dqn.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Deep Q-Learning\n",
    "\n",
    "Q関数をニューラルネットで近似したもの。ほかにもTarget Networt, Experience Replayなど工夫が入っている。\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "from collections import namedtuple, deque\n",
    "from itertools import count\n",
    "import random\n",
    "import copy\n",
    "\n",
    "import gym\n",
    "import pathlib\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "import torch.nn.functional as F\n",
    "import torchvision.transforms as T\n",
    "\n",
    "plt.rcParams['font.family'] = 'MyricaM M'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "output_type": "execute_result",
     "data": {
      "text/plain": "<torch._C.Generator at 0x1e6001270f0>"
     },
     "metadata": {},
     "execution_count": 2
    }
   ],
   "source": [
    "# ハイパラ\n",
    "BATCH_SIZE = 64\n",
    "N_EPISODES = 600\n",
    "TARGET_UPDATE_FREQ = 100\n",
    "N_HIDDEN = 64\n",
    "REPLAY_MEMORY_SIZE = 10**4\n",
    "REPLAY_START_SIZE = 500\n",
    "MIN_EPS = 0.01\n",
    "EPS_DECAY = 5000\n",
    "\n",
    "env = gym.make('CartPole-v0')\n",
    "obs_size = env.observation_space.low.size\n",
    "n_actions = env.action_space.n\n",
    "reword_threshold = env.spec.reward_threshold\n",
    "\n",
    "np.random.seed(125)\n",
    "env.seed(125)\n",
    "torch.manual_seed(125)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [],
   "source": [
    "class QNetwork(nn.Module):\n",
    "    def __init__(self, in_dim, n_actions, h_dim=64):\n",
    "        super(QNetwork, self).__init__()\n",
    "        self.h1 = nn.Linear(in_dim, h_dim)\n",
    "        self.h2 = nn.Linear(h_dim, n_actions)\n",
    "    \n",
    "    def forward(self, x):\n",
    "        h = F.relu(self.h1(x))\n",
    "        return self.h2(h)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "class EpsilonGreedy():\n",
    "    def __init__(self, n_actions):\n",
    "        self.n_actions = n_actions\n",
    "    \n",
    "    def sample(self, q_net, obs, iteration, len_mem):\n",
    "        # epsilonを線形に減ずる\n",
    "        if len_mem < REPLAY_START_SIZE:\n",
    "            epsilon = 1.0\n",
    "        else:\n",
    "            epsilon = max(MIN_EPS, np.interp(iteration, [0, EPS_DECAY], [1.0, MIN_EPS]))\n",
    "\n",
    "        # epsilon-greedyで行動を選択\n",
    "        if np.random.rand() < epsilon:\n",
    "            action = np.random.randint(low=0, high=self.n_actions)\n",
    "        else:\n",
    "            obs_t = torch.from_numpy(obs).float()\n",
    "            with torch.no_grad():\n",
    "                action = int(torch.argmax(policy_q(obs_t)))\n",
    "        \n",
    "        return action"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def ep_greedy(Q, state, num_iter, len_memory):\n",
    "    # epsilonを線形に減ずる\n",
    "    if len_memory < REPLAY_START_SIZE:\n",
    "        epsilon = 1.0\n",
    "    else:\n",
    "        epsilon = max(MIN_EPS, np.interp(num_iter, [0, EPS_DECAY], [1.0, MIN_EPS]))\n",
    "\n",
    "    # epsilon-greedyで行動を選択\n",
    "    if np.random.rand() < epsilon:\n",
    "        action = np.random.randint(low=0, high=n_actions) # ランダムに行動\n",
    "    else:\n",
    "        state_t = torch.from_numpy(state).float()\n",
    "        with torch.no_grad():\n",
    "            action = int(torch.argmax(Q(state_t)))\n",
    "    \n",
    "    return action"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "def update_policy(policy_q, target_q, optimizer, samples, gamma=0.99):\n",
    "\n",
    "    state = torch.Tensor([sample[0] for sample in samples]).float()\n",
    "    action = torch.LongTensor([[sample[1]] for sample in samples])\n",
    "    reward = torch.Tensor([sample[2] for sample in samples]).float()\n",
    "    done = torch.Tensor([sample[3] for sample in samples]).float()\n",
    "    n_state = torch.Tensor([sample[4] for sample in samples]).float()\n",
    "\n",
    "    # 現在のQnetによる、Q(s, a)の推定値\n",
    "    Q_sa = policy_q(state).gather(1, action)\n",
    "    \n",
    "    # 実際に行動した結果の実績値（target, r+gamma * max_a' Q(s', a')）\n",
    "    # - 勾配が残らないようにするためのdetach()\n",
    "    # - max(axis)の返り値は(value, indices)なので、[0]をつける\n",
    "    next_Q = target_q(n_state).max(axis=1)[0].detach() \n",
    "    target = reward + gamma * (1 - done) * next_Q \n",
    "\n",
    "    # Huber lossを計算\n",
    "    # - unsqueeze()で次元を増やす\n",
    "    loss = F.smooth_l1_loss(Q_sa, target.unsqueeze(1))\n",
    "\n",
    "    # Optimize the model\n",
    "    optimizer.zero_grad()\n",
    "    loss.backward()\n",
    "    for param in policy_q.parameters():\n",
    "        param.grad.data.clamp_(-1, 1) # 勾配クリッピング\n",
    "    optimizer.step()\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "tags": []
   },
   "outputs": [
    {
     "output_type": "stream",
     "name": "stdout",
     "text": "episodes: 0 iteration: 28 R: 28.0 average_R 28.0\nepisodes: 10 iteration: 229 R: 35.0 average_R 20.818181818181817\nepisodes: 20 iteration: 432 R: 39.0 average_R 20.571428571428573\nepisodes: 30 iteration: 653 R: 22.0 average_R 21.06451612903226\nepisodes: 40 iteration: 845 R: 22.0 average_R 20.609756097560975\nepisodes: 50 iteration: 1036 R: 27.0 average_R 20.313725490196077\nepisodes: 60 iteration: 1202 R: 12.0 average_R 19.704918032786885\nepisodes: 70 iteration: 1371 R: 28.0 average_R 19.309859154929576\nepisodes: 80 iteration: 1521 R: 18.0 average_R 18.77777777777778\nepisodes: 90 iteration: 1694 R: 15.0 average_R 18.615384615384617\nepisodes: 100 iteration: 1839 R: 14.0 average_R 18.11\nepisodes: 110 iteration: 2046 R: 14.0 average_R 18.17\nepisodes: 120 iteration: 2175 R: 11.0 average_R 17.43\nepisodes: 130 iteration: 2367 R: 20.0 average_R 17.14\nepisodes: 140 iteration: 3062 R: 52.0 average_R 22.17\nepisodes: 150 iteration: 3993 R: 81.0 average_R 29.57\nepisodes: 160 iteration: 5290 R: 200.0 average_R 40.88\nepisodes: 170 iteration: 6277 R: 78.0 average_R 49.06\nepisodes: 180 iteration: 7098 R: 87.0 average_R 55.77\nepisodes: 190 iteration: 8444 R: 154.0 average_R 67.5\nepisodes: 200 iteration: 9648 R: 108.0 average_R 78.09\nepisodes: 210 iteration: 10892 R: 137.0 average_R 88.46\nepisodes: 220 iteration: 12383 R: 200.0 average_R 102.08\nepisodes: 230 iteration: 13609 R: 102.0 average_R 112.42\nepisodes: 240 iteration: 14681 R: 90.0 average_R 116.19\nepisodes: 250 iteration: 15444 R: 55.0 average_R 114.51\nepisodes: 260 iteration: 15988 R: 54.0 average_R 106.98\nepisodes: 270 iteration: 16550 R: 75.0 average_R 102.73\nepisodes: 280 iteration: 17244 R: 74.0 average_R 101.46\nepisodes: 290 iteration: 18194 R: 105.0 average_R 97.5\nepisodes: 300 iteration: 19961 R: 200.0 average_R 103.13\nepisodes: 310 iteration: 21961 R: 200.0 average_R 110.69\nepisodes: 320 iteration: 23961 R: 200.0 average_R 115.78\nepisodes: 330 iteration: 25732 R: 133.0 average_R 121.23\nepisodes: 340 iteration: 27060 R: 137.0 average_R 123.79\nepisodes: 350 iteration: 28353 R: 144.0 average_R 129.09\nepisodes: 360 iteration: 30309 R: 200.0 average_R 143.21\nepisodes: 370 iteration: 32309 R: 200.0 average_R 157.59\nepisodes: 380 iteration: 34309 R: 200.0 average_R 170.65\nepisodes: 390 iteration: 36309 R: 200.0 average_R 181.15\nepisodes: 400 iteration: 38309 R: 200.0 average_R 183.48\nepisodes: 410 iteration: 40309 R: 200.0 average_R 183.48\nepisodes: 420 iteration: 42309 R: 200.0 average_R 183.48\nepisodes: 430 iteration: 44309 R: 200.0 average_R 185.77\nepisodes: 440 iteration: 46309 R: 200.0 average_R 192.49\nepisodes: 450 iteration: 48309 R: 200.0 average_R 199.56\nepisodes: 460 iteration: 50309 R: 200.0 average_R 200.0\nepisodes: 470 iteration: 52240 R: 200.0 average_R 199.31\nepisodes: 480 iteration: 54240 R: 200.0 average_R 199.31\nepisodes: 490 iteration: 56240 R: 200.0 average_R 199.31\nepisodes: 500 iteration: 58240 R: 200.0 average_R 199.31\nepisodes: 510 iteration: 60240 R: 200.0 average_R 199.31\nepisodes: 520 iteration: 62240 R: 200.0 average_R 199.31\nepisodes: 530 iteration: 63778 R: 46.0 average_R 194.69\nepisodes: 540 iteration: 65124 R: 200.0 average_R 188.15\nepisodes: 550 iteration: 67124 R: 200.0 average_R 188.15\nepisodes: 560 iteration: 69124 R: 200.0 average_R 188.15\nepisodes: 570 iteration: 71124 R: 200.0 average_R 188.84\nepisodes: 580 iteration: 73124 R: 200.0 average_R 188.84\nepisodes: 590 iteration: 75124 R: 200.0 average_R 188.84\n"
