Tic Tac Toe Game_Noughts and Crosses.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Tic Tac Toe Game (Noughts and Crosses)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 90,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'3.7.1'"
      ]
     },
     "execution_count": 90,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from platform import python_version\n",
    "\n",
    "from gamelearner import *\n",
    "from tictactoe import TicTacToeGame, TicTacToeExpert, \\\n",
    "                      test_player, tic_tac_toe_with_2_humans\n",
    "python_version()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Game dynamics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "TicTacToeGame()"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game = TicTacToeGame()\n",
    "game"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[1, 2]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.roles"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['X', 'O']"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.marks"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[0, 0, 0],\n",
       "       [0, 0, 0],\n",
       "       [0, 0, 0]], dtype=int8)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.state"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "game.make_move((1, (0, 2)))\n",
    "game.make_move((2, (0, 1)))\n",
    "game.make_move((1, (1, 1)))\n",
    "game.make_move((2, (2, 2)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_ O X\n",
      "_ X _\n",
      "_ _ O\n"
     ]
    }
   ],
   "source": [
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{1: 0.0}"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.get_rewards()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.check_if_game_over()\n",
    "game.game_over"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(1, (0, 2)), (2, (0, 1)), (1, (1, 1)), (2, (2, 2))]"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.moves"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.turn"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[(0, 0), (1, 0), (1, 2), (2, 0), (2, 1)]"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.available_moves()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "It is not player 2's turn.\n"
     ]
    }
   ],
   "source": [
    "try:\n",
    "    game.make_move((2, (2,0)))\n",
    "except ValueError as err:\n",
    "    print(err)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_ O X\n",
      "_ X _\n",
      "X _ O\n"
     ]
    }
   ],
   "source": [
    "game.make_move((1, (2,0)))\n",
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.game_over"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "1\n"
     ]
    }
   ],
   "source": [
    "print(game.winner)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{2: 0.0}"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.get_rewards()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [],
   "source": [
    "game.reverse_move()\n",
    "game.reverse_move()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_ O X\n",
      "_ X _\n",
      "_ _ _\n"
     ]
    }
   ],
   "source": [
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_ O X\n",
      "_ X _\n",
      "O _ _\n"
     ]
    }
   ],
   "source": [
    "game.make_move((2, (2,0)))\n",
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{1: 0.0}"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.get_rewards()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_ O X\n",
      "_ X X\n",
      "O _ _\n"
     ]
    }
   ],
   "source": [
    "game.make_move((1, (1,2)))\n",
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_ O X\n",
      "X X X\n",
      "O _ O\n"
     ]
    }
   ],
   "source": [
    "game.make_move((2, (2,2)))\n",
    "game.make_move((1, (1,0)))\n",
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(True, 1)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.game_over, game.winner"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{2: 0.0}"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.get_rewards()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Game is already over.\n"
     ]
    }
   ],
   "source": [
    "try:\n",
    "    game.make_move((2, (2,1)))\n",
    "except AssertionError as err:\n",
    "    print(err)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Playing with Human Players"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[HumanPlayer('Jill'), HumanPlayer('Jack')]"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "players = [HumanPlayer(\"Jill\"), HumanPlayer(\"Jack\")]\n",
    "players"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "TicTacToeGame()"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game = TicTacToeGame()\n",
    "game"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_ _ _\n",
      "_ _ _\n",
      "_ _ _\n",
      "Jill's turn (row, col): 0,0\n"
     ]
    }
   ],
   "source": [
    "game.show_state()\n",
    "players[0].make_move(game, role=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X _ _\n",
      "_ _ _\n",
      "_ _ _\n",
      "Jack's turn (row, col): 1,1\n"
     ]
    }
   ],
   "source": [
    "game.show_state()\n",
    "players[1].make_move(game, role=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "X _ _\n",
      "_ O _\n",
      "_ _ _\n"
     ]
    }
   ],
   "source": [
    "game.show_state()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Test the TDLearner player"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_ _ _\n",
      "_ _ _\n",
      "_ _ _\n"
     ]
    }
   ],
   "source": [
    "game = TicTacToeGame()\n",
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {},
   "outputs": [],
   "source": [
    "ex = TicTacToeExpert(\"EX\")\n",
    "td = TDLearner(\"TD\", off_policy_rate=0.0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "On-policy: True\n",
      "_ _ X\n",
      "_ _ _\n",
      "_ _ _\n"
     ]
    }
   ],
   "source": [
    "td.make_move(game, 1)  # TDLearner playing role 1\n",
    "print(\"On-policy:\", td.on_policy)\n",
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{2: 0.0}"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.get_rewards()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_ _ X\n",
      "_ O _\n",
      "_ _ _\n"
     ]
    }
   ],
   "source": [
    "ex.make_move(game, 2)\n",
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{1: 0.0}"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.get_rewards()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_ _ X\n",
