose
def __init__(
self,
# alpha=0.01, gamma=0.9, epsilon=0.9,
verbose=False):
self.states = [0, 1, 2, 3, 4, 5, 6, 7]
self.actions = [-4, 4, 1, -1, 0]
self.S = SoccerEnv().S
self.A = SoccerEnv().A
self.q_table_A = pd.DataFrame(0,
index=pd.MultiIndex.from_tuples(self.S),
columns=pd.MultiIndex.from_tuples(
self.A),
dtype=np.float64)
self.q_table_B = pd.DataFrame(0,
index=pd.MultiIndex.from_tuples(self.S),
columns=pd.MultiIndex.from_tuples(
self.A),
dtype=np.float64)
self.verbose = verbose
# "ce_Q/time_ce_a_%0.5f_adecay_%0.5f_timeout%f.png" % (start_alpha, alpha_decay_rate, timeout))
# plt.clf()
#
# q_output_A.to_csv('ce_Q/ce_A_a_%0.5f_adecay_%0.5f_timeout%f.csv' % (start_alpha, alpha_decay_rate, timeout))
# q_output_B.to_csv('ce_Q/ce_B_a_%0.5f_adecay_%0.5f_timeout%f.csv' % (start_alpha, alpha_decay_rate, timeout))
# end of game
print('game over')
# env.destroy()
if __name__ == "__main__":
seed = 1 # seed
np.random.seed(seed)
'''soccer env'''
env = SoccerEnv()
env.seed(seed)
for alpha in [
# 1,
# 0.75,
0.5,
# 0.25
]:
for alpha_decay_rate in [
# 1,
# 0.99999,
# # 0.9999,
0.9995,
# 0.999,
]: