I have these related questions here that I could really use some help on. I believe there is a related question here although I don't think it is exactly the same...

1) Let $f\in L^1(\mathbb R)$ and $\varphi$ a smooth compactly supported function. Prove that $f\ast \varphi$ is a smooth function whose derivative is $f\ast \varphi'$.

Any help would be much appreciated. I have been trying to work on this for the past couple days but all my ideas didn't lead me very far. Namely, I was trying to use the definition of a convolution. I am not sure though how the smooth and compactly supported $L^1$ function fits into this...

  • 1
    $\begingroup$ For the first part, use the definition of the convolution (and maybe the identity $f \ast g = g \ast f$), and then differentiate under the integral sign (you need to show that this is allowed). $\endgroup$ – PhoemueX Dec 7 '18 at 8:47
  • $\begingroup$ The direct approach is to find a bound for $f \ast \varphi' - f \ast (\varphi(.+y)-\varphi)/y$. The less direct one is to look at $\int f \ast \varphi'$ $\endgroup$ – reuns Dec 7 '18 at 16:44

I have an answer for the first question: recall: $$f \ast \phi= \phi \ast f=\int_\mathbb{R}{\phi(x-t)f(t)dt}$$ Note that in our case the convolution is defined everywhere since $\phi$ is compactly supported and thus bounded, and $f \in L^1$. Now look at the definition of the derivative: $$\frac{f \ast \phi (x+h) - f \ast \phi (x)}{h} = \int_\mathbb{R}{\frac{\phi(x-t+h) - \phi(x-t)}{h}f(t)dt}$$ By the mean value theorem for each fixed $t$ there exists $\theta \in [x-t+h,x-t]$ with $\phi'(\theta)=\frac{\phi(x-t+h) - \phi(x-t)}{h}$. But since it is compactly supported we know that all derevatives of $\phi$ are bounded by some constant $M$, thus we have that the absolute value of the integrand on the right is bounded (without dependence on h) by: $$\vert f(t)\phi(\theta)\vert\le M\vert f(t)\vert$$ which is an integrable function. Now apply the Dominated convergence theorem on some sequence $h_n\rightarrow 0$ and get: $$\underset{n\rightarrow \infty}{lim}\frac{f \ast \phi (x+h_n) - f \ast \phi (x)}{h_n}=\int_\mathbb{R}{\underset{n\rightarrow \infty}{lim}\frac{\phi(x-t+h_n) - \phi(x-t)}{h_n}f(t)dt}$$ $$ = \int_\mathbb{R}{\phi'(x-t) \ast f(t)dt} = \phi' \ast f (x)$$.

Great, we found the derevative of $\phi \ast f$. Now since easily $\phi'$ is again a compactly supported $C_\infty(\mathbb{R})$ function, we can apply the above argument again and again to get that $f \ast \phi$ is in $C_\infty(\mathbb{R})$.

  • $\begingroup$ Thank you very much. I really appreciate the help on this. $\endgroup$ – MathIsHard Dec 9 '18 at 6:35
  • 1
    $\begingroup$ Thanks very much for accepting my first answer on this site! $\endgroup$ – pitariver Dec 9 '18 at 7:10

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.