Math Methods I

Lia Vas

Fourier Series and Fourier Transform

Fourier Series
Recall that a function differentiable any number of times at x = a can be represented as a power
series ∞
X f (n) (a)
an (x − a)n where the coefficients are given by an =
n=0
n!
Thus, the function can be approximated by a polynomial. Since this formula involves the n-th deriva-
tive, the function f should be differentiable n-times at a. So, any function which is discontinuous
(has jumps or breaks) or not differentialbe (has corners or sharp turns) cannot be represented as a
power series. This significantly restricts functions we could expand in a power series. Some functions
frequently considered in signal processing, electrical circuits and other applications can be discontin-
uous or not differentiable. For those functions, one considers a different kind of series: the type of
series which can handle discontinuous functions.

One such the type of series is the Fourier Se-

ries. It represents a periodic function with a
period T in the form
∞
a0 X 2πnx 2πnx
f (x) = + (an cos + bn sin )
2 n=1
T T
The coefficients an and bn are called Fourier co-
efficients.
Fourier series can be used for non-periodic functions defined on a finite interval. Say that a
function f (x) is defined on an interval which we denote by (x0 , x0 + T ). Extend the function outside
this interval by copying the graph on intervals (x0 +T, x0 +2T ), (x0 +2T, x0 +3T ), (x0 +3T, x0 +4T ), . . .
right from (x0 , x0 + T ) as well as to intervals (x0 − T, x0 ), (x0 − 2T, x0 − T ), (x0 − 3T, x0 − 2T ), . . . left
from from (x0 , x0 + T ). Note that this makes the extension automatically periodic with the period
T.

To find the Fourier series of such extension, f (x) has to (1) have just a finite number of maxima
and minima and just a finite number of discontinuities on (x0 , x0 + T ), (2) the integral of |f (x)| over
(x0 , x0 + T ) must converge. In our examples, these conditions are always going to be satisfied.

1
 If these conditions are satisfied and one period of f (x) is given on an interval (x0 , x0 + T ), the
Fourier coefficients an and bn can be computed using the formulas
2 x0 +T 2 x0 +T
Z Z
2nπx 2nπx
an = f (x) cos dx bn = f (x) sin dx
T x0 T T x0 T
Obtaining the formulas for coefficients. If f (x) has a Fourier series expansion
∞
a0 X 2πnx 2πnx
f (x) = + (an cos + bn sin )
2 n=1
T T

one can prove the formulas for Fourier series coefficients an by multiplying this formula by cos 2πnx
T
and integrating over one period, say (x0 , x0 + T )). Not to confuse n we picked here from the n
appearing in the sum, make the sum have terms dependent of m where the sum start with m = 1.
For the first term (in front of the sum), think that the value of m is zero.
Z x0 +T Z x0 +T
2πnx a0 2πnx
f (x) cos dx = cos dx +
x0 T x0 2 T
∞ Z x0 +T Z x0 +T 
X 2πmx 2πnx 2πmx 2πnx
am cos cos dx + bm sin cos dx
m=1 x0 T T x0 T T
All the integrals with m = 6 n are zero and the integrals with both sin and cosine functions are
zero as well. Thus,
Z x0 +T Z x0 +T
2πnx 2πnx 2πnx
f (x) cos dx = an cos cos dx
x0 T x0 T T

This last integral can be calculated to be an T2 using the trigonometric identity cos2 α = 21 (1 + cos 2α).
Hence Z x0 +T
2 x0 +T
Z
2πnx T 2πnx
f (x) cos dx = an ⇒ an = f (x) cos dx.
x0 T 2 T x0 T
The formula for bn is proved similarly, multiplying by sin 2πnx
T
instead of cos 2πnx
T
.
At first it may seem strange that a discontinuous function can be equal to a sum of continuous
sine and cosine functions. The following figure illustrates how adding more terms of the series makes
the series match a discontinuous function better and better.

2
 When studying phenomena that are periodic in time, the term 2πT
in the above formula is usually
replaced by ω and t is used to denote the independent variable. Thus,
∞
a0 X
f (t) = + (an cos nωt + bn sin nωt)
2 n=1

If the interval (x0 , x0 + T ) is of the form (−L, L) (thus T = 2L), then the above formulas become
the following.
∞
1 L 1 L
Z Z
nπx nπx a0 X nπx nπx
an = f (x) cos dx, bn = f (x) sin dx, f (x) = + (an cos +bn sin )
L −L L L −L L 2 n=1 L L

Symmetry considerations. Note that if f (x) is even (that is f (−x) = f (x)), then bn = 0. Since
RL 1 0 1 L
bn = L1 −L f (x) sin nπx nπx
f (x) sin nπx
R R
L
dx = L −L
f (x) sin L
dx + L 0
dx. Using the substitution
R L
−1 0
u = −x for the first integral we obtain that it is equal to L L f (−u) sin −nπu du. Using that
−nπu
R0 RL R L
1 L
f (−x) = f (x), that sin L = − sin L , and that − L = 0 , we obtain L 0 f (u)(− sin nπu
nπu
L
) du =
−1 L nπu
R
L 0
f (u) sin L du. Note that this is exactly the negative of the second integral. Thus, the first
and the second integral R cancel andnπxwe obtain that bn = 0.
1 L 1 0 1 L
Similarly, an = L −L f (x) cos L dx = L −L f (x) cos nπx f (x) cos nπx
R R
L
dx + L 0R L
dx. Using the
−1 0 −nπu
substitution u = −x for the first integral we obtain that it is equal to L L f (−u) cos L du. Using
R0 RL 1 L
that f (−x) = f (x), that cos −nπu nπu
f (u) cos nπu
R
L
= cos L
, and that − L
= 0
, we obtain L 0 L
. Note
that this is equal to the second integral. Thus, the two integrals can be combined and so
∞
2 L
Z
nπx a0 X nπx
an = f (x) cos dx and f (x) = + an cos .
L 0 L 2 n=1
L

