ME 2002: Advanced Engineering Mathematics II

Spring: 2023
Instructor: Dr. I-Fan Lin

Topic III
Fourier Analysis

Fourier Series

Recall: in vector space R, any vector v can be represented by a linear combination of base vectors.
0 1 0 1 0 1
BBB1CCC BBB0CCC BBB0CCC
BBB CCC BBB CCC BBB CCC
BBB CCC 0 BBB CCC 1 BBB0CCC
BB0CCC BB0CCC B C
e1 = BBB CCC , e2 = BBB CCC , · · · , en = BBBBB ... CCCCC
B B
BBB .. CCC BBB .. CCC BBB CCC
BBB . CCC BBB . CCC BBB0CCC
@ A @ A @ A
0 0 1

n.b. {e1 , e2 , · · · , en }: orthonomal basis

0 1 0 1 0 1 0 1
BBBv1 CCC BBB1CCC BBB0CCC BBB0CCC
BBBBv2 CCCC BBB CCC
0
BBB CCC
BBB CCC
BBB CCC 1
BBB CCC
BBB0CCC
BBB CCC B
Therefore, v = BBBBv3 CCCC = v1
B C B C
C B
B C
BB0CC + v2 BBB0CCC + · · · + vn BBBBB ... CCCCC = v1 e1 + v2 e2 + · · · + vn en .
C
BBB .. CCC 8 BBBBB .. CCCCC BBB . CCC BBB CCC
BBBB .. CCCC BBBB0CCCC
1

BBB . CCC BBB . CCC

@ A @ A @ A @ A
vn 0 0 1

n.b. Coefficients vi ’s are obtained from “projection” (or inner product) of v on ei

e.g. V, I. 2
=

ei.j 0
=

if itj

k=. E if ij =

Vi V. i
=

Hence, consider functions as vectors.

vector space =) function space

Choose certain elementary functions as base functions such as sin x, cos x, x polynomial. Then, any function
can be represented as a linear combination of those base functions? And, the “coefficients” of this linear
combination are obtained from inner product?

x x2 x3
e.g. e x = 1 + + + + ···
1! 2! 3!
then 9xo, x', x2, x3, ----3 is the basis

1, ., 21.--- are coefficients

n.b. A function space is usually a space of infinite dimension. We say {'0 (x), '1 (x), · · · , 'n (x), · · · } is a
complete basis for the function space if any smooth function can be represented by a linear combination of
these base functions.

© 2023 I-Fan Lin. All Rights Reserved. 1

 Fourier Series
Let f (x) be a function defined in L  x  L (i.e. [ L, L]) and f (x) is integrable in [ L, L]. Can we find
a0 , a1 , a2 , · · · , and b0 , b1 , b2 , · · · such that
1Y
i.e.

X 1 ⇣
fexsdx exists
⇣ n⇡x ⌘ ⇣ n⇡x ⌘⌘
f (x) = a0 + an cos + bn sin (1)
n=1
L L

and the series converge to f (x) at “almost every point” of [ L, L]? yes
Let’s see some elementary integrations first.
Z L ⇣ m⇡x ⌘ ⇣ n⇡x ⌘
¿ cos cos dx = 0 cos and mix costsare orthogonal
I
L L L
Z L ⇣ m⇡x ⌘ ⇣ n⇡x ⌘
L
sin
L
sin
L
dx = 0 if m and n are distinct nonnegative integers
Z L ⇣ m⇡x ⌘ ⇣ n⇡x ⌘
¡ cos sin dx = 0 if m and n are (positive) integers.
L L L
Z L ⇣ m⇡x ⌘
¬ cos2 dx = 0·
L L
Z L ⇣ m⇡x ⌘ for any positive m
sin2 dx = ·L 0
L L

n.b. definition of “inner product” in function space

Z L
< f (x), g(x) >⌘ f (x) · g⇤ (x)dx
L

If < f (x), g(x) >= 0, then f (x) and g(x) are orthogonal.

To obtain a0 , a1 , a2 , · · · from (1)

Z L Z L 1 
X Z L ⇣ n⇡x ⌘ Z L ⇣ n⇡x ⌘
f (x)dx = a0 dx + an cos dx + bn sin dx
L L n=1 L L L L

=>
a
c).2f(xdx
=

n.b. average value of function in [ L, L]

Z L ⇣ k⇡x ⌘ Z L ⇣ k⇡x ⌘ X1  Z L ⇣ n⇡x ⌘ ⇣ k⇡x ⌘ Z L ⇣ n⇡x ⌘ ⇣ k⇡x ⌘
f (x)cos dx = a0 cos dx+ an cos cos dx+bn sin cos dx
L L L L n=1 L L L L L L

if k
fixscos(*Ydx=ach I
E
=

n =
=>

if n k
=

ais
(Yf(x)cos(dx
=>
=

Similarly,

ba t=(4 fxsin(*)dx
=

© 2023 I-Fan Lin. All Rights Reserved. 2