5
$\begingroup$

Let $f \in L^1 \cap L^\infty(\mathbb R)$ and $K \in BV(\mathbb R)$. Do these assumptions suffice to prove that for the convolution $K \ast f$ we have that $$K \ast f \in W^{1,\infty}(\mathbb R)$$ holds?

From this post, I know that this convolution is BV, but is it true that it is also $W^{1,\infty}$?

$\endgroup$

2 Answers 2

3
$\begingroup$

You can also use that for $g\in L^p(\mathbf{R})$, \begin{align*} \|\tau_h g-g\|_p\lesssim |h|, \end{align*} characterizes elements $g$ of $W^{1,p}(\mathbf{R})$ for $p>1$ and elements of $BV(\mathbf{R})$ for $p=1$. In your case you have for $h\in\mathbf{R}$, by Young's inequality (which boils down to Hölder's with the exponents involved here) \begin{align*} \|\tau_h (f\star K)-f\star K\|_\infty = \|f\star(\tau_h K-K)\|_\infty \leq \|f\|_\infty \|\tau_h K-K\|_1 \lesssim\|f\|_\infty |h|. \end{align*}

$\endgroup$
4
  • $\begingroup$ Thank you! Could you make the Lipschitz constant in the last inequality explicit? $\endgroup$
    – Hiro
    Jan 9, 2021 at 20:08
  • $\begingroup$ Well it should be $|K'|(\mathbf{R})$. $\endgroup$ Jan 9, 2021 at 21:12
  • $\begingroup$ What does it denote? The total variation of $K$? $\endgroup$
    – Hiro
    Jan 9, 2021 at 22:04
  • $\begingroup$ Yes. If $K\in W^{1,1}(\mathbf{R})$, it would be $\|K'\|_1$ $\endgroup$ Jan 9, 2021 at 22:19
2
$\begingroup$

Without loss of generality $K$ is bounded increasing left-continuous, then $K(x)=m(-\infty,x)$ for a bounded non-negative measure $m$. We have by Fubini theorem$$(K*f)(x)=\int K(y)f(x-y)dy=\int\int\mathbf{1}(z<y) f(x-y)dm(z) dy\\= \int dm(z) \int_z^\infty f(x-y)dy=\int dm(z)\int_{-\infty}^{x}f(t-z)dt=\int_{-\infty}^xdt\int f(t-z)dm(z)$$ is an "antiderivative" of a bounded $L^1$ function $x\to \int f(x-z)dm(z)$.

$\endgroup$
2
  • $\begingroup$ The function $h(x):=\int f(x-z)dm(z)$ is essentially bounded (since $f$ is bounded and $m$ is a bounded measure). Thus $\int_{-\infty}^x h(t)dt$ is Lipschitz in $x$. $\endgroup$ Jan 9, 2021 at 12:34
  • $\begingroup$ Thank you. I put the quesiton on the second derivative in a separate post: mathoverflow.net/questions/380772/… $\endgroup$
    – Hiro
    Jan 9, 2021 at 13:24

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.