dqn_fruit_merger.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "\n",
    "# **强化学习——DQN玩合成大西瓜**\n",
    "**作者：** [莱可可](https://github.com/RedContritio)<br>\n",
    "**日期：** 2022.11 <br>\n",
    "**摘要：** 使用 DQN 算法训练 AI 玩合成大西瓜"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## **一、介绍**\n",
    "本案例展示了使用DQN玩合成大西瓜的一个实现。\n",
    "\n",
    "### Deep Q-Network (DQN 算法)\n",
    "\n",
    "DQN 是基于深度学习的 Q 学习算法，主要结合了值函数近似和神经网络技术，并采用目标网络和经验回放的方法进行训练。\n",
    "- 目标网络：使用一个固定的 Q 神经网络来计算目标值，这样可以减少训练时的波动。\n",
    "- 经验回放：将经验存储在一个缓冲区中，然后从缓冲区中随机采样，这样可以减少相关性，使得训练更加稳定。\n",
    "\n",
    "\n",
    "### 合成大西瓜\n",
    "合成大西瓜玩法与《2048》《俄罗斯方块》类似，两个相同类别的水果碰撞后，合成成为更大的新水果。其中共有11种水果，最终目标是合成大西瓜。\n",
    "agent需要学会将水果从合适的位置放下，使水果能顺利合成。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "本教程的数据流方向如下所示（不含训练部分）：\n",
    "\n",
    "![数据流示意图](./resources/illustrations/dataflow.jpg)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## **二、环境配置**\n",
    "\n",
    "本教程依赖的非 python 官方包如下：\n",
    "\n",
    "- pymunk: python 2D物理引擎，用于处理物体碰撞等运动\n",
    "- opencv-python: 图形库，用于绘制界面与交互\n",
    "- numpy: 数值计算库，主要用于图像处理和模型数据处理"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": []
    }
   ],
   "source": [
    "%pip install paddlepaddle pymunk opencv-python numpy matplotlib"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "import random\n",
    "import typing\n",
    "import enum\n",
    "import collections\n",
    "from threading import Lock\n",
    "import base64\n",
    "import math\n",
    "import datetime\n",
    "\n",
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy as np\n",
    "import pymunk\n",
    "import cv2\n",
    "import paddle\n",
    "from paddle import nn\n",
    "from paddle import optimizer"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## **三、实现游戏本体**\n",
    "\n",
    "### 3.1 事件系统\n",
    "\n",
    "事件系统是对游戏与用户交互的抽象，包括鼠标点击与鼠标移动。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "class EventType(enum.Enum):\n",
    "    # 左键单击\n",
    "    LBUTTONDOWN = (1,)\n",
    "    # 鼠标移动\n",
    "    MOUSEMOVE = (2,)\n",
    "    # 右键单击\n",
    "    RBUTTONDOWN = (3,)\n",
    "\n",
    "\n",
    "class Event:\n",
    "    def __init__(self, type: EventType):\n",
    "        self.type = type\n",
    "\n",
    "\n",
    "# 鼠标事件\n",
    "class MouseEvent(Event):\n",
    "    def __init__(self, type: EventType, pos: typing.Tuple[int, int]):\n",
    "        self.pos = pos\n",
    "        super().__init__(type)\n",
    "\n",
    "\n",
    "# 事件系统基类\n",
    "class GameEventBase:\n",
    "    def __init__(self):\n",
    "        self.__events = []\n",
    "\n",
    "    # 添加一个事件\n",
    "    def add_event(self, event: Event):\n",
    "        self.__events.append(event)\n",
    "\n",
    "    # 获取（上一次获取之后）新发生的事件\n",
    "    # 该操作会清空现有事件列表\n",
    "    @property\n",
    "    def events(self) -> typing.List[Event]:\n",
    "        _events = self.__events[:]\n",
    "        self.__events.clear()\n",
    "        return _events"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.2 图像渲染\n",
    "\n",
    "游戏的图形界面绘制需要一些辅助函数，如下：\n",
    "\n",
    "- `mix`: 将 `background` 和 `foreground` 基于透明度混合（并给 `foreground` 额外乘算 `alpha` 的透明度），直接在 `background` 上修改\n",
    "- `cover`: 将 `foreground` 基于透明度覆盖到 `background` 上（并给 `foreground` 额外乘算 `alpha` 的透明度），直接在 `background` 上修改\n",
    "- `intersectRect`: 计算两个矩形框 `(x, y, w, h)` 的相交矩形框 `(ix, iy, iw, ih)`\n",
    "- `putText2`: 在 `image` 上以 `center` 为中心，绘制文字 `text` （用法同 `cv2.putText` ）\n",
    "- `putInverseColorText`: 在 `image` 上基于 `pos` 点，调用 `putTextFunc` 绘制文字 `text`，文字与 `image` 反色"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 将 background 和 foreground 基于透明度混合（并给 foreground 额外乘算 alpha 的透明度），直接在 background 上修改\n",
    "def mix(background: np.ndarray, foreground: np.ndarray, alpha: float = 1.0) -> None:\n",
    "    alpha_back = background[:, :, 3] / 255.0\n",
    "    alpha_fore = (foreground[:, :, 3] / 255.0) * alpha\n",
    "\n",
    "    for c in range(3):\n",
    "        background[:, :, c] = np.ubyte(\n",
    "            alpha_fore * foreground[:, :, c]\n",
    "            + alpha_back * background[:, :, c] * (1 - alpha_fore)\n",
    "        )\n",
    "    background[:, :, 3] = np.ubyte((1 - (1 - alpha_fore) * (1 - alpha_back)) * 255)\n",
    "\n",
    "\n",
    "# 将 foreground 基于透明度覆盖到 background 上（并给 foreground 额外乘算 alpha 的透明度），直接在 background 上修改\n",
    "def cover(background: np.ndarray, foreground: np.ndarray, alpha: float = 1.0) -> None:\n",
    "    foreground_colors = foreground[:, :, :3]\n",
    "    alpha_channel = (foreground[:, :, 3] / 255.0) * alpha\n",
    "    alpha_mask = np.dstack((alpha_channel, alpha_channel, alpha_channel))\n",
    "\n",
    "    background[:, :, :3] = (\n",
    "        background[:, :, :3] * (1 - alpha_mask) + foreground_colors * alpha_mask\n",
    "    )\n",
    "\n",
    "\n",
    "# 计算两个矩形框 (x, y, w, h) 的相交矩形框 (ix, iy, iw, ih)\n",
    "def intersectRect(rect1: typing.List[int], rect2: typing.List[int]) -> typing.List[int]:\n",
    "    l1, t1, b1, h1 = rect1\n",
    "    r1, b1 = l1 + b1, t1 + h1\n",
    "    l2, t2, b2, h2 = rect2\n",
    "    r2, b2 = l2 + b2, t2 + h2\n",
    "\n",
    "    l, r = max(l1, l2), min(r1, r2)\n",
    "    t, b = max(t1, t2), min(b1, b2)\n",
    "\n",
    "    return [l, t, max(0, r - l), max(0, b - t)]\n",
    "\n",
    "\n",
    "# 在 image 上以 center 为中心，绘制文字 text （用法同 cv2.putText ）\n",
    "def putText2(\n",
    "    image: np.ndarray,\n",
    "    text: str,\n",
    "    center: typing.List[int],\n",
    "    font_face: int = 0,\n",
    "    font_scale: float = 1.0,\n",
    "    color: typing.List[int] = (255, 255, 255),\n",
    "    thickness: int = 1,\n",
    ") -> None:\n",
    "    INNER_LINE_MARGIN = 5\n",
    "    x, y = center\n",
    "    lines = text.splitlines()\n",
    "\n",
    "    sizes = [\n",
    "        cv2.getTextSize(line, font_face, font_scale, thickness)[0] for line in lines\n",
    "    ]\n",
    "\n",
    "    h_sum = sum([size[1] for size in sizes]) + (len(sizes) - 1) * INNER_LINE_MARGIN\n",
    "    w_max = max([size[0] for size in sizes])\n",
    "\n",
    "    y_base = y - h_sum // 2\n",
    "\n",
    "    for i, (w, h) in enumerate(sizes):\n",
    "        cv2.putText(\n",
    "            image,\n",
    "            lines[i],\n",
    "            (x - w // 2, y_base),\n",
    "            font_face,\n",
    "            font_scale,\n",
    "            color,\n",
    "            thickness,\n",
    "        )\n",
    "        y_base += h + INNER_LINE_MARGIN\n",
    "\n",
    "\n",
    "# 在 image 上基于 pos 点，调用 putTextFunc 绘制文字 text，文字与 image 反色\n",
    "def putInverseColorText(\n",
    "    image: np.ndarray,\n",
    "    text: str,\n",
    "    pos: typing.List[int],\n",
    "    font_face: int = 0,\n",
    "    font_scale: float = 1.0,\n",
    "    thickness: int = 1,\n",
    "    putTextFunc: typing.Callable = putText2,\n",
    ") -> None:\n",
    "    mask = np.zeros((*image.shape[:2], 3), dtype=np.uint8)\n",
    "    putTextFunc(mask, text, pos, font_face, font_scale, (255, 255, 255), thickness)\n",
    "\n",
    "    image[:, :, :3] = mask + (2 * (0.5 - mask / 255.0)) * image[:, :, :3]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3 伪随机数生成器\n",
    "\n",
    "自定义伪随机数生成器，使用二次同余生成器（quadratic congruential generator），便于控制随机过程。\n",
    "\n",
    "二次同余生成器的算法原理如下： $y_n = (a * y_{n-1}^2 + b *y_{n-1} + c) \\mod p     ,n > 0$\n",
    "\n",
    "在本教程实现中，取 $a = 1, b = c = 0, p = p_1 * p_2$，则算法退化为 $y_n = y_{n-1}^2 \\mod (p_1 * p_2)     ,n > 0$\n",
    "\n",
