The Poisson Summation Formula, The Sampling Theorem, and Dirac Combs

The Poisson summation formula, the sampling
theorem, and Dirac combs

Jordan Bell
jordan.bell@gmail.com
Department of Mathematics, University of Toronto
April 3, 2014
Poisson summation formula
Let S(R) be the set of all infinitely differentiable functions f on R such that for
all nonnegative integers m and n,
m,n (f ) = sup |x|m |f (n) (x)|
xR
is finite. For each m and n, m,n is a seminorm. This is a countable collection of

seminorms, and S(R) is a Frchet space. (One has to prove for vk S(R) that
if for each fixed m, n the sequence m,n (fk ) is a Cauchy sequence (of numbers),
then there exists some v S(R) such that for each m, n we have m,n (f fk )
0.) I mention that S(R) is a Frchet space merely because it is satisfying to
give a structure to a set with which we are working: if I call this set Schwartz
space I would like to know what type of space it is, in the same sense that an
Lp space is a Banach space.
For f S(R), define
Z
f() =
eix f (x)dx,
R
and then
1
f (x) =
2
eix f()d.
Let T = R/2Z. For C (T), define

1
(n)
=
2
eint (t)dt,
and then
(t) =
X
nZ
int
(n)e
.
For f S(R) and 6= 0, define : T C by

X
(t) = 2
f (t + 2j),
jZ
where f (x) = f (x). We have C (T).

For n Z,
Z 2
X
c (n) = 1
eint 2
f (t + 2j)dt
2 0
jZ
X Z 2
eint f (t + 2j)dt
=
jZ
Z
=
einx f (x)dx
= fc
(n).

c
One checks that fc

() = f , so (n) = f
(t) =
. Thus for t T,
X n
eint .
f
nZ
Therefore, if f S(R) then for each t T,

2
f (t + 2j) =
jZ
X n
eint ,
f
nZ
where f (x) = f (x). This is the Poisson summation formula.
Nyquist-Shannon sampling theorem
Let f S(R). Suppose that there is some L such that f (x) = 0 if |x| > L
(namely, f is spacelimited). We can in fact choose so that all terms with j 6= 0
on the left-hand side of the Poisson summation formula are zero, which then
expresses f (t) in terms of discrete samples of the Fourier transform of f .
If there is an L such that f() = 0 for || > L, we say that f is bandlimited.
(For f : T C this correspond to f being a trigonometric polynomial.) We can
prove a dual Poisson formula and apply it to bandlimited functions, but instead
I am going to give a direct argument following Charles L. Epstein, Introduction
to the Mathematics of Medical Imaging, second ed., which is a good reference
for sampling.
Suppose that f S(R) and f () = 0 for || > L.1 For , define
fact, if f, f L1 (R) and f is bandlimited then it follows that f is infinitely differentiable,
although it does not follow that f is Schwartz.
1 In
F : T C by2

L
F () = f
.
For n Z,
Fb(n)
Z
1
ein F ()d
2

Z
1
L
ein f
d
2
Z L
t
1
ein L f(t)dt.
2L L
=
=
=
On the other hand, for x R, using the fact that f is bandlimited,

1
f (x) =
2
Z
e
ix
1
f()d =
2
Therefore for n Z,
Fb(n) = f
L
n
L
eix f()d.

.
As F C (T), for t T we have

F (t) =
Fb(n)eint =
nZ
X n
f
eint .
L
L
nZ
Hence for L L,
f() = F
X n in
=
f
e L .
L
L
nZ
For 0 < L, let : R R satisfy

(
1 || L
() =
0 || > L.
For instance, for = 0 let be the characteristic function on [L, L]. But if
> 0 we can choose to be smooth rather than a characteristic function. If
because
f() = 0 for || > L (rather than just for || > L), then f = f,
2 We are defining a periodic function F from a function f that is not periodic, but rather
which becomes 0 after L.
is equal to 1 on the support of f. Then for x R,