    }
   ],
   "source": [
    "policy_q = QNetwork(obs_size, n_actions)\n",
    "target_q = QNetwork(obs_size, n_actions)\n",
    "target_q.load_state_dict(policy_q.state_dict()) # パラメータのコピー\n",
    "target_q.eval() # target_qは更新しない\n",
    "optimizer = optim.Adam(policy_q.parameters())\n",
    "replay_mem = deque(maxlen=REPLAY_MEMORY_SIZE)\n",
    "epsilon_greedy = EpsilonGreedy(n_actions)\n",
    "Rs = deque(maxlen=100)\n",
    "hist_aveRs = []\n",
    "\n",
    "iteration = 0\n",
    "for ep in range(N_EPISODES):\n",
    "    state = env.reset()\n",
    "    done = False\n",
    "    R = 0\n",
    "    timestep = 0\n",
    "\n",
    "    while not done:\n",
    "        # epsilon-greedyで行動選択\n",
    "        action = ep_greedy(policy_q, state, iteration, len(replay_mem))\n",
    "        # action = epsilon_greedy.sample(policy_q, state, iteration, len(replay_mem))\n",
    "        next_state, reward, done, _ = env.step(action)\n",
    "        R += reward\n",
    "\n",
    "        # 状態遷移を保存\n",
    "        replay_mem.append((state, action, reward, done, next_state))\n",
    "\n",
    "        # 状態の更新\n",
    "        state = next_state\n",
    "\n",
    "        # Experience Replay\n",
    "        if len(replay_mem) >= REPLAY_START_SIZE:\n",
    "            sample_indices = random.sample(range(len(replay_mem)), BATCH_SIZE)\n",
    "            samples = [replay_mem[i] for i in sample_indices]\n",
    "            update_policy(policy_q, target_q, optimizer, samples)\n",
    "\n",
    "        # target networkの更新（pytorchのサンプルは5episodeごとに更新）\n",
    "        if iteration % TARGET_UPDATE_FREQ == 0:\n",
    "            target_q.load_state_dict(policy_q.state_dict())\n",
    "        \n",
    "        iteration += 1\n",
    "        timestep += 1\n",
    "    \n",
    "    Rs.append(R)\n",
    "    average_R = np.mean(Rs)\n",
    "    hist_aveRs.append(average_R)\n",
    "    if ep % 10 == 0:\n",
    "        print('episodes: {} iteration: {} R: {} average_R {}'.format(ep, iteration, R, average_R))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "output_type": "display_data",
     "data": {
      "text/plain": "<Figure size 700x400 with 1 Axes>",
      "image/svg+xml": "<?xml version=\"1.0\" encoding=\"utf-8\" standalone=\"no\"?>\r\n<!DOCTYPE svg PUBLIC \"-//W3C//DTD SVG 1.1//EN\"\r\n  \"http://www.w3.org/Graphics/SVG/1.1/DTD/svg11.dtd\">\r\n<!-- Created with matplotlib (https://matplotlib.org/) -->\r\n<svg height=\"273.506875pt\" version=\"1.1\" viewBox=\"0 0 439.33125 273.506875\" width=\"439.33125pt\" xmlns=\"http://www.w3.org/2000/svg\" xmlns:xlink=\"http://www.w3.org/1999/xlink\">\r\n <defs>\r\n  <style type=\"text/css\">\r\n*{stroke-linecap:butt;stroke-linejoin:round;}\r\n  </style>\r\n </defs>\r\n <g id=\"figure_1\">\r\n  <g id=\"patch_1\">\r\n   <path d=\"M 0 273.506875 \r\nL 439.33125 273.506875 \r\nL 439.33125 0 \r\nL 0 0 \r\nz\r\n\" style=\"fill:none;\"/>\r\n  </g>\r\n  <g id=\"axes_1\">\r\n   <g id=\"patch_2\">\r\n    <path d=\"M 41.53125 238.644375 \r\nL 432.13125 238.644375 \r\nL 432.13125 21.204375 \r\nL 41.53125 21.204375 \r\nz\r\n\" style=\"fill:#ffffff;\"/>\r\n   </g>\r\n   <g id=\"matplotlib.axis_1\">\r\n    <g id=\"xtick_1\">\r\n     <g id=\"line2d_1\">\r\n      <defs>\r\n       <path d=\"M 0 0 \r\nL 0 3.5 \r\n\" id=\"md5b79ef06f\" style=\"stroke:#000000;stroke-width:0.8;\"/>\r\n      </defs>\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"59.285795\" xlink:href=\"#md5b79ef06f\" y=\"238.644375\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_1\">\r\n      <!-- 0 -->\r\n      <defs>\r\n       <path d=\"M 38.578125 54.390625 \r\nQ 41.5 50.484375 43.25 44.1875 \r\nQ 45.015625 37.890625 45.015625 30.078125 \r\nQ 45.015625 22.171875 43.25 16.359375 \r\nQ 41.5 10.546875 38.671875 6.640625 \r\nQ 35.84375 2.734375 32.21875 0.828125 \r\nQ 28.609375 -1.078125 24.90625 -1.078125 \r\nQ 21.09375 -1.078125 17.484375 1.21875 \r\nQ 13.875 3.515625 11.078125 7.65625 \r\nQ 8.296875 11.8125 6.640625 17.671875 \r\nQ 4.984375 23.53125 4.984375 30.5625 \r\nQ 4.984375 37.703125 6.640625 43.609375 \r\nQ 8.296875 49.515625 11.078125 53.765625 \r\nQ 13.875 58.015625 17.484375 60.359375 \r\nQ 21.09375 62.703125 24.90625 62.703125 \r\nQ 32.71875 62.703125 38.578125 54.390625 \r\nz\r\nM 34.859375 48.53125 \r\nQ 32.90625 52.546875 30.3125 54.59375 \r\nQ 27.734375 56.640625 25 56.640625 \r\nQ 22.46875 56.640625 20.015625 54.828125 \r\nQ 17.578125 53.03125 15.671875 49.796875 \r\nQ 13.765625 46.578125 12.640625 41.984375 \r\nQ 11.53125 37.40625 11.53125 31.9375 \r\nQ 11.53125 25.78125 12.890625 20.125 \r\nz\r\nM 37.203125 41.796875 \r\nL 15.234375 13.765625 \r\nQ 17.671875 9.28125 20.265625 7.28125 \r\nQ 22.859375 5.28125 25.484375 5.28125 \r\nQ 28.21875 5.28125 30.5625 6.9375 \r\nQ 32.90625 8.59375 34.65625 11.71875 \r\nQ 36.421875 14.84375 37.453125 19.28125 \r\nQ 38.484375 23.734375 38.484375 29.296875 \r\nQ 38.484375 32.8125 38.1875 35.890625 \r\nQ 37.890625 38.96875 37.203125 41.796875 \r\nz\r\n\" id=\"MyricaMM-48\"/>\r\n      </defs>\r\n      <g transform=\"translate(56.785795 252.305312)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-48\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"xtick_2\">\r\n     <g id=\"line2d_2\">\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"118.566415\" xlink:href=\"#md5b79ef06f\" y=\"238.644375\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_2\">\r\n      <!-- 100 -->\r\n      <defs>\r\n       <path d=\"M 29.890625 62.40625 \r\nL 29.890625 0 \r\nL 23.046875 0 \r\nL 23.046875 53.71875 \r\nL 10.359375 50 \r\nL 8.796875 53.90625 \r\nL 25.09375 62.40625 \r\nz\r\n\" id=\"MyricaMM-49\"/>\r\n      </defs>\r\n      <g transform=\"translate(111.066415 252.305312)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-49\"/>\r\n       <use x=\"50\" xlink:href=\"#MyricaMM-48\"/>\r\n       <use x=\"100\" xlink:href=\"#MyricaMM-48\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"xtick_3\">\r\n     <g id=\"line2d_3\">\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"177.847034\" xlink:href=\"#md5b79ef06f\" y=\"238.644375\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_3\">\r\n      <!-- 200 -->\r\n      <defs>\r\n       <path d=\"M 7.8125 52.640625 \r\nQ 10.546875 57.421875 15.1875 60.0625 \r\nQ 19.828125 62.703125 25.203125 62.703125 \r\nQ 28.90625 62.703125 32.21875 61.328125 \r\nQ 35.546875 59.96875 38.03125 57.5625 \r\nQ 40.53125 55.171875 41.890625 52 \r\nQ 43.265625 48.828125 43.265625 45.21875 \r\nQ 43.265625 42.28125 42.4375 39.6875 \r\nQ 41.609375 37.109375 40.1875 34.765625 \r\nQ 38.765625 32.421875 36.90625 30.265625 \r\nQ 35.0625 28.125 33.109375 26.265625 \r\nL 27.15625 20.703125 \r\nQ 25.875 19.53125 24.21875 17.875 \r\nQ 22.5625 16.21875 21.046875 14.3125 \r\nQ 19.53125 12.40625 18.015625 10.296875 \r\nQ 16.5 8.203125 15.4375 6.0625 \r\nL 41.5 6.0625 \r\nQ 42 6.0625 42.53125 6.546875 \r\nQ 43.0625 7.03125 43.5625 7.03125 \r\nL 43.953125 7.03125 \r\nL 43.953125 0 \r\nL 7.328125 0 \r\nL 7.328125 4.390625 \r\nQ 11.03125 11.71875 15.328125 16.984375 \r\nQ 19.625 22.265625 23.640625 25.875 \r\nL 28.421875 30.28125 \r\nQ 30.953125 32.625 32.46875 34.578125 \r\nQ 33.984375 36.53125 34.90625 38.328125 \r\nQ 35.84375 40.140625 36.125 41.75 \r\nQ 36.421875 43.359375 36.421875 45.125 \r\nQ 36.421875 47.359375 35.5 49.359375 \r\nQ 34.578125 51.375 32.953125 53.03125 \r\nQ 31.34375 54.6875 29.296875 55.65625 \r\nQ 27.25 56.640625 25 56.640625 \r\nQ 22.359375 56.640625 20.3125 55.859375 \r\nQ 18.265625 55.078125 16.75 54.046875 \r\nQ 15.234375 53.03125 14.296875 51.859375 \r\nQ 13.375 50.6875 13.28125 49.8125 \r\nQ 13.09375 48.921875 12.890625 48.640625 \r\nz\r\n\" id=\"MyricaMM-50\"/>\r\n      </defs>\r\n      <g transform=\"translate(170.347034 252.305312)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-50\"/>\r\n       <use x=\"50\" xlink:href=\"#MyricaMM-48\"/>\r\n       <use x=\"100\" xlink:href=\"#MyricaMM-48\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"xtick_4\">\r\n     <g id=\"line2d_4\">\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"237.127653\" xlink:href=\"#md5b79ef06f\" y=\"238.644375\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_4\">\r\n      <!