    "令 $p_1$、$p_2$ 为两个大质数，则生成的随机数周期会更大。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "class PRNG:\n",
    "    def __init__(self):\n",
    "        self.s = 1234  # y_0，取为 1234\n",
    "        self.p = 999979  # p_1，第一个大质数\n",
    "        self.q = 999983  # p_2，第二个大质数\n",
    "        self.m = self.p * self.q  # p，模数\n",
    "\n",
    "    # 基于 x 的哈希结果，获得一个浮点数\n",
    "    def hash(self, x: any) -> float:\n",
    "        y = base64.encodebytes(bytes(str(x).encode(\"utf8\")))\n",
    "        z = 0\n",
    "        for i, v in enumerate(y):\n",
    "            z += v * math.pow(128, i)\n",
    "        return z\n",
    "\n",
    "    # 传入可哈希变量 seed，以此初始化 s (即 y_0)。\n",
    "    def seed(self, seed: any = datetime.datetime.now()):\n",
    "        y = 0\n",
    "        z = 0\n",
    "        while y % self.p == 0 or y % self.q == 0 or y == 0 or y == 1:\n",
    "            y = (self.hash(seed) + z) % self.m\n",
    "            z += 1\n",
    "\n",
    "        self.s = y  # 初始化 y_0\n",
    "\n",
    "        # 跳过前 10 个随机数，用来混淆\n",
    "        [self.next() for _ in range(10)]\n",
    "\n",
    "    # 获取 [0, 1) 的随机浮点数\n",
    "    def next(self) -> float:\n",
    "        self.s = (self.s * self.s) % self.m\n",
    "        return self.s / self.m\n",
    "\n",
    "    # 获取 [l = 0, r = 1) 的随机浮点数\n",
    "    def random(self, l: float = 0, r: float = 1) -> float:\n",
    "        return self.next() * (r - l) + l\n",
    "\n",
    "    # 获取 [l = 0, r = 2) 的随机整数\n",
    "    def randint(self, l: int = 0, r: int = 2) -> int:\n",
    "        return int(math.ceil(self.random(l, r)))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.4 实现水果类\n",
    "\n",
    "#### 3.4.1 水果参数\n",
    "\n",
    "定义各种水果的半径、尺寸与对应图片。\n",
    "\n",
    "（列表中首个元素无意义，用于占位 `type` `0`）"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 水果半径 r\n",
    "FRUIT_RADIUS = [int(1.5 * r) for r in [-1, 10, 15, 21, 23, 29, 35, 37, 50, 59, 60, 78]]\n",
    "# 水果尺寸 (w, h)\n",
    "FRUIT_SIZES = [(2 * r, 2 * r) for r in FRUIT_RADIUS]\n",
    "\n",
    "# 水果图像路径\n",
    "FRUIT_IMAGE_PATHS = [f\"resources/images/{i:02d}.png\" for i in range(11)]\n",
    "\n",
    "# 水果图像原图 (np.ndarray)\n",
    "FRUIT_RAW_IMAGES = [\n",
    "    cv2.imread(FRUIT_IMAGE_PATHS[i], -1) if i > 0 else None for i in range(11)\n",
    "]\n",
    "\n",
    "# 水果基于尺寸缩放后的图像 (np.ndarray)\n",
    "FRUIT_IMAGES = [\n",
    "    None if img is None else cv2.resize(img, FRUIT_SIZES[i])\n",
    "    for i, img in enumerate(FRUIT_RAW_IMAGES)\n",
    "]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 3.4.2 水果类定义\n",
    "\n",
    "定义所有水果的类 `class Fruit`。\n",
    "\n",
    "每个水果对应一个该类型实例，主要用于描述水果对象的位置。\n",
    "\n",
    "其中，水果的种类 `type` 为 $[1, 11]$，其中 $1$ 是葡萄，$11$ 是大西瓜。\n",
    "\n",
    "\n",
    "合成得分：\n",
    "- 当两个种类为 $x$ 的水果碰撞时，合成出一个种类为 $x+1$ 的新水果，并在游戏中获得 $x+1$ 分；\n",
    "- 特别的，当 $x=10$ 时，合成出的新水果为 $11$ （大西瓜），并在游戏中获得 $100$ 分。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Fruit:\n",
    "    # 初始化时，传入水果种类 type 与水果位置 (x, y)\n",
    "    def __init__(self, type: int, x: int, y: int) -> None:\n",
    "        self.type = type\n",
    "        self.r = FRUIT_RADIUS[self.type]\n",
    "        self.size = FRUIT_SIZES[self.type]\n",
    "\n",
    "        self.x, self.y = x, y\n",
    "\n",
    "    # 更新水果位置\n",
    "    def update_position(self, x: int, y: int) -> None:\n",
    "        self.x, self.y = x, y\n",
    "\n",
    "    # 在 screen 上绘制水果图像\n",
    "    def draw(self, screen: np.ndarray) -> None:\n",
    "        Fruit.paint(screen, self.type, self.x, self.y)\n",
    "\n",
    "    # Fruit.paint，绘制水果图像\n",
    "    def paint(\n",
    "        screen: np.ndarray, type: int, x: int, y: int, alpha: float = 1.0\n",
    "    ) -> None:\n",
    "        assert type > 0 and type <= 11\n",
    "        l, t = (x - FRUIT_RADIUS[type], y - FRUIT_RADIUS[type])\n",
    "        w, h = FRUIT_SIZES[type]\n",
    "\n",
    "        l, t, w, h = [int(v) for v in (l, t, w, h)]\n",
    "\n",
    "        # 计算实际绘图区域\n",
    "        il, it, iw, ih = [\n",
    "            int(v) for v in intersectRect((l, t, w, h), (0, 0, *screen.shape[1::-1]))\n",
    "        ]\n",
    "\n",
    "        cover(\n",
    "            screen[it : it + ih, il : il + iw],\n",
    "            FRUIT_IMAGES[type][it - t : it - t + ih, il - l : il - l + iw],\n",
    "            alpha=alpha,\n",
    "        )"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.5 游戏本体"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 物理引擎重力\n",
    "GRAVITY = (0, 800)\n",
    "# 游戏场景大小\n",
    "GAME_RESOLUTION = GAME_WIDTH, GAME_HEIGHT = 300, 400"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "设置背景色为 `rgb(0x41, 0x69, 0xE1)` <font color='#4169E1'>████████</font>"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "class GameCore(GameEventBase):\n",
    "    def __init__(self, gravity: typing.Tuple[int, int] = GRAVITY) -> None:\n",
    "        # 初始化游戏场景分辨率 resolution 与游戏场景宽高 width, height\n",
    "        # 其中 resolution = (width, height)\n",
    "        self.resolution = self.width, self.height = GAME_WIDTH, GAME_HEIGHT\n",
    "\n",
    "        # 水果落下时的初始 x、y 坐标\n",
    "        self.init_x = int(self.width / 2)\n",
    "        self.init_y = int(0.15 * self.height)\n",
    "\n",
    "        # 本局游戏已获得的分数\n",
    "        self.score = 0\n",
    "        # 最近一次合成的分数增量\n",
    "        self.recent_score_delta = 0\n",
    "\n",
    "        # 水果列表，每个元素为 Fruit 类型\n",
    "        self.fruits: typing.List[Fruit] = []\n",
    "        # 碰撞球列表，每个元素为 Circle 类型\n",
    "        self.balls: typing.List[pymunk.Shape] = []\n",
    "\n",
    "        # 游戏背景色 rgb(0xE1, 0x69, 0x41)\n",
    "        self.background_color = (0xE1, 0x69, 0x41, 0)\n",
    "\n",
    "        # 预渲染纯色背景图\n",
    "        self.preset_background = np.zeros((self.height, self.width, 4), dtype=np.uint8)\n",
    "        self.preset_background[:, :] = self.background_color\n",
    "\n",
    "        # 预渲染带红线的纯色背景图（红色警戒线，游戏失败的检查线）\n",
    "        self.preset_redline_screen = self.preset_background.copy()\n",
    "        cv2.line(\n",
    "            self.preset_redline_screen,\n",
    "            (0, self.init_y),\n",
    "            (self.width, self.init_y),\n",
    "            (0, 0, 255),\n",
    "            2,\n",
    "        )\n",
    "\n",
    "        # 基于纯色背景图初始化游戏画面\n",
    "        self.__screen = self.preset_background.copy()\n",
    "\n",
    "        # 逻辑互斥锁\n",
    "        self.lock = Lock()\n",
    "        # 渲染互斥锁\n",
    "        self.render_lock = Lock()\n",
    "\n",
    "        # 伪随机数生成器\n",
    "        self.prng = PRNG()\n",
    "\n",
    "        # 稳定帧数阈值，当持续 10 帧稳定时，视为稳定状态\n",
    "        # 用来避免不稳定的平衡态\n",
    "        self.stable_frame_threshold = 10\n",
    "        # 当前帧 id\n",
    "        self.current_frame_id = 0\n",
    "        # 上一稳定状态的帧 id\n",
    "        self.stable_frame_id = self.current_frame_id - self.stable_frame_threshold\n",
    "        # 游戏是否可操作（可以点击）\n",
    "        self.clickable = False\n",
    "\n",
    "        # 目前已经合成过的最大水果 type （用于生成新水果）\n",
    "        self.largest_fruit_type = 1\n",
    "        # 当前要放下的水果 type\n",
    "        self.current_fruit_type = self.create_random_fruit_type()\n",
    "\n",
    "        # 初始化物理空间与重力\n",
    "        self.space = pymunk.Space()\n",
    "        self.space.gravity = gravity\n",
    "\n",
    "        # 重置游戏状态\n",
    "        self.reset()\n",
    "\n",
    "        # 初始化物理边界\n",
    "        self.init_segment()\n",
    "        # 设置碰撞处理函数\n",
    "        self.setup_collision_handler()\n",
    "\n",
