def __init__(self, policy):
# world object, (starting state is trivial)
world = World((0,0),(1,1))
value = {}
for state in world.allStates():
value[state] = 0
discount = 0.9
delta = 1
while abs(delta) > 0.00001:
delta = 0
for state in world.allStates():
world.setState(state)
old = value[state]
# we can set the minimum to 0 since we know every value will be 0 or positive
curMax = 0
for move in world.moveList():
if world.posAfterMove(move) == (0,0):
probSum = 10
else:
probSum = 0
for nextState,prob in world.nextPreyStates():
probSum += prob*discount*value[nextState]
curMax = max(curMax,probSum)
value[state] = curMax
delta = max(delta,abs(old - curMax))
value[(0,0)] = 10
self.value = value
self.actionList = []
self.allList = []
self.bottomPolicy = policy
self.discount = discount
def isOptimal(self,state, move): world = World((0,0),(1,1)) ourMove = 0 bestMove = 0 for nmove in world.moveList(): world.setState(state) world.move(nmove) if world.position == (0,0): probSum = 10 else: probSum = 0 for nextState,prob in world.nextPreyStates(): probSum += prob*self.discount*self.value[nextState] bestMove = max(bestMove,probSum) if nmove == move: ourMove = probSum return ourMove/bestMove > 0.97