Let $f$ be a bounded tempered distribution, that is, $f\ast\varphi \in L^\infty(\mathbb R^n) $ for every Schwartz function $\varphi$. If $g \in L^1(\mathbb R^n)$, does the following definition define a tempered distribution:

$$ \langle f \ast g,\varphi \rangle = \int\limits_{{\mathbb R^n}} {(f \ast \tilde \varphi } )(x)\tilde g(x) \, dx$$

where $\varphi$ is a Schwartz function and $\tilde \varphi(x)=\varphi(-x)$?

  • $\begingroup$ I changed $<\langle f * g,\varphi >$ to $\langle f * g,\varphi \rangle$. That is standard usage. $\endgroup$ – Michael Hardy Oct 3 '12 at 0:32

Yes. Take a look on this book in the part of tempered distributions:

"Michael Eugene Taylor - Partial Differential Equations Volume I Basic Theory"

Ok lets try. Let $S$ be the space of Schwartz functions.

Note first that this number is well defined:

\begin{eqnarray} |\langle f\ast g,\phi\rangle| &\leq& \int_{\mathbb{R}^{n}}|f\ast \tilde{\phi}||\tilde{g}| \nonumber \\ &\leq& \|f\ast \tilde{\phi}\|_{\infty}\|\tilde{g}\|_{1} \nonumber \end{eqnarray}

On the other hand, as you can see in that book, $f\ast \tilde{\phi}$ is a tempered distribution, so its is continuous i.e. $$(\forall\phi)\Bigl(\phi\in S\Rightarrow |f\ast \tilde{\phi}|\leq Cp_{k}(\tilde{\phi})\Bigr)$$ where $p_{k}(\phi)$ is defined as there "in the book".


\begin{eqnarray} |\langle f\ast g,\phi\rangle| &\leq& \int_{\mathbb{R}^{n}}|f\ast \tilde{\phi}||\tilde{g}| \nonumber \\ &\leq& Cp_{k}(\tilde{\phi})\|\tilde{g}\|_{1} \nonumber \end{eqnarray}

With the last inequality you can conclude.

  • $\begingroup$ I had a look in the book as far as it was available online. But I couldn't see an explicit statement for the continuity in $\varphi \in \mathcal{S}$, as - for me - this is not obvious. $\endgroup$ – Vobo Oct 3 '12 at 12:04
  • $\begingroup$ Even though we know $S \rightarrow S'$ which is given by $\phi\rightarrow f*\phi$ is continuous, it does not say $\phi\rightarrow f*\phi$ is continuous as $S \rightarrow L^\infty$. $\endgroup$ – Hezudao Oct 3 '12 at 18:19
  • 1
    $\begingroup$ Let me try explain. A priori $f\ast \tilde{\phi}$ isnt a function, so what you have to do? You have to take its representative. What is the representative? $(f\ast \tilde{\phi})(x)$ is defined by $$(f\ast \tilde{\phi})(x)=\langle f,\tau_{x}\phi\rangle$$ where $\tau_{x}\phi=\phi(x+y)$. Now use the continuity of $f$ as a distribution. $\endgroup$ – Tomás Oct 3 '12 at 23:07
  • 1
    $\begingroup$ I know $f*\tilde\phi(x)$ is a function (actually it is smooth), and this function is bounded because of the assumption about $f$. But it is not clear if $\phi\rightarrow f*\phi$ is continuous, that is, $||f*\tilde\phi||_\infty \leq C p_k(\phi)$ where $p_k$ is some seminorm in $S$. $\endgroup$ – Hezudao Oct 4 '12 at 5:06
  • $\begingroup$ @Hezudao, the last comment of Tomas has the key argument: You know that $|<f,\tau_x\phi>| <= Cp_k(\tau_x\phi)$ for all $x$, and hence it is bounded uniformly. $\endgroup$ – Vobo Oct 4 '12 at 6:08

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.