Take the 2-minute tour ×
MathOverflow is a question and answer site for professional mathematicians. It's 100% free, no registration required.

Let $f, g \in L^{1}(\mathbb T)= L^{1} ([-\pi, \pi))$. We define, the Fourier transform of $f$ as follows: $$\hat{f}(n)=\frac{1}{2\pi}\int_{-\pi}^{\pi} f(t) e^{-int} dt, \ (n\in \mathbb Z).$$ It is well known that Fourier transform takes convolution to point wise product:indeed,by Fubini-Tonelli theorem and change of variable, we can derive, $$\widehat{(f\ast g)} (n) = \hat{f}(n) \cdot \hat{g} (n) , (n\in \mathbb Z).$$

Suppose $fg \in L^{1} (\mathbb T)$ also.

My questions:

(1) To proving $\widehat{(fg)}((n) = \hat{f} (n) \ast \hat{g} (n); (n \in \mathbb Z)$, the hypothesis $fg\in L^{1} (\mathbb T)$ is sufficient; and if yes, how to prove it ? (that is, under what conditions Fourier transform takes point wise multiplication to convolution product ) Or, we need to put some more conditions on $f$ and $g$ ?

(2) Let $f, g \ \ \text{and} \ fg \in L^{1} (\mathbb R)$ and also assume $\hat{f}, \hat{h} \in L^{1} (\mathbb R).$ Can we expect $\widehat{(fg)} = \hat{f} \ast \hat{g}$ ?

(3) Let $G$ be locally compact abelian group and $f, g, fg, \hat{f}, \hat{g} \in L^{1} (G)$. Can we expect similar result in this situation ?

My attempt: Fix $n\in \mathbb Z$. By definition we have, $\widehat{(fg)}((n)= \frac{1}{2\pi}\int_{-\pi}^{\pi} ( f(t)\cdot g(t) ) e^{-int} dt $ and $\hat{f}(n) \ast \hat{g} (n)= \sum_{k\in \mathbb Z} \hat{f} (n-k) \hat{g} (k) = \sum_{k\in \mathbb Z} \{\frac{1}{2\pi}\int_{-\pi}^{\pi} f(t) e^{-i(n-k)t} dt \} \{\frac{1}{2\pi}\int_{-\pi}^{\pi} f(y) e^{-ikt} dy\} $

Thanks,

share|improve this question
1  
On a general locally compact abelian group, the Fourier transform acts $\widehat{\cdot} : L^1(G,\mu) \to \text{C}_\infty(\widehat{G})$ where $\mu$ is a Haar measure on $G$, and Pontryagin duality exists between $G$ and $\widehat{G}$ (so that $G\cong \widehat{\widehat{G}}$). In general we have $\widehat{f*g} = \widehat{f}\cdot \widehat{g}$. If $\widehat{f},\widehat{g} \in L^1(\widehat{G},\nu)$ (where $\nu$ is the dual measure to $\mu$), then the duality allows us to say that $\widehat{\widehat{f}*\widehat{g}} = f\cdot g$. That might get you part of the way to an answer. –  Greg Zitelli Feb 11 at 17:28

1 Answer 1

Let $f,g$ be tempered distributions on $\mathbb R^n$ such that $g\in \mathscr O_M$, the so-called multipliers space: $g$ is a smooth function such that $$\forall \alpha, \exists N_\alpha\ge 0,\quad \sup_x\vert(\partial^\alpha g)(x)\vert(1+\vert x\vert)^{-N_\alpha}<+\infty. $$ Then the product $fg$ makes sense as a tempered distribution and we may define the convolution of $\hat f\ast\hat g$ as the Fourier transform of $fg$.

A dual point of view: let $F,G$ be tempered distributions on $\mathbb R^n$ such that $$ \text{supp } F\times\text{supp } G\ni(x,y)\mapsto x+y\in \mathbb R^n\text{ is proper}. $$ Then the convolution makes sense as a tempered distribution and is defined as $$ \langle F\ast G, \phi\rangle=\langle F(x)\otimes G(y), \phi(x+y)\rangle. $$ If $G$ is compactly supported, then $\hat G\in \mathscr O_M$ so that $\hat F\hat G$ makes sense from the above arguments and we have indeed in that case $ \widehat{F\ast G}=\hat F\hat G. $ We can use these remarks to answer your question (2): we have for $f,g\in L^1(\mathbb R)$ so that $fg, \hat f, \hat g\in L^1$, the following absolutely converging integrals \begin{multline} \langle\widehat{fg},\phi\rangle_{\mathscr S',\mathscr S}= \langle{fg},\widehat\phi\rangle_{\mathscr S',\mathscr S}=\int f(x) g(x)\hat \phi(x) dx =\iiint \hat f(\xi) \hat g(\eta)e^{2iπ x(\xi+\eta)}\hat \phi(x) dxd\xi d\eta \\=\iint\hat f(\xi) \hat g(\eta)\phi(\xi+\eta) d\xi d\eta \end{multline} and with absolutely converging integrals $ \langle\hat f\ast \hat g,\phi\rangle_{\mathscr S',\mathscr S}=\int (\hat f\ast \hat g)(\xi)\phi(\xi) d\xi=\iint\hat f(\xi-\eta)\hat g(\eta)\phi(\xi) d\xi d\eta=\iint\hat f(\xi) \hat g(\eta)\phi(\xi+\eta) d\xi d\eta, $ proving the sought equality. The answer to (3) should be similar.

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.