    "        super().__init__()\n",
    "\n",
    "    # 重启游戏，可指定随机数种子（影响生成水果的种类）\n",
    "    def reset(self, seed: int = None) -> None:\n",
    "        for ball in self.balls:\n",
    "            self.space.remove(ball, ball.body)\n",
    "\n",
    "        self.score = 0\n",
    "\n",
    "        self.fruits.clear()\n",
    "        self.balls.clear()\n",
    "\n",
    "        self.current_frame_id = 0\n",
    "        self.stable_frame_id = self.current_frame_id - self.stable_frame_threshold\n",
    "        self.prev_stable_frame_id = self.stable_frame_id\n",
    "        self.clickable = False\n",
    "\n",
    "        self.largest_fruit_type = 1\n",
    "        self.current_fruit_type = self.create_random_fruit_type()\n",
    "\n",
    "        # 如果传入了 seed，则使用 seed 初始化，否则随机初始化\n",
    "        self.prng.seed(**({} if seed is None else {\"seed\": seed}))\n",
    "\n",
    "        self.alive = True\n",
    "\n",
    "    # 初始化游戏场景边界（物理引擎）\n",
    "    def init_segment(self, thinkness: float = 20, friction: float = 0.6):\n",
    "        # 四边界分别向外 thinkness 个像素\n",
    "        l, t = 0 - thinkness, 0 - thinkness - self.height // 2\n",
    "        r, b = self.width + thinkness, self.height + thinkness\n",
    "\n",
    "        def create_segment(\n",
    "            p1: typing.Tuple[int, int], p2: typing.Tuple[int, int]\n",
    "        ) -> pymunk.Segment:\n",
    "            s = pymunk.Segment(self.space.static_body, p1, p2, thinkness)\n",
    "            s.friction = friction\n",
    "            return s\n",
    "\n",
    "        self.space.add(create_segment((l, t), (l, b)))  # 添加左边界\n",
    "        self.space.add(create_segment((r, t), (r, b)))  # 添加右边界\n",
    "        # 不添加顶部边界\n",
    "        # self.space.add(create_segment((l, t), (r, t)))\n",
    "        self.space.add(create_segment((l, b), (r, b)))  # 添加下边界\n",
    "\n",
    "    # 设置碰撞处理器（物理引擎）\n",
    "    def setup_collision_handler(self):\n",
    "        def collision_post_solve(arbiter: pymunk.Arbiter, space: pymunk.Space, _data):\n",
    "            # 获取逻辑互斥锁\n",
    "            with self.lock:\n",
    "                # 获取碰撞物体 s_0, s_1\n",
    "                s0, s1 = arbiter.shapes[:2]\n",
    "                # 要生成的物体类型为 type + 1 （碰撞对象类型为 type）\n",
    "                new_type = s0.collision_type + 1\n",
    "                x1, y1 = s0.body.position\n",
    "                x2, y2 = s1.body.position\n",
    "                # 新物体的坐标为碰撞两物体中，坐标更靠下的物体\n",
    "                x, y = (x1, y1) if y1 > y2 else (x2, y2)\n",
    "\n",
    "                # 删除这两个碰撞球\n",
    "                if s0 in self.balls and s1 in self.balls:\n",
    "                    self.remove_ball(space, s0)\n",
    "                    self.remove_ball(space, s1)\n",
    "\n",
    "                    # 创建新水果\n",
    "                    fruit = Fruit(new_type, x, self.init_y)\n",
    "                    self.fruits.append(fruit)\n",
    "\n",
    "                    # 创建新水果的碰撞球\n",
    "                    ball = self.create_ball(\n",
    "                        self.space, x, y, fruit.r // 10, fruit.r - 1, new_type\n",
    "                    )\n",
    "                    self.balls.append(ball)\n",
    "\n",
    "                    # 更新最大水果类型\n",
    "                    self.largest_fruit_type = max(self.largest_fruit_type, new_type)\n",
    "                    # 记录本次分数增量\n",
    "                    self.recent_score_delta = new_type if new_type < 11 else 100\n",
    "                    # 更新分数\n",
    "                    self.score += self.recent_score_delta\n",
    "\n",
    "        for collision_type in range(1, 11):\n",
    "            self.space.add_collision_handler(\n",
    "                collision_type, collision_type\n",
    "            ).post_solve = collision_post_solve\n",
    "\n",
    "    # 创建随机水果种类\n",
    "    def create_random_fruit_type(self) -> int:\n",
    "        return self.prng.randint(1, min(self.largest_fruit_type, 5))\n",
    "\n",
    "    # 创建水果（逻辑）\n",
    "    def create_fruit(self, type: int, x: int) -> Fruit:\n",
    "        return Fruit(type, x, self.init_y - FRUIT_RADIUS[type])\n",
    "\n",
    "    # 创建水果碰撞球（物理引擎）\n",
    "    def create_ball(\n",
    "        self,\n",
    "        space: pymunk.Space,\n",
    "        x: int,\n",
    "        y: int,\n",
    "        mass: int = 1,\n",
    "        radius: int = 7,\n",
    "        type: int = 1,\n",
    "    ) -> pymunk.Shape:\n",
    "        ball_moment = pymunk.moment_for_circle(mass, 0, radius)\n",
    "        ball_body = pymunk.Body(mass, ball_moment)\n",
    "        ball_body.position = x, y\n",
    "        ball_shape = pymunk.Circle(ball_body, radius)\n",
    "        ball_shape.elasticity = 0.3\n",
    "        ball_shape.friction = 0.6\n",
    "        ball_shape.collision_type = type\n",
    "        space.add(ball_body, ball_shape)\n",
    "        return ball_shape\n",
    "\n",
    "    # 移除水果碰撞球（物理引擎）\n",
    "    def remove_ball(self, space: pymunk.Space, ball: pymunk.Circle):\n",
    "        p = self.balls.index(ball)\n",
    "\n",
    "        space.remove(ball, ball.body)\n",
    "\n",
    "        self.balls.pop(p)\n",
    "        self.fruits.pop(p)\n",
    "\n",
    "    # 保存游戏截图\n",
    "    def save_screen(self, path: str = \"screenshot.png\") -> bool:\n",
    "        rgb_img = cv2.cvtColor(self.screen, cv2.COLOR_BGRA2BGR)\n",
    "        return cv2.imwrite(path, rgb_img)\n",
    "\n",
    "    # 绘制游戏界面\n",
    "    def draw(self):\n",
    "        # 双缓冲层，减少对渲染互斥锁的占用\n",
    "        backbuffer = self.preset_background.copy()\n",
    "\n",
    "        # 绘制半透明的待放置水果\n",
    "        if self.current_fruit_type > 0:\n",
    "            y = self.init_y - FRUIT_RADIUS[self.current_fruit_type]\n",
    "            Fruit.paint(\n",
    "                backbuffer,\n",
    "                self.current_fruit_type,\n",
    "                self.init_x,\n",
    "                y,\n",
    "                1 if self.clickable else 0.5,\n",
    "            )\n",
    "\n",
    "        # 绘制场地中的水果\n",
    "        for f in self.fruits:\n",
    "            f.draw(backbuffer)\n",
    "\n",
    "        # 与红线背景层合成\n",
    "        cv2.addWeighted(backbuffer, 1, self.preset_redline_screen, 0.5, 0, backbuffer)\n",
    "\n",
    "        # 绘制分数文本\n",
    "        putInverseColorText(\n",
    "            backbuffer,\n",
    "            f\"Score: {self.score}\",\n",
    "            (0, 20),\n",
    "            font_scale=0.7,\n",
    "            thickness=1,\n",
    "            putTextFunc=cv2.putText,\n",
    "        )\n",
    "\n",
    "        # 如果已经死亡，绘制死亡文本\n",
    "        if not self.alive:\n",
    "            putInverseColorText(\n",
    "                backbuffer,\n",
    "                f\"Failed\\nClick RButton to Restart\",\n",
    "                (int(self.width / 2), int(self.height / 2)),\n",
    "                font_scale=0.7,\n",
    "                thickness=2,\n",
    "            )\n",
    "\n",
    "        # 获得渲染锁，更新画面\n",
    "        with self.render_lock:\n",
    "            self.__screen[:, :, :] = backbuffer\n",
    "            return self.__screen\n",
    "\n",
    "    # 获取游戏界面\n",
    "    @property\n",
    "    def screen(self) -> np.ndarray:\n",
    "        with self.render_lock:\n",
    "            return self.__screen\n",
    "\n",
    "    # 获取特征\n",
    "    def get_features(self, width: int, height: int) -> np.ndarray:\n",
    "        \"\"\"\n",
    "        参数:\n",
    "            - width: 网格宽度\n",
