Inverse Fourier Transform for sine and cosine

wolly

Junior Member
Joined
Jul 18, 2018
Messages
116
Hi I want to find the inverse of the sine and cosine functions in the Fourier Transform

f(x)=2π0FC(ω)cos(ωt)dωf(x)=\frac{\sqrt{2}}{\sqrt{\pi}} \int_{0}^{\infty} F_{C}(ω)cos(ωt)dω
and

f(x)=2π0FC(ω)sin(ωt)dωf(x)=\frac{\sqrt{2}}{\sqrt{\pi}} \int_{0}^{\infty} F_{C}(ω)sin(ωt)dω
How can these 2 get this result?
 

Attachments

  • asgfafg.PNG
    asgfafg.PNG
    7.9 KB · Views: 7
Last edited:
Hi I want to find the inverse of the sine and cosine functions in the Fourier Transform
I think that you mean here that you want to find the inverse of a function using the sine and cosine Fourier Transforms.

How can these 2 get this result?
What do you mean?

In the picture, they want to find the inverse Fourier transform of 11+t2\displaystyle \frac{1}{1 + t^2} using the cosine Fourier Transforms.
 
What I mean îs how did 1π\frac{1}{\pi} and 2π\frac{2}{\pi} appear in that integral ?when the inverse of Fourier functions are posted above.
 
And another thing
How did This function transform into

g(z)=eizxz2+1g(z)= \frac{e^{izx}}{z^2+1}?
 
In my book This exercise îs solved by Jordan lemma and I Don't know how to apply it1000030966.jpg
 
What I mean îs how did 1π\frac{1}{\pi} and 2π\frac{2}{\pi} appear in that integral ?when the inverse of Fourier functions are posted above.
2π\displaystyle \frac{2}{\pi} is part of the inverse sine or cosine Fourier transform.

If you have \displaystyle \int_{-\infty}^{\infty}, you can take half the domain and double the integral, so =20 \displaystyle \int_{-\infty}^{\infty} = 2\int_{0}^{\infty}\ Or  20=\displaystyle \ 2\int_{0}^{\infty} = \int_{-\infty}^{\infty}


This is why 2π0=1π\displaystyle \frac{2}{\pi}\int_{0}^{\infty} = \frac{1}{\pi}\int_{-\infty}^{\infty}

And another thing
How did This function transform into

g(z)=eizxz2+1g(z)= \frac{e^{izx}}{z^2+1}?
If the domain is <z<\displaystyle -\infty < z < \infty, use the general Fourier transform:

f(x)=12πf(z)eizx dz\displaystyle f(x) = \frac{1}{2\pi}\int_{-\infty}^{\infty} f(z) e^{izx} \ dz


If you want to work in the domain is 0<z<\displaystyle 0 < z < \infty, and f(z)\displaystyle f(z) is an even function, you can use the cosine Fourier transform:

f(x)=2π0f(z)cos(zx) dz=1πf(z)cos(zx) dz\displaystyle f(x) = \frac{2}{\pi}\int_{0}^{\infty} f(z) \cos(zx) \ dz = \frac{1}{\pi}\int_{-\infty}^{\infty} f(z) \cos(zx) \ dz


If you want to work in the domain is 0<z<\displaystyle 0 < z < \infty, and f(z)\displaystyle f(z) is an odd function, you can use the sine Fourier transform:

f(x)=2π0f(z)sin(zx) dz=1πf(z)sin(zx) dz\displaystyle f(x) = \frac{2}{\pi}\int_{0}^{\infty} f(z) \sin(zx) \ dz = \frac{1}{\pi}\int_{-\infty}^{\infty} f(z) \sin(zx) \ dz


Now let us apply the idea on f(z)=1z2+1\displaystyle f(z) = \frac{1}{z^2 + 1}. To find the inverse Fourier transform of this function, we usually use the general Fourier transform.

f(x)=12π1z2+1eizx dz\displaystyle f(x) = \frac{1}{2\pi}\int_{-\infty}^{\infty} \frac{1}{z^2 + 1} e^{izx} \ dz

We know that f(z)=1z2+1\displaystyle f(z) = \frac{1}{z^2 + 1} is an even function, so we have the option to solve it by the cosine Fourier transform on the domain 0<z<\displaystyle 0 < z < \infty.

f(x)=2π01z2+1cos(zx) dz=1π1z2+1cos(zx) dz=ex\displaystyle f(x) = \frac{2}{\pi}\int_{0}^{\infty} \frac{1}{z^2 + 1} \cos(zx) \ dz = \frac{1}{\pi}\int_{-\infty}^{\infty} \frac{1}{z^2 + 1} \cos(zx) \ dz = e^{-x}

But this answer is only for x>0\displaystyle x > 0. What about x<0\displaystyle x<0? Therefore the full answer of the transform on <z<\displaystyle -\infty < z < \infty and <x<\displaystyle -\infty < x < \infty is:

F1{1z2+1}={exx>0exx<0=ex\displaystyle \mathcal{F^{-1}}\left\{\frac{1}{z^2 + 1}\right\} = \begin{cases}e^{-x} & x > 0\\e^{x} & x < 0\end{cases} = e^{-|x|}
 
In my book This exercise îs solved by Jordan lemma and I Don't know how to apply it
I will explain to you the idea of the Jordan lemma. If you solve integrals in complex analysis, you use a contour (path). This contour splits the original integral into two or more integrals. One of these integrals involves an arc (usually an arc of a semicircle). The Jordan lemma says that the integral that involves the arc =0\displaystyle = 0. That's the whole idea.

