2
$\begingroup$

Let $d\mu$ be a Gaussian measure on $\mathbb{R}$ with the center $a \in \mathbb{R}$ and variance $1$.

Let $B(a,r) \subset \mathbb{R}$ be the interval $[a-r,a+r]$.

Then, for any smooth mapping $f : \mathbb{R} \to \mathbb{R}$, is it true that \begin{equation} \frac{1}{\mu(B(a,r))}\int_{B(a,r)} f^2(x) d\mu(x) - \Bigl( \frac{1}{\mu(B(a,r))}\int_{B(a,r)} f(x) d\mu(x) \Bigr)^2 \leq \ \int_{B(a,r)} [f'(x)] ^2 d\mu(x)? \end{equation} for any $r>0$?

Due to smoothness of $f$, we observe that both sides of the above inequality tends to zero as $r \to 0^+$.

I guess that due to concentration of the Gaussian measure around the center, this inequality must hold for any $r>0$, but I cannot really find a relelvant reference to help justify my guess.

Could anyone please help me?

$\endgroup$
0

1 Answer 1

1
$\begingroup$

In "On how Poincaré inequalities imply weighted ones" which is interesting in itself in figuring out when a measure satisfies a Poincare-inequality, I came across the reference "An optimal Poincaré-Wirtinger inequality in Gauss space", they prove a Poincare inequality for the Gaussian measure over general Convex domains with a C2+accessibility boundary condition. Let

enter image description here

then enter image description here

$\endgroup$
3
  • $\begingroup$ Thank you. I read through your second link but now quite confused. The paper first assumes $N$ to be greater than $1$ but deals with one dimension at the last page. $\endgroup$
    – Isaac
    Commented Jun 22, 2023 at 3:02
  • $\begingroup$ Also, the "first Diricblet eigenvalue of the one dimensional Hermite operator" on the interval $(-a,a)$ coincides with the Gaussian measure of the interval?? I would like to divide the LHS of the Poincare inequality by the measure of the interval. $\endgroup$
    – Isaac
    Commented Jun 22, 2023 at 3:06
  • $\begingroup$ in the [3] reference i.e. (1.2) is proved for d=1 too and they get a comparison constant in terms of $1/diam(\Omega)^{2}$ that you ask. $\endgroup$ Commented Jun 22, 2023 at 4:20

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.