    "            - height: 网格高度\n",
    "        return:\n",
    "            - features: (height, width, 2) np.ndarray\n",
    "                - features[:, :, 0]: 用于记录比当前水果小的水果\n",
    "                - features[:, :, 1]: 用于记录比当前水果大的水果\n",
    "        \"\"\"\n",
    "\n",
    "        # 计算网格单元宽高\n",
    "        uw, uh = self.width / width, self.height / height\n",
    "\n",
    "        features = np.zeros((height, width, 2), dtype=np.float32)\n",
    "\n",
    "        # 辅助矩阵，分别记录 type （网格水果类型）, dr（最小距离）\n",
    "        auxilary = np.zeros((height, width, 2), dtype=np.float32)\n",
    "        auxilary[:, :, 1] = np.inf\n",
    "\n",
    "        # 更新阈值，距离 dr 大于该阈值的视为不在网格内\n",
    "        threshold = ((uw**2) + (uh**2)) // 2\n",
    "\n",
    "        for f in self.fruits:\n",
    "            r2 = f.r * f.r\n",
    "            for j in range(width):\n",
    "                x = (0.5 + j) * uw\n",
    "                for i in range(height):\n",
    "                    y = (0.5 + i) * uh\n",
    "\n",
    "                    # 计算横纵坐标差\n",
    "                    dx, dy = f.x - x, f.y - y\n",
    "                    # dr = np.sqrt(dx * dx + dy * dy) - f.r\n",
    "                    # 使用平方和代替开方，提高运算速度\n",
    "                    dr = dx * dx + dy * dy - r2\n",
    "\n",
    "                    # 如果 dr 小于阈值且小于目前的最小 dr，更新网格内水果信息\n",
    "                    if dr < threshold and dr < auxilary[i, j, 1]:\n",
    "                        auxilary[i, j, 0] = f.type\n",
    "                        auxilary[i, j, 1] = dr\n",
    "\n",
    "        # 是否为空 (True 或 False)\n",
    "        is_empty = auxilary[:, :, 0] == 0\n",
    "        # 是否和当前水果类型相同 (True 或 False)\n",
    "        is_same = auxilary[:, :, 0] == self.current_fruit_type\n",
    "\n",
    "        # 网格内水果类型 (type_1) ，当前水果 (type_0)\n",
    "        # 如果 type_1 < type_0，则值为 type_1 - type_0\n",
    "        # 如果 type_1 == type_0，则值为 1\n",
    "        # 如果 type_1 > type_0，则值为 0\n",
    "        # 如果 type_1 == 0 （网格为空），则值为 0\n",
    "        features[:, :, 0] = auxilary[:, :, 0] - self.current_fruit_type\n",
    "        features[:, :, 0] = features[:, :, 0].clip(max=0)\n",
    "        features[:, :, 0][is_same] = 1\n",
    "        features[:, :, 0][is_empty] = 0\n",
    "\n",
    "        # 网格内水果类型 (type_1) ，当前水果 (type_0)\n",
    "        # 如果 type_1 > type_0，则值为 type_0 - type_1\n",
    "        # 如果 type_1 == type_0，则值为 1\n",
    "        # 如果 type_1 < type_0，则值为 0\n",
    "        # 如果 type_1 == 0 （网格为空），则值为 0\n",
    "        features[:, :, 1] = self.current_fruit_type - auxilary[:, :, 0]\n",
    "        features[:, :, 1] = features[:, :, 1].clip(max=0)\n",
    "        features[:, :, 1][is_same] = 1\n",
    "        features[:, :, 1][is_empty] = 0\n",
    "\n",
    "        return features\n",
    "\n",
    "    # 更新游戏至稳定状态（可以进行下一动作）\n",
    "    def update_until_stable(self, fps: float = 60, max_seconds: int = 5):\n",
    "        self.set_unstable()\n",
    "\n",
    "        max_steps = int(fps * max_seconds)\n",
    "        step = 0\n",
    "\n",
    "        while (\n",
    "            self.current_frame_id <= self.stable_frame_id + self.stable_frame_threshold\n",
    "            and step < max_steps\n",
    "        ):\n",
    "            self.update(1.0 / fps)\n",
    "            step += 1\n",
    "\n",
    "        # 当超过 max_steps 步仍未稳定时，强制设置为稳定（可操作）\n",
    "        if step == max_steps:\n",
    "            self.clickable = True\n",
    "\n",
    "    # 更新游戏（一帧）\n",
    "    def update(self, time_delta: float):\n",
    "        self.current_frame_id += 1\n",
    "        # 物理引擎单步模拟\n",
    "        self.space.step(time_delta)\n",
    "\n",
    "        # 检查稳定状态\n",
    "        stable = self.check_stable()\n",
    "        if not stable:\n",
    "            self.set_unstable()\n",
    "\n",
    "        # 检查是否失败\n",
    "        self.alive = self.alive and self.check_alive()\n",
    "        # 如果已经失败，则不再进行更新，并等待右键重启游戏\n",
    "        if not self.alive:\n",
    "            for event in self.events:\n",
    "                if event.type == EventType.RBUTTONDOWN:\n",
    "                    self.reset()\n",
    "                    break\n",
    "            return\n",
    "\n",
    "        # 如果已经稳定了足够多 (stable_frame_threshold == 10) 帧并不能进行操作，则设置为可操作\n",
    "        if (\n",
    "            not self.clickable\n",
    "            and self.current_frame_id\n",
    "            > self.stable_frame_id + self.stable_frame_threshold\n",
    "        ):\n",
    "            self.prev_stable_frame_id = self.stable_frame_id\n",
    "            self.clickable = True\n",
    "\n",
    "        # 处理事件\n",
    "        for event in self.events:\n",
    "            # 左键单击，放下当前水果\n",
    "            if event.type == EventType.LBUTTONDOWN and self.clickable:\n",
    "                # 放下位置为 (x, init_y - fruit.r)\n",
    "                # 即圆心横坐标为鼠标横坐标，圆下边界与红线相切\n",
    "                x, _y = event.pos\n",
    "\n",
    "                fruit = self.create_fruit(self.current_fruit_type, x)\n",
    "                self.fruits.append(fruit)\n",
    "\n",
    "                y = self.init_y - fruit.r\n",
    "                ball = self.create_ball(\n",
    "                    self.space,\n",
    "                    x,\n",
    "                    y,\n",
    "                    (fruit.r // 10) ** 2,\n",
    "                    fruit.r - 1,\n",
    "                    self.current_fruit_type,\n",
    "                )\n",
    "                self.balls.append(ball)\n",
    "\n",
    "                # 创建新水果\n",
    "                self.current_fruit_type = self.create_random_fruit_type()\n",
    "                self.set_unstable()\n",
    "                self.clickable = False\n",
    "\n",
    "            # 鼠标移动时，移动当前水果\n",
    "            elif event.type == EventType.MOUSEMOVE:\n",
    "                self.init_x, _y = event.pos\n",
    "                self.init_x = max(\n",
    "                    self.init_x, 0 + FRUIT_RADIUS[self.current_fruit_type]\n",
    "                )\n",
    "                self.init_x = min(\n",
    "                    self.init_x, self.width - FRUIT_RADIUS[self.current_fruit_type]\n",
    "                )\n",
    "\n",
    "        assert not self.lock.locked()\n",
    "\n",
    "        # 获取逻辑锁，更新所有水果\n",
    "        with self.lock:\n",
    "            for i, ball in enumerate(self.balls):\n",
    "                x, y = ball.body.position\n",
    "                angle = ball.body.angle\n",
    "\n",
    "                # xi, yi = int(x), int(y)\n",
    "\n",
    "                self.fruits[i].update_position(x, y)\n",
    "\n",
    "    # 重置稳定状态（设置为不稳定）\n",
    "    def set_unstable(self) -> None:\n",
    "        self.stable_frame_id = self.current_frame_id + 1\n",
    "\n",
    "    # 检查游戏场景是否稳定\n",
    "    def check_stable(self) -> bool:\n",
    "        return self.max_balls_velocity_y < 20\n",
    "\n",
    "    # 获取碰撞球竖直方向速度绝对值的最大值\n",
    "    @property\n",
    "    def max_balls_velocity_y(self) -> float:\n",
    "        return (\n",
    "            max([abs(ball.body.velocity.y) for ball in self.balls])\n",
    "            if len(self.balls) > 0\n",
    "            else 0\n",
    "        )\n",
    "\n",
    "    # 检查游戏是否仍在进行\n",
    "    def check_alive(self) -> bool:\n",
    "        if self.current_frame_id > self.stable_frame_id + self.stable_frame_threshold:\n",
    "            for f in self.fruits:\n",
    "                if f.y < self.init_y:\n",
    "                    return False\n",
