def expectimax( node, depth, nextPlayer, nofMerges, maxMerging, highestMerg ):
if depth == 0:
h = state.calculateHeuristic(node, nofMerges, maxMerging, highestMerg)
return h
if nextPlayer == 'ai':
return findBestSuccessor( node, depth, nofMerges, maxMerging, highestMerg)
if nextPlayer == 'board':
return findBestAverageSuccessor( node, depth, nofMerges, maxMerging, highestMerg)
def findBestSuccessor( node, depth, nofMerges, maxMerging, highestMerg):
bestHeuristic = float('-inf')
successors, merges, maxMergings, highestMerges = state.generateMAXSuccessors(node)
if len(successors) == 0:
return state.calculateHeuristic(node, nofMerges, maxMerging, highestMerg)
for i in xrange( len(successors) ):
succHeuristic = expectimax( successors[i], depth-1, 'board', merges[i], maxMergings[i], highestMerges[i])
if succHeuristic > bestHeuristic:
bestHeuristic = succHeuristic
return bestHeuristic
def findBestAverageSuccessor( node, depth, nofMerges, maxMerging, highestMerg):
weightedAverage = 0
# successors, probabilities = state.generateSuccessorsBiased(node) #C
successors, probabilities = state.generateCHANCESuccessors(node) #2C
if len(successors) == 0:
return state.calculateHeuristic(node, nofMerges, maxMerging, highestMerg)
for i in xrange(len(successors)):
if probabilities[i] < 0.0001:
continue
successorH = expectimax( successors[i], depth-1, 'ai', nofMerges , maxMerging, highestMerg)
weightedAverage += (probabilities[i] * successorH)
return weightedAverage