-1
$\begingroup$

Given a $c$-Lipschitz function $f(X):\mathbb{R}^d \rightarrow \mathbb{R}$, and given that $X \in \mathbb{R}^d$ is a Gausssian random vector centered at $\mathbb{w} \in \mathbb{R}^d$ (not at zero) having a co-variance matrix of $\sigma^2 \mathbb{I}$, what can we say about $ P( |f(X) - {\Bbb E} f(X)|\geq t)$? I know the results are standard for the case when every component of $X$ is distributed i.i.d as $N(0,1)$, but am looking for a more general result.

Thanks very much, MK

$\endgroup$

1 Answer 1

3
$\begingroup$

Since your covariance matrix is merely a multiple of the identity matrix, the result for standard normal should be easily adapted to your case by a scaling argument. I suspect this is a homework problem. But it seems too easy for such advanced topic.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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