    "        return True\n",
    "\n",
    "    # 左键单击事件，用于放下水果\n",
    "    def click(self, pos: typing.Tuple[int, int]):\n",
    "        self.add_event(MouseEvent(EventType.LBUTTONDOWN, pos))\n",
    "\n",
    "    # 鼠标移动，用于可视化界面中的水果跟随鼠标\n",
    "    def move(self, pos: typing.Tuple[int, int]):\n",
    "        self.add_event(MouseEvent(EventType.MOUSEMOVE, pos))\n",
    "\n",
    "    # 右键单击事件，用于重启游戏\n",
    "    def rclick(self, pos: typing.Tuple[int, int]):\n",
    "        self.add_event(MouseEvent(EventType.RBUTTONDOWN, pos))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.6 游戏接口\n",
    "\n",
    "封装接口，提供强化学习环境。\n",
    "\n",
    "提供以下接口：\n",
    "\n",
    "- `reset`: 重启游戏\n",
    "- `simulate_until_stable`: 运行游戏，直到游戏结束或者可以进行下一次操作\n",
    "- `next`: 输入 $action$，进行一次模拟，并返回 $(feature, reward, alive)$ 三元组\n",
    "\n",
    "其中，$reward$ 定义为，一次动作 $action$ 后，能带来的 $score$ 提升。\n",
    "\n",
    "特别的，当 $score$ 未改变时，该动作由于减少了空间，其 $reward$ 设置为负数 $-fruit.type$ （记本次动作放下的水果为 $fruit$） 。\n",
    "\n",
    "定义模拟时，每秒帧数为 $SIMULATE\\_FPS = 60$ 帧；可进行的动作种类为 $ACTION\\_NUM = 16$ 种（在16个均匀分布的水平坐标处放下水果）。\n",
    "\n",
    "例如，当 $action = 13$ 时，下一个水果将从从红色区域正中间放下。\n",
    "\n",
    "红色区域左右边界为 $(width / ACTION\\_NUM * 13, width / ACTION\\_NUM * (13 + 1))$。\n",
    "\n",
    "由于 $width = 300$，每个 action 的区域宽度为 $width / ACTION\\_NUM = 300 / 16 = 18.75$，左右边界为 $(243.75, 262.5)$。\n",
    "\n",
    "放下水果的位置为 $x = int(13.5 * 18.75) = int(253.125) = 253$。\n",
    "\n",
    "![放下水果的辅助线](./resources/illustrations/put_aux_line.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "class GameInterface:\n",
    "    # 可选的动作数量\n",
    "    ACTION_NUM = 16\n",
    "    # 每秒的帧率\n",
    "    SIMULATE_FPS = 60\n",
    "\n",
    "    # 特征图尺寸\n",
    "    FEATURE_MAP_WIDTH, FEATURE_MAP_HEIGHT = 16, 20\n",
    "\n",
    "    def __init__(self) -> None:\n",
    "        self.game = GameCore()\n",
    "        self.action_num = GameInterface.ACTION_NUM\n",
    "        # 每个红色区域的宽度\n",
    "        self.action_segment_len = self.game.width / GameInterface.ACTION_NUM\n",
    "\n",
    "    def reset(self, seed: int = None) -> None:\n",
    "        self.game.reset(seed)\n",
    "\n",
    "    # 模拟至稳定状态\n",
    "    def simulate_until_stable(self) -> None:\n",
    "        self.game.update_until_stable(GameInterface.SIMULATE_FPS)\n",
    "\n",
    "    # 将 action 解析成放置坐标（其中 y 无意义）\n",
    "    def decode_action(self, action: int) -> typing.Tuple[int, int]:\n",
    "        x = int((action + 0.5) * self.action_segment_len)\n",
    "\n",
    "        return (x, 0)\n",
    "\n",
    "    # 输入一个 action，进行模拟，并获得特征三元组\n",
    "    def next(self, action: int) -> typing.Tuple[np.ndarray, int, bool]:\n",
    "        current_fruit = self.game.current_fruit_type\n",
    "\n",
    "        score_1 = self.game.score\n",
    "\n",
    "        self.game.click(self.decode_action(action))\n",
    "        self.simulate_until_stable()\n",
    "\n",
    "        feature = self.game.get_features(\n",
    "            GameInterface.FEATURE_MAP_WIDTH, GameInterface.FEATURE_MAP_HEIGHT\n",
    "        )\n",
    "\n",
    "        score_2 = self.game.score\n",
    "\n",
    "        score, reward, alive = self.game.score, score_2 - score_1, self.game.alive\n",
    "\n",
    "        reward = reward if reward > 0 else -current_fruit\n",
    "\n",
    "        flatten_feature = feature.flatten().astype(np.float32)\n",
    "\n",
    "        return flatten_feature, reward, alive"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## **四、实现智能体与经验池**\n",
    "\n",
    "### 4.1 构建网络\n",
    "\n",
    "定义函数 `build_model`，调用后获得构建的全连接网络。\n",
    "\n",
    "该网络含有三个全连接层，每层 `64` 个节点，并采用 `ReLU` 进行激活。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "def build_model(input_size: int, output_size: int) -> nn.Layer:\n",
    "    model_prototype = nn.Sequential(\n",
    "        nn.Linear(in_features=input_size, out_features=64),  # 全连接层\n",
    "        nn.ReLU(),\n",
    "        nn.Linear(in_features=64, out_features=64),  # 全连接层\n",
    "        nn.ReLU(),\n",
    "        nn.Linear(in_features=64, out_features=64),  # 全连接层\n",
    "        nn.ReLU(),\n",
    "        nn.Linear(in_features=64, out_features=output_size),  # 输出层\n",
    "    )\n",
    "\n",
    "    return model_prototype"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.2 构建经验池\n",
    "\n",
    "经验池可以用来持久化 $experience$ （经验），并消除各个 $experience$ 之间的相关性。\n",
    "\n",
    "每个 $experience$ 主要用于记录 $state$ （状态）、$action$ （动作） 和 $reward$ （奖励）的关联，在强化学习中，通常使用 $(state, action, new_state, reward)$ ，以表示状态转移与动作、奖励的关联。\n",
    "\n",
    "在该项目中，使用 $feature$ 表示 $state$，并加入 $alive$ （存活与否）表示游戏状态。\n",
    "\n",
    "$experience$ 的结构包括以下部分：\n",
    "- $feature$: 动作前的状态（特征）\n",
    "- $action$: 进行的动作\n",
    "- $reward$: 动作获得的奖励\n",
    "- $next_feature$: 动作后的状态（特征）\n",
    "- $alive$: 游戏是否仍能进行"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 经验池容量\n",
    "MEMORY_SIZE = 50000\n",
    "# 经验池中的最小样本数\n",
    "MEMORY_WARMUP_SIZE = 5000\n",
    "\n",
    "\n",
    "class ReplayMemory(collections.deque):\n",
    "    def __init__(self, max_size: int = MEMORY_SIZE) -> None:\n",
    "        super().__init__(maxlen=max_size)\n",
    "\n",
    "    def sample(\n",
    "        self, batch_size: int\n",
    "    ) -> typing.Tuple[np.ndarray, np.ndarray, np.ndarray, np.ndarray, np.ndarray]:\n",
    "        mini_batch = random.sample(self, batch_size)\n",
    "\n",
    "        # feature_batch, action_batch, reward_batch, next_feature_batch, alive_batch = experiences\n",
    "        experiences = list(zip(*mini_batch))\n",
    "\n",
    "        return tuple([np.array(exp) for exp in experiences])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 4.3 构建 Agent"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 4.3.1 定义 Agent 超参数"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 学习率\n",
    "LEARNING_RATE = 0.001\n",
    "# 折扣因子\n",
    "GAMMA = 0.99"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 4.3.2 构建随机 Agent\n",
    "\n",
    "构建一个随机动作的 Agent 作为 baseline，用于效果检验。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "class RandomAgent:\n",
    "    def __init__(self, action_num: int) -> None:\n",
    "        self.action_num = action_num\n",
    "\n",
    "    # 随机选择一个动作\n",
    "    def sample(self, _feature: np.ndarray) -> np.ndarray:\n",
    "        return self.predict(_feature)\n",
    "\n",
    "    # 随机选择一个动作\n",
    "    def predict(self, feature: np.ndarray) -> np.ndarray:\n",
    "        return np.random.randint(0, self.action_num, size=(1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 4.3.3 构建DQN Agent\n",
    "\n",
    "DQN 使用两个结构相同、参数不同的神经网络来训练，$policy\\_net$ 用于学习，每次训练都更新，而 $target\\_net$ 在训练过程中比较固定，定期更新，负责产生目标。\n",
    "\n",
    "优化目标为 $Q^\\pi(s_t, a_t) = r_t + Q^\\pi(s_{t+1}, \\pi(s_{t+1}))$。\n",
    "\n",
    "其中，左侧 $Q^\\pi(s_t, a_t)$ 为模型输出，即 $policy\\_net$，\n",
    "右侧 $r_t + Q^\\pi(s_{t+1}, \\pi(s_{t+1}))$ 则为目标 $target$，即 $target\\_net$。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [],
   "source": [
    "class Agent:\n",
    "    def __init__(\n",
    "        self,\n",
    "        build_model: typing.Callable,\n",
    "        feature_dim: int,\n",
    "        action_num: int,\n",
    "        e_greed: float = 0.1,\n",
    "        e_greed_decrement: float = 1e-6,\n",
    "        learning_rate: float = LEARNING_RATE,\n",
    "        loss_func: typing.Callable[\n",
    "            [paddle.Tensor, paddle.Tensor], paddle.Tensor\n",
    "        ] = nn.MSELoss(\"mean\"),\n",
    "    ) -> None:\n",
