Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

It is known that if $g_n: \mathbb{R} \rightarrow \mathbb{R}$, $n=1,2,...$, is in $C_c^{\infty}(\mathbb{R})$, $ \int_\mathbb{R} g_n(x)dx=1$, $supp(g_n) \subset (-r_n,r_n)$,where $0<r_n \rightarrow 0$, then for arbitrary locally integrable $f: \mathbb{R} \rightarrow \mathbb{R}$ the convolution $f*g_n$ is smooth and ,if $f$ continuous, $f*g_n(x) \rightrightarrows f(x)$ on compact subsets of $\mathbb{R}$.

Let now $$g_n(x)=\frac{1}{\pi} \frac{r_n}{r_n^2+x^2},$$ $x\in \mathbb{R}$, $n \in \mathbb{N}$, where $0<r_n \rightarrow 0$. How to show that for integrable $f: \mathbb{R} \rightarrow \mathbb{R}$ the convolution $f*g_n$ is smooth? (It is a part of exercise 10 in Chapter 9 of W. Rudin's book Real and complex analysis.)

Is it also true that if $f$ integrable and continuous, then $f*g_n(x) \rightrightarrows f(x)$ on compact subsets of $\mathbb{R}$?

Thanks in advance!

share|improve this question
2  
For smoothness, show by induction on $d$ that for a fixed $n$, the derivative of $g_n$ of order $d$ can be expressed as $\frac{r_n}{\pi}\frac{p_d(x)}{(r_n^2+x^2)^{d+1}}$ where $p_d$ is a polynomial of degree at most $d$. Then you will be able to show that we can apply dominated convergence theorem. –  Davide Giraudo Mar 17 '12 at 10:47

1 Answer 1

Another way of proving of smoothness is to use theorem 9.8 [Rudin, Real and complex analysis]

Let $h_\lambda(x)=\sqrt{\frac{2}{\pi}} \frac{\lambda}{\lambda^2+x^2}$, for $\lambda>0, x\in \mathbb{R}$, and let $H_\lambda(t)=e^{-\lambda |t|}$.

Then $$H_\lambda (t)=\frac{1}{\sqrt{2\pi}} \int_\mathbb{R} h_\lambda(x) e^{-itx}dx $$ $$h_\lambda(x)=\frac{1}{\sqrt{2\pi}} \int_\mathbb{R} H_\lambda (t) e^{itx} dt$$ ( the function $H_\lambda$ is the Fourier transform of $h_{\lambda}$, and the function
$h_{\lambda}$ is the inverse Fourier transform of $H_\lambda $).

Theorem. Let a function$f: \mathbb{R} \rightarrow \mathbb{R}$ be integrable. Then $$(f*h_\lambda)(x)=\frac{1}{2\pi}\int_\mathbb{R} H_\lambda (t) {\cal F} (f)(t) e^{ixt} dt, $$ where ${\cal F} (f)(t)=\frac{1}{\sqrt{2 \pi}} \int_R f(x) e^{-itx} dx$ - is the Fourier transform of $f$.

(This Theorem follows immediately by Fubini theorem and fact that $h_\lambda$ is the inverse Fourier transform of function $H_\lambda $).

The function under integral in the formula for convolution $f*h_\lambda$ is continuous with respect to $x,t$ and has derivatives of all order with respect to $x$, which (since ${\cal F}(f)$ is bounded) are bounded by integrable functions $t \mapsto C_k t^k e^{-\lambda |t|}$, where $C_k$, for $k=0,1,2...$, are constants. Hence $$(f*h_\lambda)^{(k)}(x)=\frac{1}{2 \pi} \int_R H(\lambda t) {\cal F} (f)(t) (it)^k e^{ixt} dt$$ and consequently the convolution is smooth.

In the same way, taking $h_\lambda(x)=\frac{1}{\lambda} e^{-\frac{x^2}{2\lambda^2}}$, and $H_\lambda(t)=e^{-\frac{\lambda^2 t^2}{2}}$, for $\lambda >0$, $x \in \mathbb{R}$, may show that $f* h_\lambda $ is smooth.

share|improve this answer

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

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