Fourier Transform of the Gaussian

Konstantinos G. Derpanis October 20, 2005

In this note we consider the Fourier transform1 of the Gaussian. The Gaussian function, g(x), is dened as, g(x) = where by:

x2 1 e 22 , 2

(3)

g(x)dx = 1 (i.e., normalized). The Fourier transform of the Gaussian function is given G() = e
2 2 2

(4)

Proof: We begin with dierentiating the Gaussian function: x dg(x) = 2 g(x) dx Next, applying the Fourier transform to both sides of (5) yields, iG() =
dG() d

(5)

1 dG() i 2 d

(6) (7)

G() Integrating both sides of (7) yields,

dG( ) d

= 2 .

G( )
0

d =

2 d

(8)

. 2 Since the Gaussian is normalized, the DC component G(0) = 0, thus (9) can be rewritten as, ln G() ln G(0) = ln G() = Finally, applying the exponent to each side yields, eln G() = e G() = e as desired.
1 The
2 2 2

0 2 2

(9)

2 2 2

(10)

(11) (12)

2 2 2

Fourier transform pair is given by:

Z

F () =

f (x)eix dx
Z

(1)

f (x) = where i denotes the complex unit.

1 2

F ()eix d,

(2)