If f (x) is odd, using analogous arguments, we obtain that an = 0 and

Z L ∞
2 nπx X nπx
bn = f (x) sin dx and f (x) = bn sin .
L 0 L n=1
L

Example 1. The input to an electrical circuit that switches between a high and a low state with
time period 2π can be represented by the boxcar function.

1 0≤x<π
f (x) =
−1 −π ≤ x < 0

More generally, the input to an electrical circuit that switches from a high to a low state with time
period T can be represented
 by the generalTsquare wave function with the following formula on
1 0≤x< 2
the basic period. f (x) = −T
−1 2
≤x<0
Find the Fourier series of the given square wave and the general square wave. Use the expansion
of the given square wave function to find the sum of the series
∞
X (−1)n
.
n=0
2n + 1

3
 Solutions. Graph the square wave function, then graph its periodic extension. The graphs are
on the figure below.

The periodic expansion of this function is called the square wave function.
Note that the function is odd. Thus, the coefficients an of the cosine terms are zero. Since L = π
(and the period T is 2π), the coefficients bn can be computed as
π
1 π 2 π −2 −2
Z Z
bn = f (x) sin nxdx = sin nxdx = cos nx = ((−1)n − 1).
π −π π 0 nπ 0 nπ
Note that (−1)n − 1 = 1 − 1 = 0 if n is even (say n = 2k) and (−1)n − 1 = −1 − 1 = −2 if n is odd
(say n = 2k + 1). Thus, b2k = 0 and b2k+1 = −2
nπ
4
(−2) = (2k+1)π . Hence,

∞ ∞
4 X 1 4X 1 4 1 1
f (x) = sin nx = sin(2k + 1)x = (sin x + sin 3x + sin 5x + . . .).
π n π k=0 2k + 1 π 3 5
n=odd

For the general square wave, analogously to this previousP∞ consideration you obtain that an = 0,
4 2(2k+1)πx 4 1 2(2k+1)πx
b2k = 0 and b2k+1 = (2k+1)π sin T
. Thus, f (x) = π k=0 2k+1 sin T
.
Let us use the expansion f (x) = π4 ∞ 1
P
n=0 2n+1 sin(2n + 1)x where f (x) is the square wave function
P∞ (−1)n
to obtain the sum n=0 2n+1 . To match the desired series, we need to to choose an x-value which
would make sin(2n + 1)x have value (−1) n
. Such x-value is x = π2 . Since f ( π2 ) = 1, plugging x = π2
∞
in both sides of the equation f (x) = π4 n=0 2n+1 1
P
sin(2n + 1)x produces
∞ ∞ ∞
4X 1 4 X (−1)n X (−1)n π
1= (−1)n = ⇒ = .
π n=0 2n + 1 π n=0 2n + 1 n=0
2n + 1 4

Useful formulas. In the previous example, we have seen that the formulas simplifying cos nπ
and sin nπ
2
were useful. To simplify the answers, sometimes the following identities may also be useful.

sin nπ = 0 cos nπ = (−1)n In particular, cos 2nπ = 1 and cos(2n + 1)π = −1.

If n = 2k + 1 is an odd number, sin nπ

2
= sin (2k+1)π
2
= (−1)k cos nπ
2
= cos (2k+1)π
2
=0

4
 Even and odd extensions. If an arbitrary function f (x), not necessarily even or odd, is defined
on the interval (0, L), we can extend it to an even function

f (x) 0<x<L
g(x) =
f (−x) −L < x < 0
and consider its Fourier cosine expansion:
∞
2 L
Z
nπx a0 X nπx
an = f (x) cos dx and f (x) = + an cos .
L 0 L 2 n=1
L

Similarly, we can extend f (x) to an odd function


f (x) 0<x<L
h(x) =
−f (−x) −L < x < 0

and consider its Fourier sine expansion:

∞
2 L
Z
nπx X nπx
bn = f (x) sin dx and f (x) = bn sin .
L 0 L n=1
L

x 0<x≤1
Example 2. Find the Fourier cosine expansion of f (x) = Use it to find
2−x 1<x<2
the sum of the series ∞ 1
P
n=0 (2n+1)2 .
Solution. Graph the function first. Then, extend it symmetrically with respect to y-axis and
then extend it further so that it is periodic. Note that the extension is even and that repeating the
pattern on (−1, 1) produces the entire periodic extension. So, you can take T = 2 and L = 1.

Since the function

R is even, the coefficients
R 1 bn are zero and the coefficients an can be computed
2 1 nπx
as follows. an = 1 0 f (x) cos 1 dx = 2 0 x cos nπx dx. Using integration by parts with u = x,
1 2x 1 1
v = nπ sin nπx, you obtain an = nπ sin nπx 0 + n22π2 cos nπx 0 = 0+ n22π2 (cos nπ−1) = n22π2 ((−1)n −1).
2 −4
If n = 2k is even, an = 0. If n = 2k + 1 is odd, an = (2k+1) 2 π 2 (−1 − 1) = (2k+1)2 π 2 .