      "_ O _\n",
      "X _ _\n"
     ]
    }
   ],
   "source": [
    "td.make_move(game, 1)\n",
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{2: 0.0}"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.get_rewards()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_ O X\n",
      "_ O _\n",
      "X _ _\n"
     ]
    }
   ],
   "source": [
    "ex.make_move(game, 2)\n",
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{1: 0.0}"
      ]
     },
     "execution_count": 41,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.get_rewards()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_ O X\n",
      "_ O _\n",
      "X X _\n"
     ]
    }
   ],
   "source": [
    "td.make_move(game, 1)\n",
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{2: 0.0}"
      ]
     },
     "execution_count": 43,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.get_rewards()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "metadata": {},
   "outputs": [],
   "source": [
    "while not game.game_over:\n",
    "    ex.make_move(game, 2)\n",
    "    if not game.game_over:\n",
    "        td.make_move(game, 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 45,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "O O X\n",
      "X O _\n",
      "X X O\n"
     ]
    }
   ],
   "source": [
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 46,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(True, 2)"
      ]
     },
     "execution_count": 46,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.game_over, game.winner"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 47,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{}"
      ]
     },
     "execution_count": 47,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Value function before reward update\n",
    "td.value_function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 48,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{2: 1.0, 1: 0.0}"
      ]
     },
     "execution_count": 48,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "terminal_rewards = game.get_terminal_rewards()\n",
    "terminal_rewards"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 49,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{b'-OSSO-SSO': 0.45}"
      ]
     },
     "execution_count": 49,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "td.update_terminal(game, terminal_rewards[1])\n",
    "\n",
    "# Value function after reward update\n",
    "td.value_function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 50,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[b'--S------', b'--S-O-S--', b'-OS-O-SS-', b'-OSSO-SSO']"
      ]
     },
     "execution_count": 50,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "td.saved_game_states[game]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "# If you inform td that game is over it will\n",
    "# reset its state memory\n",
    "td.gameover(game, 1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "False"
      ]
     },
     "execution_count": 52,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game in td.saved_game_states"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Simulate game against random player"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "game.reset()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_ _ _\n",
      "_ _ X\n",
      "_ _ _\n"
     ]
    }
   ],
   "source": [
    "td.make_move(game, 1)\n",
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "_ O X\n",
      "X O X\n",
      "X O O\n"
     ]
    }
   ],
   "source": [
    "while not game.game_over:\n",
    "    random_move = random.choice(game.available_moves())\n",
    "    game.make_move((2, random_move))\n",
    "    if game.game_over:\n",
    "        break\n",
    "    td.make_move(game, 1)\n",
    "    rewards = game.get_rewards()\n",
    "    if rewards.get(1, False):\n",
    "        # Rewards during game are zero for tic-tac-toe\n",
    "        # so it is not actually necessary to update\n",
    "        td.update(game, 1, rewards[1])\n",
    "game.show_state()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(True, 2)"
      ]
     },
     "execution_count": 56,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.game_over, game.winner"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{2: 1.0, 1: 0.0}"
      ]
     },
     "execution_count": 57,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "terminal_rewards = game.get_terminal_rewards()\n",
    "terminal_rewards"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{b'-OSSO-SSO': 0.45, b'-OSS-SSOO': 0.45}"
      ]
     },
     "execution_count": 58,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "td.update_terminal(game, terminal_rewards[1])\n",
    "\n",
    "# Value function after reward update\n",
    "td.value_function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{b'-OSSO-SSO': 0.45, b'-OSS-SSOO': 0.45}"
      ]
     },
     "execution_count": 59,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "td.value_function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[b'-----S---', b'---S-S-O-', b'---S-SSOO', b'-OSS-SSOO']"
      ]
     },
     "execution_count": 60,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# TD learner keeps a list of past states\n",
    "td.saved_game_states[game]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 61,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Inform TD that game is over\n",
    "td.gameover(game, 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Using the game controller"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 62,
   "metadata": {},
   "outputs": [],
   "source": [
    "game = TicTacToeGame()\n",
    "players = [TicTacToeExpert(\"EXP1\"), TicTacToeExpert(\"EXP2\")]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GameController(TicTacToeGame(), [TicTacToeExpert('EXP1'), TicTacToeExpert('EXP2')])"
      ]
     },
     "execution_count": 63,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ctrl = GameController(game, players)\n",
    "ctrl"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 64,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{TicTacToeExpert('EXP1'): 1, TicTacToeExpert('EXP2'): 2}"
      ]
     },
     "execution_count": 64,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ctrl.player_roles"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 65,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Game of Tic Tac Toe with 2 players ['EXP1', 'EXP2']\n",
