Induced Matrix Norms 诱导矩阵范数证明

意义

向量的范数可以看做是对向量长度的度量,诱导矩阵通过与向量范数类似的方法来达到相同的目的。

An m × n matrix can be viewed as a vector in an mn-dimensional space: each of the mn entries of the matrix is an independent coordinate. Any mn-dimensional norm can therefore be used for measuring the ‘size’ of such a matrix. In dealing with a space of matrices, these are the induced matrix norms, defined in terms of the behavior of a matrix as an operator between its normed domain and range space.

形式

Given vector norms || · ||(m) , || · ||(n) and on the domain and the range of A ∈ Rm×n, respectively, the induced matrix norm || A ||(m,n) is the smallest number C for which the following inequilty holds for any x ∈ Rn.

\[\|\mathbf{A} \mathbf{x}\|_{(m)} \leq C\|\mathbf{x}\|_{(n)}\]

The induced matrix norm can be defined equivalently as following:

\[\|\mathbf{A}\|_{(m, n)}=\sup _{\mathbf{x} \in \mathbb{R}^{n}, \mathbf{x} \neq 0} \frac{\|\mathbf{A} \mathbf{x}\|_{(m)}}{\|\mathbf{x}\|_{(n)}}=\sup _{\mathbf{x} \in \mathbb{R}^{n},\|\mathbf{x}\|=1}\|\mathbf{A} \mathbf{x}\|_{(m)}\]

注意

The (m) and (n) in subscript are only represented the dimensionality of vectors, rather than Lm or Ln norm. The notations || · ||(m) and || · ||(n) can be any the same type of vector norm.

证明

证明主要的内容是以下两种矩阵范数形式:

1-范数,列和范数 \(\|\mathbf{A}\|_{1}=\max_{j} \sum_{i=1}^{m}\left|A_{i j}\right| = \max _{1 \leq j \leq n}\left\|\mathbf{a}_{j}\right\|_{1}\)

∞-范数,行和范数 \(\|\mathbf{A}\|_{\infty}=\max_{i} \sum_{j=1}^{n}\left|A_{i j}\right| = \max _{1 \leq i \leq m}\left\|\mathbf{a}_{(i)}\right\|_{1}\)

2-范数的形式为

\[\|\mathbf{A}\|_{2}=\max \mu\]

表示最大的奇异值(sigular value)。

1-范数形式的证明


将矩阵 A 记作:

\[\mathbf{A}=\left[\begin{array}{cccc} | & | & \cdots & | \\ \mathbf{a}_{1} & \mathbf{a}_{2} & \cdots & \mathbf{a}_{n} \\ | & | & \cdots & | \end{array}\right]\]

涉及到的两向量的 1-范数分别为:

\[\|x\|_{1}=\sum_{i=1}^{n}\left|x_{i}\right| \\ \\ \|\mathbf{A} \mathbf{x}\|_{1} = \left\|\sum_{i=1}^{n} x_{i} \mathbf{a}_{i}\right\|_{1}\]

根据向量范数的三角不等式:

\[\|a+b\| \leq \|a\|+\|b\|\]

有:

\[\left\|\sum_{i=1}^{n} x_{i} \mathbf{a}_{i}\right\|_{1} \leq \sum_{i=1}^{n}\left|x_{i}\right||| \mathbf{a}_{i}|| \leq \sum_{i=1}^{n}\left|x_{i}\right|\left(\max _{1 \leq j \leq n}\left\|\mathbf{a}_{j}\right\|_{1}\right)\]

所以:

\[\frac{\|\mathbf{A} \mathbf{x}\|_{1}}{\|\mathbf{x}\|_{1}} \leq \max _{1 \leq j \leq n}\left\|\mathbf{a}_{j}\right\|_{1}\]

∞-范数形式的证明


涉及到的两向量的 ∞-范数分别为:

\[\|x\|_{\infty}= \max_i \left|x_{i}\right| \\ \\ \|\mathbf{A} \mathbf{x}\|_{\infty} = \left\|\sum_{i=1}^{n} x_{i} \mathbf{a}_{i}\right\|_{\infty} = \max_i\left|\sum_{j=1}^{n}A_{ij}x_{j}\right|\]

注意一下上面符号的下标表示的含义。

并且有:

\[\begin{aligned} \left|\sum_{j=1}^{n} A_{i j} x_{j}\right| &\leq \sum_{j=1}^{n}\left|A_{i j} x_{j}\right|=\sum_{j=1}^{n}\left|A_{i j} \| x_{j}\right| \\ &\leq \sum_{j=1}^{n}\left|A_{i j}\right| \max _{i}\left|x_{i}\right| \end{aligned}\]

所以:

\[\frac{\|\mathbf{A} \mathbf{x}\|_{\infty}}{\|\mathbf{x}\|_{\infty}} = \frac{\max_i\left|\sum_{j=1}^{n}A_{ij}x_{j}\right|}{\max_i \left|x_{i}\right|} \leq \max_i \sum_{j=1}^n|A_{ij}|\]