def getAction(self, state):
GameState.getAndResetExplored()
studentAction = (self.studentAgent.getAction(state), len(GameState.getAndResetExplored()))
print studentAction
optimalActions = self.optimalActions[self.stepCount]
altDepthActions = self.altDepthActions[self.stepCount]
partialPlyBugActions = self.partialPlyBugActions[self.stepCount]
studentOptimalAction = False
curRightStatesExplored = False;
for i in range(len(optimalActions)):
if studentAction[0] in optimalActions[i][0]:
studentOptimalAction = True
else:
self.actionsConsistentWithOptimal[i] = False
if studentAction[1] == int(optimalActions[i][1]):
curRightStatesExplored = True
if not curRightStatesExplored and self.wrongStatesExplored < 0:
self.wrongStatesExplored = 1
for i in range(len(altDepthActions)):
if studentAction[0] not in altDepthActions[i]:
self.actionsConsistentWithAlternativeDepth[i] = False
for i in range(len(partialPlyBugActions)):
if studentAction[0] not in partialPlyBugActions[i]:
self.actionsConsistentWithPartialPlyBug[i] = False
if not studentOptimalAction:
self.suboptimalMoves.append((state, studentAction[0], optimalActions[0][0][0]))
self.stepCount += 1
random.seed(self.seed + self.stepCount)
return optimalActions[0][0][0]
def getAction(self, state):
GameState.getAndResetExplored()
studentAction = (self.studentAgent.getAction(state), sum(map(hash, GameState.getAndResetExplored())))
optimalActions = self.optimalActions[self.stepCount]
altDepthActions = self.altDepthActions[self.stepCount]
partialPlyBugActions = self.partialPlyBugActions[self.stepCount]
studentOptimalAction = False
curRightStatesExplored = False;
for i in range(len(optimalActions)):
if studentAction[0] in optimalActions[i][0]:
studentOptimalAction = True
else:
self.actionsConsistentWithOptimal[i] = False
if studentAction[1] == int(optimalActions[i][1]):
curRightStatesExplored = True
if not curRightStatesExplored and self.wrongStatesExplored < 0:
self.wrongStatesExplored = 1
for i in range(len(altDepthActions)):
if studentAction[0] not in altDepthActions[i]:
self.actionsConsistentWithAlternativeDepth[i] = False
for i in range(len(partialPlyBugActions)):
if studentAction[0] not in partialPlyBugActions[i]:
self.actionsConsistentWithPartialPlyBug[i] = False
if not studentOptimalAction:
self.suboptimalMoves.append((state, studentAction[0], optimalActions[0][0][0]))
self.stepCount += 1
random.seed(self.seed + self.stepCount)
return optimalActions[0][0][0]
def getAction(self, state):
# survey agents
GameState.getAndResetExplored()
optimalActionLists = []
for agent in self.solutionAgents:
optimalActionLists.append(
(
agent.getBestPacmanActions(state)[0],
len(GameState.getAndResetExplored()),
)
)
alternativeDepthLists = [
agent.getBestPacmanActions(state)[0]
for agent in self.alternativeDepthAgents
]
partialPlyBugLists = [
agent.getBestPacmanActions(state)[0] for agent in self.partialPlyBugAgents
]
# record responses
self.optimalActionLists.append(optimalActionLists)
self.alternativeDepthLists.append(alternativeDepthLists)
self.partialPlyBugLists.append(partialPlyBugLists)
self.stepCount += 1
random.seed(self.seed + self.stepCount)
return optimalActionLists[0][0][0]
def processGames(self):
maxNumTraining = DatatypeUtils.stringToInteger(self.view.numTrainingVar.get())
increment = DatatypeUtils.stringToInteger(self.view.incrementVar.get())
#from pympler import tracker
#memory_tracker = tracker.SummaryTracker()
for numTraining in range(0, maxNumTraining, increment):
self.processGame(numTraining)
GameState.getAndResetExplored()
def getAction(self, state):
# survey agents
GameState.getAndResetExplored()
optimalActionLists = []
for agent in self.solutionAgents:
optimalActionLists.append((agent.getBestPacmanActions(state)[0], sum(map(hash, GameState.getAndResetExplored()))))
alternativeDepthLists = [agent.getBestPacmanActions(state)[0] for agent in self.alternativeDepthAgents]
partialPlyBugLists = [agent.getBestPacmanActions(state)[0] for agent in self.partialPlyBugAgents]
# record responses
self.optimalActionLists.append(optimalActionLists)
self.alternativeDepthLists.append(alternativeDepthLists)
self.partialPlyBugLists.append(partialPlyBugLists)
self.stepCount += 1
random.seed(self.seed + self.stepCount)
return optimalActionLists[0][0][0]
def getAction(self, state):
GameState.getAndResetExplored()
studentAction = (self.studentAgent.getAction(state),
len(GameState.getAndResetExplored()))
optimalActions = self.optimalActions[self.stepCount]
studentOptimalAction = False
curRightStatesExplored = False
for i in range(len(optimalActions)):
if studentAction[0] in optimalActions[i][0]:
studentOptimalAction = True
else:
self.actionsConsistentWithOptimal[i] = False
if studentAction[1] == int(optimalActions[i][1]):
curRightStatesExplored = True
if not curRightStatesExplored and self.wrongStatesExplored < 0:
self.wrongStatesExplored = 1
if not studentOptimalAction:
self.suboptimalMoves.append(
(state, studentAction[0], optimalActions[0][0][0]))
self.stepCount += 1
random.seed(self.seed + self.stepCount)
return optimalActions[0][0][0]
