def monte_play(game_state): ans=0.0 numRuns = 10 if(game_state.maxValue() >= 16): numRuns = 20 elif(game_state.maxValue() >= 32): numRuns = 50 elif(game_state.maxValue() >= 256): numRuns = 75 elif(game_state.maxValue() >= 512): numRuns = 120 for i in range(numRuns): # print(i) tmp=TwentyFortyEight() tmp.cells=game_state.cells tmp.score=game_state.score count = 0 while(tmp.canMove()): # count += 1 # print(count) dir=random.choice([1,2,3,4]) tmp.move(dir) tmp.new_tile() #print(game_state.cells) ans+=tmp.score1() ans /= numRuns return ans
def run_2048_new_tile_test(): """ Checks if new_tile() creates a new tile of either number 2 with 90% possibility or a 4 with 10% possibility. """ twos = 0 fours = 0 NUM_COUNTS = 1000 for count in range(NUM_COUNTS): o_game_board = TwentyFortyEight(2, 2) o_game_board.new_tile() twos += sum([row.count(2) for row in o_game_board.arr_grid]) fours += sum([row.count(4) for row in o_game_board.arr_grid]) print "twos: %d / %d - %4.1f percent" % (twos, NUM_COUNTS, (twos * 100.0) / NUM_COUNTS) print "fours: %d / %d - %4.1f percent" % (fours, NUM_COUNTS, (fours * 100.0) / NUM_COUNTS)
class Environment(object): def __init__(self): self.S = TwentyFortyEight() self.S.make_tables() self.score = 0 self.S.new_tile() print("New Episode") def reset(self): self.S.score = 0 self.S.cells = 0 self.score = 0 self.S.new_tile() return self.S.vectorize_state() def step(self, action): score_prev = self.S.score1() cells = self.S.cells self.S.move(action + 1) r = self.S.score1() - score_prev self.score = self.S.score1() if not cells == self.S.cells: self.S.new_tile() if not self.S.canMove(): return self.S.vectorize_state(), r, True return self.S.vectorize_state(), r, False def seed(self, a): return
# x.new_tile() # # print("GAME ENDs") # # print(x.maxValue()) # print("Score:"+str(x.score)+"\t Max Tile:"+str(x.maxValue())) # # x.__str__() # # print(x.get_available_moves()) occ = np.zeros(16) for i in range(500): x=TwentyFortyEight() x.make_tables() # x.print_tables() # print("Generated Tables") x.new_tile() # avail_moves = x.get_available_moves() while(True): # x.__str__() # print("-----------------------") temp = x.vectorize_state(); tempx = np.zeros((1, 256)) for j in range(16): tempx[0, j*16 + temp[j]] = 1 probs=y.eval(feed_dict={x1: tempx})[0] # print(probs) # dir = eminimax(x,2) # dir=minimax_alpha_beta(x,6) # dir=monte_carlo(x) avail_moves = x.get_available_moves()