def receive_reward(self, reward):
for chain in sublists(self.chain):
# Consider the previous moves to be the current state
state = chain[1:]
action = chain[0]
self.update_chain(state, action, reward)
self.on_frame_end()
def receive_reward(self, reward):
for chain in sublists(self.chain):
# Consider the previous moves to be the current state
state = chain[1:]
action = chain[0]
self.update_chain(state, action, reward)
self.on_frame_end()
def get_greedy_action(self, available_actions):
# Do a tree search in the previously seen states
# that match the current state
best_action = None
best_value = None
for state in sublists(self.chain):
lhstate = listhash(state)
if lhstate in self.q:
s = self.q[lhstate]
for a in available_actions:
if a in s:
val = s[a]
if val > best_value:
best_action = a
best_value = val
return best_action
def get_greedy_action(self, available_actions):
# Do a tree search in the previously seen states
# that match the current state
best_action = None
best_value = None
for state in sublists(self.chain):
lhstate = listhash(state)
if lhstate in self.q:
s = self.q[lhstate]
for a in available_actions:
if a in s:
val = s[a]
if val > best_value:
best_action = a
best_value = val
return best_action