    "        self.policy_net = build_model(feature_dim, action_num)\n",
    "        self.target_net = build_model(feature_dim, action_num)\n",
    "        # 特征维度（输入向量维度）\n",
    "        self.feature_dim = feature_dim\n",
    "        # 可选动作数量（输出向量维度）\n",
    "        self.action_num = action_num\n",
    "\n",
    "        # e-greedy 策略的 e-greed\n",
    "        self.e_greed = e_greed\n",
    "        self.e_greed_decrement = e_greed_decrement\n",
    "\n",
    "        # 损失函数与优化器\n",
    "        self.loss_func = loss_func\n",
    "        self.optimizer = optimizer.Adam(\n",
    "            parameters=self.policy_net.parameters(), learning_rate=learning_rate\n",
    "        )\n",
    "\n",
    "        # 当前步数\n",
    "        self.global_step = 0\n",
    "        # 每 update_target_steps (200) 步进行 target_net 的更新\n",
    "        self.update_target_steps = 200\n",
    "\n",
    "    # 基于 e_greed 策略，随机选择一个动作或基于当前模型选择一个动作\n",
    "    def sample(self, feature: np.ndarray) -> np.ndarray:\n",
    "        if np.random.uniform() < self.e_greed:\n",
    "            action = np.random.randint(0, self.action_num, size=(1))\n",
    "        else:\n",
    "            action = self.predict(feature)\n",
    "\n",
    "        self.e_greed = max(0, self.e_greed - self.e_greed_decrement)\n",
    "\n",
    "        return action\n",
    "\n",
    "    # 基于当前模型选择一个动作\n",
    "    def predict(self, feature: np.ndarray) -> np.ndarray:\n",
    "        with paddle.no_grad():\n",
    "            action = self.policy_net(paddle.to_tensor(feature)).argmax()\n",
    "        return action.numpy()\n",
    "\n",
    "    # 学习，更新 policy_net，并根据 global_step 决定是否更新 target_net\n",
    "    def learn(\n",
    "        self,\n",
    "        feature: np.ndarray,\n",
    "        action: int,\n",
    "        reward: float,\n",
    "        next_feature: np.ndarray,\n",
    "        alive: bool,\n",
    "    ):\n",
    "        # 更新 target_net\n",
    "        if self.global_step % self.update_target_steps == 0:\n",
    "            self.target_net.load_dict(self.policy_net.state_dict())\n",
    "            pass\n",
    "\n",
    "        self.global_step += 1\n",
    "\n",
    "        feature_batch = paddle.to_tensor(feature, dtype=\"float32\")\n",
    "        action_batch = paddle.to_tensor(action, dtype=\"int32\")\n",
    "        reward_batch = paddle.to_tensor(reward, dtype=\"float32\")\n",
    "        next_feature_batch = paddle.to_tensor(next_feature, dtype=\"float32\")\n",
    "        alive_batch = paddle.to_tensor(alive, dtype=\"float32\")\n",
    "\n",
    "        output_policy = paddle.squeeze(self.policy_net(feature_batch))\n",
    "        action_batch = paddle.squeeze(action_batch)\n",
    "\n",
    "        action_batch_onehot = nn.functional.one_hot(action_batch, self.action_num)\n",
    "\n",
    "        # policy_net 的输出与 action_batch_onehot 点乘，得到 q 值 policy_q_value\n",
    "        policy_q_value = paddle.sum(\n",
    "            paddle.multiply(output_policy, action_batch_onehot), axis=1\n",
    "        )\n",
    "\n",
    "        with paddle.no_grad():\n",
    "            # 计算 target_net 的输出 output_target_next\n",
    "            output_target_next = paddle.squeeze(self.target_net(next_feature_batch))\n",
    "            # 计算 target_next_q_value (action)\n",
    "            target_next_q_value = paddle.max(output_target_next, axis=1)\n",
    "\n",
    "        # 计算 target_q_value\n",
    "        target_q_value = paddle.squeeze(reward_batch) + GAMMA * paddle.squeeze(\n",
    "            target_next_q_value\n",
    "        ) * paddle.squeeze(alive_batch)\n",
    "\n",
    "        # 计算损失\n",
    "        loss = self.loss_func(policy_q_value, target_q_value)\n",
    "\n",
    "        # 反向传播损失用于优化\n",
    "        self.optimizer.clear_grad()\n",
    "        loss.backward()\n",
    "\n",
    "        self.optimizer.step()\n",
    "\n",
    "        return loss.item()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## **五、实现训练方法**\n",
    "\n",
    "### 5.1 定义训练超参数"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "LEARN_FREQUENCY = 1  # 每 LEARN_FREQUENCY (1) 步学习一次\n",
    "BATCH_SIZE = 32  # 每次学习的 batch 大小"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 5.2 运行一局\n",
    "\n",
    "基于传入的 `env` （游戏环境）、`agent` （智能体） 进行一局游戏，直至游戏结束。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [],
   "source": [
    "def run_episode(\n",
    "    env: GameInterface, agent: Agent, memory: ReplayMemory, episode_id: int, debug=False\n",
    "):\n",
    "    # 初始化环境\n",
    "    env.reset()\n",
    "\n",
    "    step, rewards_sum = 0, 0\n",
    "    # 随机选择第一个动作\n",
    "    action = np.random.randint(0, env.action_num)\n",
    "    # 执行第一个动作，开始模拟以获取特征\n",
    "    feature, _, alive = env.next(action)\n",
    "\n",
    "    assert alive\n",
    "\n",
    "    while alive:\n",
    "        step += 1\n",
    "\n",
    "        # 获取 action\n",
    "        action = agent.sample(feature)\n",
    "        # 执行 action，获取下一个状态\n",
    "        next_feature, reward, alive = env.next(action)\n",
    "\n",
    "        # 如果动作导致游戏结束，则 reward 设为 -1000\n",
    "        reward = reward if alive else -1000\n",
    "\n",
    "        # 记录到经验池中\n",
    "        memory.append((feature, action, reward, next_feature, alive))\n",
    "\n",
    "        # 当经验池中的数据量不小于 MEMORY_WARMUP_SIZE 时，开始学习\n",
    "        # 由于 LEARN_FREQUENCY = 1，所以每步都学习\n",
    "        if (\n",
    "            len(memory) >= MEMORY_WARMUP_SIZE\n",
    "            and agent.global_step % LEARN_FREQUENCY == 0\n",
    "        ):\n",
    "            (\n",
    "                feature_batch,\n",
    "                action_batch,\n",
    "                reward_batch,\n",
    "                next_feature_batch,\n",
    "                alive_batch,\n",
    "            ) = memory.sample(BATCH_SIZE)\n",
    "\n",
    "            _loss = agent.learn(\n",
    "                feature_batch,\n",
    "                action_batch,\n",
    "                reward_batch,\n",
    "                next_feature_batch,\n",
    "                alive_batch,\n",
    "            )\n",
    "\n",
    "        reward_sum = np.sum(reward)\n",
    "        # reward 和加上本次的 reward\n",
    "        rewards_sum += reward_sum\n",
    "\n",
    "        feature = next_feature\n",
    "\n",
    "    # 返回本局游戏的总 reward\n",
    "    return rewards_sum"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## **六、开始训练**"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.1 初始化训练环境"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 特征图的维度（宽、高）\n",
    "feature_map_height = GameInterface.FEATURE_MAP_HEIGHT\n",
    "feature_map_width = GameInterface.FEATURE_MAP_WIDTH\n",
    "\n",
    "# 动作数量\n",
    "action_dim = GameInterface.ACTION_NUM\n",
    "# 输入特征维度\n",
    "feature_dim = feature_map_height * feature_map_width * 2\n",
    "# e-greed 的初始值\n",
    "e_greed = 0.5\n",
    "# e-greed 的衰减量\n",
    "e_greed_decrement = 1e-6\n",
    "\n",
    "# 创建环境\n",
    "env = GameInterface()\n",
    "\n",
    "# 创建经验池\n",
    "memory = ReplayMemory(MEMORY_SIZE)\n",
    "\n",
    "# 创建智能体\n",
    "agent = Agent(build_model, feature_dim, action_dim, e_greed, e_greed_decrement)\n",
    "\n",
    "FINAL_PARAM_PATH = \"final.pdparams\"  # 模型保存路径"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.2 尝试加载已有模型\n",
    "\n",
    "本次演示时，不使用预训练权重。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [],
   "source": [
    "# if os.path.exists(FINAL_PARAM_PATH):\n",
    "#     print(\"Load final param.\")\n",
