def get_possible_nxt_prms(self, params=None):
if params is None:
params = self.get_all_params()
count = 0
nxt_prms = {}
for wk_r, wk_c, wr_r, wr_c, bk_r, bk_c,white_plays in params:
board = ChessBoard(wk=King(wk_r, wk_c, Piece.WHITE),
wr=Rook(wr_r, wr_c, Piece.WHITE),
bk=King(bk_r, bk_c, Piece.BLACK),
white_plays=white_plays)
if not board.valid:
continue
nxt_pos = {}
for nxt_moves in board.get_possible_moves():
q = 0
if nxt_moves.state == ChessBoard.BLACK_KING_CHECKMATE:
q = 100
elif nxt_moves.state == ChessBoard.DRAW:
q = -100
nxt_pos[nxt_moves.board_id()] = q
nxt_prms[(wk_r,wk_c,wr_r,wr_c,bk_r,bk_c,white_plays)] = nxt_pos
count += 1
if count % 1000 == 0:
print (count)
return nxt_prms
def learn(self, iterations):
for i in range(0, iterations):
board = ChessBoard.get_random_chessboard()
self.parameter_set.calculate_parameters(board)
rank = self.parameter_set.learning_move()
if self.parameter_set.rank_higher(self.knowledge, rank):
self.knowledge = self.parameter_set.update_knowledge(self.knowledge)
if __name__ == '__main__':
draw = 0
win = 0
total = 1000.0
lit = []
for t in range(0,10):
draw = 0
win = 0
for i in range(0,int(total)):
board = ChessBoard.get_random_chessboard()
if i % 1000 == 0:
print i
if play_random(board) is ChessBoard.DRAW:
draw += 1
else:
win += 1
lit.append((win,draw))
print lit
print np.mean(np.array(lit))
print np.std(np.array(lit))