Since we are dividing by n already when finding v, the above formula does not apply to n = 0 so
R1 R1 2
1
a0 must be computed separately as a0 = 21 0 x cos 0dx = 2 0 xdx = 2 x2 = 1.
0

5
 1 4
P∞ 1
So, f (x) = 2
− π2 k=0 (2k+1)2 cos(2k + 1)πx.
Alternatively, you can considerR the function as defined on the basic
R 2 period [-2,2] and take
2 nπx
R1 nπx nπx
T = 4 and L = 2. In this case, an = 0 f (x) cos 2 dx = 0 x cos 2 dx + 1 (2 − x) cos 2 dx. Using
integration by parts, obtain the same formula for an as above and the same Fourier series as above.
To find the sum of ∞ 1
P
n=0 (2n+1)2 , note that when x = 0 the function f (x) is equal to f (0) = 0
and its Fourier cosine expansion is equal to 12 − π42 ∞
P 1 1 4
P∞ 1
k=0 (2k+1)2 cos(0) = 2 − π 2 k=0 (2k+1)2 . Rename
the index k to be n if you want and obtain the equation. Then solve for the required sum.
∞ ∞ ∞
1 4 X 1 4 X 1 1 X 1 π2
0= − 2 ⇒ = ⇒ = .
2 π n=0 (2n + 1)2 π 2 n=0 (2n + 1)2 2 n=0
(2n + 1)2 8

Example 3. Consider the function f (x) = x2 for 0 ≤ x ≤ 2.

(a) Sketch the graphs of the following (1) the periodic extension of f (x), (2) the even periodic
extension of f (x), (3) the odd periodic extension of f (x) and write the integrals computing the
coefficients of the corresponding Fourier series in all three cases.

(b) Find the Fourier cosine expansion for f (x).

(c) Using part (b), show the following

∞ ∞
X (−1)n+1 π2 X 1 π2
= and =
n=1
n2 12 n=1
n2 6

Solution. (a) The periodic extension of f (x) is neither even nor odd. You can obtain the graph
of it by replicating f (x) on intervals . . . [−4, −2], [−2, 0], [0, 2], [2, 4], [4, 6] . . . of length T = 2. The

coefficients of the corresponding Fourier series can

be calculated by
Z 2 Z 2
2
an = x cos nπx dx, bn = x2 sin nπx dx
0 0

a0
P∞
and f (x) = 2
+ n=1 an cos nπx + bn sin nπx.

The even extension of f (x) is obtained by extending f (x) from [0, 2] to [−2, 2] by defining
f (x) = (−x)2 = x2 on [−2, 0]. Thus, the result is the function x2 defined on interval [−2, 2].

Then, replicate this function on intervals

. . . [−6, −2], [−2, 2], [2, 6], [6, 10], . . . of length T =
4. Thus, T = 4 and L = 2. The coefficients of the
cosine Fourier series can be calculated by
Z 2
nπx
an = x2 cos dx bn = 0.
0 2

6
 The odd extension of f (x) is obtained by extending f (x) from [0, 2] to [−2, 2] by defining f (x) =
−(x)2 = −x2 on [−2, 0]. Thus, the result is the function
 2
x 0≤x≤2
2
−x −2 ≤ x < 0
defined on interval [−2, 2]. Replicate this function
on intervals . . . [−6, −2], [−2, 2], [2, 6], [6, 10], . . .
of length T = 4. Thus, T = 4 and L = 2. The co-
efficients of the corresponding since Fourier series
can be calculated by
Z 2
nπx
bn = x2 sin dx an = 0.
0 2
R2
(b) By part (a), the coefficients an can be calculated by an = 0 x2 cos nπx 2
dx. Using integration
2
by parts with u = x2 and v = cos nπx 2
sin nπx 2 2
x sin nπx
R
2
dx = nπ 2
, we obtain that an = nπ 2 0
−
4
R2 nπx
nπ 0
x sin 2 dx. The first term is zero and the second term requires another integration by parts,
2 nπx 2
this time with u = x and v = sin nπx dx = −2 cos nπx . Thus an = n28π2 x cos nπx − n16
R
2 nπ 2 2 0 3 π 3 sin 2 0
.
The first term is zero inn the lower bound and the second term is zero at both bounds. Thus
16(−1)
an = n162 π 2 cos nπ = n2 π 2
. Note that this formula works just for n > 0 so a0 has to be computed
R2 2 3
2
separately by a0 = 0 x dx = x3 = 83 .
0
(−1)n
Thus, the Fourier series is x2 = 34 + π162 ∞ nπx
P
n=1 n2 cos 2 .
(c) For both sums, we need to find convenient values of x which reduce the right side of the
formula ∞
4 16 X (−1)n nπx
x2 = + 2 cos
3 π n=1 n2 2
to the series we need to sum. For the first one, it seems that the value of cosine should be 1 so take
x = 0. Plugging x = 0 in the formula above produces
∞ ∞ ∞
4 16 X (−1)n 4 16 X (−1)n π 2 X (−1)n
0= + 2 ⇒ − = ⇒ − =
3 π n=1 n2 3 π 2 n=1 n2 12 n=1 n2

