Wiener's lemma explained

In mathematics, Wiener's lemma is a well-known identity which relates the asymptotic behaviour of the Fourier coefficients of a Borel measure on the circle to its atomic part. This result admits an analogous statement for measures on the real line. It was first discovered by Norbert Wiener.

Statement

\mu

on the unit circle

T

, let

\mua=\sumjcj\delta

zj
be its atomic part (meaning that

\mu(\{zj\})=cj0

and

\mu(\{z\})=0

for

z\not\in\{zj\}

. Then

\limN\toinfty

1
2N+1
N|\widehat\mu(n)|
\sum
n=-N
2=\sum
j|c
2,
j|
where

\widehat{\mu}(n)=\intTz-nd\mu(z)

is the

n

-th Fourier coefficient of

\mu

.

\mu

on the real line

R

and called

\mua=\sumjcj\delta

xj
its atomic part, we have

\limR\toinfty

1
2R
R|\widehat\mu(\xi)|
\int
-R
2d\xi=\sum
j|c
2,
j|
where

\widehat{\mu}(\xi)=\intRe-2\pid\mu(x)

is the Fourier transform of

\mu

.

Proof

\nu

is a complex measure on the circle then
1
2N+1
N\widehat{\nu}(n)=\int
\sum
T

fN(z)d\nu(z),

with
f
N(z)=1
2N+1
N
\sum
n=-N

z-n

. The function

fN

is bounded by

1

in absolute value and has

fN(1)=1

, while
f
N(z)=zN+1-z-N
(2N+1)(z-1)
for

z\inT\setminus\{1\}

, which converges to

0

as

N\toinfty

. Hence, by the dominated convergence theorem,

\limN\toinfty

1
2N+1
N\widehat{\nu}(n)=\int
\sum
T

1\{1\

}(z)\,d\nu(z)=\nu(\).We now take

\mu'

to be the pushforward of

\overline\mu

under the inverse map on

T

, namely

\mu'(B)=\overline{\mu(B-1)}

for any Borel set

B\subseteqT

. This complex measure has Fourier coefficients

\widehat{\mu'}(n)=\overline{\widehat{\mu}(n)}

. We are going to apply the above to the convolution between

\mu

and

\mu'

, namely we choose

\nu=\mu*\mu'

, meaning that

\nu

is the pushforward of the measure

\mu x \mu'

(on

T x T

) under the product map

:T x T\toT

. By Fubini's theorem

\widehat{\nu}(n)=\intT x T(zw)-nd(\mu x \mu')(z,w) =\intT\intTz-nw-nd\mu'(w)d\mu(z)=\widehat{\mu}(n)\widehat{\mu'}(n)=|\widehat{\mu}(n)|2.

So, by the identity derived earlier,

\limN\toinfty

1
2N+1
N|\widehat{\mu}(n)|
\sum
n=-N
2=\nu(\{1\})=\int
T x T

1\{zw=1\

}\,d(\mu\times\mu')(z,w).By Fubini's theorem again, the right-hand side equals

\intT\mu'(\{z-1\})d\mu(z)=\intT\overline{\mu(\{z\})}d\mu(z)=\sumj|\mu(\{z

2=\sum
j|c
2.
j|
1
2R
R\widehat\nu(\xi)d\xi=\int
\int
R

fR(x)d\nu(x)

(which follows from Fubini's theorem), where
f
R(x)=1
2R
R
\int
-R

e-2\pid\xi

.We observe that

|fR|\le1

,

fR(0)=1

and
f
R(x)=e2\pi-e-2\pi
4\piiRx
for

x0

, which converges to

0

as

R\toinfty

. So, by dominated convergence, we have the analogous identity

\limR\toinfty

1
2R
R\widehat\nu(\xi)d\xi=\nu(\{0\}).
\int
-R

Consequences

\mu

on the circle is diffuse (i.e.

\mua=0

) if and only if

\limN\toinfty

1
2N+1
N|\widehat\mu(n)|
\sum
n=-N

2=0

.

\mu

on the circle is a Dirac mass if and only if

\limN\toinfty

1
2N+1
N|\widehat\mu(n)|
\sum
n=-N

2=1

. (Here, the nontrivial implication follows from the fact that the weights

cj

are positive and satisfy

1=\sumj

2\le\sum
c
j

cj\le1

, which forces
2=c
c
j
and thus

cj=1

, so that there must be a single atom with mass

1

.