-- 300 -->\r\n      <defs>\r\n       <path d=\"M 41.109375 46.6875 \r\nQ 41.109375 42.1875 38.5625 38.46875 \r\nQ 36.03125 34.765625 31.84375 33.203125 \r\nQ 36.625 31.546875 39.546875 27.09375 \r\nQ 42.484375 22.65625 42.484375 17.09375 \r\nQ 42.484375 13.375 41.15625 10.109375 \r\nQ 39.84375 6.84375 37.25 4.296875 \r\nQ 34.671875 1.765625 31 0.34375 \r\nQ 27.34375 -1.078125 22.5625 -1.078125 \r\nQ 13.375 -1.078125 6.9375 5.953125 \r\nL 12.203125 12.3125 \r\nQ 12.59375 11.921875 12.78125 10.9375 \r\nQ 12.984375 9.96875 13.484375 9.375 \r\nL 14.9375 8.109375 \r\nQ 15.71875 7.328125 18.15625 6.296875 \r\nQ 20.609375 5.28125 23.25 5.28125 \r\nQ 25.984375 5.28125 28.21875 6.296875 \r\nQ 30.46875 7.328125 32.171875 9.078125 \r\nQ 33.890625 10.84375 34.859375 13.1875 \r\nQ 35.84375 15.53125 35.84375 18.0625 \r\nQ 35.84375 23.4375 31.78125 26.609375 \r\nQ 27.734375 29.78125 21.09375 29.78125 \r\nQ 20.3125 29.78125 19.578125 29.734375 \r\nQ 18.84375 29.6875 17.96875 29.59375 \r\nL 17.96875 35.15625 \r\nQ 22.5625 35.15625 25.6875 36.125 \r\nQ 28.8125 37.109375 30.765625 38.71875 \r\nQ 32.71875 40.328125 33.546875 42.46875 \r\nQ 34.375 44.625 34.375 47.078125 \r\nQ 34.375 48.828125 33.59375 50.53125 \r\nQ 32.8125 52.25 31.390625 53.46875 \r\nQ 29.984375 54.6875 27.984375 55.515625 \r\nQ 25.984375 56.34375 23.640625 56.34375 \r\nQ 17.1875 56.34375 12.890625 51.46875 \r\nL 8.984375 55.765625 \r\nQ 15.046875 62.5 23.921875 62.5 \r\nQ 27.546875 62.5 30.671875 61.234375 \r\nQ 33.796875 59.96875 36.078125 57.765625 \r\nQ 38.375 55.5625 39.734375 52.734375 \r\nQ 41.109375 49.90625 41.109375 46.6875 \r\nz\r\n\" id=\"MyricaMM-51\"/>\r\n      </defs>\r\n      <g transform=\"translate(229.627653 252.305312)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-51\"/>\r\n       <use x=\"50\" xlink:href=\"#MyricaMM-48\"/>\r\n       <use x=\"100\" xlink:href=\"#MyricaMM-48\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"xtick_5\">\r\n     <g id=\"line2d_5\">\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"296.408272\" xlink:href=\"#md5b79ef06f\" y=\"238.644375\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_5\">\r\n      <!-- 400 -->\r\n      <defs>\r\n       <path d=\"M 31.34375 62.40625 \r\nL 37.203125 62.40625 \r\nL 37.203125 23.4375 \r\nL 45.21875 23.4375 \r\nL 45.21875 17.1875 \r\nL 37.203125 17.1875 \r\nL 37.203125 0 \r\nL 29.890625 0 \r\nL 29.890625 17.28125 \r\nL 4.78125 17.28125 \r\nL 4.78125 22.359375 \r\nz\r\nM 29.984375 51.265625 \r\nL 11.71875 23.4375 \r\nL 29.984375 23.4375 \r\nz\r\n\" id=\"MyricaMM-52\"/>\r\n      </defs>\r\n      <g transform=\"translate(288.908272 252.305312)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-52\"/>\r\n       <use x=\"50\" xlink:href=\"#MyricaMM-48\"/>\r\n       <use x=\"100\" xlink:href=\"#MyricaMM-48\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"xtick_6\">\r\n     <g id=\"line2d_6\">\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"355.688892\" xlink:href=\"#md5b79ef06f\" y=\"238.644375\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_6\">\r\n      <!-- 500 -->\r\n      <defs>\r\n       <path d=\"M 10.640625 62.3125 \r\nL 42.1875 62.3125 \r\nL 42.1875 55.953125 \r\nL 16.5 55.953125 \r\nL 15.53125 38.09375 \r\nQ 20.40625 40.53125 25.6875 40.53125 \r\nQ 29.6875 40.53125 33.15625 39.0625 \r\nQ 36.625 37.59375 39.15625 34.859375 \r\nQ 41.703125 32.125 43.109375 28.265625 \r\nQ 44.53125 24.421875 44.53125 19.625 \r\nQ 44.53125 14.84375 43.015625 10.984375 \r\nQ 41.5 7.125 38.859375 4.390625 \r\nQ 36.234375 1.65625 32.671875 0.28125 \r\nQ 29.109375 -1.078125 24.90625 -1.078125 \r\nQ 19.234375 -1.078125 14.34375 1.40625 \r\nQ 9.46875 3.90625 6.453125 8.5 \r\nL 12.59375 13.09375 \r\nQ 13.09375 12.796875 13.140625 11.90625 \r\nQ 13.1875 11.03125 13.28125 10.84375 \r\nQ 13.671875 10.15625 14.6875 9.125 \r\nQ 15.71875 8.109375 17.28125 7.21875 \r\nQ 18.84375 6.34375 20.84375 5.65625 \r\nQ 22.859375 4.984375 25.296875 4.984375 \r\nQ 27.734375 4.984375 29.921875 6.046875 \r\nQ 32.125 7.125 33.734375 9.03125 \r\nQ 35.359375 10.9375 36.375 13.671875 \r\nQ 37.40625 16.40625 37.40625 19.921875 \r\nQ 37.40625 23.4375 36.375 26.171875 \r\nQ 35.359375 28.90625 33.640625 30.8125 \r\nQ 31.9375 32.71875 29.640625 33.6875 \r\nQ 27.34375 34.671875 24.609375 34.671875 \r\nQ 21.390625 34.671875 18.3125 33.15625 \r\nQ 15.234375 31.640625 12.890625 28.8125 \r\nL 8.5 30.5625 \r\nz\r\n\" id=\"MyricaMM-53\"/>\r\n      </defs>\r\n      <g transform=\"translate(348.188892 252.305312)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-53\"/>\r\n       <use x=\"50\" xlink:href=\"#MyricaMM-48\"/>\r\n       <use x=\"100\" xlink:href=\"#MyricaMM-48\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"xtick_7\">\r\n     <g id=\"line2d_7\">\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"414.969511\" xlink:href=\"#md5b79ef06f\" y=\"238.644375\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_7\">\r\n      <!-- 600 -->\r\n      <defs>\r\n       <path d=\"M 28.71875 62.984375 \r\nQ 36.53125 62.984375 42.390625 57.328125 \r\nL 37.5 51.953125 \r\nQ 36.625 52.734375 36.328125 53.703125 \r\nQ 36.03125 54.6875 34.859375 55.078125 \r\nQ 33.40625 56.0625 31.84375 56.453125 \r\nQ 30.28125 56.84375 28.90625 56.84375 \r\nQ 26.953125 56.84375 24.265625 55.859375 \r\nQ 21.578125 54.890625 19.28125 52.296875 \r\nQ 17 49.703125 15.328125 44.96875 \r\nQ 13.671875 40.234375 13.578125 32.71875 \r\nQ 15.71875 36.234375 19.234375 38.234375 \r\nQ 22.75 40.234375 26.765625 40.234375 \r\nQ 30.28125 40.234375 33.40625 38.765625 \r\nQ 36.53125 37.3125 38.875 34.625 \r\nQ 41.21875 31.9375 42.578125 28.171875 \r\nQ 43.953125 24.421875 43.953125 19.828125 \r\nQ 43.953125 15.234375 42.53125 11.375 \r\nQ 41.109375 7.515625 38.71875 4.734375 \r\nQ 36.328125 1.953125 33.046875 0.484375 \r\nQ 29.78125 -0.984375 26.078125 -0.984375 \r\nQ 21.390625 -0.984375 17.765625 1.0625 \r\nQ 14.15625 3.125 11.71875 6.9375 \r\nQ 9.28125 10.75 8 16.15625 \r\nQ 6.734375 21.578125 6.734375 28.421875 \r\nQ 6.734375 36.8125 8.25 43.25 \r\nQ 9.765625 49.703125 12.6875 54.09375 \r\nQ 15.625 58.5 19.671875 60.734375 \r\nQ 23.734375 62.984375 28.71875 62.984375 \r\nz\r\nM 13.96875 25.6875 \r\nQ 13.875 25.09375 13.875 24.5 \r\nQ 13.875 23.921875 13.875 23.34375 \r\nQ 13.875 19.4375 14.84375 16.109375 \r\nQ 15.828125 12.796875 17.53125 10.40625 \r\nQ 19.234375 8.015625 21.53125 6.59375 \r\nQ 23.828125 5.171875 26.375 5.171875 \r\nQ 30.671875 5.171875 33.890625 8.875 \r\nQ 37.109375 12.59375 37.109375 19.734375 \r\nQ 37.109375 23.53125 36.078125 26.21875 \r\nQ 35.0625 28.90625 33.546875 30.65625 \r\nQ 32.03125 32.421875 30.078125 33.25 \r\nQ 28.125 34.078125 26.171875 34.078125 \r\nQ 22.5625 34.078125 19.234375 31.640625 \r\nQ 15.921875 29.203125 13.96875 25.6875 \r\nz\r\n\" id=\"MyricaMM-54\"/>\r\n      </defs>\r\n      <g transform=\"translate(407.469511 252.305312)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-54\"/>\r\n       <use x=\"50\" xlink:href=\"#MyricaMM-48\"/>\r\n       <use x=\"100\" xlink:href=\"#MyricaMM-48\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"text_8\">\r\n     <!-- episode -->\r\n     <defs>\r\n      <path d=\"M 25.875 46.78125 \r\nQ 29.5 46.78125 32.90625 45.5 \r\nQ 36.328125 44.234375 38.859375 41.546875 \r\nQ 41.40625 38.875 42.96875 34.765625 \r\nQ 44.53125 30.671875 44.53125 25.09375 \r\nQ 44.53125 24.3125 44.484375 23.4375 \r\nQ 44.4375 22.5625 44.34375 21.78125 \r\nL 11.8125 21.78125 \r\nQ 12.015625 17 13.46875 13.765625 \r\nQ 14.9375 10.546875 17.140625 8.546875 \r\nQ 19.34375 6.546875 22.015625 5.65625 \r\nQ 24.703125 4.78125 27.4375 4.78125 \r\nQ 34.578125 4.78125 39.15625 9.859375 \r\nL 43.171875 5.953125 \r\nQ 37.40625 -1.078125 27.25 -0.984375 \r\nQ 22.265625 -0.984375 18.0625 0.578125 \r\nQ 13.875 2.15625 10.984375 5.125 \r\nQ 8.109375 8.109375 6.5 12.546875 \r\nQ 4.890625 17 4.890625 22.65625 \r\nQ 4.890625 28.421875 6.546875 32.953125 \r\nQ 8.203125 37.5 11.03125 40.53125 \r\nQ 13.875 43.5625 17.671875 45.171875 \r\nQ 21.484375 46.78125 25.875 46.78125 \r\nz\r\nM 12.109375 27.4375 \r\nL 37.203125 27.4375 \r\nQ 37.203125 27.828125 37.25 28.171875 \r\nQ 37.3125 28.515625 37.3125 29 \r\nQ 37.3125 31.546875 36.328125 33.78125 \r\nQ 35.359375 36.03125 33.796875 37.640625 \r\nQ 32.234375 39.265625 30.078125 40.28125 \r\nQ 27.9375 41.3125 25.390625 41.3125 \r\nQ 20.703125 41.3125 16.890625 