Multiply both sides with −1 to have

∞ ∞
π 2 X (−1)(−1)n π 2 X (−1)n+1
= ⇒ =
12 n=1 n2 12 n=1 n2

For the second sum, note that the value of cosine should be (−1)n so that, multiplied by another
(−1)n , the numerator becomes (−1)2n = 1. The value x = 2 would produce cos nπ(2) 2
= cos nπ =
n
(−1) . Plugging x = 2 in the expansion in part (b) produces
∞ ∞ ∞ ∞
24 16 X (−1)n n 4 16 X (−1)2n 8 16 X 1 π2 X 1
2 = + 2 (−1) ⇒ 4 = + ⇒ = ⇒ =
3 π n=1 n2 3 π 2 n=1 n2 3 π 2 n=1 n2 6 n=1
n2

7
 Complex Fourier Series. The complex form of Fourier series is the following:
∞ ∞
1 x0 +T
  Z
X 2nπix
X 2nπx 2nπx −2nπix
f (x) = cn e T = cn cos + i sin where cn = f (x)e T dx.
n=−∞ n=−∞
T T T x0

If f (x) is a real function, the coefficients cn satisfy the relations cn = 12 (an − ibn ) and c−n =
1
(a + ibn ) for n > 0. Thus, c−n = cn for all n > 0. In addition,
2 n

an = cn + c−n and bn = i(cn − c−n ) for n > 0 and a0 = 2c0 .

These coefficients cn are further associated to f (x) by Parseval’s Theorem. This theorem is
related to conservation law and states that
∞ ∞
1 x0 +T a20 1 X 2
Z X
2 2
|f (x)| dx = |cn | = + (an + b2n ).
T x0 n=−∞
4 2 n=1

Note that the integral on the left side computes the average value of |f (x)|2 over one period. The
proof of this theorem can be your project topic.
Symmetry Considerations. If f (x) is either even or odd function defined on interval (−L, L),
the value of |f (x)|2 on (−L, 0) is the same as the value of |f (x)|2 on (0, L). Thus,
1
RL RL
2L −L
|f (x)|2 dx = L1 0 |f (x)|2 dx. In this case, Parseval’s Theorem has the following form.
L ∞ ∞
a2 1 X 2
Z
1 X
2
|f (x)| dx = |cn | = 0 +
2
(an + b2n ).
L 0 n=−∞
4 2 n=1

Example 4. The voltage in an electronic oscillator is represented as a sawtooth function f (t) = t

for 0 ≤ t ≤ 1 that keeps repeating with the period of 1. (a) Sketch this function and represent it
using a complex Fourier series. (b) Use the Fourier series expansion and Parseval’s Theorem to find
the sum of the series ∞
X 1
.
n=1
n2

Solution. Graph the function first, note

its periodic extension
P∞ and the period T = 1.
Thus, f (t) = n=−∞ cn e2nπit where
Z 1 1 1
t 1 1 1 1
cn = te−2nπit = e−2nπit + 2 2 e−2nπit = e−2nπi + 2 2 e−2nπi − 2 2 .
0 −2nπi 0 4n π 0 −2nπi 4n π 4n π

Note that e−2nπi = cos(−2nπ) + i sin(−2nπ) = 1. Thus cn = −2nπi 1

+ 0 = 2nπ i
. This formula applies
R1
just to n 6= 0 so compute c0 separately as c0 = 0 t dt = 12 . Hence, f (t) = 12 + ∞ i 2nπit
P
n=−∞,n6=0 2nπ e .
−i
This sum can be expressed using formulas with an and bn also. Since c−n = 2nπ = cn , we have
that an = 0 for n > 0, a0 = 2c0 = 1, and bn = −1
nπ
.
By Parseval’s Theorem,
Z 1 ∞ ∞ ∞
2 1 1X 1 1 1 1 X 1 X 1 π2
x dx = + ⇒ = + ⇒ = .
0 4 2 n=1 n2 π 2 3 4 2π 2 n=1 n2 n=1
n2 6

8
 2
Example 5. Using the Fourier cosine series expansion P∞ for1 f (x) = x for 0 < x < 2 from Example
3 and Parseval’s Theorem, find the sum of the series n=1 n4 .
Solution. Recall that T = 4, L = 2 and that the function is even. In Example 3 we have find
16(−1)n
that an = n2 π2 for n > 0, a0 = 38 . Thus, Parseval’s Theorem applied to this function results in the
following
∞ ∞ 
1 2 4 a20 1 X 2 16 162 X 1
Z 
2
x dx = + (a + bn ) = + 4 +0 .
2 0 4 2 n=1 n 9 2π n=1 n4
R2
Note that integral on the left side is 12 0 x4 dx = 16
5
. Dividing the equation above by 16 produces
∞ ∞
1 1 8 X 1 X 1 π4
= + 4 ⇒ = .
5 9 π n=1 n4 n=1
n4 90

Practice Problems.
1. Note that the boxcar function from Example 1 represents the odd extension of the function
f (x) = 1 for 0 ≤ x < π. Consider the even extension of this function and find its Fourier cosine
expansion.

