find the gradient of trace of the matrix

$\begingroup$

Prove that $\nabla_A Tr(AA^T) = 2A$, where A is any square matrix

I did simple derivative with product rule,but i don't know where i messed up,

I started with $\frac{\partial}{\partial A} Tr(AA^T)=Tr(\frac {\partial A}{\partial A}A^T+A\frac {\partial A^T}{\partial A})$ Thanks in Advance..

$\endgroup$

1 Answer

$\begingroup$

I'm going to work with square matrices, but it should be easy enough to change the dimensions as you please. Write $A = [\vec x_1 | \cdots | \vec x_n ]$ where each $\vec x_i = [x_{1i}, \ldots, x_{ni}]$. Notice then that the $i^{th}$ diagonal element of $AA^T$ is just $\vec x_i \cdot \vec x_i = x_{1i}^2 + \cdots + x_{ni}^2$. Thus the trace of $AA^T$ is the sum of every element of $A$; namely $$ \operatorname{Tr}(AA^T) = \sum_{i=1}^n \sum_{j=1}^n x_{ij}^2.$$ Now taking $\nabla_A$ will give the desired result.

$\endgroup$

Your Answer

Sign up or log in

Sign up using Google Sign up using Facebook Sign up using Email and Password

Post as a guest

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

You Might Also Like