37.984375 \r\nQ 13.09375 34.671875 12.109375 27.4375 \r\nz\r\n\" id=\"MyricaMM-101\"/>\r\n      <path d=\"M 5.953125 45.703125 \r\nL 13.28125 45.703125 \r\nL 13.28125 38.671875 \r\nQ 15.625 42.390625 19.484375 44.53125 \r\nQ 23.34375 46.6875 27.25 46.6875 \r\nQ 31.0625 46.6875 34.46875 45.265625 \r\nQ 37.890625 43.84375 40.53125 40.90625 \r\nQ 43.171875 37.984375 44.734375 33.59375 \r\nQ 46.296875 29.203125 46.296875 23.25 \r\nQ 46.296875 17.28125 44.625 12.640625 \r\nQ 42.96875 8.015625 40.28125 4.9375 \r\nQ 37.59375 1.859375 34.125 0.34375 \r\nQ 30.671875 -1.171875 27.15625 -1.171875 \r\nQ 22.953125 -1.171875 19.328125 0.734375 \r\nQ 15.71875 2.640625 13.375 5.953125 \r\nL 13.375 -16.703125 \r\nL 5.953125 -16.703125 \r\nz\r\nM 13.28125 19.046875 \r\nQ 13.484375 11.625 16.703125 8.390625 \r\nQ 19.921875 5.171875 24.609375 5.171875 \r\nQ 27.15625 5.171875 29.640625 6 \r\nQ 32.125 6.84375 34.21875 8.796875 \r\nQ 36.328125 10.75 37.546875 13.921875 \r\nQ 38.765625 17.09375 38.765625 21.875 \r\nQ 38.765625 30.671875 35.15625 35.546875 \r\nQ 31.546875 40.4375 24.90625 40.625 \r\nQ 22.953125 40.625 20.796875 39.890625 \r\nQ 18.65625 39.15625 17.046875 37.5 \r\nQ 15.4375 35.84375 14.359375 32.953125 \r\nQ 13.28125 30.078125 13.28125 25.875 \r\nz\r\n\" id=\"MyricaMM-112\"/>\r\n      <path d=\"M 10.640625 45.703125 \r\nL 28.90625 45.703125 \r\nL 28.90625 5.765625 \r\nL 39.65625 5.765625 \r\nL 39.65625 0 \r\nL 9.96875 0 \r\nL 9.96875 5.765625 \r\nL 21.6875 5.765625 \r\nL 21.6875 39.9375 \r\nL 10.640625 39.9375 \r\nz\r\nM 21.78125 64.0625 \r\nQ 23.25 65.53125 25.390625 65.625 \r\nQ 27.546875 65.625 29.046875 64.0625 \r\nQ 30.5625 62.5 30.5625 60.359375 \r\nQ 30.5625 58.203125 29.046875 56.734375 \r\nQ 27.546875 55.28125 25.390625 55.28125 \r\nQ 23.25 55.28125 21.78125 56.78125 \r\nQ 20.3125 58.296875 20.3125 60.359375 \r\nQ 20.3125 62.59375 21.78125 64.0625 \r\nz\r\n\" id=\"MyricaMM-105\"/>\r\n      <path d=\"M 43.171875 39.65625 \r\nL 38.96875 33.296875 \r\nQ 38.484375 33.40625 38.484375 34.078125 \r\nQ 38.484375 34.765625 38.28125 35.0625 \r\nQ 36.53125 37.40625 33.203125 39.359375 \r\nQ 29.890625 41.3125 25.6875 41.3125 \r\nQ 23.828125 41.3125 22.0625 40.875 \r\nQ 20.3125 40.4375 18.9375 39.59375 \r\nQ 17.578125 38.765625 16.796875 37.546875 \r\nQ 16.015625 36.328125 16.015625 34.671875 \r\nQ 16.015625 33.5 16.59375 32.5625 \r\nQ 17.1875 31.640625 18.5 30.703125 \r\nQ 19.828125 29.78125 22.171875 28.90625 \r\nQ 24.515625 28.03125 28.328125 26.859375 \r\nQ 36.921875 24.515625 40.484375 21.390625 \r\nQ 44.046875 18.265625 44.046875 13.671875 \r\nQ 44.046875 10.640625 42.71875 7.953125 \r\nQ 41.40625 5.28125 38.90625 3.265625 \r\nQ 36.421875 1.265625 32.90625 0.140625 \r\nQ 29.390625 -0.984375 25 -0.984375 \r\nQ 13.484375 -0.984375 5.671875 6.15625 \r\nL 9.765625 13.375 \r\nQ 10.0625 13.09375 10.15625 12.453125 \r\nQ 10.25 11.8125 10.890625 10.890625 \r\nQ 11.53125 9.96875 13.140625 8.84375 \r\nQ 14.75 7.71875 17.96875 6.34375 \r\nQ 19.4375 5.671875 21.390625 5.328125 \r\nQ 23.34375 4.984375 25.203125 4.984375 \r\nQ 27.15625 4.984375 29.25 5.46875 \r\nQ 31.34375 5.953125 32.953125 6.921875 \r\nQ 34.578125 7.90625 35.59375 9.265625 \r\nQ 36.625 10.640625 36.625 12.3125 \r\nQ 36.625 14.9375 34.234375 16.6875 \r\nQ 31.84375 18.453125 25.390625 20.40625 \r\nQ 21 21.6875 19.625 22.171875 \r\nQ 16.015625 23.53125 14.40625 24.609375 \r\nQ 12.796875 25.6875 11.46875 27.140625 \r\nQ 10.15625 28.609375 9.375 30.46875 \r\nQ 8.59375 32.328125 8.59375 34.46875 \r\nQ 8.59375 36.921875 9.90625 39.109375 \r\nQ 11.234375 41.3125 13.578125 43.015625 \r\nQ 15.921875 44.734375 19.140625 45.75 \r\nQ 22.359375 46.78125 26.171875 46.78125 \r\nQ 36.625 46.875 43.171875 39.65625 \r\nz\r\n\" id=\"MyricaMM-115\"/>\r\n      <path d=\"M 25.390625 46.6875 \r\nQ 29.59375 46.6875 33.34375 45.015625 \r\nQ 37.109375 43.359375 39.890625 40.28125 \r\nQ 42.671875 37.203125 44.328125 32.765625 \r\nQ 46 28.328125 46 22.75 \r\nQ 46 17.1875 44.390625 12.796875 \r\nQ 42.78125 8.40625 39.984375 5.28125 \r\nQ 37.203125 2.15625 33.390625 0.484375 \r\nQ 29.59375 -1.171875 25.296875 -1.171875 \r\nQ 20.90625 -1.171875 17.046875 0.625 \r\nQ 13.1875 2.4375 10.296875 5.5625 \r\nQ 7.421875 8.6875 5.703125 13.078125 \r\nQ 4 17.484375 4 22.5625 \r\nQ 4 27.640625 5.75 32.171875 \r\nQ 7.515625 36.71875 10.4375 39.84375 \r\nQ 13.375 42.96875 17.28125 44.828125 \r\nQ 21.1875 46.6875 25.390625 46.6875 \r\nz\r\nM 38.765625 22.5625 \r\nQ 38.765625 26.859375 37.640625 30.21875 \r\nQ 36.53125 33.59375 34.625 35.890625 \r\nQ 32.71875 38.1875 30.265625 39.40625 \r\nQ 27.828125 40.625 25.09375 40.625 \r\nQ 22.359375 40.625 19.921875 39.40625 \r\nQ 17.484375 38.1875 15.625 35.890625 \r\nQ 13.765625 33.59375 12.6875 30.375 \r\nQ 11.625 27.15625 11.625 23.046875 \r\nQ 11.625 19.046875 12.6875 15.71875 \r\nQ 13.765625 12.40625 15.625 10.015625 \r\nQ 17.484375 7.625 19.96875 6.296875 \r\nQ 22.46875 4.984375 25.296875 4.984375 \r\nQ 28.125 4.984375 30.5625 6.15625 \r\nQ 33.015625 7.328125 34.859375 9.5625 \r\nQ 36.71875 11.8125 37.734375 15.078125 \r\nQ 38.765625 18.359375 38.765625 22.5625 \r\nz\r\n\" id=\"MyricaMM-111\"/>\r\n      <path d=\"M 37.109375 38.96875 \r\nL 37.109375 66.5 \r\nL 44.828125 66.5 \r\nQ 44.828125 65.921875 44.328125 65.140625 \r\nQ 43.84375 64.359375 43.84375 63.671875 \r\nL 43.953125 4.203125 \r\nQ 43.953125 1.765625 44.4375 0 \r\nL 37.203125 0 \r\nQ 36.8125 1.171875 36.8125 3.421875 \r\nL 36.8125 7.328125 \r\nQ 34.578125 3.515625 30.765625 1.265625 \r\nQ 26.953125 -0.984375 23.140625 -0.984375 \r\nQ 19.34375 -0.984375 15.96875 0.484375 \r\nQ 12.59375 1.953125 10.046875 4.984375 \r\nQ 7.515625 8.015625 6 12.546875 \r\nQ 4.5 17.09375 4.5 23.34375 \r\nQ 4.5 29.6875 6.203125 34.078125 \r\nQ 7.90625 38.484375 10.640625 41.359375 \r\nQ 13.375 44.234375 16.796875 45.546875 \r\nQ 20.21875 46.875 23.734375 46.875 \r\nQ 28.515625 46.875 31.984375 44.71875 \r\nQ 35.453125 42.578125 37.109375 38.96875 \r\nz\r\nM 15.828125 37.59375 \r\nQ 11.53125 33.6875 11.53125 24.421875 \r\nQ 11.53125 15.828125 14.796875 10.546875 \r\nQ 18.0625 5.28125 24.515625 5.28125 \r\nQ 25.984375 5.28125 27.6875 5.859375 \r\nQ 29.390625 6.453125 30.90625 7.515625 \r\nQ 32.421875 8.59375 33.59375 10.109375 \r\nQ 34.765625 11.625 35.359375 13.484375 \r\nQ 36.421875 17.09375 36.53125 23.140625 \r\nQ 36.53125 26.953125 36.234375 29.484375 \r\nQ 35.9375 32.03125 35.25 33.828125 \r\nQ 34.578125 35.640625 33.59375 36.765625 \r\nQ 32.625 37.890625 31.0625 38.765625 \r\nQ 29.296875 39.84375 27.734375 40.421875 \r\nQ 26.171875 41.015625 24.8125 41.015625 \r\nQ 21.875 41.015625 19.828125 40.1875 \r\nQ 17.78125 39.359375 15.828125 37.59375 \r\nz\r\n\" id=\"MyricaMM-100\"/>\r\n     </defs>\r\n     <g transform=\"translate(219.33125 264.636562)scale(0.1 -0.1)\">\r\n      <use xlink:href=\"#MyricaMM-101\"/>\r\n      <use x=\"50\" xlink:href=\"#MyricaMM-112\"/>\r\n      <use x=\"100\" xlink:href=\"#MyricaMM-105\"/>\r\n      <use x=\"150\" xlink:href=\"#MyricaMM-115\"/>\r\n      <use x=\"200\" xlink:href=\"#MyricaMM-111\"/>\r\n      <use x=\"250\" xlink:href=\"#MyricaMM-100\"/>\r\n      <use x=\"300\" xlink:href=\"#MyricaMM-101\"/>\r\n     </g>\r\n    </g>\r\n   </g>\r\n   <g id=\"matplotlib.axis_2\">\r\n    <g id=\"ytick_1\">\r\n     <g id=\"line2d_8\">\r\n      <defs>\r\n       <path d=\"M 0 0 \r\nL -3.5 0 \r\n\" id=\"m78bdbc8a25\" style=\"stroke:#000000;stroke-width:0.8;\"/>\r\n      </defs>\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"41.53125\" xlink:href=\"#m78bdbc8a25\" y=\"220.17097\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_9\">\r\n      <!-- 25 -->\r\n      <g transform=\"translate(24.53125 223.501439)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-50\"/>\r\n       <use x=\"50\" xlink:href=\"#MyricaMM-53\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"ytick_2\">\r\n     <g id=\"line2d_9\">\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"41.53125\" xlink:href=\"#m78bdbc8a25\" y=\"193.159119\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_10\">\r\n      <!-- 50 -->\r\n      <g transform=\"translate(24.53125 196.489587)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-53\"/>\r\n       <use x=\"50\" xlink:href=\"#MyricaMM-48\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"ytick_3\">\r\n     <g id=\"line2d_10\">\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"41.53125\" xlink:href=\"#m78bdbc8a25\" y=\"166.147267\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_11\">\r\n      <!