1−x 0≤x<1
2. Find the Fourier series of f (x) =
1+x −1 < x < 0

x 0<x≤1
3. Find the Fourier sine expansion of f (x) = . Use it to find the sum of
2−x 1<x<2
the series ∞
X 1
.
n=0
(2n + 1)2
Solutions.
1. The even extension of f (x) is f (x) = 1 for −π < x < π. The periodic extension of this
function is constant function equal to 1 for every x value. Note that this is already in the
form of a Fourier series with bn = 0 for every n, an = 0 for n >R 1 and a0 = 2. Computing the
π
Fourier Rcoefficients would give you the same answer: an = π2 0 cos nx dx = 0 if n > 1 and
π
a0 = π2 0 dx = 2. Thus, f (x) = 22 + ∞
P
n=1 0 = 1. This answerPshould not be surprising since
this function is already in the form of a Fourier series 1 = 1 + ∞ n=0 (0 cos nx + 0 sin nx).

2. Graph the function and note it is even.

R1
Thus, bn = 0. Since T = 2 and L = 1 aRn = 2 0 (1 − x) cos(nπx) dx. Using the inte-
1
gration by parts with u = 1 − x and v = cos(nπx)dx = nπ sin(nπx), obtain that an =
2 1 1 1
(1 − x) sin(nπx) 0 + nπ 0 sin(nπx) dx = 0 − n2 π2 cos(nπx) 0 = − n22π2 ((−1)n − 1). This last
2 2
R
nπ

9
 expression is 0 if n is even and equal to n22π2 = (2k+1)
2
2 π 2 if n = 2k + 1 is odd. Note that the
2 n
formula − n2 π2 ((−1) −1) does not compute a0 because of the n in the denominator so calculate
R1 2
1
a0 from the formula a0 = 2 0 (1 − x) dx = x − x2 = 12 . This gives you the Fourier series
0
expansion
∞
1 X 4
f (x) = + cos(2k + 1)πx.
4 k=0 (2k + 1)2 π 2

3. Extend the function symmetrically about the origin so that it is odd.

Thus T = 4 and L = 2. Since this extension is odd, an = 0. Compute bn as

Z 2 Z 1 Z 2
nπx nπx nπx
bn = f (x) sin dx = x sin dx + (2 − x) sin dx =
0 2 0 2 1 2
 1   2
−2x nπx 4 nπx −2(2 − x) nπx 4 nπx
cos + 2 2 sin + cos − 2 2 sin =
nπ 2 nπ 2 0 nπ 2 nπ 2 1
−2 nπ 4 nπ 2 nπ 4 nπ 8 nπ
cos + 2 2 sin + cos + 2 2 sin = 2 2 sin .
nπ 2 nπ 2 nπ 2 nπ 2 nπ 2
8(−1)k
This last expression is 0 if n is even. If n = 2k + 1 is odd, this is (2k+1)2 π 2
. So,
∞
8 X (−1)k (2k + 1)πx
f (x) = 2 2
sin .
π k=0 (2k + 1) 2

(−1)n
To find the sum of the series ∞
P
n=0 2n+1 , note that when x = 1 the function f (1) is equal to 1
(−1)n (−1)n
and its Fourier sine expansion is equal to π82 ∞ (2n+1)π
= π82 ∞ n
P P
n=0 (2n+1)2 sin 2 n=0 (2n+1)2 (−1) =
8
P ∞ 1
π2 n=0 (2n+1)2 so
∞ ∞
8 X 1 X 1 π2
= 1 ⇒ = .
π 2 n=0 (2n + 1)2 n=0
(2n + 1) 2 8

10
 Fourier Transformation

The Fourier transform is an integral operator meaning that it is defined via an integral and that it
maps one function to the other. If you took a differential equations course, you may have encountered
the Laplace transform which is another integral operator.
2nπit
Recall that the complex Fourier series is f (t) = ∞
P
n=−∞ cn e . The sequence cn can be regarded
T

as a function of n and is called Fourier spectrum of f (t). We can think of c(n) being another
representation of f (t), meaning that f (t) and c(n) are different representations of the same object.
Indeed: given f (t) the coefficients c(n) can be computed and, conversely, given c(n), the Fourier
series with coefficients c(n) defines a function f (t). We can plot c(n) as a function of n (and get a
set of infinitely many equally spaced points). In this case we think of c as a function of n, the wave
number.
We can also think of c as a function of ω = 2nπ T
, the frequency. If T is large, then ω is small and
the function T cn becomes a continuous function of ω for T → ∞. If we let T → ∞, the requirement
that f (t) is periodic can be waved since the period becomes (−∞, ∞).
The Fourier Transform F (ω) of f (t) is the limit of the continuous function √12π T cn when
T → ∞. Since
Z T /2 Z T /2
1 1 −2nπit 1
√ T cn = √ f (t)e T dt = √ f (t)e−iωt dt,
2π 2π −T /2 2π −T /2

we have that Z ∞
1 1
√ T cn → F (ω) = √ f (t)e−iωt dt
2π 2π −∞

Let us consider now the Fourier series expansion of f (t) with the substitution ω = 2nπ
T
. Two
2π 2π T dω
consecutive n values are length 1 apart so dn = 1. Thus, dω = T dn ⇒ dω = T and 2π = 1. Hence

∞ ∞ ∞ Z ∞
2nπit
ωit T dω 1 Tc 1
X X X
f (t) = cn e T = cn e =√ √ n eωit dω → √ F (ω)eωit dω.
n=−∞
2π 2π 2π 2π −∞
T ω/(2π)=−∞ T ω/(2π)=−∞

