def noncapturewin(pieces):
# If a lion is already captured, then this function reverts to being
# the capturewin function.
lioncapture = capturewin(pieces)
if lioncapture is not None : return lioncapture
# If Player A's lion has reached the furthest rank.
if pieces['lion'][0][1][0] == 1:
selectedactions = selection(b, pieces)
for action in selectedactions:
state = perform(action, pieces)
if state is not None:
if capturewin(state) == b:
return b
return a
# If Player B's lion has reached the furthest rank.
if pieces['lion'][1][1][0] == 4:
selectedactions = selection(a, pieces)
for action in selectedactions:
state = perform(action, pieces)
if state is not None:
if capturewin(state) == a:
return a
return b
def alphabeta(pieces, depth, alpha=-math.inf, beta=+math.inf):
if depth <= 0: return
selected = selection(b, pieces)
# Randomly permute the children of the root node before alpha-beta
# pruning begins.
random.shuffle(selected)
possibilities = []
# Determine all candidate moves for Player B.
for action in selected:
possibilities.append(perform(action, pieces))
possibilities = [i for i in possibilities if i is not None]
# Apply alpha-beta pruning to the root node.
maximinlist = []
for possibility in possibilities:
value = minimizer(possibility, depth - 1, alpha, beta)
maximinlist.append(value)
alpha = max(alpha, value)
maxvalue = max(maximinlist)
maxindex = maximinlist.index(maxvalue)
return possibilities[maxindex]
def enter(side, command, pieces):
try:
action = translate(side, command, pieces)
except:
print("Press enter to exit the game.")
input()
quit()
return perform(action, pieces)
def maximizer(pieces, depth, alpha, beta):
# Skew the score if one side has at least two more pieces.
skew = 0
majority = piecemajority(pieces)
if majority == a: skew = -0.01
elif majority == b: skew = 0.01
# If one side wins, provide a score. Otherwise, if the depth
# limit has been reached, provide the score skew.
if capturewin(pieces) == b: return 1
elif noncapturewin(pieces) == b: return 1
elif capturewin(pieces) == a: return -1
elif noncapturewin(pieces) == a: return -1
elif depth == 0: return skew
# Determine all candidate moves for Player B at this node.
selected = selection(b, pieces)
possibilities = []
for action in selected:
possibilities.append(perform(action, pieces))
possibilities = [i for i in possibilities if i is not None]
# Apply alpha-beta pruning to this node.
maximinlist = []
for possibility in possibilities:
value = minimizer(possibility, depth - 1, alpha, beta)
maximinlist.append(value)
alpha = max(alpha, value)
if beta <= alpha:
break
return max(maximinlist) + skew