1
$\begingroup$

For $a\in R^d$, let $T_af(x)=f(x-a),$ for all $f\in L^p(R^d), 1\leq p<\infty$ and all $x\in R^d$. I need example that this operator doesn't converge uniformly when $a\rightarrow 0$. I know that this operator has strong (norm) limit.

$\endgroup$
0
$\begingroup$

Let $d=1,p=1$ and $f_n(x)=nx^{n}$ for $0<x<1$, $0$ for $x \notin (0,1)$. Then $\|f_n\| <1$ for all $n$ . Suppose $\int_{\mathbb R}|f_n(x-a)-f_n(x)|\, dx \to 0$ uniformly in $n$ as $a\to 0$. Then, given $\epsilon \in (0,e^{-1}-e^{-2}) $ there exists $b>0$ such that $\int_{\mathbb R}|f_n(x-a)-f_n(x)|\, dx <\epsilon $ for all $a \in (0,b)$ for all $n$. Put $a=\frac 1 n$ where $n$ is so large that $\frac 1 n <b$. After a little computation we get $\frac n {n+1} ((1-\frac 1n)^{1+n}-(\frac 1 n)^{1+n}-(1-\frac 2 n)^{1+n}) <\epsilon $. Letting $n \to \infty$ we get $e^{-1}-e^{-2}<\epsilon$. This is a contradiction.

$\endgroup$
  • $\begingroup$ Why you said that $sup$ converge to $\infty$ for each $a\in(0,1)$ when $a$ need to converge to 0? $\endgroup$ – Hana Jan 15 at 7:25
  • $\begingroup$ If we had uniform convergence then there would be $a_0$ such that $\int |f_n(x-a)-f_n(x)|\, dx <1$ for all $n$ for all $a<a_0$. In my example you can get a contradiction to this by fixing $a$, say $a=a_0/2$ and letting $n \to \infty$. $\endgroup$ – Kavi Rama Murthy Jan 15 at 7:28
  • $\begingroup$ Thanks, I understand that now. But I cannot get that limit is infinity: $\int_a^{1-a}|n(x-a)^n-nx^n|= \frac{n}{n+1}((1 - a)^{1+n} - a^{1 + n}-(1-2a)^{1+n})$ this converge to 0 for $a\in(0,1)$ when $n\rightarrow\infty$ or I made some mistake. $\endgroup$ – Hana Jan 15 at 16:03
  • $\begingroup$ @Hana I was a bit careless with details but the sequence $\{f_n\}$ does work. Please look at my revised answer. $\endgroup$ – Kavi Rama Murthy Jan 15 at 23:21
  • $\begingroup$ Thanks! Whether the uniform convergence of this operator $T_a$ depends on p? I.e. whether the uniform convergence will be valid for some $p$? $\endgroup$ – Hana Jan 17 at 19:25

Your Answer

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

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