-- 75 -->\r\n      <defs>\r\n       <path d=\"M 43.171875 58.40625 \r\nQ 36.8125 44.046875 31.390625 29.546875 \r\nQ 25.984375 15.046875 21.390625 0 \r\nL 13.484375 0 \r\nQ 18.265625 14.265625 23.484375 28.078125 \r\nQ 28.71875 41.890625 34.46875 55.671875 \r\nL 7.90625 55.671875 \r\nL 7.90625 62.3125 \r\nL 43.171875 62.3125 \r\nz\r\n\" id=\"MyricaMM-55\"/>\r\n      </defs>\r\n      <g transform=\"translate(24.53125 169.477736)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-55\"/>\r\n       <use x=\"50\" xlink:href=\"#MyricaMM-53\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"ytick_4\">\r\n     <g id=\"line2d_11\">\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"41.53125\" xlink:href=\"#m78bdbc8a25\" y=\"139.135416\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_12\">\r\n      <!-- 100 -->\r\n      <g transform=\"translate(19.53125 142.465885)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-49\"/>\r\n       <use x=\"50\" xlink:href=\"#MyricaMM-48\"/>\r\n       <use x=\"100\" xlink:href=\"#MyricaMM-48\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"ytick_5\">\r\n     <g id=\"line2d_12\">\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"41.53125\" xlink:href=\"#m78bdbc8a25\" y=\"112.123565\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_13\">\r\n      <!-- 125 -->\r\n      <g transform=\"translate(19.53125 115.454034)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-49\"/>\r\n       <use x=\"50\" xlink:href=\"#MyricaMM-50\"/>\r\n       <use x=\"100\" xlink:href=\"#MyricaMM-53\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"ytick_6\">\r\n     <g id=\"line2d_13\">\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"41.53125\" xlink:href=\"#m78bdbc8a25\" y=\"85.111714\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_14\">\r\n      <!-- 150 -->\r\n      <g transform=\"translate(19.53125 88.442183)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-49\"/>\r\n       <use x=\"50\" xlink:href=\"#MyricaMM-53\"/>\r\n       <use x=\"100\" xlink:href=\"#MyricaMM-48\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"ytick_7\">\r\n     <g id=\"line2d_14\">\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"41.53125\" xlink:href=\"#m78bdbc8a25\" y=\"58.099863\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_15\">\r\n      <!-- 175 -->\r\n      <g transform=\"translate(19.53125 61.430331)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-49\"/>\r\n       <use x=\"50\" xlink:href=\"#MyricaMM-55\"/>\r\n       <use x=\"100\" xlink:href=\"#MyricaMM-53\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"ytick_8\">\r\n     <g id=\"line2d_15\">\r\n      <g>\r\n       <use style=\"stroke:#000000;stroke-width:0.8;\" x=\"41.53125\" xlink:href=\"#m78bdbc8a25\" y=\"31.088011\"/>\r\n      </g>\r\n     </g>\r\n     <g id=\"text_16\">\r\n      <!-- 200 -->\r\n      <g transform=\"translate(19.53125 34.41848)scale(0.1 -0.1)\">\r\n       <use xlink:href=\"#MyricaMM-50\"/>\r\n       <use x=\"50\" xlink:href=\"#MyricaMM-48\"/>\r\n       <use x=\"100\" xlink:href=\"#MyricaMM-48\"/>\r\n      </g>\r\n     </g>\r\n    </g>\r\n    <g id=\"text_17\">\r\n     <!-- reward -->\r\n     <defs>\r\n      <path d=\"M 6.546875 45.703125 \r\nL 20.21875 45.703125 \r\nL 19.921875 36.921875 \r\nQ 22.171875 41.40625 26.171875 44.046875 \r\nQ 30.171875 46.6875 35.0625 46.6875 \r\nQ 42 46.6875 46.875 41.796875 \r\nL 43.453125 34.859375 \r\nL 42.484375 36.140625 \r\nQ 39.546875 40.53125 34.578125 40.53125 \r\nQ 31.640625 40.53125 29 38.859375 \r\nQ 26.375 37.203125 24.3125 34.65625 \r\nQ 22.265625 32.125 21.09375 29.25 \r\nQ 19.921875 26.375 19.921875 23.046875 \r\nL 19.921875 5.375 \r\nL 29.5 5.375 \r\nL 29.5 -0.09375 \r\nL 6.34375 -0.09375 \r\nL 6.34375 5.375 \r\nL 12.796875 5.375 \r\nL 12.796875 39.9375 \r\nL 6.546875 39.9375 \r\nz\r\n\" id=\"MyricaMM-114\"/>\r\n      <path d=\"M 1.5625 45.703125 \r\nL 8.890625 45.703125 \r\nL 8.890625 45.3125 \r\nQ 8.890625 44.734375 8.6875 43.953125 \r\nQ 8.5 43.171875 8.5 42.484375 \r\nQ 8.5 42.09375 8.59375 41.890625 \r\nL 15.328125 8.984375 \r\nL 23.140625 41.40625 \r\nL 27.546875 41.40625 \r\nL 37.203125 9.078125 \r\nQ 38.484375 17.875 39.359375 23.578125 \r\nQ 40.234375 29.296875 40.671875 33.046875 \r\nQ 41.109375 36.8125 41.25 39.25 \r\nQ 41.40625 41.703125 41.40625 44.234375 \r\nL 41.40625 45.703125 \r\nL 48.140625 45.703125 \r\nQ 46.6875 34.078125 44.734375 22.703125 \r\nQ 42.78125 11.328125 40.53125 -0.09375 \r\nL 33.6875 -0.09375 \r\nL 25.296875 29.5 \r\nL 17.28125 -0.09375 \r\nL 10.84375 -0.09375 \r\nz\r\n\" id=\"MyricaMM-119\"/>\r\n      <path d=\"M 9.078125 40.328125 \r\nQ 15.140625 46.78125 25.390625 46.6875 \r\nQ 33.796875 46.6875 38.625 42.4375 \r\nQ 43.453125 38.1875 43.453125 28.328125 \r\nL 43.453125 0 \r\nL 36.71875 0 \r\nL 36.71875 4.984375 \r\nQ 29.390625 -1.078125 20.3125 -1.078125 \r\nQ 16.703125 -1.078125 13.859375 0 \r\nQ 11.03125 1.078125 9.03125 2.875 \r\nQ 7.03125 4.6875 5.953125 6.984375 \r\nQ 4.890625 9.28125 4.890625 11.625 \r\nQ 4.890625 18.5625 10.84375 22.515625 \r\nQ 16.796875 26.46875 28.03125 26.859375 \r\nL 36.625 26.859375 \r\nL 36.625 28.609375 \r\nQ 36.625 35.453125 33.734375 38.140625 \r\nQ 30.859375 40.828125 24.90625 40.828125 \r\nQ 17.484375 40.828125 12.59375 35.75 \r\nz\r\nM 36.921875 21.390625 \r\nL 35.15625 21.390625 \r\nQ 33.796875 21.390625 32.5625 21.4375 \r\nQ 31.34375 21.484375 30.078125 21.484375 \r\nQ 24.609375 21.484375 21.09375 20.75 \r\nQ 17.578125 20.015625 15.578125 18.703125 \r\nQ 13.578125 17.390625 12.84375 15.71875 \r\nQ 12.109375 14.0625 12.109375 12.203125 \r\nQ 12.109375 8.984375 14.890625 6.6875 \r\nQ 17.671875 4.390625 21.78125 4.390625 \r\nQ 24.8125 4.390625 27.203125 5.359375 \r\nQ 29.59375 6.34375 31.390625 7.65625 \r\nQ 33.203125 8.984375 34.28125 10.453125 \r\nQ 35.359375 11.921875 35.9375 12.984375 \r\nQ 36.921875 14.9375 36.921875 18.359375 \r\nz\r\n\" id=\"MyricaMM-97\"/>\r\n     </defs>\r\n     <g transform=\"translate(13.860938 144.924375)rotate(-90)scale(0.1 -0.1)\">\r\n      <use xlink:href=\"#MyricaMM-114\"/>\r\n      <use x=\"50\" xlink:href=\"#MyricaMM-101\"/>\r\n      <use x=\"100\" xlink:href=\"#MyricaMM-119\"/>\r\n      <use x=\"150\" xlink:href=\"#MyricaMM-97\"/>\r\n      <use x=\"200\" xlink:href=\"#MyricaMM-114\"/>\r\n      <use x=\"250\" xlink:href=\"#MyricaMM-100\"/>\r\n     </g>\r\n    </g>\r\n   </g>\r\n   <g id=\"line2d_16\">\r\n    <path clip-path=\"url(#pe82774c53d)\" d=\"M 59.285795 216.929548 \r\nL 59.878602 214.228363 \r\nL 61.064214 220.711207 \r\nL 61.65702 222.115823 \r\nL 62.249826 222.151839 \r\nL 62.842633 222.794978 \r\nL 64.028245 225.693393 \r\nL 64.621051 226.221625 \r\nL 65.213857 224.689316 \r\nL 65.806664 225.213182 \r\nL 66.992276 224.492866 \r\nL 67.585082 224.997087 \r\nL 68.177888 225.303222 \r\nL 69.363501 226.353683 \r\nL 69.956307 225.857675 \r\nL 70.549113 225.951506 \r\nL 71.141919 224.955926 \r\nL 71.734725 224.934878 \r\nL 72.327532 224.58682 \r\nL 72.920338 224.943064 \r\nL 73.513144 224.147114 \r\nL 74.10595 224.617536 \r\nL 74.698756 224.652936 \r\nL 75.291563 224.87875 \r\nL 75.884369 224.939959 \r\nL 76.477175 224.45685 \r\nL 77.069981 224.423158 \r\nL 77.662787 224.627925 \r\nL 78.255594 224.263675 \r\nL 78.8484 224.588202 \r\nL 81.219625 225.08997 \r\nL 82.998043 224.914515 \r\nL 85.369268 225.525319 \r\nL 85.962074 225.761249 \r\nL 86.55488 225.803228 \r\nL 87.147686 225.978518 \r\nL 87.740493 225.396936 \r\nL 88.333299 225.378855 \r\nL 88.926105 225.234368 \r\nL 89.518911 225.44867 \r\nL 90.111717 224.961751 \r\nL 91.890136 225.457575 \r\nL 92.482942 225.44065 \r\nL 93.075748 225.647856 \r\nL 93.668555 225.609966 \r\nL 96.039779 226.173604 \r\nL 96.632586 226.012283 \r\nL 97.225392 226.138511 \r\nL 97.818198 226.080836 \r\nL 99.00381 226.383696 \r\nL 100.189423 226.453155 \r\nL 100.782229 226.319019 \r\nL 103.74626 226.795982 \r\nL 104.339066 226.738007 \r\nL 104.931872 226.861598 \r\nL 105.524678 226.749552 \r\nL 106.117485 226.883415 \r\nL 108.488709 227.013972 \r\nL 109.081516 227.124138 \r\nL 109.674322 227.106106 \r\nL 110.267128 226.852522 \r\nL 119.752027 227.64785 \r\nL 120.344833 227.46417 \r\nL 