INTEGRAL TRANSFORMS
Prove the Wiener–Kinchin theorem,
̃C(k)=
√
2 π[ ̃f(k)]∗ ̃g(k). (13.42)
Following a method similar to that for the convolution offandg, let us consider the
Fourier transform of (13.40):
C ̃(k)=√^1
2 π
∫∞
−∞
dz e−ikz
{∫∞
−∞
f∗(x)g(x+z)dx
}
=
1
√
2 π
∫∞
−∞
dx f∗(x)
{∫∞
−∞
g(x+z)e−ikzdz
}
.
Making the substitutionu=x+zinthesecondintegralweobtain
C ̃(k)=√^1
2 π
∫∞
−∞
dx f∗(x)
{∫∞
−∞
g(u)e−ik(u−x)du
}
=
1
√
2 π
∫∞
−∞
f∗(x)eikxdx
∫∞
−∞
g(u)e−ikudu
=
1
√
2 π
×
√
2 π[ ̃f(k)]∗×
√
2 π ̃g(k)=
√
2 π[ ̃f(k)]∗ ̃g(k).
Thus the Fourier transform of the cross-correlation offandgis equal to
the product of [ ̃f(k)]∗and ̃g(k) multiplied by
√
2 π. This a statement of the
Wiener–Kinchin theorem. Similarly we can derive the converse theorem
F
[
f∗(x)g(x)
]
=
1
√
2 π
̃f⊗ ̃g.
If we now consider the special case wheregis taken to be equal tofin (13.40)
then, writing the LHS asa(z), we have
a(z)=
∫∞
−∞
f∗(x)f(x+z)dx; (13.43)
this is called theauto-correlation functionoff(x). Using the Wiener–Kinchin
theorem (13.42) we see that
a(z)=
1
√
2 π
∫∞
−∞
̃a(k)eikzdk
=
1
√
2 π
∫∞
−∞
√
2 π[ ̃f(k)]∗ ̃f(k)eikzdk,
so thata(z) is the inverse Fourier transform of
√
2 π| ̃f(k)|^2 , which is in turn called
theenergy spectrumoff.
13.1.9 Parseval’s theorem
Using the results of the previous section we can immediately obtainParseval’s
theorem. The most general form of this (also called themultiplication theorem)is