How do I calculate the Fourier Transform of a/(2pi) 1/(a^2 + x^2)a2π1a2+x2?

Define mathcal(F)_(omega)[f(x)] = hat(f)(omega) as the Fourier transform of f(x):

hat(f)(omega) -= mathcal(F)_(omega)[f(x)] = 1/(sqrt(2pi)) int_(-oo)^(oo) f(x)e^(-iomegax)dx

and define the inverse Fourier transform mathcal(F)_(x)^(-1)[hat(f)(omega)] of hat(f)(omega) as:

f(x) -= mathcal(F)_(x)^(-1)[hat(f)(omega)] = 1/(sqrt(2pi)) int_(-oo)^(oo) hat(f)(omega) e^(iomegax)dx

How would I evaluate a Fourier Transform with:

f(x) = a/(2pi) 1/(a^2 + x^2)?

1 Answer
Oct 9, 2017

int_(-oo)^(oo) \ e^(i omega x )/(a^2+x^2) \ dx = (pi e^(-omega a ))/(a)

Explanation:

I never studied Fourier Transforms, but at the end of the day it is just an integral transform similar to Laplace, so:

I won't worry about the a/(2pi) weight of the function or the 1/sqrt(2pi) weight of the transform, and just calculate

I_x = int_(-oo)^(oo) \ e^(i omega x )/(a^2+x^2) \ dx where x in RR.

In order to compute this definite integral, consider the following complex variable function over a domain CC:

f(z) = e^(i omega z )/(a^2+z^2)

And its associated contour integral:

I_z = oint_C \ f(z) \ dz

Where C is the following semi-circular contour in the complex plane with radius R gt 0:

enter image source here

We will restrict R to enclose the poles in the upper quadrants once we have analyzed the poles of f(z). The denominator of the integrand is a^2+z^2 , and so we have simples poles when z^2+a^2 = 0 => z=+-ai.

![www.wolframalpha.com](https://d2jmvrsizmvf4x.cloudfront.net/SfyBzgPtRuaf5MXXDM3V_gif%26s%3D18)

We need only be concerned with the poles in Q1 and Q2, that (providing we make R large enough) lie within our contour C, that is the pole z=ai, assuming that a is positive.

So then:

oint_C \ f(z) \ dz = int_(-R)^(R) \ f(x) \ dx + int_(gamma_R) \ f(z) \ dz

As C encloses a single pole:

alpha = ai (assuming a gt 0)

Then by the residue theorem:

oint_C \ f(z) \ dz = 2pii xx ( ("sum of the residues of the"), ("poles of " f(z) " within "C) )
" " = 2pii \ { res_(z=alpha) \ f(z)

And we can calculate the residues as follows:

res_(z=alpha) = lim_(z rarr alpha) (z-alpha) f(z)
" " = lim_(z rarr ai) (z-ai) ( e^(i omega z )/(a^2+z^2) )
" " = lim_(z rarr ai) (z-ai) ( e^(i omega z )/((z-ai)(z+ai)) )
" " = lim_(z rarr ai) ( e^(i omega z )/(z+ai) )
" " = e^(i omega ai )/(ai+ai)
" " = e^(-omega a )/(2ai)

Thus:

oint_C \ f(z) \ dz = 2pii e^(-omega a )/(2ai) = (pi e^(-omega a ))/(a)

Now, as we let R rarr oo we get:

oint_C \ f(z) \ dz = int_(-oo)^(oo) \ f(x) \ dx + int_(gamma_R) \ f(z) \ dz

As is often the case with contour integrals, we find that:

lim_(R rarr oo) int_(gamma_R) \ f(z) \ dz = 0

So we will end up with the result:

int_(-oo)^(oo) \ e^(i omega x )/(a^2+x^2) \ dx = oint_C \ f(z) \ dz
" " = (pi e^(-omega a ))/(a)