chapter35.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 第三十五讲：期末复习\n",
    "\n",
    "依然是从以往的试题入手复习知识点。\n",
    "\n",
    "1. *已知$m\\times n$矩阵$A$，有$Ax=\\begin{bmatrix}1\\\\0\\\\0\\end{bmatrix}$无解；$Ax=\\begin{bmatrix}0\\\\1\\\\0\\end{bmatrix}$仅有唯一解，求关于$m,n,rank(A)$的信息。*\n",
    "\n",
    "    首先，最容易判断的是$m=3$；而根据第一个条件可知，矩阵不满秩，有$r<m$；根据第二个条件可知，零空间仅有零向量，也就是矩阵消元后没有自由变量，列向量线性无关，所以有$r=n$。\n",
    "     \n",
    "    综上，有$m=3>n=r$。\n",
    "     \n",
    "    *根据所求写出一个矩阵$A$的特例*：$A=\\begin{bmatrix}0&0\\\\1&0\\\\0&1\\end{bmatrix}$。\n",
    "     \n",
    "    *$\\det A^TA\\stackrel{?}{=}\\det AA^T$*：不相等，因为$A^TA$可逆而$AA^T$不可逆，所以行列式不相等。（但是对于方阵，$\\det AB=\\det BA$恒成立。）\n",
    "     \n",
    "    *$A^TA$可逆吗？*是，因为$r=n$，矩阵列向量线性无关，即列满秩。\n",
    "     \n",
    "    *$AA^T$正定吗？*否，因为$AA^T$是$3\\times n$矩阵与$n\\times 3$矩阵之积，是一个三阶方阵，而$AA^T$秩为$2$，所以不是正定矩阵。（不过$AA^T$一定是半正定矩阵。）\n",
    "     \n",
    "    *求证$A^Ty=c$至少有一个解*：因为$A$的列向量线性无关，所以$A^T$的行向量线性无关，消元后每行都有主元，且总有自由变量，所以零空间中有非零向量，零空间维数是$m-r$（可以直接从$\\dim N\\left(A^T\\right)=m-r$得到结论）。\n",
    "\n",
    "2. *设$A=\\Bigg[v_1\\ v_2\\ v_3\\Bigg]$，对于$Ax=v_1-v_2+v_3$，求$x$。*\n",
    "    \n",
    "    按列计算矩阵相乘，有$x=\\begin{bmatrix}1\\\\-1\\\\1\\end{bmatrix}$。\n",
    "    \n",
    "    *若Ax=v_1-v_2+v_3=0，则解是唯一的吗？为什么。*如果解释唯一的，则零空间中只有零向量，而在此例中$x=\\begin{bmatrix}1\\\\-1\\\\1\\end{bmatrix}$就在零空间中，所以解不唯一。\n",
    "    \n",
    "    *若$v_1,v_2,v_3$是标准正交向量，那么怎样的线性组合$c_1v_1+c_2v_2$能够最接近$v_3$？*此问是考察投影概念，由于是正交向量，所以只有$0$向量最接近$v_3$。\n",
    "    \n",
    "3. *矩阵$A=\\begin{bmatrix}.2&.4&.3\\\\.4&.2&.3\\\\.4&.4&.4\\end{bmatrix}$，求稳态。*\n",
    "\n",
    "    这是个马尔科夫矩阵，前两之和为第三列的两倍，奇异矩阵总有一个特征值为$0$，而马尔科夫矩阵总有一个特征值为$1$，剩下一个特征值从矩阵的迹得知为$-.2$。\n",
    "    \n",
    "    再看马尔科夫过程，设从$u(0)$开始，$u_k=A^ku_0, u_0=\\begin{bmatrix}0\\\\10\\\\0\\end{bmatrix}$。先代入特征值$\\lambda_1=0,\\ \\lambda_2=1,\\ \\lambda_3=-.2$查看稳态$u_k=c_1\\lambda_1^kx_1+c_2\\lambda_2^kx_2+c_3\\lambda_3^kx_3$，当$k\\to\\infty$，第一项与第三项都会消失，剩下$u_\\infty=c_2x_2$。\n",
    "    \n",
    "    到这里我们只需求出$\\lambda_2$对应的特征向量即可，带入特征值求解$(A-I)x=0$，有$\\begin{bmatrix}-.8&.4&.3\\\\.4&-.8&.3\\\\.4&.4&-.6\\end{bmatrix}\\begin{bmatrix}?\\\\?\\\\?\\end{bmatrix}=\\begin{bmatrix}0\\\\0\\\\0\\end{bmatrix}$，可以消元得，也可以直接观察得到$x_2=\\begin{bmatrix}3\\\\3\\\\4\\end{bmatrix}$。\n",
    "    \n",
    "    剩下就是求$c_2$了，可以通过$u_0$一一解出每个系数，但是这就需要解出每一个特征值。另一种方法，我们可以通过马尔科夫矩阵的特性知道，对于马尔科夫过程的每一个$u_k$都有其分量之和与初始值分量之和相等，所以对于$x_2=\\begin{bmatrix}3\\\\3\\\\4\\end{bmatrix}$，有$c_2=1$。所以最终结果是$u_\\infty=\\begin{bmatrix}3\\\\3\\\\4\\end{bmatrix}$。\n",
    "\n",
    "4. *对于二阶方阵，回答以下问题：*\n",
    "    \n",
    "    *求投影在直线$a=\\begin{bmatrix}4\\\\-3\\end{bmatrix}$上的投影矩阵*：应为$P=\\frac{aa^T}{a^Ta}$。\n",
    "    \n",
    "    *已知特征值$\\lambda_1=2,\\ x_1=\\begin{bmatrix}1\\\\2\\end{bmatrix}\\quad \\lambda_2=3,\\ x_2=\\begin{bmatrix}2\\\\1\\end{bmatrix}$求原矩阵$A$*：从对角化公式得$A=S\\Lambda S^{-1}=\\begin{bmatrix}1&2\\\\2&1\\end{bmatrix}\\begin{bmatrix}0&0\\\\0&3\\end{bmatrix}\\begin{bmatrix}1&2\\\\2&1\\end{bmatrix}^{-1}$，解之即可。\n",
