本节是本书的第一章第一节,主要回顾一些线性代数的基础知识,在流形分析中用到的线性代数知识是非常基本而经典的,比如线性空间、内积、矩阵、线性变换、矩阵的秩、转置。
本节的六个定理,Theorem 1.1 说明空间维数和基向量个数的关系,Theorem 1.2说明任何低于空间维数的线性无关向量组都可以扩充为一组基,Theorem 1.3说明矩阵乘积的上限模(sup norm)不大于两个矩阵上限模的乘积再乘以二者的公共维度,是一个卡上界的定理,在后面关于 R n \mathbf R^n Rn上微积分的讨论中比较有用。Theorem 1.4说明了线性变换对目标空间的一种“遍历性”,可以变换至任意的向量组。Theorem 1.5是矩阵的行秩和列秩相等。Theorem 1.6说明初等行变换不影响矩阵的秩。
Exercises
Exercise 1. Let V V V be a vector space with inner product ⟨ x , y ⟩ \langle x,y\rangle ⟨x,y⟩ and norm ∥ x ∥ = ⟨ x , x ⟩ \| x \|=\langle x,x\rangle ∥x∥=⟨x,x⟩.
( a ) Prove the Cauchy-Schwarz inequality ⟨ x , y ⟩ ≤ ∥ x ∥ ∥ y ∥ \langle x,y\rangle\leq\| x \| \| y \| ⟨x,y⟩≤∥x∥∥y∥.
( b ) Prove that ∥ x + y ∥ ≤ ∥ x ∥ + ∥ y ∥ \|x+y\|\leq\|x\|+\|y\| ∥x+y∥≤∥x∥+∥y∥.
( c ) Prove that ∥ x − y ∥ ≥ ∥ x ∥ − ∥ y ∥ \|x-y\|\geq\|x\|-\|y\| ∥x−y∥≥∥x∥−∥y∥.
Solution:
( a ) If x = 0 x=0 x=0 or y = 0 y=0 y=0 the inequality holds. Suppose x , y ≠ 0 x,y\neq 0 x,y=0 and let c = 1 / ∥ x ∥ c=1/\|x\| c=1/∥x∥ and d = 1 / ∥ y ∥ d=1/\|y\| d=1/∥y∥, since ∥ c x − d y ∥ ≥ 0 \| cx-dy\|\geq 0 ∥cx−dy∥≥0, we expand it to get
c 2 ∥ x ∥ 2 − 2 c d ⟨ x , y ⟩ + d 2 ∥ y ∥ 2 ≥ 0 ⟹ c d ⟨ x , y ⟩ ≤ 1 c^2\|x\|^2-2cd\langle x,y\rangle+d^2\|y\|^2 \geq 0 \implies cd\langle x,y\rangle\leq 1 c2∥x∥2−2cd⟨x,y⟩+d2∥y∥2≥0⟹cd⟨x,y⟩≤1
( b ) We have
∥ x + y ∥ 2 = ⟨ x + y , x + y ⟩ = ∥ x ∥ 2 + 2 ⟨ x , y ⟩ + ∥ y ∥ 2 ≤ ∥ x ∥ 2 + 2 ∥ x ∥ ∥ y ∥ + ∥ y ∥ 2 = ( ∥ x ∥ + ∥ y ∥ ) 2 \begin{aligned} \|x+y\|^2&=\langle x+y,x+y\rangle =\|x\|^2+2\langle x,y\rangle +\|y\|^2 \\&\leq \|x\|^2+2\|x\|\|y\| +\|y\|^2=(\|x\|+\|y\|)^2 \end{aligned} ∥x+y∥2=⟨x+y,x+y⟩=∥x∥2+2⟨x,y⟩+∥y∥2≤∥x∥2+2∥x∥∥y∥+∥y∥2=(∥x∥+∥y∥)2
( c ) We have
∥ x − y ∥ 2 = ⟨ x − y , x − y ⟩ = ∥ x ∥ 2 − 2 ⟨ x , y ⟩ + ∥ y ∥ 2 ≥ ∥ x ∥ 2 − 2 ∥ x ∥ ∥ y ∥ + ∥ y ∥ 2 = ( ∥ x ∥ − ∥ y ∥ ) 2 \begin{aligned} \|x-y\|^2&=\langle x-y,x-y\rangle =\|x\|^2-2\langle x,y\rangle +\|y\|^2 \\&\geq \|x\|^2-2\|x\|\|y\| +\|y\|^2=(\|x\|-\|y\|)^2 \end{aligned} ∥x−y∥2=⟨x−y,x−y⟩=∥x∥2−2⟨x,y⟩+∥y∥2≥∥x∥2−2∥x∥∥y∥+∥y∥2=(∥x∥−∥y∥)2
Exercise 2. If A A A is an n n n by m m m matrix and B B B is an m m m by p p p matrix, show that
∣ A ⋅ B ∣ ≤ m ∣ A ∣ ∣ B ∣ |A\cdot B|\leq m|A| |B| ∣A⋅B∣≤m∣A∣∣B∣
Solution: We denote A = ( a i j ) A=(a_{ij}) A=(aij) and B = ( b j k ) B=(b_{jk}) B=(bjk), where 1 ≤ i ≤ n , 1 ≤ j ≤ m 1\leq i\leq n,1\leq j\leq m 1≤i≤n,1≤j≤m and 1 ≤ k ≤ p 1\leq k\leq p 1≤k≤p. Obciously a i j ≤ ∣ A ∣ a_{ij}\leq |A| aij≤∣A∣ and b j k ≤ ∣ B ∣ b_{jk}\leq |B| bjk≤∣B∣. Now let c i k c_{ik} cik be the i k ik ik-th element of the product A ⋅ B A\cdot B A⋅B, we have
c i k = ∑ j = 1 m a i j b j k ≤ m ∣ A ∣ ∣ B ∣ ⟹ ∣ A ⋅ B ∣ ≤ m ∣ A ∣ ∣ B ∣ c_{ik}=\sum_{j=1}^ma_{ij}b_{jk}\leq m|A||B|\implies |A\cdot B|\leq m|A| |B| cik=