2 - Orthogonal Functions and Fourier Series

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 10

Orthogonal Functions and

Fourier Series

ENGR 311 - Dr. R. Tadayon 1

Orthogonal Functions
• Consider two vectors of and ⃗ in 3D space. Recall that the inner (dot) product
of these two vectors is defined as follows:
·⃗= ,⃗ = + + = cos
• If these two vector are orthogonal, their inner product is zero.
• Using the same terminology, we introduce two orthogonal functions.
• Consider two functions and . The inner product of and on
interval , is a number obtained as follows:

, =

ENGR 311 - Dr. R. Tadayon 2

1
Orthogonal Functions
• Similar to two vectors, two functions are said to be orthogonal if their inner
product is zero.

, = =0

• Two orthogonal vectors are gematrically perpendicular. But the term orthogonal
and the above condition have no geometric significance.
• A set of real-valued functions , , , … is said to be orthogonal
on an interval , if:

, = = 0, ≠

ENGR 311 - Dr. R. Tadayon 3

Orthogonal Functions
• In the vector calculus, the norm of a vector is defined as = · = , .
• Similarly, the norm of function is defined as follows:

= , =

• If , , , … is an infinite orthogonal set of functions on the


interval [a, b], and if = 1 for = 0, 1, 2, …, then , , ,…
is said to be an orthonormal set on the interval [a, b].
• Example 1

ENGR 311 - Dr. R. Tadayon 4

2
Orthogonal Series Expansion
• Suppose , , , … is an infinite orthogonal set of functions on the interval [a, b].
We hope to express a function as follows if we can find coefficients % , % , % , …:
(

=% +% + ⋯+ % +⋯= '%
)
• Multiplying both sides by and integrating over the interval [a, b]:

=% +% +⋯+% +⋯

+
= 0+0+⋯+% +0+⋯ → % =
+
( 8
+ 0 1 43 1 71
Therefore: 0 1 = ' 23 43 1 where 23 = 9
43 :
3)5
ENGR 311 - Dr. R. Tadayon 5

Orthogonal Series with respect to a Weight Function


• A set of real-valued functions , , , … is said to be orthogonal
with respect to a weight function ; > 0 on an interval , if:

; = 0, ≠

• And for a function = ∑() % :

+ ; + ;
% = = is the norm -
+ ; a general definition.
• To express = ∑() % , must be a complete set. i.e., cannot
be orthogonal to any .

ENGR 311 - Dr. R. Tadayon 6

3
Fourier Series
• We saw how we can expand a function using an infinite orthogonal set of
functions. Now, we shall expand function in terms of a special orthogonal set
of trigonometric functions (Fourier series).
• Fourier series are infinite series that are very powerful to represent general
periodic functions, as an example, in terms of simple ones (sines and cosines -
trigonometric system).
• Similar to Taylor series in calculus, Fourier series are very powerful. For instance,
they can handle many discontinuous periodic functions we face in applications.
• Fourier series provide the solution of ODEs in connection with forced oscillations
and approximate discontinuous periodic functions. Also, Fourier analysis can be
used to solve some partial differential equations (PDEs).

ENGR 311 - Dr. R. Tadayon 7

Periodic Functions
• A function is said to be periodic if for a T > 0, +> = . Here, is
defined for all real x except at some points. T is the period of .
• Function sin is a periodic function defined everywhere. Also, function tan is
another periodic function that is not defined at some points (±kπ/2, k: an integer).
• If and D have period T, then with any constants a and b, + D
has the period T.
• Example 2
• We can prove that the following set of trigonometric functions is orthogonal on
the interval [-p, p]:
E 2E 3E E 2E 3E
1, cos , cos , cos , … , sin , sin , sin ,…
F F F F F F
ENGR 311 - Dr. R. Tadayon 8

4
Fourier Series Expansion
• Now, we would like to expand a function defined on −F, F using the
I I I I I I
orthogonal series 1, cos , cos , cos , … , sin , sin , sin ,… :
J J J J J J
(
E E
= +' cos + sin
2 F F
)

• To find the coefficient :


(
J J J
E J
E
= +' cos + sin
KJ 2 KJ KJ F KJ F
)

0 0

ENGR 311 - Dr. R. Tadayon 9

Fourier Series Expansion


L M
Therefore: 95 = 0 1 71
M KM

• To find the coefficient :


J
E
cos
KJ F
(
J
E J
E E J
E E
= cos +' cos cos + sin cos
2 KJ F KJ F F KJ F F
)
0 if m ≠ n
0 =R
p if m = n
0

L M
3Q
Therefore: 93 = 0 1 NOP 1 71
M KM M
ENGR 311 - Dr. R. Tadayon 10

10

5
Fourier Series Expansion
• To find the coefficient :
J
E
sin
KJ F
(
J
E J
E E J
E E
= sin +' cos sin + sin sin
2 KJ F KJ F F KJ F F
)
0 if m ≠ n
=R
0 0
p if m = n
L M
3Q
Therefore: 83 = 0 1 PST 1 71
M KM M
• The presented trigonometric series with coefficients , , and is said to be the
Fourier series of the function .

