Cauchy-Schwarz inequality applied to Trace of two products $\mathbf{Tr}(A'B)$ has the form

$$ \mathbf{Tr}(A'B) \leq \sqrt{\mathbf{Tr}(A'A)} \sqrt{\mathbf{Tr}(B'B)} $$

I saw many places where people use this inequality. But did not see a formal proof. Is it difficult to prove ? Anyone can give a simple proof ?

closed as off-topic by user21820, amWhy, KReiser, user10354138, Cesareo Nov 21 at 2:11

This question appears to be off-topic. The users who voted to close gave this specific reason:

  • "This question is missing context or other details: Please improve the question by providing additional context, which ideally includes your thoughts on the problem and any attempts you have made to solve it. This information helps others identify where you have difficulties and helps them write answers appropriate to your experience level." – user21820, amWhy, KReiser, user10354138, Cesareo
If this question can be reworded to fit the rules in the help center, please edit the question.

up vote 16 down vote accepted

$Tr(A-tB)'(A-tB) \geq 0$ for all $t$ real. Expand this, use the fact that $Tr(M')=Tr(M)$ (so $Tr(A'B)=Tr(B'A))$ and minimize the left side over $t$. You will get the inequality you want. (This is the standard proof of C-S inequality)

The Cauchy-Schwarz inequality is valid for any inner product, so you just need to show $\operatorname{\textbf{Tr}}A'B$ is an inner product. It's clearly bilinear (or sesquilinear if by $'$ you meant a complex adjoint), with $$\operatorname{\textbf{Tr}}A'A=\sum_i (A'A)_{i}=\sum_{ij}A'_{ij}A_{ji}.$$Depending on whether you're working with the real or complex case, this quantity is either $\sum_{ij}A_{ji}^2$ or $\sum_{ij}|A_{ji}|^2$. Either way it's non-negative, completing the proof.

  • 3
    The point here is that while $\mathbf{Tr}(A,B)$ looks complicated, it is really nothing more than a plain vanilla inner product with n^2-element vectors, so there's nothing special in proving C-S here. – einpoklum Nov 20 at 19:13

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