{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import torch"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'1.0.0'"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "torch.__version__"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "x = torch.tensor([[1, 2, 3], [4, 5, 6]])\n",
    "y = torch.tensor([[7, 8, 9], [10, 11, 12]])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<torch._C.Generator at 0x1b55d9d1db0>"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "torch.manual_seed(42)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[0.8823, 0.9150, 0.3829],\n",
      "        [0.9593, 0.3904, 0.6009]])\n"
     ]
    }
   ],
   "source": [
    "print(torch.rand([2, 3]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "ename": "TypeError",
     "evalue": "add(): argument 'other' (position 1) must be Tensor, not numpy.ndarray",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mTypeError\u001b[0m                                 Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-7-2e651424c109>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[0;32m      1\u001b[0m \u001b[0mxnp\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mnp\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0marray\u001b[0m\u001b[1;33m(\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;33m[\u001b[0m\u001b[1;36m1\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m2\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m3\u001b[0m\u001b[1;33m]\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;33m[\u001b[0m\u001b[1;36m4\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m5\u001b[0m\u001b[1;33m,\u001b[0m \u001b[1;36m6\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[1;32m----> 2\u001b[1;33m \u001b[0mf2\u001b[0m \u001b[1;33m=\u001b[0m \u001b[0mxnp\u001b[0m \u001b[1;33m+\u001b[0m \u001b[0my\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;31mTypeError\u001b[0m: add(): argument 'other' (position 1) must be Tensor, not numpy.ndarray"
     ]
    }
   ],
   "source": [
    "xnp = np.array([[1, 2, 3], [4, 5, 6]])\n",
    "f2 = xnp + y"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "えっ？？？  \n",
    "numpy_arrayとTensorって計算できるの？？？？？\n",
    "\n",
    "-> torchのversionが1.0.0だとエラー吐くらしい．  \n",
    "その前のversionだと通るっぽい"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "1. p.10 installing-pytorchまで\n",
    "2. p.13 終わりまで\n",
    "3. p.14-18\n",
    "4. p.19-22(loading dataの前まで)(伊藤)\n",
    "5. p.22(loading dataから)-26(Data Loaderまで)\n",
    "6. p.26-(DataLoader以降最後まで)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 輪講chapter1\n",
    "## Slicing and indexing and reshaping(P.19)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[1, 2, 3],\n",
       "        [4, 5, 6]])"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* 一個目の要素を表示"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([1, 2, 3])\n"
     ]
    }
   ],
   "source": [
    "print(x[0])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* 二個目の要素の0~1を表示  \n",
    "slicingは0始まりで，最後を含まないので注意  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([4, 5])\n"
     ]
    }
   ],
   "source": [
    "print(x[1][0:2])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "`view()`関数を使って既存のtensorをreshapeされたcopyを作れるよ．  \n",
    "3つの例を示すね  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([1, 2, 3, 4, 5, 6])\n"
     ]
    }
   ],
   "source": [
    "print(x.view(-1))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[1, 2],\n",
      "        [3, 4],\n",
      "        [5, 6]])\n"
     ]
    }
   ],
   "source": [
    "print(x.view(3, 2))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[1],\n",
      "        [2],\n",
      "        [3],\n",
      "        [4],\n",
      "        [5],\n",
      "        [6]])\n"
     ]
    }
   ],
   "source": [
    "print(x.view(6, 1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "(3, 2)と(6, 1)はわかりやすいけど，-1は何をやっているんだろう．  \n",
    "必要な列数がわかっているけど，何行のデータが来るかわからないときに便利だよ．  \n",
