Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I have been working on the following question for a couple of days, and I am not getting anywhere. Suppose $A\in M_{n}$ with singular values $\sigma_{1}\geq \sigma_{2} \geq \cdots \sigma_n$. How can I show the following? $$\sigma_{1}\leq 1 \iff\begin{pmatrix}I_{n} & A \\ A^{*} & I_{n} \\ \end{pmatrix} \geq 0$$

share|cite|improve this question
up vote 3 down vote accepted

(i) $\sigma_1 \leq 1 \implies B = \begin{pmatrix}I_{n} & A \\ A^{*} & I_{n} \\ \end{pmatrix} \geq 0$:

Consider any non-zero $x \in \mathbb{C}^{2n \times 1}$. Let $x = \begin{pmatrix} x_1 \\ x_2 \end{pmatrix}$ where $x_1,x_2 \in \mathbb{C}^{n \times 1}$. Then $$x^* Bx = \|x_1\|^2 + \|x_2\|^2 + x_1^*Ax_2 + x_2^*A^*x_1 = \|x_1\|^2 + \|x_2\|^2 + 2 \text{Re}(x_1^*Ax_2)$$ We have that $$|x_1^*Ax_2| \leq \|x_1\|\cdot\|Ax_2\| \leq \sigma_1 \|x_1\|\cdot\|x_2\|.$$ We also have that $$-|x_1^*Ax_2| \leq \text{Re}(x_1^*Ax_2) \leq |x_1^*Ax_2|.$$Hence, we get that $$x_1^*Ax_2 + x_2^*A^*x_1 = 2 \text{Re}(x_1^*Ax_2) \in \left[ -2 \sigma_1 \|x_1\|\cdot\|x_2\|,2 \sigma_1 \|x_1\|\cdot\|x_2\| \right].$$ Hence, $$x^* Bx \geq \|x_1\|^2 + \|x_2\|^2 - 2\sigma_1\|x_1\|\cdot\|x_2\| \geq 0 \qquad(\text{since }\sigma_1 \leq 1)$$

(ii) $B = \begin{pmatrix}I_{n} & A \\ A^{*} & I_{n} \\ \end{pmatrix} \geq 0 \implies \sigma_1 \leq 1$:

Since $B$ is given to be positive semi-definite, any non-zero $x \in \mathbb{C}^{2n \times 1}$ satisfies $x^*Bx \geq 0$. Consider $x = \begin{pmatrix} -u \\ v \end{pmatrix}$ where $u \in \mathbb{C}^{n \times 1}$ is the first left singular vector of $A$ and $v \in \mathbb{C}^{n \times 1}$ is the first right singular vector of $A$. Then $$0 \leq x^* Bx = \|u\|^2 + \|v\|^2 - u^*Av - v^*A^*u = 2 - 2 \sigma_1.$$ Hence, we get that $$\sigma_1 \leq 1.$$

share|cite|improve this answer

This is a typical application the nonstrict Schur Complement Formula.

Basically the following are equivalent (two of them are trivially true): $$ \begin{pmatrix} I &A\\ A^* &I \end{pmatrix} \succeq 0 \iff \begin{align} I &\succeq 0\\ I - A^* I A &\succeq 0\\ A(I-I) &= 0 \end{align} $$ Hence, from the second condition we infer $I\succeq A^*A$, which is another version of the desired property.

share|cite|improve this answer
Nice demonstration. – user1551 Nov 30 '11 at 13:49

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.