{ "cells": [ { "cell_type": "markdown", "metadata": {}, "source": [ "# 2.8 奇异值分解" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ "奇异值分解(`singular value decomposition, SVD`)提供了另一种分解矩阵的方式,将其分解为奇异向量和奇异值。\n", "\n", "与特征值分解相比,奇异值分解更加通用,所有的实矩阵都可以进行奇异值分解,而特征值分解只对某些方阵可以。\n", "\n", "奇异值分解的形式为:\n", "\n", "$$\n", "\\bf A=UDV^\\top\n", "$$\n", "\n", "若 $\\bf A$ 是 $m\\times n$ 的,那么 $\\bf U$ 是 $m\\times m$ 的,其列向量称为左奇异向量,而 $\\bf V$ 是 $n\\times n$ 的,其列向量称为右奇异向量,而 $\\bf D$ 是 $m\\times n$ 的一个对角矩阵,其对角元素称为矩阵 $\\bf A$ 的奇异值。\n", "\n", "事实上,左奇异向量是 $\\bf AA^\\top$ 的特征向量,而右奇异向量是 $\\bf A^\\top A$ 的特征向量,非零奇异值是 $\\bf A^\\top A$ 的非零特征值的平方。" ] } ], "metadata": { "kernelspec": { "display_name": "Python 2", "language": "python", "name": "python2" }, "language_info": { "codemirror_mode": { "name": "ipython", "version": 2 }, "file_extension": ".py", "mimetype": "text/x-python", "name": "python", "nbconvert_exporter": "python", "pygments_lexer": "ipython2", "version": "2.7.6" } }, "nbformat": 4, "nbformat_minor": 0 }