ENGR 311 - Dr. R. Tadayon 11

11

Fourier Series Expansion


• Summarizing the results, the Fourier series of a function defined on the interval
−F, F is given by:
(
E E
= +' cos + sin
2 F F
)

1 J
=
F KJ
1 J
E
= cos ← Euler Formulas
F KJ F
1 J
E
= sin
F KJ F

ENGR 311 - Dr. R. Tadayon 12

12

6
Fourier Series
• Convergence of Fourier Series Theorem (with no proof): Let and [ be
piecewise continuous on interval −F, F . Then for all x in interval −F, F , the Fourier
series of converges to at a point of continuous, and it converges to the
following value at a discontinuous point:
+ \ K

2
• Sequence of Partial Sums: One should use several terms of Fourier series to
approximate function . How many terms of the series should be used?
• Examples 3 & 4
• In the Fourier series expansion of , 2F is the fundamental period. Therefore, a
Fourier series gives also the periodic extension of outside the interval [-p, p].

ENGR 311 - Dr. R. Tadayon 13

13

Gibbs Phenomenon
• Using several terms of
−1 for − E < < 0
=R + 2E =
Fourier series expansion, we
face error especially close to 1 for 0 < < E
the discontinuous points. 1.5
S1

• Having stronger waves close S2 1.0

S3
to the discontinuous points Exact 0.5

that do not smooth out is S20


0.0
-4 -3 -2 -1 0 1 2 3 4
called Gibbs phenomenon. -0.5

-1.0

-1.5

ENGR 311 - Dr. R. Tadayon 14

14

7
Fourier Cosine and Sine Series
• Consider a function whose domain contains –x and x.
• If − = , is an even function.
• If − =− , is an odd function.
• Examples of even functions:
=1 = = =
= cos( ) = cosℎ( )
• Examples of odd functions:
= = = \

= sin( ) = sinℎ( )

ENGR 311 - Dr. R. Tadayon 15

15

Fourier Cosine and Sine Series


• Properties of Even/Odd Functions:
• The product of two even functions is even.
• The product of two odd functions is even.
• The product of an even function and an odd function is odd.
• The sum (difference) of two even functions is even.
• The sum (difference) of two odd functions is odd.

• If is even: +K = 2+ .

• If is odd: +K = 0.

ENGR 311 - Dr. R. Tadayon 16

16

8
Fourier Cosine and Sine Series
• Consider the Fourier series expansion of function on the interval −F, F :
(
E E
= +' cos + sin
2 F F
)
1 J
1 J
E 1 J
E
= ; = cos ; = sin
F KJ F KJ F F KJ F
• Based on the properties of odd/even functions (odd x odd = even, even x even = even, odd x
even = odd):
J I
• If is odd: =0 =0 = + sin
J J

J J I
• If is even: = + = + cos =0
J J J

ENGR 311 - Dr. R. Tadayon 17

17

Fourier Cosine and Sine Series


• The Fourier series of an even function defined on the interval −F, F is called cosine
Fourier series:
(
E 2 J
2 J
E
= +' cos ; = ; = cos
2 F F F F
)

• The Fourier series of an odd function defined on the interval −F, F is called sine
Fourier series:
(
E 2 J
E
=' sin ; = sin
F F F
)

• Example 5

ENGR 311 - Dr. R. Tadayon 18

18

9
Half-Range Expansions
• So far, we have dealt with functions defined on an interval (-p, p)
with the origin as midpoint. However, in most practical situations, we
face functions that are defined on an interval (0, L).
1. To express a function f (x) defined on an interval (0, L) as a Fourier
series, we can extend it as a function of period L. Generally, this
Fourier series would contain both cosine and sine terms.
2. To have simpler Fourier series, we can extend the graph to the left
of y axis from – L to 0. This is called Half-Range Expansion:
A. Reflect the graph about the y-axis onto (-L, 0); the function is
now even on the interval (– L, L) with period 2b, or
B. Reflect the graph through the origin onto (-L, 0); the function
is now odd on the interval (– L, L) with period 2b.

ENGR 311 - Dr. R. Tadayon 19

19

Half-Range Expansions
• Practically, we do not need to make the reflections described in 2(A) and 2(B),
half-range expansions. Since the cosine series and sine series are defined for
function on the interval [0, p], we just need to replace p by L when is
defined on the interval [0, L].
• For approach 1, the original Fourier series expansion can be used with setting
F = b/2 and replacing the lower and upper limits of each integral by zero and L,
respectively (it can be proved).
• Examples 6 & 7:

ENGR 311 - Dr. R. Tadayon 20

20

10

You might also like