def max_value_first(board, alpha, beta, depth):
v = float("-inf")
moves = mmUtil.get_moves(board)
# Get number of extra processors after first level
num_extra = numCPU - len(moves)
# Ensure num_extra is positive
if num_extra < 0:
num_extra = 0
pool = p.MyPool(numCPU - num_extra)
func = partial(t.find_min, board, activePlayer, alpha, beta, depth)
# Make array of tuples for each node of tree at the next depth
# each tuple: (move, # of moves available)
next_nodes = []
for move in moves:
temp_board = mmUtil.make_move(board, move, activePlayer)
num_next = mmUtil.get_moves(temp_board)
next_nodes.append((move, len(num_next)))
# Sort by decreasing number of moves available
sorted(next_nodes, key=itemgetter(1), reverse=True)
# Create data array for number of processors allocated for each move
data = []
# Calculate the base number of extra processors that each move will use
base = num_extra / len(moves)
# For each tuple add the base number of extra processors to the default amount (1)
for node in next_nodes:
data.append((node[0], 1 + base))
# Allocate remaining processor with moves that have the highest branching factor
for i in range((num_extra % len(moves))):
node = data[i]
data[i] = (node[0], node[1] + 1)
v_mins = pool.map_async(func, data)
pool.close()
pool.join()
mov_val = max(v_mins.get(), key=itemgetter(1)) # (move, value)
print v_mins.get()
print mov_val
return mov_val[::-1] # (value, move)
def max_value_first(board, alpha, beta, depth):
v = float('-inf')
moves = mmUtil.get_moves(board)
# Get number of extra processors after first level
num_extra = numCPU - len(moves)
# Ensure num_extra is positive
if num_extra < 0:
num_extra = 0
pool = p.MyPool(numCPU - num_extra)
func = partial(t.find_min, board, activePlayer, alpha, beta, depth)
# Make array of tuples for each node of tree at the next depth
# each tuple: (move, # of moves available)
next_nodes = []
for move in moves:
temp_board = mmUtil.make_move(board, move, activePlayer)
num_next = mmUtil.get_moves(temp_board)
next_nodes.append((move, len(num_next)))
# Sort by decreasing number of moves available
sorted(next_nodes, key=itemgetter(1), reverse=True)
# Create data array for number of processors allocated for each move
data = []
# Calculate the base number of extra processors that each move will use
base = num_extra / len(moves)
# For each tuple add the base number of extra processors to the default amount (1)
for node in next_nodes:
data.append((node[0], 1 + base))
# Allocate remaining processor with moves that have the highest branching factor
for i in range((num_extra % len(moves))):
node = data[i]
data[i] = (node[0], node[1] + 1)
v_mins = pool.map_async(func, data)
pool.close()
pool.join()
mov_val = max(v_mins.get(), key=itemgetter(1)) # (move, value)
print v_mins.get()
print mov_val
return mov_val[::-1] # (value, move)
def min_value(board, alpha, beta, depth):
if mmUtil.terminal_test(board, activePlayer, depth, startTime, maxTime, maxDepth):
return mmUtil.utility(board, activePlayer)
v = float("inf")
for a in mmUtil.get_moves(board):
v = min(v, max_value(mmUtil.make_move(board, a, opponentPlayer), alpha, beta, depth + 1))
if v <= alpha:
return v
beta = min(beta, v)
return v
def min_value(board, alpha, beta, depth):
if mmUtil.terminal_test(board, activePlayer, depth, startTime, maxTime,
maxDepth):
return mmUtil.utility(board, activePlayer)
v = float('inf')
for a in mmUtil.get_moves(board):
v = min(
v,
max_value(mmUtil.make_move(board, a, opponentPlayer), alpha, beta,
depth + 1))
if v <= alpha:
return v
beta = min(beta, v)
return v
def max_value_first(board, alpha, beta, depth):
v = float("-inf")
move = 3
for a in mmUtil.get_moves(board):
vMin = min_value(mmUtil.make_move(board, a, activePlayer), alpha, beta, depth + 1)
if v >= vMin:
move = move
else:
v = vMin
move = a
if v >= beta:
return (v, move)
alpha = max(alpha, v)
return (v, move)
def max_value_first(board, alpha, beta, depth):
v = float('-inf')
move = 3
for a in mmUtil.get_moves(board):
vMin = min_value(mmUtil.make_move(board, a, activePlayer), alpha, beta,
depth + 1)
if v >= vMin:
move = move
else:
v = vMin
move = a
if v >= beta:
return (v, move)
alpha = max(alpha, v)
return (v, move)
def find_min(board, activePlayer, alpha, beta, depth, mov_cpu): print "[" + str(depth) + "] Data for move " + str(mov_cpu[0]) + ": " + str(mov_cpu) a = mov_cpu[0] numCPU = mov_cpu[1] if numCPU == 1: return (a, ai.min_value(mmUtil.make_move(board,a,activePlayer), alpha, beta, depth+1)) else: moves = mmUtil.get_moves(board) data = [] for move in moves: data.append((move, 1)) pool = multiprocessing.Pool(processes = numCPU) func = partial(find_max, board, activePlayer, alpha, beta, depth+1) v_maxs = pool.map_async(func, data) pool.close() pool.join() mov_val = min(v_maxs.get(), key=itemgetter(1)) # (move, value) return mov_val