      "_ _ _\n",
      "_ _ _\n",
      "_ _ _\n",
      "EXP1's turn (row, col): (1, (0, 0))\n",
      "X _ _\n",
      "_ _ _\n",
      "_ _ _\n",
      "EXP2's turn (row, col): (2, (1, 1))\n",
      "X _ _\n",
      "_ O _\n",
      "_ _ _\n",
      "EXP1's turn (row, col): (1, (2, 2))\n",
      "X _ _\n",
      "_ O _\n",
      "_ _ X\n",
      "EXP2's turn (row, col): (2, (2, 1))\n",
      "X _ _\n",
      "_ O _\n",
      "_ O X\n",
      "EXP1's turn (row, col): (1, (0, 1))\n",
      "X X _\n",
      "_ O _\n",
      "_ O X\n",
      "EXP2's turn (row, col): (2, (0, 2))\n",
      "X X O\n",
      "_ O _\n",
      "_ O X\n",
      "EXP1's turn (row, col): (1, (2, 0))\n",
      "X X O\n",
      "_ O _\n",
      "X O X\n",
      "EXP2's turn (row, col): (2, (1, 0))\n",
      "X X O\n",
      "O O _\n",
      "X O X\n",
      "EXP1's turn (row, col): (1, (1, 2))\n",
      "X X O\n",
      "O O X\n",
      "X O X\n",
      "Game over!\n",
      "Draw\n"
     ]
    }
   ],
   "source": [
    "ctrl.play()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Competition between computer players"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [],
   "source": [
    "game = TicTacToeGame()\n",
    "players = [TDLearner('TD1'), TicTacToeExpert('EXPERT')]\n",
    "ctrl = GameController(game, players)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 70,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Game of Tic Tac Toe with 2 players ['TD1', 'EXPERT']\n",
      "_ _ _\n",
      "_ _ _\n",
      "_ _ _\n",
      "TD1's turn (row, col): (1, 0)\n",
      "_ _ _\n",
      "X _ _\n",
      "_ _ _\n",
      "EXPERT's turn (row, col): (2, (1, 1))\n",
      "TD1 got 0.0 reward.\n",
      "_ _ _\n",
      "X O _\n",
      "_ _ _\n",
      "TD1's turn (row, col): (0, 0)\n",
      "X _ _\n",
      "X O _\n",
      "_ _ _\n",
      "EXPERT's turn (row, col): (2, (2, 0))\n",
      "TD1 got 0.0 reward.\n",
      "X _ _\n",
      "X O _\n",
      "O _ _\n",
      "TD1's turn (row, col): (2, 1)\n",
      "X _ _\n",
      "X O _\n",
      "O X _\n",
      "EXPERT's turn (row, col): (2, (0, 2))\n",
      "TD1 got 0.0 reward.\n",
      "TD1 got 0.0 reward.\n",
      "X _ O\n",
      "X O _\n",
      "O X _\n",
      "Game over!\n",
      "EXPERT won in 6 moves\n"
     ]
    }
   ],
   "source": [
    "ctrl.play()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Player TD1\n",
      "  played: 1, wins: 0, losses: 1\n",
      "\n",
      "Player EXPERT\n",
      "  played: 1, wins: 1, losses: 0\n"
     ]
    }
   ],
   "source": [
    "for player in players:\n",
    "    print(\"\\nPlayer %s\" % str(player.name))\n",
    "    print(\"  played: %d, wins: %d, losses: %d\" % (player.games_played, player.games_won, player.games_lost))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Train a TD Learner"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "td_learner = TDLearner('TD1')\n",
    "computer_players = [td_learner, TicTacToeExpert('EXPERT')]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Training 2 computer players...\n",
      "0 games completed\n",
      "100 games completed\n",
      "200 games completed\n",
      "300 games completed\n",
      "400 games completed\n",
      "500 games completed\n",
      "600 games completed\n",
      "700 games completed\n",
      "800 games completed\n",
      "900 games completed\n",
      "\n",
      "Results:\n",
      "TD1: won 0, lost 598, drew 402\n",
      "EXPERT: won 598, lost 0, drew 402\n"
     ]
    }
   ],
   "source": [
    "game = TicTacToeGame()\n",
    "train_computer_players(game, computer_players)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 74,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Training 2 computer players...\n",
      "0 games completed\n",
      "100 games completed\n",
      "200 games completed\n",
      "300 games completed\n",
      "400 games completed\n",
      "500 games completed\n",
      "600 games completed\n",
      "700 games completed\n",
      "800 games completed\n",
      "900 games completed\n",
      "\n",
      "Results:\n",
      "TD1: won 0, lost 246, drew 754\n",
      "EXPERT: won 246, lost 0, drew 754\n"
     ]
    }
   ],
   "source": [
    "train_computer_players(game, computer_players)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "Training 2 computer players...\n",
      "0 games completed\n",
      "100 games completed\n",
      "200 games completed\n",
      "300 games completed\n",
      "400 games completed\n",
      "500 games completed\n",
      "600 games completed\n",
      "700 games completed\n",
      "800 games completed\n",
      "900 games completed\n",
      "\n",
      "Results:\n",
      "TD1: won 0, lost 202, drew 798\n",
      "EXPERT: won 202, lost 0, drew 798\n"
     ]
    }
   ],
   "source": [
    "train_computer_players(game, computer_players)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Now play against the computer!"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 76,
   "metadata": {},
   "outputs": [],
   "source": [
    "game = TicTacToeGame()\n",
    "players = [HumanPlayer(\"You\"), td_learner]\n",
    "ctrl = GameController(game, players)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 77,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Game of Tic Tac Toe with 2 players ['You', 'TD1']\n",
      "_ _ _\n",
      "_ _ _\n",
      "_ _ _\n",
      "You's turn (row, col): 0,0\n",
      "TD1 got 0.0 reward.\n",
      "X _ _\n",
      "_ _ _\n",
      "_ _ _\n",
      "TD1's turn (row, col): (0, 2)\n",
      "X _ O\n",
      "_ _ _\n",
      "_ _ _\n",
      "You's turn (row, col): 1,1\n",
      "TD1 got 0.0 reward.\n",
      "X _ O\n",
      "_ X _\n",
      "_ _ _\n",
      "TD1's turn (row, col): (2, 2)\n",
      "X _ O\n",
      "_ X _\n",
      "_ _ O\n",
      "You's turn (row, col): 1,2\n",
      "TD1 got 0.0 reward.\n",
      "X _ O\n",
      "_ X X\n",
      "_ _ O\n",
      "TD1's turn (row, col): (1, 0)\n",
      "X _ O\n",
      "O X X\n",
      "_ _ O\n",
      "You's turn (row, col): 2,1\n",
      "TD1 got 0.0 reward.\n",
      "X _ O\n",
      "O X X\n",
      "_ X O\n",
      "TD1's turn (row, col): (2, 0)\n",
      "X _ O\n",
      "O X X\n",
      "O X O\n",
      "You's turn (row, col): 0,1\n",
      "TD1 got 0.0 reward.\n",
      "TD1 got 0.0 reward.\n",
      "You you won!\n",
      "X X O\n",
      "O X X\n",
      "O X O\n",
      "Game over!\n",