Z
1
f (x) =
eix f(xi)()d
2 R
Z
X n in
1
ix
e
f
e L ()d
=
2 R
L
L
nZ

Z
in
1 X
n
=
f
eix+ L ()d
2L
L
R
nZ

1 X
n
n
=
f
2 x +
2L
L
L
nZ

X
n
n
f
=
x+
.
L
L
L
nZ

can be chosen so that x + n
L quickly goes to 0 as n . This means that
fewer terms of the above series need to be calculated to get a good approximation
of f (x). For the characteristic function on [L, L], we have
Z L
1
sin(xL)
L
(x) =
eix d =
= sinc(xL),
2 L
x
where sinc(x) =
sin x
x .
Then we obtain, for x R,

X n
n
sinc x +
,
f (x) =
f
L
L
nZ
or
f (x) =

X n
n
f
sinc x
.
L
L
nZ
Thus if f is bandlimited, we can sample f at a sequence of points and using

these values reconstruct f (x) for any x R. This fact is the Nyquist-Shannon
sampling theorem, and the actual formula above is the Whittaker-Shannon interpolation formula.
A pleasant exposition of sampling is given in Sampling: What Nyquist Didnt Say, and What to Do About It, by Tim Wescott, 2010.
Dirac combs, pulse trains
Define a tempered distribution X on R by

X
X(t) =
(t n).
nZ
This is called a Dirac comb or pulse train. If f S(R),

X
Xf =
f (n).
nZ
The convolution of f and X is the function

X
f (t n).
(X f )(t) =
nZ
Define

1
t
X
.
T
T
Because (t/a) = |a|(t) (the scaling property of the Dirac delta distribution;
we have an absolute value sign because if a is negative then doing a change of
variables in the definition of (t/a) we get two negative signs), if T > 0 then

X
X 1
t
XT (t) =
n =
(t nT ),
T
T
XT (t) =
nZ
nZ
and we have
XT f =
f (nT )
nZ
and
(XT f )(t) =
f (t nT ).
nZ
XT (t + T ) = XT (t), so XT is a periodic distribution on R. It follows that

XT is a distribution on T. Folland talks about periodic distributions around
p. 298 of his Real Analysis, second
ed.
R
1
Let Cc (R) such that R (x)dx = 2
, and let = [0,2) . If we have
a 2-periodic distribution F on R and C (T), we define F to be F ()
( is viewed in this instance as a 2-periodic function on R). For C (T)
we have, viewing as a 2-periodic function on R (thus (2n) = (0)),
X
X2 () =
(t 2n)()
nZ
(2n)(2n)
nZ
= (0)
XZ
nZ
(x)[0,2) (2n x)dx
Z
= (0)
(x)
R
[0,2) (2n x)dx
nZ
Z
= (0)
(x)
R
(0)
.
2
Thus as a distribution on T, for C (T) we have X2 =

1
int
[
X
= 2
. Thus the Fourier series of X2 is
2 (n) = X2 e
1 X int
e .
2
nZ
(0)
2 .
For n Z,

The Poisson Summation Formula, The Sampling Theorem, and Dirac Combs

Uploaded by

Copyright:

Available Formats

The Poisson Summation Formula, The Sampling Theorem, and Dirac Combs

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Poisson Summation Formula, The Sampling Theorem, and Dirac Combs

Uploaded by

Copyright:

Available Formats

The Poisson summation formula, the sampling

theorem, and Dirac combs

Poisson summation formula

is finite. For each m and n, m,n is a seminorm. This is a countable collection of

Let T = R/2Z. For C (T), define

For f S(R) and 6= 0, define : T C by

where f (x) = f (x). We have C (T).

One checks that fc

Therefore, if f S(R) then for each t T,

where f (x) = f (x). This is the Poisson summation formula.

Nyquist-Shannon sampling theorem

On the other hand, for x R, using the fact that f is bandlimited,

As F C (T), for t T we have

For 0 < L, let : R R satisfy

is equal to 1 on the support of f. Then for x R,

Then we obtain, for x R,

Thus if f is bandlimited, we can sample f at a sequence of points and using

Dirac combs, pulse trains

Define a tempered distribution X on R by

This is called a Dirac comb or pulse train. If f S(R),

The convolution of f and X is the function

XT (t + T ) = XT (t), so XT is a periodic distribution on R. It follows that

(x)[0,2) (2n x)dx

[0,2) (2n x)dx

Thus as a distribution on T, for C (T) we have X2 =

You might also like

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.