    "    \n",
    "    *$A$是一个实矩阵，且对任意矩阵$B$，$A$都不能分解成$A=B^TB$，给出$A$的一个例子*：我们知道$B^TB$是对称的，所以给出一个非对称矩阵即可。\n",
    "    *矩阵$A$有正交的特征向量，但不是对称的，给出一个$A$的例子*：我们在三十三讲提到过，反对称矩阵，因为满足$AA^T=A^TA$而同样具有正交的特征向量，所以有$A=\\begin{bmatrix}0&1\\\\-1&0\\end{bmatrix}$或旋转矩阵$\\begin{bmatrix}\\cos\\theta&-\\sin\\theta\\\\\\sin\\theta&\\cos\\theta\\end{bmatrix}$，这些矩阵都具有复数域上的正交特征向量组。\n",
    "    \n",
    "5. *最小二乘问题，因为时间的关系直接写出计算式和答案，$\\begin{bmatrix}1&0\\\\1&1\\\\1&2\\end{bmatrix}\\begin{bmatrix}C\\\\D\\end{bmatrix}=\\begin{bmatrix}3\\\\4\\\\1\\end{bmatrix}(Ax=b)$，解得$\\begin{bmatrix}\\hat C\\\\\\hat D\\end{bmatrix}=\\begin{bmatrix}\\frac{11}{3}\\\\-1\\end{bmatrix}$。*\n",
    "\n",
    "    *求投影后的向量$p$*：向量$p$就是向量$b$在矩阵$A$列空间中的投影，所以$p=\\begin{bmatrix}p_1\\\\p_2\\\\p_3\\end{bmatrix}=\\begin{bmatrix}1&0\\\\1&1\\\\1&2\\end{bmatrix}\\begin{bmatrix}\\hat C\\\\\\hat D\\end{bmatrix}$。\n",
    "    \n",
    "    *求拟合直线的图像*：$x=0,1,2$时$y=p_1,p_2,p_2$所在的直线的图像，$y=\\hat C+\\hat Dx$即$y=\\frac{11}{3}-x$。"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAd8AAAFVCAYAAACuK+XmAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nYII=\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x104190320>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "%matplotlib inline\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn import linear_model\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "\n",
    "x = np.array([0, 1, 2]).reshape((-1,1))\n",
    "y = np.array([3, 4, 1]).reshape((-1,1))\n",
    "predict_line = np.array([-1, 4]).reshape((-1,1))\n",
    "\n",
    "regr = linear_model.LinearRegression()\n",
    "regr.fit(x, y)\n",
    "ey = regr.predict(x)\n",
    "\n",
    "fig = plt.figure()\n",
    "plt.axis('equal')\n",
    "plt.axhline(y=0, c='black')\n",
    "plt.axvline(x=0, c='black')\n",
    "\n",
    "plt.scatter(x, y, c='r')\n",
    "plt.scatter(x, regr.predict(x), s=20, c='b')\n",
    "plt.plot(predict_line, regr.predict(predict_line), c='g', lw='1')\n",
    "[ plt.plot([x[i], x[i]], [y[i], ey[i]], 'r', lw='1') for i in range(len(x))]\n",
    "\n",
    "plt.draw()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "plt.close(fig)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "* 接上面的题目\n",
    "\n",
    "    *求一个向量$b$使得最小二乘求得的$\\begin{bmatrix}\\hat C\\\\\\hat D\\end{bmatrix}=\\begin{bmatrix}0\\\\0\\end{bmatrix}$*：我们知道最小二乘求出的向量$\\begin{bmatrix}\\hat C\\\\\\hat D\\end{bmatrix}$使得$A$列向量的线性组合最接近$b$向量（即$b$在$A$列空间中的投影），如果这个线性组合为$0$向量（即投影为$0$），则$b$向量与$A$的列空间正交，所以可以取$b=\\begin{bmatrix}1\\\\-2\\\\1\\end{bmatrix}$同时正交于$A$的两个列向量。\n",
    "\n",
    "# MIT线性代数的全部课程到此结束"
   ]
  }
 ],
 "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": "3.5.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}