Thus, for a given function F (ω) of ω, the above expression computes a function f (t) of t. This formula
computes the Inverse Fourier Transform f (t) for a given function F (ω).
We can still think of f (t) and F (ω) being the same representations of the same object:

R∞
F (ω) = √1
2π −∞
f (t)e−iωt dt F (ω) is the Fourier transform of f (t)
F (ω) ! the coefficients of the Fourier series for f (t)
R∞
f (t) = √1 F (ω)eiωt dω
f (t) is the inverse Fourier transform of F (ω)
2π −∞
f (t) ! the expansion having coefficients F (ω)
√ √
The constant 2π in the numerator of the formula ω = 2nπ T
can be written as 2π = 2π · 2π so that
it can be evenly distributed in both formulas above resulting in each of the two formulas having the

11
same constant √12π in front of the integral. This makes the formulas for Fourier and inverse Fourier
transforms symmetric and certain calculations easier.
One of the most important applications of Fourier transform is in signal processing with various
applications in cryptography, acoustics, optics and other areas. In this case, one can think of f (t) as
of a signal which is measured in time and is represented as a function of time t but which needs to
be represented as a function of frequency, not time. The Fourier transform produces exactly that –
a representation of f (t) as a function F (ω) of frequency ω. The function F (ω) is also known as the
frequency spectrum of the signal.
Moreover, Fourier transform provides information on the amplitude and phase of a source signal
at various frequencies. The transform F (ω) of a signal f (t) can be written in polar coordinates as
F = |F |eθi . The modulus |F | represents the amplitude of the signal at respective frequency ω, while
θ computes the phase shift at frequency ω.
The Fourier transform is not limited to functions of time and temporal frequencies. It can be
used to analyze spatial frequencies. If the independent variable in f (t) stands for space instead of
time, x is usually used instead of t. In some cases, the independent variable of the inverse transform
is denoted by k.
t ←→ x and ω ←→ k
We consider some applications in magnetic resonance imaging in the last section of this note.
Besides generalizing Fourier series, the mathematical importance of the Fourier transform lies in
the following.

1. If the initial function f (t) has the properties that are not desirable in a particular application
(e.g. discontinuous, not smooth), we can consider the function F (ω) instead which is possible
better behaved.

2. Fourier transform represents a function that is not necessarily periodic and that is defined on
an
R ∞infinite interval. The only requirement for the Fourier transform to exist is that the integral
−∞
|f (t)|dt is convergent.

3. Fourier transform can be used similarly to Laplace transform: in converts a differential equation
into an ordinary (algebraic) equation that is easier to solve. After solving it, the solution of
the original differential equation can be obtained by using the inverse Fourier transform.

Example 1. Find the Fourier transform FaA (ω) of the boxcar function faA (t).

 0 t ≤ −a
faA (t) = A −a < t < a
0 t≥a

Express your answer as real function.
The boxcar function faA (t) is said to be nor-
1
malized if A = 2a so that the total area under the
function is 1. We use fa to denote the normalized
boxcar function and Fa for its Fourier transform.
Consider how changes in value of a impact the
shape and values of Fa (ω).

12
 R∞ R∞
Solution. FaA (ω) = √12π −∞ faA (t)e−iωt dt. Write the integral −∞ as a sum of the integrals
R −a R a R ∞
−∞
+ −a + a to match the three branches of the function. Since faA (t) = 0 for t < −a and t > a,
the first and the last integral of faA (t)e−iωt are zero. For the middle branch, faA (t) = A so we have
that Z a a
1 −iωt 1 A −iωt −A
A
Fa (ω) = √ Ae dt = √ e =√ (e−aiω − eaiω ).
2π −a 2π −iω −a 2πiω
Use the Euler’s formula to obtain e−aiω − eaiω = (cos(−aω) + i sin(−aω) − cos aω − i sin aω) . Using
that cos(−aω) = cos aω since cosine is even and that sin(−aω) = − sin aω since sine is an odd
function, the cosine terms cancel and we obtain that

−A 2A sin aω
FaA (ω) = √ (−2i sin aω) = √ .
2πiω 2π ω
sin x
The function x
is known as the sinc func-
tion.
sin x
sinc x =
x
Note that limx→0 sinc x = 1 and that
limx→∞ sinc x = 0.
sin aω sin aω
Since ω
=a aω
, we can represent

2aA
the Fourier transform of the boxcar function faA (t) as the sinc function FaA (ω) = √
2π
sinc aω.

1
If the boxcar function is normalized, A = 2a so Fa (ω) = √12π sinc aω. Since sinc(0)=1, the peak of
this function has value √12π .
Let us compare the graphs of fa and Fa for several different values of a. On the graphs below,
you can notice that larger values of a correspond to fa having smaller height and being more spread
out. In this case, Fa has a narrow, sharp peak at ω = 0 and converges to 0 faster.

13
 If a is small, fa has large height and very narrow base. In this case, Fa has very spread out peak
around ω = 0 and the convergence to 0 is much slower.
In the limiting cases a → 0 represents a constant function and a → ∞ represents the Dirac
delta function δ(t). This function, known also as the impulse function is used to represent
phenomena of an impulsive nature. For example, voltages that act over a very short period of time.
It is defined by
δ(t) = lim fa (t).
a→0

Since the area under fa (t) is 1, the area under δ(t) is 1 as well. Thus, δ(t) is characterized by the
following properties:
R∞
(1) δ(t) = 0 for all values of t 6= 0 (2) −∞
δ(t)dt = 1.

