当前位置: 首页>>代码示例>>Python>>正文


Python Cube.move方法代码示例

本文整理汇总了Python中cube.Cube.move方法的典型用法代码示例。如果您正苦于以下问题:Python Cube.move方法的具体用法?Python Cube.move怎么用?Python Cube.move使用的例子?那么恭喜您, 这里精选的方法代码示例或许可以为您提供帮助。您也可以进一步了解该方法所在cube.Cube的用法示例。


在下文中一共展示了Cube.move方法的2个代码示例,这些例子默认根据受欢迎程度排序。您可以为喜欢或者感觉有用的代码点赞,您的评价将有助于系统推荐出更棒的Python代码示例。

示例1: Environment

# 需要导入模块: from cube import Cube [as 别名]
# 或者: from cube.Cube import move [as 别名]
class Environment(object):

    # Initialize a random cube
    def __init__(self, N):
        self.N = N
        self.cube = Cube(N=N)

    def suffle(self, rand_nb=None, fixed_action=None):
        if rand_nb is not None:
            moves = self.cube.randomize(rand_nb)
            return moves
        self.perform_action(fixed_action)
        return fixed_action

    # Make a move and get a reward:
    # 0 is the cube is not finish
    # 1 is the cube is done
    def perform_action(self, action):
        [f, l, d] = action
        self.cube.move(f, l, d)
        return self.reward()

    def reward(self,):
        return self.cube.finish()

        # if self.cube.finish():
        #     return 1.
        # for i in range(6):
        #     if np.array_equal(self.cube.stickers[i, :, :], i * np.ones((self.N, self.N))):
        #         return  0.1
        # return 0.

    # Select a random_action
    def random_action(self,):
        f = np.random.randint(6)
        l = np.random.randint(self.N)
        d = 1 + np.random.randint(3)
        return [f, l, d]

    def get_state(self,):
        return self.cube.stickers
开发者ID:EloiZ,项目名称:DeepCube,代码行数:43,代码来源:environment.py

示例2: enumerate

# 需要导入模块: from cube import Cube [as 别名]
# 或者: from cube.Cube import move [as 别名]
        # Show good examples in the replay memory with probability
        # "good_examples"
        r = np.random.uniform(0., 1., 1)
        if r < args.good_examples:
            for i, move in enumerate(reversed(moves)):
                r_move = reverse_action(move)

                # Make the move
                if i == 0:
                    x_t = np.copy(env.get_state())
                    y_t = np.copy(env.get_state())
                else:
                    x_t = np.copy(x_tp1)
                    y_t = np.copy(x_tp1)
                cube = Cube(stickers=y_t)
                cube.move(r_move[0], r_move[1], r_move[2])
                x_tp1 = np.copy(cube.stickers)

                # Compute reward
                reward = args.gamma ** (len(moves) - i - 1)

                # Shift action
                r_move[1] += 6
                r_move[2] += 6 + N -1

                fill_replay_memory(replay_memory,
                                   args.max_replay_memory,
                                   [x_t, r_move, reward, x_tp1])

        finish_episode = False
        t = 0
开发者ID:EloiZ,项目名称:DeepCube,代码行数:33,代码来源:qlearning.py


注:本文中的cube.Cube.move方法示例由纯净天空整理自Github/MSDocs等开源代码及文档管理平台,相关代码片段筛选自各路编程大神贡献的开源项目,源码版权归原作者所有,传播和使用请参考对应项目的License;未经允许,请勿转载。