Now let us apply it step by step.

g(z)=eizxz2+1g(z)= \frac{e^{izx}}{z^2+1}?
First step is that we have to define our complex function.

Let our complex function be g(z)=eizxz2+1\displaystyle g(z)= \frac{e^{izx}}{z^2+1}.

We realize that this function has two poles z=i\displaystyle z = i and z=i\displaystyle z = -i

Step 2. We will choose a contour. Let our contour be γ\displaystyle \gamma. We will let this contour contains only one of the poles. What about the top part of a semicircle of radius R\displaystyle R and R>1\displaystyle R > 1. This means that the pole z=i\displaystyle z = i will be inside the contour (the semicircle) and we can compute its residue.

This is our contour: We will walk from z=R\displaystyle z = -R to z=R\displaystyle z = R (real line), then from z=R\displaystyle z = R back to z=R\displaystyle z = -R through the semicircular arc which we will call γR\displaystyle \gamma_R. As you can see, our contour is perfectly closed.

Then, our integral is:

γg(z) dz=RRg(z) dz+γRg(z) dz\displaystyle \int_{\gamma} g(z) \ dz = \int_{-R}^{R} g(z) \ dz + \int_{\gamma_R} g(z) \ dz


Step 3. Apply the residue theorem.

γg(z) dz=2πiRes(g,i)\displaystyle \int_{\gamma} g(z) \ dz = 2\pi i \cdot \text{Res}(g,i)

This means that:

RRg(z) dz+γRg(z) dz=2πiRes(g,i)\displaystyle \int_{-R}^{R} g(z) \ dz + \int_{\gamma_R} g(z) \ dz = 2\pi i \cdot \text{Res}(g,i)


Step 4. Apply the limit to both sides.

limRRRg(z) dz+limRγRg(z) dz=limR2πiRes(g,i)\displaystyle \lim_{R \rightarrow \infty}\int_{-R}^{R} g(z) \ dz + \lim_{R \rightarrow \infty} \int_{\gamma_R} g(z) \ dz = \lim_{R \rightarrow \infty} 2\pi i \cdot \text{Res}(g,i)


Step 5. Use the Jordan lemma.

limRγRg(z) dz=0\displaystyle \lim_{R \rightarrow \infty} \int_{\gamma_R} g(z) \ dz = 0

Then,

limRRRg(z) dz=limR2πiRes(g,i)\displaystyle \lim_{R \rightarrow \infty}\int_{-R}^{R} g(z) \ dz = \lim_{R \rightarrow \infty} 2\pi i \cdot \text{Res}(g,i)

Or

limRRRg(z) dz=2πiRes(g,i)\displaystyle \lim_{R \rightarrow \infty}\int_{-R}^{R} g(z) \ dz = 2\pi i \cdot \text{Res}(g,i)


Step 6. Calculate the residue.

Res(g,i)=ex2i\displaystyle \text{Res}(g,i) = \frac{e^{-x}}{2i}

Then,

limRRRg(z) dz=πex\displaystyle \lim_{R \rightarrow \infty}\int_{-R}^{R} g(z) \ dz = \pi e^{-x}

Or

γg(z) dz=limRRRg(z) dz=πex\displaystyle \int_{\gamma} g(z) \ dz = \lim_{R \rightarrow \infty}\int_{-R}^{R} g(z) \ dz = \pi e^{-x}

Or

eizxz2+1 dz=πex\displaystyle \int_{-\infty}^{\infty} \frac{e^{izx}}{z^2 + 1} \ dz = \pi e^{-x}

Since we have chosen the contour as the upper half of the semicircle, we were actually calculating only x>0\displaystyle x > 0.

In other words, we were calculating the cosine Fourier transform.

Then,

cos(xz)z2+1 dz=Re(eizxz2+1 dz)=πex\displaystyle \int_{-\infty}^{\infty}\frac{\cos(xz)}{z^2 + 1} \ dz = \text{Re}\left(\int_{-\infty}^{\infty} \frac{e^{izx}}{z^2 + 1} \ dz\right) = \pi e^{-x}

Or

1πcos(xz)z2+1 dz=ex\displaystyle \frac{1}{\pi}\int_{-\infty}^{\infty}\frac{\cos(xz)}{z^2 + 1} \ dz = e^{-x}

The same result we have obtained in post #8\displaystyle 8.

Now if you wanna calculate x<0\displaystyle x < 0 using the Jordan lemma, you do the same calculations we have done but you choose your contour to be the bottom part of the semicircle.

You will get: eizxz2+1 dz=πex\displaystyle \int_{-\infty}^{\infty} \frac{e^{izx}}{z^2 + 1} \ dz = \pi e^{x}

And by combining the two solutions together, you will get again:

f(x)=F1{1z2+1}={exx>0exx<0=ex\displaystyle f(x) = \mathcal{F^{-1}}\left\{\frac{1}{z^2 + 1}\right\} =\begin{cases}e^{-x} & x > 0\\e^{x} & x < 0\end{cases} = e^{-|x|}
 
Last edited:
Top