      "You won in 9 moves\n",
      "Press enter to play again or q to quit: q\n",
      "Results\n",
      "\n",
      "Player You\n",
      "  played: 1, wins: 1, losses: 0\n",
      "\n",
      "Player TD1\n",
      "  played: 3001, wins: 0, losses: 1047\n"
     ]
    }
   ],
   "source": [
    "while True:\n",
    "    game.reset()\n",
    "    ctrl.play()\n",
    "    text = input(\"Press enter to play again or 'q' to quit: \")\n",
    "    if text.strip().lower() == 'q':\n",
    "        break\n",
    "\n",
    "print(\"Results\")        \n",
    "for player in players:\n",
    "    print(\"\\nPlayer %s\" % str(player.name))\n",
    "    print(\"  played: %d, wins: %d, losses: %d\" % (player.games_played, player.games_won, player.games_lost))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## TD Parameter Optimization"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 78,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = {\n",
    "    'learning_rate': [],\n",
    "    'off_policy_rate': [],\n",
    "    'initial_values': [],\n",
    "    'Wins (1000)': [],\n",
    "    'Losses (1000)': [],\n",
    "    'Wins (2000)': [],\n",
    "    'Losses (2000)': []\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 79,
   "metadata": {},
   "outputs": [],
   "source": [
    "initial_values = [0.0, 0.5, 1.0]\n",
    "learning_rates = [0.25, 0.1]\n",
    "off_policy_rates = [0.0, 0.1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 81,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "\n",
      "lr: 0.25, opr: 0.0, v0: 0.0\n",
      "Training against expert...\n",
      "\n",
      "lr: 0.25, opr: 0.0, v0: 0.5\n",
      "Training against expert...\n",
      "\n",
      "lr: 0.25, opr: 0.0, v0: 1.0\n",
      "Training against expert...\n",
      "\n",
      "lr: 0.25, opr: 0.1, v0: 0.0\n",
      "Training against expert...\n",
      "\n",
      "lr: 0.25, opr: 0.1, v0: 0.5\n",
      "Training against expert...\n",
      "\n",
      "lr: 0.25, opr: 0.1, v0: 1.0\n",
      "Training against expert...\n",
      "\n",
      "lr: 0.1, opr: 0.0, v0: 0.0\n",
      "Training against expert...\n",
      "\n",
      "lr: 0.1, opr: 0.0, v0: 0.5\n",
      "Training against expert...\n",
      "\n",
      "lr: 0.1, opr: 0.0, v0: 1.0\n",
      "Training against expert...\n",
      "\n",
      "lr: 0.1, opr: 0.1, v0: 0.0\n",
      "Training against expert...\n",
      "\n",
      "lr: 0.1, opr: 0.1, v0: 0.5\n",
      "Training against expert...\n",
      "\n",
      "lr: 0.1, opr: 0.1, v0: 1.0\n",
      "Training against expert...\n"
     ]
    }
   ],
   "source": [
    "game = TicTacToeGame()\n",
    "for lr in learning_rates:\n",
    "    for opr in off_policy_rates:\n",
    "        for v0 in initial_values:\n",
    "            print(\"\\nlr: %s, opr: %s, v0: %s\" % (str(lr), str(opr), str(v0)))\n",
    "            print(\"Training against expert...\")\n",
    "            td_learner = TDLearner('TD', learning_rate=lr, \n",
    "                                   off_policy_rate=opr, initial_value=v0)\n",
    "            computer_players = [td_learner, TicTacToeExpert('EXPERT')]\n",
    "            train_computer_players(game, computer_players, \n",
    "                                   iterations=1000, show=False)\n",
    "            results['learning_rate'].append(lr)\n",
    "            results['off_policy_rate'].append(opr)\n",
    "            results['initial_values'].append(v0)\n",
    "            results['Wins (1000)'].append(td_learner.games_won)\n",
    "            results['Losses (1000)'].append(td_learner.games_lost)\n",
    "            train_computer_players(game, computer_players, \n",
    "                                   iterations=1000, show=False)\n",
    "            results['Wins (2000)'].append(td_learner.games_won)\n",
    "            results['Losses (2000)'].append(td_learner.games_lost)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>learning_rate</th>\n",
       "      <th>off_policy_rate</th>\n",
       "      <th>initial_values</th>\n",
       "      <th>Wins (1000)</th>\n",
       "      <th>Losses (1000)</th>\n",
       "      <th>Wins (2000)</th>\n",
       "      <th>Losses (2000)</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.25</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>613</td>\n",
       "      <td>0</td>\n",
       "      <td>1046</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.25</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0</td>\n",
       "      <td>483</td>\n",
       "      <td>0</td>\n",
       "      <td>555</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.25</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>621</td>\n",
       "      <td>0</td>\n",
       "      <td>960</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.25</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>874</td>\n",
       "      <td>0</td>\n",
       "      <td>1475</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.25</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0</td>\n",
       "      <td>597</td>\n",
       "      <td>0</td>\n",
       "      <td>824</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>0.25</td>\n",
       "      <td>0.1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>690</td>\n",
       "      <td>0</td>\n",
       "      <td>1133</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>0.10</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>843</td>\n",
       "      <td>0</td>\n",
       "      <td>1422</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>0.10</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0</td>\n",
       "      <td>493</td>\n",
       "      <td>0</td>\n",
       "      <td>550</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>0.10</td>\n",
       "      <td>0.0</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>616</td>\n",
       "      <td>0</td>\n",
       "      <td>981</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>0.10</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.0</td>\n",
       "      <td>0</td>\n",
       "      <td>845</td>\n",
       "      <td>0</td>\n",
       "      <td>1409</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>0.10</td>\n",
       "      <td>0.1</td>\n",
       "      <td>0.5</td>\n",
       "      <td>0</td>\n",
       "      <td>595</td>\n",
       "      <td>0</td>\n",
       "      <td>838</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>0.10</td>\n",
