3

Let $f:\mathbb R \to \mathbb R$ such that $$f(x)= \frac{\sin \pi x}{x (x^{2}-1)}$$ for $x\in \mathbb R - \{ 0, -1, 1 \}$ and $f(x):= \pi $ for $x=0$ and $f(x)=-\frac{\pi}{2}$ for $x= -1, 1$.

Let $g:\mathbb R \to \mathbb R$ such that $$g(x): = |f(x)|.$$

As $g\in L^{1}(\mathbb R)$, we get $\hat{g} \in C_{0}(\mathbb R)$.

My questions are: (1) Can we compute the Fourier transform of $g$ ? (2) Is it true that $\int_{\mathbb R} |\hat {g} (\xi)| d\xi = \infty $, that is, $\hat{g} \notin L^{1} (\mathbb R)$ ?


Fourier transform of $f$ :

By complex analysis method one can compute the Fourier transform of cardinal sine function or sinc function : $$\int_ {\mathbb R}\frac{\sin (ax)}{ax} e^{-2\pi i x \xi} dx = \frac{\pi}{a}; (|\xi|\leq \frac{a}{2\pi}); (*).$$

Note that, $$\frac{\sin (\pi x )}{x (1-x^{2})}= \pi \frac{\sin (\pi x)}{\pi x }+ \frac{\pi}{2} \frac{\sin (\pi (x+1))}{\pi (x+1)}+\frac{\pi}{2}\frac{\sin (\pi (x-1))}{\pi (x-1))} .$$

We define $f_{1}:\mathbb R \to \mathbb R$ such that $f_{1}(x)= \frac{\sin (\pi x)}{\pi x}$ for $x\not = 0$ and f(0)=1. Also we define, $f_{1}, f_{2}:\mathbb R \to \mathbb R$ such that $f_{1}(x):= f(x+1)$ and $f_{2}(x):= f(x-1)$.

By $(*)$ we get,

$$ \hat{f_{1}(\xi)}=\begin{cases} 1 & \text{if} \ |\xi| < \frac{1}{2},\\ \frac{1}{2} & \text{if} \ |\xi| = \frac{1}{2}, \\ 0, & \textbf{if} \ |\xi|> \frac{1}{2}. \end {cases}$$

By translation property of Fourier transform we get, $$ \hat{f_{2}(\xi)}=\begin{cases} e^{2\pi i \xi} & \text{if} \ |\xi| < \frac{1}{2},\\ \frac{e^{2\pi i \xi}}{2} & \text{if} \ |\xi| = \frac{1}{2}, \\ 0, & \text {if} \ |\xi|> \frac{1}{2}; \end {cases}$$ and $$ \hat{f_{3}(\xi)}=\begin{cases} e^{-2\pi i \xi} & \text{if} \ |\xi| < \frac{1}{2},\\ \frac{e^{-2\pi i \xi}}{2} & \text{if} \ |\xi| = \frac{1}{2}, \\ 0, & \text {if} \ |\xi|> \frac{1}{2}. \end {cases}$$

Now, by using the linearity of the Fourier transform, we get, $\hat{f}(\xi)= \pi \hat{f}_{1}(\xi)+ \frac{\pi}{2}\hat{f}_{2}(\xi ) + \frac{\pi}{2} \hat{f}_{3}(\xi);$ next by putting above values, $$ \hat{f(\xi)}=\begin{cases} \pi (1 + \cos (2 \pi \xi)) & \text{if} \ |\xi| < \frac{1}{2},\\ 0, & \text {if} \ |\xi|\ge \frac{1}{2}. \end {cases}$$

Clearly, $\hat{f} \in L^{1} (\mathbb R)$.

Fourier transform of $|\frac{\sin x}{x}|$ :

Fourier transform of $\left|\frac{\sin x}{x}\right|$


Thanks to Math fraternity;-)

Inquisitive
  • 3,807
  • Your use of the phrase "such that" is off. – Pedro Jan 07 '14 at 03:43
  • Is there a question in there somewhere? – Igor Rivin Jan 07 '14 at 04:15
  • The questions seem to be directly above the "Fourier Transform of f" line. – Lost Jan 07 '14 at 04:19
  • @Lost whereas the answers are below the line. So why are we being shown this? Is the thought that we have nothing more interesting to read? – Igor Rivin Jan 07 '14 at 04:55
  • @IgorRivin I don't see anything wrong with the post. A question (in two parts) is stated about about $\hat g$. Some partial results about related functions are presented, appropriately separated from the question. They do not yield an answer. Is not "show what you tried" pretty much the motto of Math.SE these days? – Post No Bulls Jan 07 '14 at 05:26
  • @IgorRivin - your guess is as good as mine. – Lost Jan 08 '14 at 23:59

1 Answers1

2

By coincidence, I happened to see this post: Sobolev spaces and integrability of Fourier transforms

It seems to me that your function is in $W^{1,2}$, hence it's Fourier transform is in $L^1$.

To see that $g'$ is in $L^2$, you can see that its derivative is discontinuous at the integers (except $0, \pm 1$), but they are simple discontinuities in that the left and right limits exist. Also, the derivative decays like $1/|x|^3$ as $x \to \infty$.

Stephen Montgomery-Smith
  • 26,430
  • 2
  • 35
  • 64
  • 1
    Also, this argument also shows that the Fourier transform of $|\sin(x)/x|$ is in $L^1$. I know that the plot given in the answer to http://math.stackexchange.com/questions/523886/fourier-transform-of-left-frac-sin-xx-right shows singularities at $\frac1{2\pi}\mathbb Z$, but I believe these will be singularities like $\log(1/x)$ at $x = 0$, and hence integrable. – Stephen Montgomery-Smith Jan 07 '14 at 06:14
  • Prof. S. M ; thanks a lot;- ) – Inquisitive Jan 07 '14 at 13:39
  • Is it true that, $h(x)=|\frac{\sin x} {x}| \in W^{1, 2}$ ? (see, $h'(x)=\frac {x\cos x + \sin x} {x^{2}}$ for $x\not = n \pi$ , $n\in \mathbb Z$; $h'\in L^{2}(\mathbb R)$ ? ) – Inquisitive Jan 07 '14 at 15:51
  • Well, $h'(x) = \pm \frac{x\cos x + \sin(x)}{x^2}$. At $x=0$, the singularities cancel. Away from $x=0$, $h'(x)$ decays like $1/|x|$, hence $|h'(x)|^2$ decays like $1/|x|^2$. – Stephen Montgomery-Smith Jan 07 '14 at 16:08