def playBestMove():
global gamestate
global depth_to_consider
global move
fuseki_match, fuseki_move = op.make_move(gamestate)
if fuseki_match:
print('Found a cool fuseki move!')
move = fuseki_move[:]
gamestate = do_move(gamestate, fuseki_move, True)
return
print('Thinking hard about this one...')
p_predicts = sess1.run(res,
feed_dict={
pre_x: np.copy(gamestate),
keep_prob: 1.0
})[0]
printDistribution(p_predicts)
p_indices = [[
i, j
] for i, j in itertools.product(*[range(19), range(19)])
if not np.any(gamestate[i][j])]
p_indices.sort(key=lambda x: p_predicts[x[0]][x[1]], reverse=True)
i = 0
while bc.suicide(np.copy(gamestate), p_indices[i]):
i += 1
bestMove = p_indices[i]
move = bestMove[:]
gamestate = do_move(gamestate, bestMove, True)
def playMove():
global gamestate
global depth_to_consider
fuseki_match, fuseki_move = op.make_move(gamestate)
if fuseki_match:
print('Found a cool fuseki move!')
gamestate = do_move(gamestate, fuseki_move, True)
return
print('Thinking hard about this one...')
gametree = growTree(GameTree(name=gamestate), num_moves_considered, depth_to_consider)
print('I\'ve made a game tree!')
direction = tMiniMax(gametree)
gamestate = gametree.children[direction].name
def playMove():
global gamestate
global depth_to_consider
fuseki_match, fuseki_move = op.make_move(gamestate)
if fuseki_match:
print('Found a cool fuseki move!')
gamestate = do_move(gamestate, fuseki_move, True)
return
print('Thinking hard about this one...')
gametree = growTree(GameTree(name=gamestate), num_moves_considered, depth_to_consider)
print('I\'ve made a game tree!')
direction = tMiniMax(gametree)
gamestate = gametree.children[direction].name
def playBestMove():
global gamestate
global depth_to_consider
fuseki_match, fuseki_move = op.make_move(gamestate)
if fuseki_match:
print('Found a cool fuseki move!')
gamestate = do_move(gamestate, fuseki_move, True)
return
print('Thinking hard about this one...')
p_predicts = sess1.run(res, feed_dict={pre_x: np.copy(gamestate), keep_prob: 1.0})[0]
p_indices = [[i, j] for i, j in itertools.product(*[range(19), range(19)]) if not np.any(gamestate[i][j])]
p_indices.sort(key=lambda x: p_predicts[x[0]][x[1]], reverse=True)
bestMove = p_indices[0]
gamestate = do_move(gamestate, bestMove, True)
def playBestMove():
global gamestate
fuseki_match, fuseki_move = op.make_move(gamestate)
if fuseki_match:
gamestate = do_move(gamestate, fuseki_move)
# print('Found a cool fuseki move!')
return True
p_predicts = sess.run(res, feed_dict={pre_x: np.copy(gamestate), keep_prob: 1.0})[0]
p_indices = [[i, j] for i, j in itertools.product(*[range(19), range(19)]) if not np.any(gamestate[i][j])]
p_indices.sort(key=lambda x: p_predicts[x[0]][x[1]], reverse=True)
i = 0
while i < len(p_indices) and bc.suicide(np.copy(gamestate), p_indices[i]):
i += 1
if i == len(p_indices):
gamestate = flip(gamestate)
return False
bestMove = p_indices[i]
gamestate = do_move(gamestate, bestMove)
return True
def showBoard():
result = " a b c d e f g h i j k l m n o p q r s \n"
for i in range(19):
result += str(i).zfill(2) + ' '
for j in range(19):
if gamestate[i][j][0] == 1:
result += '0 ' enumerate(
else:
result += '+ '
result += '\n'
return result
def do_move(gs, pos, isBlackMove):
if not isBlackMove:
gs = flip(gs)
gs = bc.boardchange(np.copy(gs), pos)
if isBlackMove:
gs = flip(gs)
return gs
def updateTree(root, width, depth):
if depth == 0:
root.name[1] = sess2.run(res_v, feed_dict={pre_x_v: root.name[0], keep_prob_v: 1.0})[0][0]
root.children = None
return root
if root.children is not None and root.children != []:
root.children = [updateTree(c,width,depth-1) for c in root.children]
return root
p_predicts = sess1.run(res, feed_dict={pre_x: root.name[0], keep_prob: 1.0})[0]
p_indices = [[i, j] for i, j in itertools.product(*[range(19), range(19)]) if not np.any(root.name[0][i][j])]
p_indices.sort(key=lambda x: p_predicts[x[0]][x[1]], reverse=True)
root.children = [updateTree(GameTree(name=do_move(np.copy(root.name[0]), x, True)), width, depth-1) for x in p_indices[:width]]
return root
def moveTree(direction):
global gametree
gametree = gametree.children[direction]
def matchTreeToState():
global gamestate
global gametree
if gametree.children is None or gametree.children == []:
gametree = GameTree(name=[gamestate, None])
return
for c in gametree.children:
if gamestate == c.name[0]:
gametree = c
return
gametree = GameTree(name=[gamestate, None])
def playMove():
global gamestate
global depth_to_consider
global num_moves_considered
global gametree
fuseki_match, fuseki_move = op.make_move(gamestate)
if fuseki_match:
print('Found a cool fuseki move!')
gamestate = do_move(gamestate, fuseki_move, True)
matchTreeToState()
return
print('Thinking hard about this one...')
gametree = updateTree(gametree, num_moves_considered, depth_to_consider)
print('I\'ve made a game tree!')
moveTree(tMiniMax(gametree))
gamestate = np.copy(gametree.name[0])
def tMin(tree):
if tree.children is None or tree.children == []]:
return tree.name[1]
import numpy as np import opening as op pos = np.zeros((19, 19, 3), dtype=np.int) pos[3][3][1] = 1 print(op.make_move(pos))