122.123252 227.615436 \r\nL 122.716058 227.334513 \r\nL 123.90167 227.323708 \r\nL 124.494477 227.550608 \r\nL 125.680089 227.734288 \r\nL 126.865701 227.863945 \r\nL 128.051314 227.87475 \r\nL 128.64412 227.831531 \r\nL 131.015345 228.382573 \r\nL 131.608151 228.523034 \r\nL 132.200957 228.51223 \r\nL 132.793763 228.760739 \r\nL 135.164988 228.566253 \r\nL 136.3506 228.663496 \r\nL 136.943407 228.382573 \r\nL 137.536213 228.577058 \r\nL 138.129019 227.799117 \r\nL 139.314631 226.967152 \r\nL 139.907438 226.200015 \r\nL 140.500244 225.011494 \r\nL 141.09305 224.125505 \r\nL 141.685856 223.552854 \r\nL 142.278662 223.228712 \r\nL 142.871469 221.467539 \r\nL 143.464275 221.175811 \r\nL 144.057081 220.17097 \r\nL 144.649887 219.436248 \r\nL 145.8355 218.258531 \r\nL 146.428306 217.426566 \r\nL 147.021112 217.178057 \r\nL 147.613918 215.81666 \r\nL 148.206724 215.233204 \r\nL 148.79953 213.774564 \r\nL 149.392337 213.428812 \r\nL 151.170755 210.900503 \r\nL 151.763561 209.701176 \r\nL 153.54198 205.044333 \r\nL 154.134786 203.013042 \r\nL 154.727592 201.932568 \r\nL 155.320399 200.355076 \r\nL 155.913205 199.166554 \r\nL 157.098817 197.513429 \r\nL 158.28443 195.914328 \r\nL 159.470042 194.715001 \r\nL 160.062848 194.174764 \r\nL 163.619685 189.679992 \r\nL 164.212491 188.94527 \r\nL 164.805298 188.523885 \r\nL 165.99091 186.924783 \r\nL 167.176522 185.725457 \r\nL 167.769329 184.850273 \r\nL 169.547747 180.8093 \r\nL 170.73336 176.897984 \r\nL 171.326166 175.752682 \r\nL 171.918972 174.250823 \r\nL 174.290197 170.555602 \r\nL 174.883003 169.431909 \r\nL 175.475809 167.400617 \r\nL 176.661421 165.01277 \r\nL 177.847034 162.808603 \r\nL 178.43984 162.07388 \r\nL 179.032646 160.939383 \r\nL 180.811065 158.249002 \r\nL 181.996677 156.142078 \r\nL 182.589483 154.251248 \r\nL 184.367902 150.437175 \r\nL 184.960708 149.324287 \r\nL 185.553514 147.91967 \r\nL 186.146321 145.899184 \r\nL 187.331933 142.863052 \r\nL 189.110352 138.930126 \r\nL 189.703158 136.88803 \r\nL 190.295964 135.699509 \r\nL 191.481576 132.663377 \r\nL 194.445607 127.379859 \r\nL 195.63122 125.715929 \r\nL 196.816832 123.78188 \r\nL 198.002444 122.582554 \r\nL 198.595251 122.182778 \r\nL 199.188057 122.247607 \r\nL 199.780863 122.614968 \r\nL 200.373669 122.474506 \r\nL 201.559282 121.642541 \r\nL 202.152088 122.658187 \r\nL 202.744894 122.236802 \r\nL 203.3377 122.139559 \r\nL 203.930506 122.171974 \r\nL 204.523313 122.031512 \r\nL 206.301731 122.398873 \r\nL 206.894537 123.176815 \r\nL 207.487343 123.457738 \r\nL 208.08015 124.268093 \r\nL 210.451374 125.575467 \r\nL 211.044181 126.483065 \r\nL 211.636987 127.649977 \r\nL 212.229793 129.119422 \r\nL 212.822599 130.016215 \r\nL 213.415405 131.593707 \r\nL 214.008212 132.090725 \r\nL 214.601018 133.311661 \r\nL 215.193824 134.132821 \r\nL 215.78663 134.456964 \r\nL 216.379436 135.062029 \r\nL 216.972243 135.86158 \r\nL 217.565049 136.023651 \r\nL 218.157855 135.937213 \r\nL 218.750661 136.153308 \r\nL 219.343467 136.185722 \r\nL 219.936274 136.391012 \r\nL 220.52908 136.877226 \r\nL 221.121886 137.028492 \r\nL 221.714692 136.931249 \r\nL 222.900304 137.32022 \r\nL 223.493111 137.374244 \r\nL 224.085917 137.277001 \r\nL 225.271529 137.557924 \r\nL 226.457142 137.309415 \r\nL 227.049948 137.331025 \r\nL 227.642754 137.774019 \r\nL 228.23556 138.411499 \r\nL 229.421173 140.259109 \r\nL 230.013979 141.166707 \r\nL 230.606785 141.307169 \r\nL 231.199591 141.836601 \r\nL 232.97801 141.091074 \r\nL 234.163622 139.221854 \r\nL 234.756428 139.211049 \r\nL 235.349235 138.508741 \r\nL 239.498878 131.831412 \r\nL 240.68449 129.80012 \r\nL 241.277296 129.097812 \r\nL 241.870103 128.957351 \r\nL 243.055715 127.585149 \r\nL 244.241327 125.780757 \r\nL 244.834134 125.186496 \r\nL 245.42694 125.186496 \r\nL 246.019746 124.775916 \r\nL 247.205358 123.565785 \r\nL 248.390971 122.085536 \r\nL 248.983777 122.085536 \r\nL 249.576583 121.307594 \r\nL 250.169389 120.907819 \r\nL 250.762196 120.335168 \r\nL 252.540614 117.547545 \r\nL 253.13342 116.964089 \r\nL 253.726226 117.245012 \r\nL 254.319033 116.531899 \r\nL 254.911839 116.196952 \r\nL 255.504645 116.250976 \r\nL 256.097451 115.807982 \r\nL 256.690257 115.689129 \r\nL 257.87587 114.954407 \r\nL 259.061482 114.057614 \r\nL 260.247095 113.938761 \r\nL 261.432707 112.944725 \r\nL 262.025513 111.90747 \r\nL 262.618319 112.901506 \r\nL 263.211126 113.38772 \r\nL 264.989544 110.686535 \r\nL 266.767963 107.704426 \r\nL 267.360769 106.775218 \r\nL 269.139187 102.085961 \r\nL 270.917606 97.256242 \r\nL 278.03128 78.261508 \r\nL 279.216893 75.463081 \r\nL 280.402505 72.38373 \r\nL 281.588118 69.693349 \r\nL 283.366536 65.511915 \r\nL 285.737761 60.314834 \r\nL 288.701792 54.501884 \r\nL 291.073017 50.666201 \r\nL 292.258629 49.088709 \r\nL 292.851435 48.937443 \r\nL 312.414039 48.883419 \r\nL 313.006846 47.565241 \r\nL 313.599652 47.187075 \r\nL 315.37807 45.058541 \r\nL 316.563683 43.837605 \r\nL 320.713326 38.435235 \r\nL 321.306132 38.175921 \r\nL 322.491745 34.4807 \r\nL 324.86297 32.535847 \r\nL 325.455776 32.168485 \r\nL 326.641388 31.088011 \r\nL 334.940675 31.088011 \r\nL 335.533481 31.833538 \r\nL 371.694659 31.833538 \r\nL 374.065883 38.521673 \r\nL 374.65869 39.775023 \r\nL 375.844302 39.775023 \r\nL 376.437108 40.423307 \r\nL 377.622721 43.891629 \r\nL 394.221294 43.891629 \r\nL 394.8141 43.146102 \r\nL 414.376705 43.146102 \r\nL 414.376705 43.146102 \r\n\" style=\"fill:none;stroke:#1f77b4;stroke-linecap:square;stroke-width:1.5;\"/>\r\n   </g>\r\n   <g id=\"patch_3\">\r\n    <path d=\"M 41.53125 238.644375 \r\nL 41.53125 21.204375 \r\n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\r\n   </g>\r\n   <g id=\"patch_4\">\r\n    <path d=\"M 432.13125 238.644375 \r\nL 432.13125 21.204375 \r\n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\r\n   </g>\r\n   <g id=\"patch_5\">\r\n    <path d=\"M 41.53125 238.644375 \r\nL 432.13125 238.644375 \r\n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\r\n   </g>\r\n   <g id=\"patch_6\">\r\n    <path d=\"M 41.53125 21.204375 \r\nL 432.13125 21.204375 \r\n\" style=\"fill:none;stroke:#000000;stroke-linecap:square;stroke-linejoin:miter;stroke-width:0.8;\"/>\r\n   </g>\r\n   <g id=\"text_18\">\r\n    <!-- 過去100エピソードの報酬の平均値 -->\r\n    <defs>\r\n     <path d=\"M 56.546875 44.625 \r\nL 56.546875 59.46875 \r\nL 78.125 59.46875 \r\nL 78.125 67.390625 \r\nL 43.84375 67.390625 \r\nL 43.84375 44.625 \r\nz\r\nM 60.75 44.625 \r\nL 78.125 44.625 \r\nL 78.125 55.765625 \r\nL 60.75 55.765625 \r\nz\r\nM 50.390625 15.234375 \r\nL 50.390625 9.375 \r\nL 46.296875 9.375 \r\nL 46.296875 34.46875 \r\nL 75.78125 34.46875 \r\nL 75.78125 15.234375 \r\nz\r\nM 50.390625 18.84375 \r\nL 71.875 18.84375 \r\nL 71.875 30.859375 \r\nL 50.390625 30.859375 \r\nz\r\nM 84.1875 14.75 \r\nL 84.1875 40.921875 \r\nL 37.890625 40.921875 \r\nL 37.890625 4.109375 \r\nL 33.6875 4.109375 \r\nL 33.6875 44.625 \r\nL 39.65625 44.625 \r\nL 39.65625 71.1875 \r\nL 82.515625 71.1875 \r\nL 82.515625 44.625 \r\nL 88.375 44.625 \r\nL 88.375 14.75 \r\nQ 88.375 8.203125 87.203125 6.484375 \r\nQ 86.03125 4.78125 81.25 4.78125 \r\nQ 77.640625 4.78125 71 5.28125 \r\nL 71 9.28125 \r\nQ 77.15625 8.796875 80.171875 8.796875 \r\nQ 83.109375 8.796875 83.640625 9.578125 \r\nQ 84.1875 10.359375 84.1875 14.75 \r\nz\r\nM 28.125 54.59375 \r\nL 25.09375 51.859375 \r\nQ 17.875 60.0625 10.359375 67.671875 \r\nL 13.375 70.40625 \r\nQ 21.1875 62.5 28.125 54.59375 \r\nz\r\nM 9.671875 38.765625 \r\nL 26.5625 38.765625 \r\nL 26.5625 10.640625 \r\nQ 30.375 3.71875 37.15625 1.21875 \r\nQ 43.953125 -1.265625 59.578125 -1.265625 \r\nL 89.65625 -1.265625 \r\nL 89.453125 -5.171875 \r\nL 59.375 -5.171875 \r\nQ 44.234375 -5.171875 36.5625 -2.734375 \r\nQ 28.90625 -0.296875 24.8125 6.0625 \r\nQ 19.828125 -0.296875 11.140625 -6.84375 \r\nL 8.890625 -3.21875 \r\nQ 17.390625 3.21875 22.265625 9.28125 \r\nL 22.171875 9.28125 \r\nL 22.171875 34.859375 \r\nL 9.671875 34.859375 \r\nz\r\n\" id=\"MyricaMM-36942\"/>\r\n     <path d=\"M 64.453125 23.25 \r\nL 68.265625 25.59375 \r\nQ 80.28125 10.546875 89.453125 -4.5 \r\nL 85.546875 -6.546875 \r\nQ 82.8125 -2.15625 80.078125 1.765625 \r\nQ 45.21875 -3.03125 13.484375 -4 \r\nL 13.1875 0 \r\nQ 22.171875 0.390625 26.65625 0.59375 \r\nQ 34.859375 16.40625 41.40625 34.375 \r\nL 10.453125 34.375 \r\nL 10.453125 38.375 \r\nL 