    "#     agent.policy_net.set_state_dict(paddle.load(FINAL_PARAM_PATH))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.3 预热经验池\n",
    "\n",
    "先在经验池中填充经验，避免前期训练时选取经验相关度过高。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Warm up.\n",
      "Memory size is 5007.\n"
     ]
    }
   ],
   "source": [
    "print(\"Warm up.\")\n",
    "\n",
    "while len(memory) < MEMORY_WARMUP_SIZE:\n",
    "    run_episode(env, agent, memory, -1)\n",
    "\n",
    "print(f\"Memory size is {len(memory)}.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.4 创建评价函数\n",
    "\n",
    "#### 6.4.1 评价函数\n",
    "\n",
    "传入环境和随机数种子，返回游戏得分与总奖励。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [],
   "source": [
    "def evaluate(\n",
    "    env: GameInterface, agent: Agent, seed: int = None\n",
    ") -> typing.Tuple[float, float]:\n",
    "    env.reset(seed)  # 基于 seed 初始化环境\n",
    "    action = np.random.randint(0, env.action_num)  # 选择随机动作\n",
    "    feature, _, alive = env.next(action)  # 执行动作，获取特征\n",
    "    rewards_sum = 0\n",
    "\n",
    "    while alive:\n",
    "        action = agent.predict(feature)\n",
    "        feature, reward, alive = env.next(action)\n",
    "\n",
    "        reward_sum = np.sum(reward)\n",
    "        # 计算 reward 和\n",
    "        rewards_sum += reward_sum\n",
    "\n",
    "    # 返回本局游戏分数 score 和奖励 reward\n",
    "    return env.game.score, rewards_sum"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "#### 6.4.2 与随机智能体对比\n",
    "\n",
    "进行一定轮次的游戏，并比较双方的平均得分。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 初始化用于评估的伪随机数生成器\n",
    "evaluate_random = PRNG()\n",
    "evaluate_random.seed(\"RedContritio\")\n",
    "\n",
    "# 进行 50 局游戏，对分数、奖励取平均，降低偶然性\n",
    "EVALUATE_TIMES = 50\n",
    "\n",
    "# 与随机智能体对比\n",
    "def compare_with_random(env: GameInterface, agent: Agent, action_count: int) -> None:\n",
    "    random_agent = RandomAgent(action_count)\n",
    "\n",
    "    scores1, rewards1 = [], []\n",
    "    scores2, rewards2 = [], []\n",
    "\n",
    "    for _ in range(EVALUATE_TIMES):\n",
    "        # 选择环境初始化的随机数种子\n",
    "        seed = evaluate_random.random()\n",
    "\n",
    "        # 评估 DQN 智能体\n",
    "        score1, reward1 = evaluate(env, agent, seed)\n",
    "        scores1.append(score1)\n",
    "        rewards1.append(reward1)\n",
    "\n",
    "        # 评估随机智能体\n",
    "        score2, reward2 = evaluate(env, random_agent, seed)\n",
    "        scores2.append(score2)\n",
    "        rewards2.append(reward2)\n",
    "\n",
    "    print(\n",
    "        f\"[DQN Agent]\\t:\\tmean_score: {np.mean(scores1)},\\tmean_reward: {np.mean(rewards1)}\"\n",
    "    )\n",
    "    print(\n",
    "        f\"[Random Agent]\\t:\\tmean_score: {np.mean(scores2)},\\tmean_reward: {np.mean(rewards2)}\"\n",
    "    )\n",
    "\n",
    "    return [np.mean(scores1), np.mean(rewards1)], [np.mean(scores2), np.mean(rewards2)]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.5 训练模型\n",
    "\n",
    "进行 2000 局训练，每训练 200 局进行一次评价。特别的，在开始训练前进行一次评价。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Start training.\n",
      "Episode: 0, e_greed: 0.49495300000013503\n",
      "[DQN Agent]\t:\tmean_score: 76.34,\tmean_reward: 22.24\n",
      "[Random Agent]\t:\tmean_score: 150.04,\tmean_reward: 68.12\n",
      "Episode: 200, e_greed: 0.4856550000003838\n",
      "[DQN Agent]\t:\tmean_score: 165.44,\tmean_reward: 83.56\n",
      "[Random Agent]\t:\tmean_score: 167.26,\tmean_reward: 80.94\n",
      "Episode: 400, e_greed: 0.4763610000006325\n",
      "[DQN Agent]\t:\tmean_score: 192.22,\tmean_reward: 100.96\n",
      "[Random Agent]\t:\tmean_score: 153.32,\tmean_reward: 66.52\n",
      "Episode: 600, e_greed: 0.46702500000088226\n",
      "[DQN Agent]\t:\tmean_score: 197.26,\tmean_reward: 105.44\n",
      "[Random Agent]\t:\tmean_score: 160.68,\tmean_reward: 72.7\n",
      "Episode: 800, e_greed: 0.45757400000113513\n",
      "[DQN Agent]\t:\tmean_score: 183.18,\tmean_reward: 91.92\n",
      "[Random Agent]\t:\tmean_score: 163.68,\tmean_reward: 77.58\n",
      "Episode: 1000, e_greed: 0.4481560000013871\n",
      "[DQN Agent]\t:\tmean_score: 189.1,\tmean_reward: 99.46\n",
      "[Random Agent]\t:\tmean_score: 153.24,\tmean_reward: 70.5\n",
      "Episode: 1200, e_greed: 0.4386110000016425\n",
      "[DQN Agent]\t:\tmean_score: 199.06,\tmean_reward: 104.68\n",
      "[Random Agent]\t:\tmean_score: 149.84,\tmean_reward: 65.22\n",
      "Episode: 1400, e_greed: 0.42925800000189274\n",
      "[DQN Agent]\t:\tmean_score: 145.36,\tmean_reward: 70.54\n",
      "[Random Agent]\t:\tmean_score: 148.64,\tmean_reward: 66.72\n",
      "Episode: 1600, e_greed: 0.41988700000214346\n",
      "[DQN Agent]\t:\tmean_score: 193.4,\tmean_reward: 101.98\n",
      "[Random Agent]\t:\tmean_score: 161.54,\tmean_reward: 77.1\n",
      "Episode: 1800, e_greed: 0.41034600000239874\n",
      "[DQN Agent]\t:\tmean_score: 201.82,\tmean_reward: 107.8\n",
      "[Random Agent]\t:\tmean_score: 165.48,\tmean_reward: 81.4\n",
      "Episode: 2000, e_greed: 0.40094500000265026\n",
      "[DQN Agent]\t:\tmean_score: 194.7,\tmean_reward: 100.16\n",
      "[Random Agent]\t:\tmean_score: 152.68,\tmean_reward: 68.44\n"
     ]
    }
   ],
   "source": [
    "max_episode = 2000\n",
    "episode_per_save = max_episode // 10\n",
    "\n",
    "# 记录 DQN 智能体和随机智能体的表现\n",
    "history1, history2 = [], []\n",
    "# 记录横坐标（局数）\n",
    "x_history = []\n",
    "\n",
    "print(\"Start training.\")\n",
    "for episode_id in range(0, max_episode + 1):\n",
    "    total_reward = run_episode(env, agent, memory, episode_id)\n",
    "\n",
    "    if episode_id % episode_per_save == 0:\n",
    "        print(f\"Episode: {episode_id}, e_greed: {agent.e_greed}\")\n",
    "\n",
    "        his1, his2 = compare_with_random(env, agent, action_dim)\n",
    "        history1.append(his1)\n",
    "        history2.append(his2)\n",
    "        x_history.append(episode_id)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.6 展示训练历史"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1000x500 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(10, 5))\n",
    "\n",
    "plt.plot(x_history, [score for (score, _) in history1], label=\"DQN score\")\n",
    "plt.plot(x_history, [reward for (_, reward) in history1], label=\"DQN reward\")\n",
    "plt.plot(x_history, [score for (score, _) in history2], label=\"Random score\")\n",
    "plt.plot(x_history, [reward for (_, reward) in history2], label=\"Random reward\")\n",
    "\n",
    "plt.xlabel(\"episode\")\n",
    "plt.ylabel(\"score/reward\")\n",
    "\n",
    "plt.title(\"DQN Agent VS Random Agent\")\n",
    "\n",
    "plt.legend()\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "其中，可以看出，DQN 智能体在 400 轮后（奖励 $reward$）优于随机智能体，因此可以认为该模型学习到了游戏特点。\n",
    "\n",
    "在 1400 局时的评估中，$score_{DQN} < score_{Random}$，即模型尚有部分情况处理能力较差。\n",
    "\n",
    "但由于 $score$ 和 $reward$ 存在差异，在 200 轮后，有 $reward_{DQN} > score_{Random}$。\n",
    "\n",
    "由于奖励 $reward$ 与分数 $score$ 相关，因此两曲线正相关。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.7 保存模型参数"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [],
   "source": [
    "paddle.save(agent.policy_net.state_dict(), FINAL_PARAM_PATH)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## **七、可视化与模型评估**\n",
    "\n",
    "### 7.1 特征可视化\n",
    "\n",
    "通过直观观察特征，进一步理解特征对智能体决策的作用。\n",
    "\n",
    "#### 7.1.1 特征可视化函数"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [],
   "source": [
    "def visualize_feature(\n",
    "    feature: np.ndarray, game_resolution: typing.Tuple[int, int]\n",