       "      <td>0.1</td>\n",
       "      <td>1.0</td>\n",
       "      <td>0</td>\n",
       "      <td>656</td>\n",
       "      <td>0</td>\n",
       "      <td>1128</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    learning_rate  off_policy_rate  initial_values  Wins (1000)  \\\n",
       "0            0.25              0.0             0.0            0   \n",
       "1            0.25              0.0             0.5            0   \n",
       "2            0.25              0.0             1.0            0   \n",
       "3            0.25              0.1             0.0            0   \n",
       "4            0.25              0.1             0.5            0   \n",
       "5            0.25              0.1             1.0            0   \n",
       "6            0.10              0.0             0.0            0   \n",
       "7            0.10              0.0             0.5            0   \n",
       "8            0.10              0.0             1.0            0   \n",
       "9            0.10              0.1             0.0            0   \n",
       "10           0.10              0.1             0.5            0   \n",
       "11           0.10              0.1             1.0            0   \n",
       "\n",
       "    Losses (1000)  Wins (2000)  Losses (2000)  \n",
       "0             613            0           1046  \n",
       "1             483            0            555  \n",
       "2             621            0            960  \n",
       "3             874            0           1475  \n",
       "4             597            0            824  \n",
       "5             690            0           1133  \n",
       "6             843            0           1422  \n",
       "7             493            0            550  \n",
       "8             616            0            981  \n",
       "9             845            0           1409  \n",
       "10            595            0            838  \n",
       "11            656            0           1128  "
      ]
     },
     "execution_count": 83,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.DataFrame(results)\n",
    "df"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'After 1000 iterations:': 1, 'After 2000 iterations:': 7}"
      ]
     },
     "execution_count": 84,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "good_results = {\n",
    "    'After 1000 iterations:': df['Losses (1000)'].idxmin(),\n",
    "    'After 2000 iterations:': df['Losses (2000)'].idxmin()\n",
    "}\n",
    "good_results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 85,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "learning_rate        0.1\n",
       "off_policy_rate      0.0\n",
       "initial_values       0.5\n",
       "Wins (1000)          0.0\n",
       "Losses (1000)      493.0\n",
       "Wins (2000)          0.0\n",
       "Losses (2000)      550.0\n",
       "Name: 7, dtype: float64"
      ]
     },
     "execution_count": 85,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_result = good_results['After 2000 iterations:']\n",
    "df.loc[best_result]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Look at learning rates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 86,
   "metadata": {},
   "outputs": [],
   "source": [
    "td_players = [TDLearner(\"TD %d\" % i) for i in range(4)]\n",
    "rd = RandomPlayer(\"Random\")\n",
    "ex = TicTacToeExpert(\"Expert\")\n",
    "\n",
    "opponents = [rd, ex, td_players[3]]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 87,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Select parameters for test\n",
    "for td in td_players:\n",
    "    td.learning_rate = df.loc[best_result]['learning_rate']\n",
    "    td.initial_values = df.loc[best_result]['initial_values']\n",
    "    td.off_policy_rate = df.loc[best_result]['off_policy_rate']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 88,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "([TDLearner('TD 0'), TDLearner('TD 1'), TDLearner('TD 2')],\n",
       " [RandomPlayer('Random'), TicTacToeExpert('Expert'), TDLearner('TD 3')])"
      ]
     },
     "execution_count": 88,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "td_players[0:3], opponents"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Start time: 2019-06-02 17:03:27.899094\n",
      "100  0.04  0.01  0.02  0.94\n",
      "200  0.01  0.06  0.02  0.96\n",
      "300  0.01  0.07  0.00  0.98\n",
      "400  0.01  0.10  0.04  0.86\n",
      "500  0.03  0.19  0.03  0.88\n",
      "600  0.05  0.29  0.03  1.00\n",
      "700  0.03  0.36  0.06  0.92\n",
      "800  0.02  0.33  0.03  0.96\n",
      "900  0.02  0.39  0.05  0.96\n",
      "1000  0.02  0.41  0.03  0.98\n",
      "1100  0.02  0.28  0.03  0.96\n",
      "1200  0.03  0.33  0.03  0.96\n",
      "1300  0.03  0.37  0.02  0.94\n",
      "1400  0.05  0.26  0.04  0.96\n",
      "1500  0.04  0.27  0.06  0.96\n",
      "1600  0.06  0.36  0.02  0.98\n",
      "1700  0.02  0.37  0.03  0.94\n",
      "1800  0.03  0.38  0.06  0.94\n",
      "1900  0.04  0.34  0.03  0.94\n",
      "2000  0.03  0.32  0.01  1.00\n",
      "2100  0.05  0.39  0.04  0.96\n",
      "2200  0.02  0.29  0.03  0.96\n",
      "2300  0.08  0.22  0.15  0.94\n",
      "2400  0.06  0.47  0.08  0.92\n",
      "2500  0.00  0.35  0.10  0.94\n",
      "2600  0.08  0.36  0.12  0.94\n",
      "2700  0.08  0.51  0.10  0.94\n",
      "2800  0.07  0.33  0.07  0.98\n",
      "2900  0.03  0.36  0.16  0.92\n",
      "3000  0.03  0.33  0.11  0.92\n",
      "3100  0.03  0.35  0.07  0.94\n",
      "3200  0.07  0.41  0.19  0.94\n",
      "3300  0.03  0.27  0.12  0.94\n",
      "3400  0.07  0.39  0.18  0.94\n",
      "3500  0.06  0.33  0.22  0.96\n",
      "3600  0.07  0.38  0.18  0.98\n",
      "3700  0.06  0.47  0.12  0.96\n",
      "3800  0.04  0.37  0.20  0.96\n",
      "3900  0.08  0.48  0.13  0.98\n",
      "4000  0.05  0.50  0.19  0.94\n",
      "4100  0.08  0.27  0.17  0.94\n",
      "4200  0.06  0.46  0.19  0.90\n",
      "4300  0.03  0.41  0.29  0.94\n",
      "4400  0.09  0.46  0.31  0.94\n",
      "4500  0.08  0.45  0.27  0.94\n",
      "4600  0.05  0.37  0.26  0.92\n",
      "4700  0.06  0.43  0.34  0.94\n",
      "4800  0.10  0.26  0.37  0.94\n",
      "4900  0.09  0.32  0.38  0.94\n",
      "5000  0.08  0.30  0.39  0.94\n",
      "5100  0.13  0.32  0.33  0.92\n",
      "5200  0.12  0.35  0.50  0.94\n",
      "5300  0.19  0.39  0.52  1.00\n",
      "5400  0.12  0.32  0.39  0.98\n",