    "-1を使うと，全体の要素数から適切な行数(や列数)を計算してくれるよ．  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([[1, 2],\n",
      "        [3, 4],\n",
      "        [5, 6]])\n"
     ]
    }
   ],
   "source": [
    "print(x.view(3, -1))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "重要な操作として，軸の入れ替えがあるよ．  \n",
    "これは`tensor.transpose()`メソッド(2軸の入れ替えの場合)で実現できるよ．  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[1, 2, 3],\n",
       "        [4, 5, 6]])"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[1, 4],\n",
       "        [2, 5],\n",
       "        [3, 6]])"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x.transpose(0, 1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "ここで注意したいのは，`transpose()`は一度に二軸の入れ替えしかできないよ．  \n",
    "2軸より大きい軸同士を入れ替えたいときは，`permute()`メソッドを使うといいよ．  \n",
    "軸の番号を指定することで利用できるよ．  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "a = torch.ones(1, 2, 3, 4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[[[1., 1., 1., 1.],\n",
       "          [1., 1., 1., 1.],\n",
       "          [1., 1., 1., 1.]],\n",
       "\n",
       "         [[1., 1., 1., 1.],\n",
       "          [1., 1., 1., 1.],\n",
       "          [1., 1., 1., 1.]]]])"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "a"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([4, 3, 2, 1])"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 2回に分けて全ての軸を入れ替えた場合\n",
    "a.transpose(0, 3).transpose(1, 2).size()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([4, 3, 2, 1])"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# permuteを使うと一回で済むよ\n",
    "a.permute(3, 2, 1, 0).size()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "二次元までのtensorだとフラットなtableで表せるよ．  \n",
    "より高次元になると表せないよ．  \n",
    "magic of deep learning　では問題にならないよ．  \n",
    "実世界の特徴量はデータ構造の次元にエンコードされています．  \n",
    "なので，"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## In place operations(p.21-22)\n",
    "上書きする関数と，しない関数の違いを理解することは大事だよ．  \n",
    "例えば，`transpose(x)`だと処理後の変数が戻り値として返ってきたけれど，元のxは変更されませんでした．  \n",
    "今までの例で使ったのは，上書きしないものばかりでした．  \n",
    "別に `x = x.transpose(x)` と書いてもいいんだけど，より便利な方法として上書きする関数を使う方法があるよ．  \n",
    "大体アンダースコアをつけると上書き版になるよ．  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[1, 2, 3],\n",
       "        [4, 5, 6]])"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[1, 4],\n",
       "        [2, 5],\n",
       "        [3, 6]])"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x.transpose_(1, 0)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[1, 4],\n",
       "        [2, 5],\n",
       "        [3, 6]])"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "x"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([[ 7,  8,  9],\n",
       "        [10, 11, 12]])"
      ]
     },
     "execution_count": 38,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "ename": "RuntimeError",
     "evalue": "The size of tensor a (3) must match the size of tensor b (2) at non-singleton dimension 1",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mRuntimeError\u001b[0m                              Traceback (most recent call last)",
      "\u001b[1;32m<ipython-input-39-e46502eb7b4e>\u001b[0m in \u001b[0;36m<module>\u001b[1;34m()\u001b[0m\n\u001b[1;32m----> 1\u001b[1;33m \u001b[0my\u001b[0m\u001b[1;33m.\u001b[0m\u001b[0madd_\u001b[0m\u001b[1;33m(\u001b[0m\u001b[0mx\u001b[0m\u001b[1;33m*\u001b[0m\u001b[1;36m2\u001b[0m\u001b[1;33m)\u001b[0m\u001b[1;33m\u001b[0m\u001b[0m\n\u001b[0m",
      "\u001b[1;31mRuntimeError\u001b[0m: The size of tensor a (3) must match the size of tensor b (2) at non-singleton dimension 1"
     ]
    }
   ],
   "source": [
    "y.add_(x*2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "いや，上で上書きさせといて，下はそれかよ  \n",
    "copyしとくとか書いとけや  "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "anaconda-cloud": {},
  "kernelspec": {
   "display_name": "Python [conda env:Anaconda3]",
   "language": "python",
   "name": "conda-env-Anaconda3-py"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.5.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 1
}