Since no ordinary function satisfies both of these properties, δ is not a function in the usual sense
of the word. It is an example of a generalized function or a distribution. An alternate definition of
the Dirac delta function can be found on Wikipedia.
We have seen that Fa for a small a is very spread out and almost completely flat. So, in the
limiting case when a = 0, it becomes a constant. Thus, the Fourier transform of δ(t) can be obtained
as limit of Fa (ω) when a → 0. We have seen that this is a constant function passing √12π . Thus,

the Fourier transform of the delta function δ(t) is the constant function F (ω) = √1 .
2π

Conversely, when a → ∞ fa → √12π . The Fourier transform of the constant function √12π is the
limit of Fa for a → ∞. We have seen that this limit function is zero at all nonzero values of ω. So,
we can relate this limit to δ(ω). The exact relation will be clear after the next example.
Example
 2. Find the inverse Fourier transforms of the boxcar function
A, −a < ω < a
GA
a (ω) = and of the Dirac delta function δ(ω).
0, otherwise
Solution. The inverse Fourier transform gaA (t) of GA
a (ω) can be computed by

14
 Ra
gaA (t) = √1
2π −a
Aeiωt dω = √ A (eiat
2πit
− e−iat )
To express this answer as a real function, use the Euler’s formula and symmetries of sine and
cosine functions similarly as in Example 1 to obtain the following.
A A
gaA (t) = √2πit (cos at + i sin at − cos(−at) − i sin(−at)) = √2πit (cos at + i sin at − cos at + i sin at) =
√ A (2i sin at) = √2A sin at = √ 2aA sin at 2aA sin at
= √ 2aA
= √ sinc at. Thus, the inverse transform of the
2πit 2πt 2πt a 2π at 2π
general boxcar function is the sinc function above. This also illustrates that

the Fourier transform of the sinc function above is the general boxcar function.

Taking the limit when a → 0 in the normalized case, we obtain that the inverse Fourier of the
delta function is the limit of √12π sinc at when a → 0 which is the constant function √12π . Thus, the
inverse Fourier transform of δ(ω) to √12π . Hence, we have that

the Fourier transform of the constant function f (t) = √1 is the delta function δ(ω).
2π

Example 3. Consider the Gaussian probability function (a.k.a. the ”bell curve”) f (t) =
−at2
Ne where N and a are constants. N determines the height of the peak and a determines how
fast it decreases after reaching the peak. The Fourier transform of f (t) can be found to be F (ω) =
2
√N e−ω /4a . This is another Gaussian probability function. Examine how changes in a impact the
2a
graph of the transform. This consideration is related to Heisenberg uncertainty principle and has
applications in quantum mechanics.
Solution. If a is small, f is flattened. In this case the presence of a in the denominator of the
exponent of F (ω) will cause the F to be sharply peaked and the presence of a in the denominator of
√N , will cause F to have high peak value. If a is large, f is sharply peaked and F is flattened and
2a
with small peak value.
The following table summarizes our current conclusions.

Function in time domain Fourier Transform in frequency domain

boxcar function sinc function

sinc function boxcar function

delta function constant function

constant function delta function

Gaussian function Gaussian function

15
 Symmetry considerations
The table on the right displays the formulas q R
of Fourier and inverse Fourier transform for even 2 ∞
f (t) even F (ω) = π 0 f (t) cos ωt dt
and odd functions. q R
∞
Fourier sine and cosine functions. Sup- F (ω) even f (t) = π2 0 F (ω) cos ωt dω
pose that f (t) is defined just for t > 0. We can q R
extend f (t) so that it is even. Then we get the ∞
f (t) odd F (ω) = π2 0 f (t) sin ωt dt
formula for F (ω) by using the formulas for even q R
∞
function above. F (ω) is then called Fourier co- F (ω) odd f (t) = π2 0 F (ω) sin ωt dω
sine transformation.

Similarly, if we extend f (t) so that it is odd, we get the formula for F (ω) same as for an odd
function above. F (ω) is then called Fourier sine transformation.

Fourier series and Fourier transform

We have seen that the Fourier transform of a boxcar function is a sinc function so that the inverse
Fourier transform of a sinc function is a boxcar function. The following series of figures illustrates
the connection between a boxcar function f (t), the sinc function F (ω) (the inverse Fourier transform
of f (t)), and the coefficients of the Fourier series expansion of the boxcar function f (t).
The length of the base of the box determines the sampling period T. The spacing ω0 = 2π T
in the
frequency-domain determines the value of ω since ω = 2nπ T
⇒ ω = nω 0 .
In the following figures, sn denotes the Fourier coefficient of the complex Fourier series. We
illustrate how adding sufficiently many terms of the Fourier series results in the formation of the
shape of the initial boxcar function. We consider the boxcar function with a = 12 (so T = 1, ω0 = 2π
and sn = sinc ω2 ). The first graph on each figure displays the Fourier transform, the sinc function,
in the frequency-domain. The highlighted frequency on the first graph determines the value of n.
The second graph displays the harmonic function corresponding to n-th term of the Fourier series of
the function f (t) in the time-domain. The third graph is the sum of the first n terms of the Fourier
series of f (t).
Note that as n → ∞ the sum of Fourier series terms converge to the boxcar function.

