9
$\begingroup$

In what follows, $g:\lbrack 0,\infty)\to \mathbb{R}$ will be a continuous function obeying the following constraints: $g(x)=1$ for $x\leq 1-\delta$, $g(x)=0$ for $x\geq 1+\delta$, and $g(1-t)=1-g(1+t)$ for $0\leq t\leq \delta$. (Otherwise put: $g'(t)$ is a symmetric function around $1$ of support $\lbrack 1-\delta,1+\delta\rbrack$.) Other than that, we are free to choose $g$, and for that matter $0<\delta<1$ as well.

Write $G(s)$ for the Mellin transform of $g$.

Let $T>0$, $0<\sigma<1$ be given. Let $\epsilon = \max_{t\geq T} |\sigma+ i t| \left|G(\sigma+it)\right|$. How small can $$\frac{\int_{1}^{1+\delta} |g(t)|^2 dt}{(1-\epsilon)^2}$$ be? Is the minimum reached by a nice function $g$?

(I take there has to be a non-zero infimum, by the uncertainty principle. It is also easy to get a crude upper bound, by choosing any particular $g$ (such as, say, $g(1+t) = \frac{1}{2} - \frac{t}{\delta}$ for $-\delta\leq t\leq \delta$).)

$\endgroup$
1
  • $\begingroup$ Yes it should be $g$. $\endgroup$ Nov 14, 2017 at 0:03

0

Your Answer

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

Browse other questions tagged or ask your own question.