      "5500  0.13  0.26  0.56  0.96\n",
      "5600  0.08  0.29  0.48  0.96\n",
      "5700  0.13  0.39  0.49  0.92\n",
      "5800  0.15  0.35  0.53  0.96\n",
      "5900  0.06  0.33  0.51  0.96\n",
      "6000  0.11  0.39  0.52  0.98\n",
      "6100  0.10  0.29  0.59  0.98\n",
      "6200  0.16  0.37  0.54  0.98\n",
      "6300  0.15  0.38  0.67  0.94\n",
      "6400  0.14  0.30  0.58  0.94\n",
      "6500  0.10  0.46  0.64  0.96\n",
      "6600  0.25  0.22  0.64  0.94\n",
      "6700  0.18  0.31  0.69  0.98\n",
      "6800  0.25  0.37  0.71  0.98\n",
      "6900  0.15  0.22  0.90  0.98\n",
      "7000  0.14  0.31  0.69  0.94\n",
      "7100  0.16  0.38  0.67  0.98\n",
      "7200  0.26  0.35  0.66  0.92\n",
      "7300  0.13  0.37  0.79  0.96\n",
      "7400  0.21  0.37  0.75  0.92\n",
      "7500  0.26  0.38  0.79  0.96\n",
      "7600  0.16  0.24  0.81  0.98\n",
      "7700  0.16  0.39  0.83  1.00\n",
      "7800  0.19  0.35  0.81  0.96\n",
      "7900  0.20  0.32  0.76  0.96\n",
      "8000  0.24  0.37  0.88  0.94\n",
      "8100  0.26  0.40  0.81  0.98\n",
      "8200  0.08  0.36  0.67  0.94\n",
      "8300  0.27  0.40  0.67  0.98\n",
      "8400  0.19  0.47  0.79  0.92\n",
      "8500  0.25  0.41  0.94  0.96\n",
      "8600  0.22  0.27  0.70  0.98\n",
      "8700  0.25  0.22  0.85  0.98\n",
      "8800  0.29  0.42  0.77  0.98\n",
      "8900  0.24  0.31  0.80  0.96\n",
      "9000  0.39  0.29  0.90  0.96\n",
      "9100  0.28  0.46  0.94  0.94\n",
      "9200  0.22  0.29  0.76  0.96\n",
      "9300  0.36  0.39  0.84  0.96\n",
      "9400  0.29  0.41  0.78  0.96\n",
      "9500  0.28  0.32  0.90  0.98\n",
      "9600  0.32  0.45  0.90  0.94\n",
      "9700  0.30  0.40  0.88  0.98\n",
      "9800  0.35  0.54  0.79  0.92\n",
      "9900  0.33  0.39  0.82  1.00\n",
      "10000  0.31  0.45  0.82  0.90\n",
      "End time: 0:03:18.076224\n"
     ]
    }
   ],
   "source": [
    "import datetime\n",
    "\n",
    "# This can take a long time!\n",
    "start_time = datetime.datetime.now()\n",
    "print(\"Start time:\", start_time)\n",
    "\n",
    "scores = {}\n",
    "index = []\n",
    "n_iter = 100\n",
    "\n",
    "def player_key(p1, p2):\n",
    "    return \"%s vs. %s\" % (p1.name, p2.name)\n",
    "\n",
    "for td_player, opponent in zip(td_players[0:3], opponents):\n",
    "    scores[player_key(td_player, opponent)] = [test_player(td_player)]\n",
    "scores[ex.name] = [test_player(ex)]\n",
    "\n",
    "index.append(td_player.games_played)\n",
    "\n",
    "game = TicTacToeGame()\n",
    "iteration = 0\n",
    "iterations = [0]\n",
    "\n",
    "for epoch in range(100):\n",
    "    \n",
    "    for td_player, opponent in zip(td_players[0:3], opponents):\n",
    "        train_computer_players(game, [td_player, opponent], \n",
    "                               iterations=n_iter, show=False)\n",
    "        key = \"%s vs. %s\" % (td_player.name, opponent.name)\n",
    "        scores[key].append(test_player(td_player))\n",
    "    scores[ex.name].append(test_player(ex))\n",
    "\n",
    "    index.append(td_player.games_played)\n",
    "    iteration += n_iter\n",
    "    iterations.append(iteration)\n",
    "    print(iteration, \"%5.2f %5.2f %5.2f %5.2f\" % tuple(s[-1] for s in \n",
    "                                                     scores.values()))\n",
    "\n",
    "print(\"End time:\", datetime.datetime.now() - start_time)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "pd.DataFrame(scores, index=iterations).plot()\n",
    "plt.title(\"Learning curves against different opponents\")\n",
    "plt.xlabel(\"Games played\")\n",
    "plt.ylabel(\"Score\")\n",
    "plt.ylim(0,1)\n",
    "plt.grid()\n",
    "plt.savefig(\"learning_rates.pdf\")\n",
    "plt.savefig(\"images/learning_rates.png\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>TD 0 vs. Random</th>\n",
       "      <th>TD 1 vs. Expert</th>\n",
       "      <th>TD 2 vs. TD 3</th>\n",
       "      <th>Expert</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>9600</th>\n",
       "      <td>0.322560</td>\n",
       "      <td>0.446400</td>\n",
       "      <td>0.900000</td>\n",
       "      <td>0.94</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9700</th>\n",
       "      <td>0.301760</td>\n",
       "      <td>0.403200</td>\n",
       "      <td>0.880000</td>\n",
       "      <td>0.98</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9800</th>\n",
       "      <td>0.346368</td>\n",
       "      <td>0.536256</td>\n",
       "      <td>0.787528</td>\n",
       "      <td>0.92</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9900</th>\n",
       "      <td>0.330096</td>\n",
       "      <td>0.388800</td>\n",
       "      <td>0.820000</td>\n",
       "      <td>1.00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10000</th>\n",
       "      <td>0.306544</td>\n",
       "      <td>0.448000</td>\n",
       "      <td>0.823200</td>\n",
       "      <td>0.90</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       TD 0 vs. Random  TD 1 vs. Expert  TD 2 vs. TD 3  Expert\n",
       "9600          0.322560         0.446400       0.900000    0.94\n",
       "9700          0.301760         0.403200       0.880000    0.98\n",
       "9800          0.346368         0.536256       0.787528    0.92\n",
       "9900          0.330096         0.388800       0.820000    1.00\n",
       "10000         0.306544         0.448000       0.823200    0.90"
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.DataFrame(scores, index=index).tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[10000, 10000, 10000, 10000]"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "[p.games_played for p in td_players]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Games lost against expert: 3\n"
     ]
    }
   ],
   "source": [
    "games_lost = td_players[1].games_lost\n",
    "players = [td_players[1], TicTacToeExpert(\"EXPERT\")]\n",
    "train_computer_players(game, players, iterations=1000, show=False)\n",
    "games_lost = td_players[1].games_lost - games_lost\n",
    "print(\"Games lost against expert:\", games_lost)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "909"
      ]
     },
     "execution_count": 97,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(td_players[1].value_function)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>Mean</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>b'OOSSOOSSS'</th>\n",