16
17
 In the previous examples, the function in frequency-domain was a continuous function. In appli-
cations, it is impossible to collect infinitely many infinitely dense samples. As a consequence, certain
error may come from sampling just finite number of points taken over a finite interval. Thus, it is
relevant to keep in mind what effects in time (resp. frequency) domain may have finite and discrete
samples of frequencies (resp. time).

frequency-domain function in time-domain effects in time-domain

infinite continuous set of frequencies non-periodic function none

finite continuous set of frequencies non-periodic function Gibbs ringing, blurring

infinite number of discrete frequencies periodic function aliasing

finite number of discrete frequencies periodic function aliasing, Gibbs ringing,

blurring

Assume that the sample consists of frequencies and that all the values lie on the graph of a sinc

18
function. The goal is to reconstruct the boxcar function in the time-domain. The following four
figures illustrate each of the four scenarios from the table.

Application in Magnetic Resonance Imaging

The Fourier transform is prominently used in

Magnetic Resonance Imaging (MRI). The fist fig-
ure represents a typical MRI scanner. The scan-
ner samples spacial frequencies and creates the
Fourier transform of the image we would like
to obtain. The inverse Fourier transform con-
verts the measured signal (input) into the recon-
structed image (output).
In the following two figures, the image on the right represents the amplitude of the scanned input
signal. The image on the left represents the output - the image of a human wrist, more precisely,
the density of protons in a human wrist. The whiter area on the image correspond to the fat rich
regions. The darker area on the image correspond to the water rich regions.

19
 Each point on the input images corresponds
to certain frequency. Two smaller figures on the
right side represent the components of the inverse
transforms at two highlighted frequencies. The
output image is created by combining many such
images - one for each sampled frequency in fact.
The last figure represent the output with high (first pair of images) and low frequencies (second
pair of images) removed.
If high frequencies are removed (low-pass filter), the image becomes blurred and only shows the
rough shape of the object.
If low frequencies are removed (high-pass filter), the image is sharp but intensity variations are
lost.

Practice Problems.

1. Find the Fourier transform of f (t) = e−t , t > 0, f (t) = 0 otherwise.

2. Find Fourier cosine transformation of the function from the previous problem.

3. Find cosine Fourier transform of f (t) = 2t − 3 for 0 < t < 3/2, f (x) = 0 otherwise.

20
4. If
a
fa (t) =
+ t2 a2
the graph of f has a peak at 0. Since f (0) = a1 , the height is conversely proportional to a. The
Fourier transform of f can shown to be
p
Fa (ω) = π/2e−a|ω| .
Graph Fa (ω) for several values of a and make conclusion how values of a impact the graph of
Fa .

R∞ 1−ω 0≤ω≤1
5. Solve the equation 0 f (t) cos tω dt =
0 ω>1
Solutions.
R∞ ∞
−1
1. F (ω) = √1
2π 0
e−t e−iωt dt = √
2π(1+iω)
e−(1+iω)t
= √ 1
2π(1+iω)
. Rationalizing the denominator,
0
1−iω
you obtain F (ω) = √2π(1+ω 2)

q R
∞
2. F (ω) = π2 0 e−t cos ωt dt. Using two integration by parts with u = e−t , one obtains that
q ∞ q R∞
F (ω) = π2 ω1 e−t sin ωt + π2 ω1 0 e−t sin ωt dt =
0q
q ∞ R ∞ −t q
2 1 −t
0 − π ω2 e cos ωt − π ω2 0 e cos ωt dt. Note that the first term is equal to π2 ω12 and
2 1
0
the second is − ω12 F (ω). Hence,
r r
2 1 1 1 2 1
F (ω) = − F (ω) ⇒ F (ω)(1 + ) = .
π ω2 ω2 ω2 π ω2
Multiply by ω 2 to get
r r
2 2 1
F (ω)(ω 2 + 1) = ⇒ F (ω) = 2
.
π π ω +1
q R q 3/2
2 3/2 2 2t−3 2 3/2
R
3. F (ω) = π 0
(2t − 3) cos ωt dt = π
( ω
sin ωt − ω 0
sin ωt dt) =
0
q 3/2 q √
2
π
(0 + 2
ω2
cos ωt ) = π2 ω22 (cos 3ω 2
− 1) = ω22 √2π (cos 3ω
2
− 1).
0

4. If a is small, then fa has a larger peak. In this case Fa is more spread out and flattened. If a
is large, fa is spread out and the height of the peak is not large. In this case, Fa has a sharp
peak and converges to zero fast.
q
5. Use inverse cosine Fourier transform. First multiply with π2 to match the definition of the
transform. Then the left side is exactly the Fourier cosine transform ofq
f (t). So we can get f (t)
as the inverse transform of the left side of the equation multiplied by π2 .
q q R
2 2 1 2 1−ω 1 1 1 1
sin ωt dt) = π2 (0 − 1
R
f (t) = π π 0
(1 − ω) cos ωt dω = π
( t
sin ωt 0
+ t 0 t2
cos ωt 0 ) =
2 2
πt2
(− cos t + 1) = πt2
(1 − cos t).

21