47.65625 38.375 \r\nL 47.65625 58.015625 \r\nL 16.796875 58.015625 \r\nL 16.796875 62.015625 \r\nL 47.65625 62.015625 \r\nL 47.65625 73.140625 \r\nL 52.34375 73.140625 \r\nL 52.34375 62.015625 \r\nL 83.203125 62.015625 \r\nL 83.203125 58.015625 \r\nL 52.34375 58.015625 \r\nL 52.34375 38.375 \r\nL 89.546875 38.375 \r\nL 89.546875 34.375 \r\nL 46.1875 34.375 \r\nQ 40.046875 16.890625 31.84375 0.875 \r\nQ 54.203125 2.4375 77.640625 5.46875 \r\nQ 70.40625 16.015625 64.453125 23.25 \r\nz\r\n\" id=\"MyricaMM-21435\"/>\r\n     <path d=\"M 16.5 65.4375 \r\nL 83.5 65.4375 \r\nL 83.5 61.03125 \r\nL 52.4375 61.03125 \r\nL 52.4375 7.328125 \r\nL 86.328125 7.328125 \r\nL 86.328125 3.03125 \r\nL 13.671875 3.03125 \r\nL 13.671875 7.328125 \r\nL 47.5625 7.328125 \r\nL 47.5625 61.03125 \r\nL 16.5 61.03125 \r\nz\r\n\" id=\"MyricaMM-12456\"/>\r\n     <path d=\"M 25.09375 67.09375 \r\nL 25.09375 39.84375 \r\nQ 48.046875 43.5625 71.09375 50.984375 \r\nQ 67.484375 54.984375 67.484375 60.453125 \r\nQ 67.484375 66.796875 71.6875 71.09375 \r\nQ 75.984375 75.390625 82.078125 75.390625 \r\nQ 88.1875 75.390625 92.484375 71.09375 \r\nQ 96.875 66.796875 96.875 60.453125 \r\nQ 96.875 54.6875 92.484375 50.296875 \r\nQ 88.1875 46 82.125 46 \r\nQ 78.125 46 75 47.75 \r\nQ 50.59375 39.65625 25.09375 35.546875 \r\nL 25.09375 20.3125 \r\nQ 25.09375 9.96875 28.359375 6.734375 \r\nQ 31.640625 3.515625 42 3.515625 \r\nQ 63.09375 3.515625 81.9375 4.78125 \r\nL 82.234375 0.78125 \r\nQ 61.234375 -0.6875 41.796875 -0.6875 \r\nQ 29.5 -0.6875 25.140625 3.65625 \r\nQ 20.796875 8.015625 20.796875 20.40625 \r\nL 20.796875 67.09375 \r\nz\r\nM 71.296875 56.25 \r\nQ 77.546875 50 82.03125 50 \r\nQ 86.53125 50 89.703125 53.125 \r\nQ 92.875 56.25 92.875 60.453125 \r\nQ 92.875 65.234375 89.703125 68.359375 \r\nQ 86.53125 71.484375 82.03125 71.484375 \r\nQ 77.546875 71.484375 74.421875 68.359375 \r\nQ 71.296875 65.234375 71.296875 60.453125 \r\nQ 71.296875 56.25 77.546875 50 \r\nz\r\n\" id=\"MyricaMM-12500\"/>\r\n     <path d=\"M 85.84375 69.53125 \r\nQ 84.28125 46.578125 77.921875 32.5625 \r\nQ 71.578125 18.5625 59.421875 10.59375 \r\nQ 47.265625 2.640625 26.5625 -1.375 \r\nL 25.59375 2.9375 \r\nQ 45.21875 6.734375 56.59375 14.203125 \r\nQ 67.96875 21.6875 73.734375 34.859375 \r\nQ 79.5 48.046875 81.15625 69.828125 \r\nz\r\nM 14.84375 66.609375 \r\nL 19.234375 68.359375 \r\nQ 26.765625 52.546875 33.796875 36.53125 \r\nL 29.203125 34.859375 \r\nQ 21.875 52.046875 14.84375 66.609375 \r\nz\r\n\" id=\"MyricaMM-12477\"/>\r\n     <path d=\"M 15.71875 37.59375 \r\nL 17.484375 41.609375 \r\nL 35.15625 31.84375 \r\nL 83.40625 31.84375 \r\nL 83.40625 27.828125 \r\nL 33.59375 27.828125 \r\nz\r\n\" id=\"MyricaMM-12540\"/>\r\n     <path d=\"M 54.890625 67.09375 \r\nL 58.796875 69.140625 \r\nQ 64.9375 59.578125 68.953125 52.046875 \r\nL 65.046875 50.09375 \r\nQ 60.15625 58.796875 54.890625 67.09375 \r\nz\r\nM 68.359375 71.1875 \r\nL 72.265625 73.34375 \r\nQ 78.03125 64.265625 82.515625 55.859375 \r\nL 78.71875 53.90625 \r\nQ 74.90625 61.03125 68.359375 71.1875 \r\nz\r\nM 31.25 71.484375 \r\nL 31.25 43.453125 \r\nQ 57.125 37.3125 85.15625 26.765625 \r\nL 83.59375 22.46875 \r\nQ 56.15625 32.8125 31.25 38.671875 \r\nL 31.25 -4.984375 \r\nL 26.5625 -4.984375 \r\nL 26.5625 71.484375 \r\nz\r\n\" id=\"MyricaMM-12489\"/>\r\n     <path d=\"M 59.765625 2.734375 \r\nQ 71.578125 3.8125 78.359375 12.203125 \r\nQ 85.15625 20.609375 85.15625 34.28125 \r\nQ 85.15625 47.171875 76.265625 55.5625 \r\nQ 67.390625 63.96875 53.8125 63.96875 \r\nL 53.328125 63.96875 \r\nQ 47.171875 1.078125 27.640625 1.078125 \r\nQ 21.1875 1.078125 15.765625 9.078125 \r\nQ 10.359375 17.09375 10.359375 28.71875 \r\nQ 10.359375 46.09375 22.46875 57.125 \r\nQ 34.578125 68.171875 53.8125 68.171875 \r\nQ 69.4375 68.171875 79.546875 58.59375 \r\nQ 89.65625 49.03125 89.65625 34.28125 \r\nQ 89.65625 18.75 81.78125 9.125 \r\nQ 73.921875 -0.484375 60.546875 -1.5625 \r\nz\r\nM 48.828125 63.765625 \r\nQ 33.5 62.203125 24.171875 52.6875 \r\nQ 14.84375 43.171875 14.84375 28.71875 \r\nQ 14.84375 18.953125 19.046875 12.203125 \r\nQ 23.25 5.46875 27.640625 5.46875 \r\nQ 29.203125 5.46875 30.90625 6.390625 \r\nQ 32.625 7.328125 34.625 9.5625 \r\nQ 36.625 11.8125 38.578125 16.109375 \r\nQ 40.53125 20.40625 42.375 26.5625 \r\nQ 44.234375 32.71875 45.890625 42.28125 \r\nQ 47.5625 51.859375 48.828125 63.765625 \r\nz\r\n\" id=\"MyricaMM-12398\"/>\r\n     <path d=\"M 24.90625 28.609375 \r\nL 34.96875 28.609375 \r\nQ 38.1875 37.5 40.625 48.34375 \r\nL 20.609375 48.34375 \r\nQ 22.859375 39.453125 24.90625 28.609375 \r\nz\r\nM 28.125 24.8125 \r\nL 9.765625 24.8125 \r\nL 9.765625 28.609375 \r\nL 21 28.609375 \r\nQ 19.046875 38.671875 16.609375 48.34375 \r\nL 9.1875 48.34375 \r\nL 9.1875 52.15625 \r\nL 28.125 52.15625 \r\nL 28.125 63.578125 \r\nL 12.015625 63.578125 \r\nL 12.015625 67.1875 \r\nL 28.125 67.1875 \r\nL 28.125 74.515625 \r\nL 32.515625 74.515625 \r\nL 32.515625 67.1875 \r\nL 48.53125 67.1875 \r\nL 48.53125 63.578125 \r\nL 32.515625 63.578125 \r\nL 32.515625 52.15625 \r\nL 51.46875 52.15625 \r\nL 51.46875 48.34375 \r\nL 44.625 48.34375 \r\nQ 42.578125 38.1875 39.15625 28.609375 \r\nL 51.46875 28.609375 \r\nL 51.46875 24.8125 \r\nL 32.515625 24.8125 \r\nL 32.515625 12.203125 \r\nL 49.3125 12.203125 \r\nL 49.3125 8.40625 \r\nL 32.515625 8.40625 \r\nL 32.515625 -8.015625 \r\nL 28.125 -8.015625 \r\nL 28.125 8.40625 \r\nL 11.03125 8.40625 \r\nL 11.03125 12.203125 \r\nL 28.125 12.203125 \r\nz\r\nM 62.109375 36.328125 \r\nL 65.71875 37.984375 \r\nQ 68.84375 25.78125 75.875 14.546875 \r\nQ 83.296875 27.15625 85.15625 41.5 \r\nL 58.40625 41.5 \r\nL 58.40625 -7.234375 \r\nL 54.296875 -7.234375 \r\nL 54.296875 70.609375 \r\nL 90.046875 70.609375 \r\nL 90.046875 68.65625 \r\nQ 89.359375 60.640625 88.671875 57.265625 \r\nQ 87.984375 53.90625 86.90625 52.875 \r\nQ 85.84375 51.859375 83.5 51.859375 \r\nQ 79.203125 51.859375 67.671875 52.34375 \r\nL 67.671875 56.0625 \r\nQ 78.03125 55.671875 82.125 55.671875 \r\nQ 83.203125 55.671875 83.6875 56.109375 \r\nQ 84.1875 56.546875 84.671875 58.890625 \r\nQ 85.15625 61.234375 85.640625 66.796875 \r\nL 58.40625 66.796875 \r\nL 58.40625 45.40625 \r\nL 89.265625 45.40625 \r\nL 89.265625 41.5 \r\nQ 87.203125 25.203125 78.328125 10.75 \r\nQ 84.28125 2.546875 92 -3.328125 \r\nL 90.140625 -7.125 \r\nQ 82.515625 -1.5625 76.078125 7.125 \r\nQ 70.21875 -1.171875 63.09375 -7.03125 \r\nL 60.75 -3.421875 \r\nQ 67.875 2.4375 73.4375 10.75 \r\nQ 65.625 22.5625 62.109375 36.328125 \r\nz\r\n\" id=\"MyricaMM-22577\"/>\r\n     <path d=\"M 14.546875 25.296875 \r\nQ 17.875 29.203125 18.796875 33.734375 \r\nQ 19.734375 38.28125 19.734375 50.390625 \r\nL 14.546875 50.390625 \r\nz\r\nM 38.1875 2.15625 \r\nL 38.1875 12.796875 \r\nL 14.546875 12.796875 \r\nL 14.546875 2.15625 \r\nz\r\nM 19.734375 65.828125 \r\nL 8.59375 65.828125 \r\nL 8.59375 69.734375 \r\nL 43.453125 69.734375 \r\nL 43.453125 65.828125 \r\nL 32.71875 65.828125 \r\nL 32.71875 54.203125 \r\nL 41.890625 54.203125 \r\nL 41.890625 -6.640625 \r\nL 38.1875 -6.640625 \r\nL 38.1875 -1.5625 \r\nL 14.75 -1.5625 \r\nL 14.75 -6.640625 \r\nL 10.640625 -6.640625 \r\nL 10.640625 54.203125 \r\nL 19.734375 54.203125 \r\nz\r\nM 23.53125 65.828125 \r\nL 23.53125 54.203125 \r\nL 28.8125 54.203125 \r\nL 28.8125 65.828125 \r\nz\r\nM 38.1875 29 \r\nL 34.078125 29 \r\nQ 30.765625 29 29.78125 30.078125 \r\nQ 28.8125 31.15625 28.8125 35.0625 \r\nL 28.8125 50.390625 \r\nL 23.53125 50.390625 \r\nQ 23.53125 36.328125 22.265625 31.390625 \r\nQ 21 26.46875 16.109375 21 \r\nL 14.546875 22.359375 \r\nL 14.546875 16.5 \r\nL 38.1875 16.5 \r\nz\r\nM 38.1875 32.515625 \r\nL 38.1875 50.390625 \r\nL 32.71875 50.390625 \r\nL 32.71875 36.03125 \r\nQ 32.71875 33.40625 33.15625 32.953125 \r\nQ 33.59375 32.515625 35.640625 32.515625 \r\nz\r\nM 46.484375 23.25 \r\nL 42.578125 24.3125 \r\nQ 46.6875 37.5 47.75 51.265625 \r\nL 51.65625 50.59375 \r\nQ 50.59375 36.328125 46.484375 23.25 \r\nz\r\nM 53.71875 37.015625 \r\nL 53.71875 70.609375 \r\nL 58.015625 70.609375 \r\nL 58.015625 35.84375 \r\nQ 58.015625 6.453125 47.171875 -7.03125 \r\nL 43.65625 -4.203125 \r\nQ 53.71875 8.296875 53.71875 37.015625 \r\nz\r\nM 62.5 51.5625 \r\nQ 66.609375 38.765625 68.359375 26.859375 \r\nL 64.65625 25.59375 \r\nQ 62.890625 37.984375 58.890625 50 \r\nz\r\nM 69.34375 -1.265625 \r\nL 