矩阵分析(九)Gram矩阵

2021-04-01 00:14:30 浏览数 (1)

欧氏空间

$V$是$mathbb{R}$上的线性空间,定义映射

$$ sigma: Vtimes V to mathbb{R} $$

对于$alpha, beta in V$,将$sigma(alpha, beta)$记为$left<alpha, betaright>$,若$sigma$满足:

  1. 对称性:$left<alpha.betaright>=left<beta, alpharight>$
  2. (右)齐次性:$left<alpha, kbetaright>=kleft<alpha,betaright>$
  3. (右)可加性:$left<alpha, beta gammaright>=left<alpha,betaright> left<alpha, gammaright>$
  4. 非负性:$left<alpha,alpharight>≥0$,且$left<alpha,alpharight>=0Leftrightarrowalpha=0$

则称$sigma$为$V$上的(实)内积,当$V$是有限维时,称其为欧氏空间($mathbb{R}^n$为标准欧氏空间)

实际上$alpha$是一个向量,$beta$是一个向量,$left<alpha, betaright>$表示向量$alpha$与向量$beta$的内积,结果是一个实数

实内积的性质
  1. (左)齐次性:$left<kalpha, betaright>=kleft<alpha,betaright>$
  2. (左)可加性:$left<alpha beta, gammaright>=left<alpha,gammaright> left<beta, gammaright>$
  3. $left<k_1alpha_1 ··· k_salpha_s,betaright>=k_1left<alpha_1,betaright> ···k_sleft<alpha_s,betaright>$
  4. $left<alpha,k_1beta_1 ··· k_sbeta_sright>=k_1left<alpha,beta_1right> ···k_sleft<alpha,beta_sright>$

复内积

$V$是$mathbb{C}$上的线性空间,定义映射

$$ sigma: Vtimes V to mathbb{C} $$

对于$alpha, beta in V$,将$sigma(alpha, beta)$记为$left<alpha, betaright>$,若$sigma$满足:

  1. 共轭对称性:$left<alpha.betaright>=overline{left<beta, alpharight>}$
  2. (右)齐次性:$left<alpha, kbetaright>=kleft<alpha,betaright>$
  3. (右)可加性:$left<alpha, beta gammaright>=left<alpha,betaright> left<alpha, gammaright>$
  4. 非负性:$left<alpha,alpharight>≥0$,且$left<alpha,alpharight>=0Leftrightarrowalpha=0$

则称$sigma$为$V$上的(复)内积,当$V$是有限维时,称其为酉空间($mathbb{R}^n$为标准欧氏空间)

复内积的性质
  1. (左)齐次性:$left<kalpha, betaright>=bar{k}left<alpha,betaright>$
  2. (左)可加性:$left<alpha beta, gammaright>=left<alpha,gammaright> left<beta, gammaright>$
  3. $left<k_1alpha_1 ··· k_salpha_s,betaright>=overline{k_1}left<alpha_1,betaright> ···overline{k_s}left<alpha_s,betaright>$
  4. $left<alpha,k_1beta_1 ··· k_sbeta_sright>=k_1left<alpha,beta_1right> ···k_sleft<alpha,beta_sright>$

线性组合的内积的矩阵表示

$alpha_1,...,alpha_s;beta_1,...,beta_t$是$mathbb{C}$上的内积空间$V$中的两个向量组,则

$$ begin{aligned} left<k_1alpha_1 ··· k_salpha_s,l_1beta_1 ··· l_tbeta_tright>\ =(overline{k_1},...,overline{k_s})begin{bmatrix}left<alpha_1,beta_1right>&cdots &left<alpha_1,beta_tright>\ vdots & ddots &vdots \left<alpha_s,beta_1right> &cdots & left<alpha_s,beta_tright>end{bmatrix}begin{bmatrix}l_1\ vdots \ l_tend{bmatrix} end{aligned} $$


Gram矩阵

$alpha_1,...,alpha_s;beta_1,...,beta_t$是$mathbb{C}$上的内积空间$V$中的两个向量组,则

$$ begin{bmatrix}left<alpha_1,beta_1right>&cdots &left<alpha_1,beta_tright>\ vdots & ddots &vdots \left<alpha_s,beta_1right> &cdots & left<alpha_s,beta_tright>end{bmatrix} $$

称为$alpha_1,...,alpha_s;beta_1,...,beta_t$的协Gram矩阵,记为$G(alpha_1,...,alpha_s;beta_1,...,beta_t)$

$alpha_1,...,alpha_s$是$mathbb{C}$上的内积空间$V$中的一个向量组,则

$$ begin{bmatrix}left<alpha_1,beta_1right>&cdots &left<alpha_1,beta_tright>\ vdots & ddots &vdots \left<alpha_s,beta_1right> &cdots & left<alpha_s,beta_tright>end{bmatrix} $$

称为$alpha_1,...,alpha_s$的Gram矩阵,记为$G(alpha_1,...,alpha_s)$

$alpha_1,...,alpha_s$是$mathbb{C}^n$中的一个向量组,记$A=(alpha_1,...,alpha_s)$,则

$$ G(alpha_1,...,alpha_s)=A^HA $$

其中,$A^H=(bar{A})^T=overline{(A^T)}$

$alpha_1,...,alpha_s$是$mathbb{R}^n$中的一个向量组,记$A=(alpha_1,...,alpha_s)$,则

$$ G(alpha_1,...,alpha_s)=A^TA $$

$alpha_1,...,alpha_s;beta_1,...,beta_t$是$mathbb{C}$上的内积空间$V$中的两个向量组,如果$alpha_1,...,alpha_s$可由$beta_1,...,beta_t$线性表出,且

$$ (alpha_1,...,alpha_s)=(beta_1,...,beta_t)A $$

$$ G(alpha_1,...,alpha_s)=A^HG(beta_1,...,beta_t)A $$

Gram矩阵的性质
  1. $Rank(G)=rank(alpha_1,...,alpha_s)$
  2. Hermite性:$G^H=G$
  3. 非负性:$forall xin mathbb{C}^s$,复二次型$x^HGx≥0$,并且$G$正定$Leftrightarrow alpha_1,...,alpha_s$线性无关

0 人点赞