       "      <td>0.950761</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.734279</td>\n",
       "      <td>0.760852</td>\n",
       "      <td>0.815298</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b'OSOSSSOOS'</th>\n",
       "      <td>0.858785</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.760852</td>\n",
       "      <td>0.760852</td>\n",
       "      <td>0.793496</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b'SOOOSSOSS'</th>\n",
       "      <td>0.907349</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.734279</td>\n",
       "      <td>0.734279</td>\n",
       "      <td>0.791969</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b'OOSSSSOSO'</th>\n",
       "      <td>0.872907</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.760852</td>\n",
       "      <td>0.734279</td>\n",
       "      <td>0.789346</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b'OOSSSOSSO'</th>\n",
       "      <td>0.858785</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.760852</td>\n",
       "      <td>0.734279</td>\n",
       "      <td>0.784639</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b'SOOSSSOSO'</th>\n",
       "      <td>0.858785</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.734279</td>\n",
       "      <td>0.760852</td>\n",
       "      <td>0.784639</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b'SSSSOOOOS'</th>\n",
       "      <td>0.907349</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.734279</td>\n",
       "      <td>0.704755</td>\n",
       "      <td>0.782128</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b'SSSOOSSOO'</th>\n",
       "      <td>0.924953</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.704755</td>\n",
       "      <td>0.704755</td>\n",
       "      <td>0.778154</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b'OSOSSOOSS'</th>\n",
       "      <td>0.843095</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.704755</td>\n",
       "      <td>0.784766</td>\n",
       "      <td>0.777539</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b'SOSSOOSSO'</th>\n",
       "      <td>0.907349</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.704755</td>\n",
       "      <td>0.704755</td>\n",
       "      <td>0.772286</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                     0   1         2         3      Mean\n",
       "b'OOSSOOSSS'  0.950761 NaN  0.734279  0.760852  0.815298\n",
       "b'OSOSSSOOS'  0.858785 NaN  0.760852  0.760852  0.793496\n",
       "b'SOOOSSOSS'  0.907349 NaN  0.734279  0.734279  0.791969\n",
       "b'OOSSSSOSO'  0.872907 NaN  0.760852  0.734279  0.789346\n",
       "b'OOSSSOSSO'  0.858785 NaN  0.760852  0.734279  0.784639\n",
       "b'SOOSSSOSO'  0.858785 NaN  0.734279  0.760852  0.784639\n",
       "b'SSSSOOOOS'  0.907349 NaN  0.734279  0.704755  0.782128\n",
       "b'SSSOOSSOO'  0.924953 NaN  0.704755  0.704755  0.778154\n",
       "b'OSOSSOOSS'  0.843095 NaN  0.704755  0.784766  0.777539\n",
       "b'SOSSOOSSO'  0.907349 NaN  0.704755  0.704755  0.772286"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "combined_values = pd.concat(\n",
    "    [pd.Series(td.value_function) for td in td_players], \n",
    "    axis=1, \n",
    "    sort=True\n",
    ")\n",
    "combined_values['Mean'] = combined_values.mean(axis=1)\n",
    "combined_values = combined_values.sort_values(by='Mean', ascending=False)\n",
    "combined_values.head(10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>Mean</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>b'SOS-OOOSS'</th>\n",
       "      <td>0.36450</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.36450</td>\n",
       "      <td>0.4050</td>\n",
       "      <td>0.37800</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b'-OSSOSOSO'</th>\n",
       "      <td>0.36450</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.40500</td>\n",
       "      <td>0.3645</td>\n",
       "      <td>0.37800</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b'SSOO-SOOS'</th>\n",
       "      <td>0.36450</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.36450</td>\n",
       "      <td>0.3645</td>\n",
       "      <td>0.36450</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b'OSSSOOSO-'</th>\n",
       "      <td>0.32805</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.36450</td>\n",
       "      <td>0.3645</td>\n",
       "      <td>0.35235</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>b'SOSOO-SSO'</th>\n",
       "      <td>0.36450</td>\n",
       "      <td>NaN</td>\n",
       "      <td>0.32805</td>\n",
       "      <td>0.3645</td>\n",
       "      <td>0.35235</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                    0   1        2       3     Mean\n",
       "b'SOS-OOOSS'  0.36450 NaN  0.36450  0.4050  0.37800\n",
       "b'-OSSOSOSO'  0.36450 NaN  0.40500  0.3645  0.37800\n",
       "b'SSOO-SOOS'  0.36450 NaN  0.36450  0.3645  0.36450\n",
       "b'OSSSOOSO-'  0.32805 NaN  0.36450  0.3645  0.35235\n",
       "b'SOSOO-SSO'  0.36450 NaN  0.32805  0.3645  0.35235"
      ]
     },
     "execution_count": 99,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "combined_values.tail()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "bins = np.linspace(0.0, 1.0, 21)\n",
    "for td in td_players:\n",
    "    h = pd.Series(td.value_function).hist(bins=bins)\n",
    "    plt.xlim(0, 1.0)\n",
    "    plt.title(\"Distribution of state-values - %s\" % td.name)\n",
    "    plt.xlabel(\"Value\")\n",
    "    plt.ylabel(\"Frequency\")\n",
    "    plt.savefig(\"images/values_%s_hist.png\" % td.name)\n",
    "    plt.savefig(\"images/values_%s_hist.pdf\" % td.name)\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "number_of_values = pd.Series(\n",
    "    [len(td.value_function) for td in td_players],\n",
    "    index=pd.Index([td.name for td in td_players], name='Player')\n",
    ")\n",
    "\n",
    "number_of_values.plot.bar()\n",
    "plt.title(\"Number of values stored in TD value functions\")\n",
    "plt.grid()\n",
    "plt.savefig(\"number_of_values.pdf\")\n",
    "plt.savefig(\"images/number_of_values.png\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Player\n",
       "TD 0    4219\n",
       "TD 1     909\n",
       "TD 2    4630\n",
       "TD 3    4678\n",
       "dtype: int64"
      ]
     },
     "execution_count": 102,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "number_of_values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.8062897554999999,\n",