69.34375 68.65625 \r\nL 73.640625 68.65625 \r\nL 73.640625 -1.265625 \r\nz\r\nM 78.21875 51.5625 \r\nQ 82.125 39.265625 84.078125 26.859375 \r\nL 80.46875 25.59375 \r\nQ 78.8125 37.015625 74.515625 50 \r\nz\r\nM 85.0625 -6.0625 \r\nL 85.0625 70.609375 \r\nL 89.265625 70.609375 \r\nL 89.265625 -6.0625 \r\nz\r\n\" id=\"MyricaMM-37228\"/>\r\n     <path d=\"M 87.203125 65.71875 \r\nL 52.34375 65.71875 \r\nL 52.34375 29 \r\nL 90.921875 29 \r\nL 90.921875 25 \r\nL 52.34375 25 \r\nL 52.34375 -7.515625 \r\nL 47.65625 -7.515625 \r\nL 47.65625 25 \r\nL 9.078125 25 \r\nL 9.078125 29 \r\nL 47.65625 29 \r\nL 47.65625 65.71875 \r\nL 12.796875 65.71875 \r\nL 12.796875 69.734375 \r\nL 87.203125 69.734375 \r\nz\r\nM 32.71875 35.84375 \r\nQ 25.390625 48.4375 18.265625 58.296875 \r\nL 21.96875 60.453125 \r\nQ 28.8125 51.265625 36.421875 37.890625 \r\nz\r\nM 81.84375 58.5 \r\nQ 75.984375 46.578125 67.28125 35.75 \r\nL 63.671875 37.984375 \r\nQ 72.359375 48.828125 77.9375 60.359375 \r\nz\r\n\" id=\"MyricaMM-24179\"/>\r\n     <path d=\"M 23.828125 71.390625 \r\nL 23.828125 53.125 \r\nL 34.96875 53.125 \r\nL 34.96875 49.21875 \r\nL 23.828125 49.21875 \r\nL 23.828125 10.640625 \r\nQ 29.78125 13.1875 36.625 16.703125 \r\nL 37.40625 12.703125 \r\nQ 23.4375 5.46875 8.984375 0.78125 \r\nL 8.203125 4.890625 \r\nQ 14.65625 6.9375 19.34375 8.796875 \r\nL 19.34375 49.21875 \r\nL 8.59375 49.21875 \r\nL 8.59375 53.125 \r\nL 19.34375 53.125 \r\nL 19.34375 71.390625 \r\nz\r\nM 34.1875 37.5 \r\nL 31.15625 40.4375 \r\nQ 42.578125 54.59375 47.5625 74.703125 \r\nL 51.765625 74.125 \r\nQ 50.6875 69.625 48.53125 63.375 \r\nL 90.140625 63.375 \r\nQ 89.9375 51.171875 89.546875 42.234375 \r\nQ 89.15625 33.296875 88.71875 25.96875 \r\nQ 88.28125 18.65625 87.84375 13.8125 \r\nQ 87.40625 8.984375 86.671875 5.3125 \r\nQ 85.9375 1.65625 85.203125 -0.296875 \r\nQ 84.46875 -2.25 83.296875 -3.421875 \r\nQ 82.125 -4.59375 81.09375 -4.9375 \r\nQ 80.078125 -5.28125 78.21875 -5.28125 \r\nQ 71.6875 -5.28125 60.359375 -4.5 \r\nL 60.25 -0.390625 \r\nQ 69.4375 -1.171875 76.859375 -1.171875 \r\nQ 79.109375 -1.171875 80.28125 0.390625 \r\nQ 81.453125 1.953125 82.625 8.15625 \r\nQ 83.796875 14.359375 84.421875 26.265625 \r\nQ 85.0625 38.1875 85.640625 59.375 \r\nL 46.875 59.375 \r\nQ 41.5 46.296875 34.1875 37.5 \r\nz\r\nM 45.515625 41.21875 \r\nL 45.515625 45.015625 \r\nL 74.125 45.015625 \r\nL 74.125 41.21875 \r\nz\r\nM 41.609375 10.359375 \r\nQ 57.71875 15.828125 75.390625 25.59375 \r\nL 76.953125 21.96875 \r\nQ 58.984375 12.015625 42.875 6.546875 \r\nz\r\n\" id=\"MyricaMM-22343\"/>\r\n     <path d=\"M 21 49.125 \r\nL 21 -7.515625 \r\nL 16.703125 -7.515625 \r\nL 16.703125 40.328125 \r\nQ 12.59375 32.125 7.71875 25.6875 \r\nL 5.375 30.078125 \r\nQ 18.453125 48.140625 23.921875 73.34375 \r\nL 28.125 72.5625 \r\nQ 25.484375 59.671875 21 49.125 \r\nz\r\nM 50.484375 36.71875 \r\nL 50.484375 26.859375 \r\nL 81.84375 26.859375 \r\nL 81.84375 36.71875 \r\nz\r\nM 50.484375 40.328125 \r\nL 81.84375 40.328125 \r\nL 81.84375 49.90625 \r\nL 50.484375 49.90625 \r\nz\r\nM 30.859375 62.796875 \r\nL 30.859375 66.609375 \r\nL 61.421875 66.609375 \r\nQ 62.203125 71.578125 62.5 74.21875 \r\nL 66.890625 73.828125 \r\nQ 66.5 71.296875 65.828125 66.609375 \r\nL 91.703125 66.609375 \r\nL 91.703125 62.796875 \r\nL 65.234375 62.796875 \r\nQ 63.96875 55.375 63.578125 53.71875 \r\nL 86.234375 53.71875 \r\nL 86.234375 9.078125 \r\nL 50.484375 9.078125 \r\nL 46.1875 9.078125 \r\nL 46.1875 53.71875 \r\nL 58.984375 53.71875 \r\nQ 60.15625 59.078125 60.84375 62.796875 \r\nz\r\nM 81.84375 12.796875 \r\nL 81.84375 23.25 \r\nL 50.484375 23.25 \r\nL 50.484375 12.796875 \r\nz\r\nM 36.328125 51.765625 \r\nL 36.328125 1.171875 \r\nL 91.703125 1.171875 \r\nL 91.703125 -2.640625 \r\nL 36.328125 -2.640625 \r\nL 36.328125 -7.515625 \r\nL 31.9375 -7.515625 \r\nL 31.9375 51.765625 \r\nz\r\n\" id=\"MyricaMM-20516\"/>\r\n    </defs>\r\n    <g transform=\"translate(143.83125 15.204375)scale(0.12 -0.12)\">\r\n     <use xlink:href=\"#MyricaMM-36942\"/>\r\n     <use x=\"100\" xlink:href=\"#MyricaMM-21435\"/>\r\n     <use x=\"200\" xlink:href=\"#MyricaMM-49\"/>\r\n     <use x=\"250\" xlink:href=\"#MyricaMM-48\"/>\r\n     <use x=\"300\" xlink:href=\"#MyricaMM-48\"/>\r\n     <use x=\"350\" xlink:href=\"#MyricaMM-12456\"/>\r\n     <use x=\"450\" xlink:href=\"#MyricaMM-12500\"/>\r\n     <use x=\"550\" xlink:href=\"#MyricaMM-12477\"/>\r\n     <use x=\"650\" xlink:href=\"#MyricaMM-12540\"/>\r\n     <use x=\"750\" xlink:href=\"#MyricaMM-12489\"/>\r\n     <use x=\"850\" xlink:href=\"#MyricaMM-12398\"/>\r\n     <use x=\"950\" xlink:href=\"#MyricaMM-22577\"/>\r\n     <use x=\"1050\" xlink:href=\"#MyricaMM-37228\"/>\r\n     <use x=\"1150\" xlink:href=\"#MyricaMM-12398\"/>\r\n     <use x=\"1250\" xlink:href=\"#MyricaMM-24179\"/>\r\n     <use x=\"1350\" xlink:href=\"#MyricaMM-22343\"/>\r\n     <use x=\"1450\" xlink:href=\"#MyricaMM-20516\"/>\r\n    </g>\r\n   </g>\r\n  </g>\r\n </g>\r\n <defs>\r\n  <clipPath id=\"pe82774c53d\">\r\n   <rect height=\"217.44\" width=\"390.6\" x=\"41.53125\" y=\"21.204375\"/>\r\n  </clipPath>\r\n </defs>\r\n</svg>\r\n",
      "image/png": "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\n"
     },
     "metadata": {
      "needs_background": "light"
     }
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(7, 4), dpi=100)\n",
    "ax.plot(hist_aveRs)\n",
    "ax.set_title('過去100エピソードの報酬の平均値')\n",
    "ax.set_xlabel('episode')\n",
    "ax.set_ylabel('reward')\n",
    "plt.savefig('images/dqn_train.png')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "env.close()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "tags": []
   },
   "source": [
    "## テスト"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "output_type": "error",
     "ename": "NameError",
     "evalue": "name 'epsilong_greedy' is not defined",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-10-9e3213d495b8>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m\u001b[0m\n\u001b[0;32m      7\u001b[0m     \u001b[0menv\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mrender\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m      8\u001b[0m \u001b[1;33m\u001b[0m\u001b[0m\n\u001b[1;32m----> 9\u001b[1;33m     \u001b[0maction\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mepsilong_greedy\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0msample\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mpolicy_q\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mstate\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0miteration\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mlen\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mreplay_mem\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m\u001b[0;32m     10\u001b[0m     \u001b[0mn_state\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mreward\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0mdone\u001b[0m\u001b[1;33m,\u001b[0m \u001b[0m_\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0menv\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0mstep\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0maction\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0;32m     11\u001b[0m     \u001b[0mR\u001b[0m \u001b[1;33m+=\u001b[0m \u001b[0mreward\u001b[0m\u001b[1;33m\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n",
      "\u001b[1;31mNameError\u001b[0m: name 'epsilong_greedy' is not defined"
     ]
    }
   ],
   "source": [
    "state = env.reset()\n",
    "done = False\n",
    "R = 0\n",
    "timestep = 0\n",
    "\n",
    "while not done and timestep < env.spec.max_episode_steps:\n",
    "    env.render()\n",
    "\n",
    "    action = epsilong_greedy.sample(policy_q, state, iteration, len(replay_mem))\n",
    "    n_state, reward, done, _ = env.step(action)\n",
    "    R += reward\n",
    "\n",
    "    # 状態の更新\n",
    "    state = n_state\n",
    "\n",
    "env.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "# モデルの保存\n",
    "# torch.save(policy_q.state_dict(), './data/model/dqn_cartpole.pth')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "env.close()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.8.2-final"
  },
  "orig_nbformat": 2,
  "kernelspec": {
   "name": "python3",
   "display_name": "Python 3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}