Fourier S eries Expansion of Functions in Two or

More Dimensions
by S tefan H ollos , Ric hard Hollos
Exstrom Lab oratories LLC
Longmont, C olorado, US A
http: // www. exstrom. com
Email: s t e f an@e xs t ro m. c o m
May 4, 2004

Abstract
The goal of this article i s to look at the Fouri er series expansion of periodi c functions
i n two or more dimensi ons. We start by qui ckly revi ewing the expansi on of one di mensi onal periodi c functi ons since conceptually there is very little di fference between expansi ons i n one di mension and expansi ons i n hi gher dimensi ons.

1 Expansion in O ne D imension
C onsider a function f ( x) that is periodic with period equal to a : f ( x + a) = f ( x) . The Fourier
series expansion of f ( x) is then
X
ck e i k x
(1)
f ( x) =
k

To find the values of k , we impose the periodicity condition

X
f( x + a) =
ck e i k ( x + a )

( 2)

ck e i k x e i k a

= f ( x)
This means we must have
eik a = 1

or

k a= 2 m

m = integer

The allowable values of k are then

k = 2 m/ a

m=

, 2, 1 , 0, 1 , 2,

( 3)

Equation ( 1 ) can then b e written as

f ( x) =

ck e i 2 m x / a

( 4)

m=
0

To determine the expansion coefficients c k multiply b oth sides of equation ( 4) by e i k x , k 0 =

2 m 0 /a, and integrate over one period of the function
Za/2
X Za/2
0
ik 0x
ei( k k ) x d x
( 5)
ck
f ( x) e
dx =
a/2

a/2

S ec t io n 2

The integral on the right hand side is

Za/2
0
0
ei ( k k ) x
ei( k k ) xd x =
i ( k k 0)
a/2

 a/2




a/2

 a/2

e

0
i 2 ( m m ) / a  a/2
i 2 ( m m 0) x/ a

( 6)

a sin( ( m m 0 ) )
( m m 0)
= a m , m 0
= a k , k 0

S ubstituting this back into eq. ( 5 ) we see that the coefficients c k are given by
Z
1 a/2
ck =
f ( x) e i k x d x
a a/2

( 7)

2 Expansion in Two or More D imensions

We now consider a function f ( r ) in two or more dimensions with the following periodicity:
f ( r + R ) = f( r )

( 8)

In three dimensions the vector R is expressed as

R = n1 a1 + n2 a2 + n3 a3

( 9)

The vectors a 1 , a 2 , a 3 are three linearly independent vectors, not necessarily orthonormal, and
the variables n 1 , n 2 , n 3 take on all integer values. For a given coordinate system the vectors in
eq. ( 9) define a set of points called a Bravais lattice, and are called lattice vectors.
The Fourier series expansion of f ( r ) is in terms of plane waves e i k r
X
f( r ) =
ck e i k r
( 1 0)

The set of allowable wave vectors k is determined by the periodicity condition

X
f( r + R ) =
ck e i k ( r + R )

(11)

ck e i k r e i k R

= f( r )
This means we must have

eik R = 1

or

k R = 2 m

m = integer

To satisfy this condition k must by a reciprocal lattice vector and in three dimensions it is
expressed as
k = m1 b1 + m2 b2 + m3 b3

( 1 2)

The variables m 1 , m 2 , m 3 take on all integer values and the vectors b 1 , b 2 , b 3 are defined as
b1 = 2

a2 a3
a1 ( a2 a3)

b2 = 2

a3 a1
a1 ( a2 a3)

b3 = 2

a1 a2
a1 ( a2 a3)

With this definition b i a j = 2 i j and

k R = 2 ( m1 n1 + m2 n2 + m3 n3) = 2 m

m = integer

( 1 3)

E xp an s io n in T wo o r M o re D imens io ns

To find the expansion coefficients c k we proceed as in the one dimensional case by multiplying
0
b oth sides of eq. ( 1 0) by e i k r and integrating over one period of the function in all directions.
The region of integration corresponds to what is called a primitive unit cell of the Bravais lattice. The integration can b e carried out over any of the primitive unit cells of the lattice ( for a
definition of the primitive unit cell of a Bravais lattice see Ashcroft and Mermin, p. 71 ) .
Z
X Z
0
ik 0 r
ei( k k ) r d r
dr =
ck
( 1 4)
f( r ) e

The vector r can b e written in the a 1 , a 2 , a 3 basis as

a1
a
a
+ x2 2 + x3 3
a1
a2
a3

r = x1

( 15)

where a i = | a i | and x i is a real variable. Note that r is in general not a lattice vector. We then
have
d r = d x1 d x2 d x3
k k 0 = ( m 1 m 10 ) b 1 + ( m 2 m 20 ) b 2 + ( m 3 m 30 ) b 3


x 1 ( m 1 m 10 ) x 2 ( m 2 m 20 ) x 3 ( m 3 m 30 )
0
(k k ) r = 2
+
+
a1
a2
a3
The integral on the right hand side of eq. ( 1 4) is then
Za1 /2
Za2/2
Za3/2
0
0
0
ei 2 x1 ( m1 m1 ) / a 1 d x 1
e i 2 x 2 ( m2 m2 ) / a 2 d x 2
e i 2 x 3 ( m3 m3 ) / a 3 d x3
a1 /2

a2/2

( 1 6)

a3/2

This is the product of three integrals each similar in form to eq. ( 6)

Zaj/2
0
a j sin( ( m j m j0 ) )
ei2 xj( mj mj) /aj d x j =
( m j m j0 )
a j/2

( 1 7)

= a j m j , m j0

The integral is then

ei ( k k

)r

d r = a 1 a 2 a 3 k , k 0

( 1 8)

= Vk , k 0

Where V = a 1 a 2 a 3 is the volume of the primitive unit cell. S ubstituting this back into eq. ( 1 4)
we see that the expansion coefficients are given by
Z
1
ck =
f( r ) e i k r d r
( 1 9)
V V