       " 0.9324574141163504,\n",
       " 0.8970544339526756,\n",
       " 0.843094701955,\n",
       " 0.87290670858355,\n",
       " 0.9392116727047154,\n",
       " 0.9324574141163504,\n",
       " 0.8970544339526756,\n",
       " 0.9676945905538666,\n",
       " 0.885616037725195,\n",
       " 0.9166140915016672,\n",
       " 0.9601167784615637,\n",
       " 0.704755)"
      ]
     },
     "execution_count": 103,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "v = td_players[0].value_function\n",
    "\n",
    "# These should be high values\n",
    "(v[b'SSS---OO-'], v[b'SSS----OO'], v[b'SSSOO----'], v[b'SSS-OO---'], \n",
    " v[b'SSS---O-O'], v[b'SSSO-O---'], v[b'SSSO--O--'], v[b'SSS-O--O-'],\n",
    " v[b'SSS--O--O'], v[b'SSSO---O-'], v[b'SSSO----O'], v[b'SSS-O-O--'],\n",
    " v[b'SSS-O---O'])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Prepare some tests"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [],
   "source": [
    "from importlib import reload\n",
    "import gamelearner, tictactoe\n",
    "reload(gamelearner)\n",
    "reload(tictactoe)\n",
    "TDLearner = gamelearner.TDLearner\n",
    "TicTacToeGame = tictactoe.TicTacToeGame\n",
    "GameController = gamelearner.GameController\n",
    "RandomPlayer = gamelearner.RandomPlayer"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [],
   "source": [
    "game = TicTacToeGame()\n",
    "roles = game.roles"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {},
   "outputs": [],
   "source": [
    "learning_rate=0.25\n",
    "gamma=1.0\n",
    "off_policy_rate=0.1\n",
    "initial_value=0.5\n",
    "use_afterstates=True\n",
    "\n",
    "td = TDLearner(\n",
    "    learning_rate=learning_rate, \n",
    "    gamma=gamma,\n",
    "    off_policy_rate=off_policy_rate,\n",
    "    initial_value=initial_value,\n",
    "    use_afterstates=use_afterstates\n",
    ")\n",
    "assert len(td.value_function) == 0"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Simulate value_functions\n",
    "value_functions = {\n",
    "    1: {},\n",
    "    2: {}\n",
    "}"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [],
   "source": [
    "# First play a series of set-moves\n",
    "\n",
    "planned_moves = [\n",
    "    (1, (0, 0)),\n",
    "    (2, (2, 2)),\n",
    "    (1, (0, 1)),\n",
    "    (2, (2, 1)),\n",
    "    (1, (0, 2))\n",
    "]\n",
    "game_states = {\n",
    "    1: [],\n",
    "    2: []\n",
    "}\n",
    "afterstates = {\n",
    "    1: [],\n",
    "    2: []   \n",
    "}\n",
    "\n",
    "def get_value(role, state_key):\n",
    "    \n",
    "    value = value_functions[role].get(state_key, None)\n",
    "    if value is None:\n",
    "        value = initial_value\n",
    "    \n",
    "    return value\n",
    "\n",
    "game = TicTacToeGame()\n",
    "for move in planned_moves:\n",
    "    role, position = move\n",
    "    state_key = game.generate_state_key(game.state, role)\n",
    "    game_states[role].append(state_key)\n",
    "    for position in game.available_moves():\n",
    "        possible_move = (role, position)\n",
    "        next_state = game.next_state(game.state, possible_move)\n",
    "        next_state_key = game.generate_state_key(next_state, role)\n",
    "        #value_functions[role][next_state_key] = initial_value\n",
    "        if possible_move == move:\n",
    "            afterstates[role].append(next_state_key)\n",
    "            value_functions[role][next_state_key] = \\\n",
    "                get_value(role, next_state_key)\n",
    "            \n",
    "    game.make_move(move)\n",
    "    if game.game_over:\n",
    "        opponent = [r for r in roles if r != role][0]\n",
    "        if game.winner:\n",
    "            if game.winner == role:\n",
    "                rewards = {role: 1.0, opponent: 0.0}\n",
    "            else:\n",
    "                rewards = {role: 0.0, opponent: 1.0}\n",
    "            for r in roles:\n",
    "                state = afterstates[r][-1]\n",
    "                value_functions[r][state] = \\\n",
    "                    value_functions[r][state] + \\\n",
    "                    learning_rate*(rewards[r] - \\\n",
    "                                   value_functions[r][state])\n",
    "    if len(afterstates[role]) > 1:\n",
    "        states = afterstates[role]\n",
    "        # import pdb; pdb.set_trace()\n",
    "        value_functions[role][states[-2]] = \\\n",
    "                value_functions[role][states[-2]] + \\\n",
    "                learning_rate*(0.0 + gamma*value_functions[role][states[-1]] -\\\n",
    "                               value_functions[role][states[-2]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{1: [b'---------', b'S-------O', b'SS-----OO'],\n",
       " 2: [b'O--------', b'OO------S']}"
      ]
     },
     "execution_count": 110,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game_states"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{1: [b'S--------', b'SS------O', b'SSS----OO'],\n",
       " 2: [b'O-------S', b'OO-----SS']}"
      ]
     },
     "execution_count": 111,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "afterstates"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{1: {b'S--------': 0.5, b'SS------O': 0.53125, b'SSS----OO': 0.625},\n",
       " 2: {b'O-------S': 0.5, b'OO-----SS': 0.375}}"
      ]
     },
     "execution_count": 112,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "value_functions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 113,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(True, 1)"
      ]
     },
     "execution_count": 113,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "game.game_over, game.winner"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 114,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 114,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "correct_values = {\n",
    "    1: {b'S--------': 0.5, b'SS------O': 0.53125, b'SSS----OO': 0.625},\n",
    "    2: {b'O-------S': 0.5, b'OO-----SS': 0.375}\n",
    "}\n",
    "checks = []\n",
    "for state_key, value in correct_values.items():\n",
    "    checks.append(value_functions[state_key] == \n",
    "                  correct_values[state_key])\n",
    "all(checks)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "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": "3.7.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}