    ") -> np.ndarray:\n",
    "    game_w, game_h = game_resolution\n",
    "    feature_img = np.zeros((game_h, game_w * 2, 3), dtype=np.uint8)  # 初始化特征图\n",
    "\n",
    "    uw, uh = game_w / feature.shape[1], game_h / feature.shape[0]  # 网格宽高\n",
    "\n",
    "    _v2c = (\n",
    "        lambda v: 255 if v > 0 else (0 if v == 0 else int(-v / 13.0 * 255.0))\n",
    "    )  # 特征值转颜色（单通道）\n",
    "    value2color = (\n",
    "        lambda v: (_v2c(v), _v2c(v), _v2c(v)) if v >= 0 else (127, _v2c(v), _v2c(v))\n",
    "    )  # 特征值转颜色（三通道）\n",
    "\n",
    "    for i in range(feature.shape[0]):\n",
    "        for j in range(feature.shape[1]):\n",
    "            # 对特征的每个值，基于值对应的颜色来渲染相应的网格\n",
    "            feature_img[\n",
    "                int(i * uh) : int((i + 1) * uh), int(j * uw) : int((j + 1) * uw)\n",
    "            ] = value2color(feature[i, j, 0])\n",
    "            feature_img[\n",
    "                int(i * uh) : int((i + 1) * uh),\n",
    "                int(j * uw + game_w) : int((j + 1) * uw + game_w),\n",
    "            ] = value2color(feature[i, j, 1])\n",
    "\n",
    "            # 绘制特征值文本到对应网格中\n",
    "            putText2(\n",
    "                feature_img,\n",
    "                f\"{int(feature[i, j, 0])}\",\n",
    "                (int((j + 0.5) * uw), int((i + 0.5) * uh)),\n",
    "                font_scale=0.3,\n",
    "                color=(0, 0, 255),\n",
    "            )\n",
    "            putText2(\n",
    "                feature_img,\n",
    "                f\"{int(feature[i, j, 1])}\",\n",
    "                (int((j + 0.5) * uw + game_w), int((i + 0.5) * uh)),\n",
    "                font_scale=0.3,\n",
    "                color=(0, 0, 255),\n",
    "            )\n",
    "\n",
    "    # 绘制网格线（横向）\n",
    "    for i in range(feature.shape[0]):\n",
    "        cv2.line(\n",
    "            feature_img, (0, int(i * uh)), (game_w * 2, int(i * uh)), (255, 0, 0), 1\n",
    "        )\n",
    "    # 绘制最下面的网格线\n",
    "    cv2.line(feature_img, (0, game_h - 1), (game_w * 2, game_h - 1), (255, 0, 0), 1)\n",
    "\n",
    "    # 绘制左半部分网格线（纵向）\n",
    "    for j in range(feature.shape[1]):\n",
    "        cv2.line(feature_img, (int(j * uw), 0), (int(j * uw), game_h), (0, 255, 0), 1)\n",
    "    # 绘制左半部分的右边界线\n",
    "    cv2.line(feature_img, (game_w - 1, 0), (game_w - 1, game_h), (0, 255, 0), 1)\n",
    "\n",
    "    # 绘制右半部分网格线（纵向）\n",
    "    for j in range(feature.shape[1]):\n",
    "        cv2.line(\n",
    "            feature_img,\n",
    "            (int(j * uw + game_w), 0),\n",
    "            (int(j * uw + game_w), game_h),\n",
    "            (0, 255, 0),\n",
    "            1,\n",
    "        )\n",
    "\n",
    "    # 绘制右边界线\n",
    "    cv2.line(\n",
    "        feature_img,\n",
    "        (game_w - 1 + game_w, 0),\n",
    "        (game_w - 1 + game_w, game_h),\n",
    "        (0, 255, 0),\n",
    "        1,\n",
    "    )\n",
    "\n",
    "    # 绘制两部分中间的分隔线（较粗）\n",
    "    cv2.line(feature_img, (game_w, 0), (game_w, game_h), (255, 255, 0), 3)\n",
    "\n",
    "    return feature_img"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7.1.2 特征可视化\n",
    "\n",
    "首先开始一局游戏并进行一些操作。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "score: 69, alive: True\n"
     ]
    }
   ],
   "source": [
    "# 初始化环境\n",
    "env.reset(11235813)\n",
    "\n",
    "max_steps = 30\n",
    "# 预设动作\n",
    "actions = [(i % GameInterface.ACTION_NUM) for i in range(max_steps)]\n",
    "feature, _, alive = env.next(0)\n",
    "rewards_sum = 0\n",
    "\n",
    "step = 0\n",
    "\n",
    "while alive and step < max_steps:\n",
    "    action = actions[step]\n",
    "    feature, reward, alive = env.next(action)\n",
    "\n",
    "    step += 1\n",
    "\n",
    "# 渲染游戏画面\n",
    "env.game.draw()\n",
    "\n",
    "print(f\"score: {env.game.score}, alive: {env.game.alive}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "随后分别显示游戏画面与特征可视化的结果。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<Figure size 7200x4800 with 0 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 640x480 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(12, 8), dpi=600, layout=\"tight\")\n",
    "\n",
    "fig, (ax1, ax2) = plt.subplots(1, 2, gridspec_kw={\"width_ratios\": [1, 2]})\n",
    "\n",
    "ax1.set_title(\"Game Screen\")\n",
    "ax1.axis(\"off\")\n",
    "\n",
    "rgb_img = cv2.cvtColor(env.game.screen, cv2.COLOR_BGRA2BGR)\n",
    "\n",
    "# 左侧显示游戏界面\n",
    "ax1.imshow(rgb_img[:, :, ::-1])\n",
    "\n",
    "ax2.set_title(\"Feature\")\n",
    "ax2.axis(\"off\")\n",
    "\n",
    "rgb_img = visualize_feature(\n",
    "    env.game.get_features(\n",
    "        GameInterface.FEATURE_MAP_WIDTH, GameInterface.FEATURE_MAP_HEIGHT\n",
    "    ),\n",
    "    env.game.resolution,\n",
    ")\n",
    "\n",
    "# 右侧显示特征图\n",
    "ax2.imshow(rgb_img[:, :, ::-1])\n",
    "\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "特征图中，分为左右两部分，使用白色表示和当前水果同类的区域，数值为 1。\n",
    "\n",
    "左侧用于记录比当前水果小的水果，右侧记录比当前水果大的水果，不满足条件且不为同类的水果在所属特征图中不可见。\n",
    "\n",
    "以上图为例，当前水果为红色的樱桃，类别为 `2`。\n",
    "\n",
    "因此特征图中（左右部分分别）有四个区域为白色，对应左图中的四颗樱桃 `2`。\n",
    "\n",
    "左部记录比当前水果小的：当前图中仅葡萄 `1`（在最右下角）比樱桃 `2` 小，因此其余部分均为黑色 `0` （空白）。\n",
    "\n",
    "右部记录比当前水果大的：当前图中大多数水果均大于樱桃 `2`，差距越大的水果颜色越浅，可以看到特征图的右部和游戏界面整体表现一致。"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 7.2 评估训练结果\n",
    "\n",
    "基于给定的 `env` 和 `agent`，进行一定数量局数的游戏。\n",
    "\n",
    "在游戏结束后，返回每局的平均得分和平均总奖励。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[DQN Agent]\t:\tmean_score: 205.01,\tmean_reward: 109.82,\n",
      "\t\t\tmax_score: 352,\tmax_reward: 222,\n",
      "\t\t\tmin_score: 90,\tmin_reward: 16\n",
      "[Random Agent]\t:\tmean_score: 153.215,\tmean_reward: 69.67,\n",
      "\t\t\tmax_score: 322,\tmax_reward: 194,\n",
      "\t\t\tmin_score: 57,\tmin_reward: -16\n"
     ]
    }
   ],
   "source": [
    "# 评估 200 局，降低偶然性影响\n",
    "EVALUATE_TIMES = 200\n",
    "\n",
    "random_agent = RandomAgent(GameInterface.ACTION_NUM)\n",
    "\n",
    "scores1, rewards1 = [], []\n",
    "scores2, rewards2 = [], []\n",
    "\n",
    "for _ in range(EVALUATE_TIMES):\n",
    "    seed = evaluate_random.random()\n",
    "\n",
    "    score1, reward1 = evaluate(env, agent, seed)\n",
    "    scores1.append(score1)\n",
    "    rewards1.append(reward1)\n",
    "\n",
    "    score2, reward2 = evaluate(env, random_agent, seed)\n",
    "    scores2.append(score2)\n",
    "    rewards2.append(reward2)\n",
    "\n",
    "print(\n",
    "    f\"\"\"[DQN Agent]\\t:\\tmean_score: {np.mean(scores1)},\\tmean_reward: {np.mean(rewards1)},\n",
    "\\t\\t\\tmax_score: {np.max(scores1)},\\tmax_reward: {np.max(rewards1)},\n",
    "\\t\\t\\tmin_score: {np.min(scores1)},\\tmin_reward: {np.min(rewards1)}\"\"\"\n",
    ")\n",
    "print(\n",
    "    f\"\"\"[Random Agent]\\t:\\tmean_score: {np.mean(scores2)},\\tmean_reward: {np.mean(rewards2)},\n",
    "\\t\\t\\tmax_score: {np.max(scores2)},\\tmax_reward: {np.max(rewards2)},\n",
    "\\t\\t\\tmin_score: {np.min(scores2)},\\tmin_reward: {np.min(rewards2)}\"\"\"\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "可以看到，DQN 智能体分数显著优于随机智能体的分数，因此认为该网络能学习到游戏策略。"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3.10.7 ('venv': venv)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.7"
  },
  "orig_nbformat": 4,
  "vscode": {
   "interpreter": {
    "hash": "7e4b57045e20a141e4d93711a695d68b49f01d14e09382e54c9e39218953d712"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}