Chap.

4 Partial Differential Equation

(I) Fourier Series Expansion
(1) Fourier Series
(i) Definition:
The Fourier expansion of f defined on the interval (-P,P) is given by
a0 n n
f ( x) = + [an cos x + bn sin x]
2 n=1 p p
1 p
p p
, where a0 = f ( x)dx

1 p n
an =
p p
f ( x) cos
p
xdx

1 p n
bn =
p p
f ( x) sin
p
xdx Fourier coefficient of f .

Example: Expand f ( x) = 0, < x < 0

f ( x) = 2 x,0 x < in a Fourier series
Here ( p, p) = ( , )
1 p 1
a0 =
p p
f ( x)dx = f ( x)dx =
2
1 p n 1 n 1 (1) n
p p
an = f ( x ) cos dx = f ( x ) cos xdx =
p n 2
1 n 1

bn = f ( x) sin xdx =
n
1 (1) n

1
f ( x) = + [ cos nx + sin nx]
4 n =1 n 2
n

(ii) Odd and even functions:

Even, if f ( x) = f ( x) , e.g. x 2 , cos x
Odd, if f ( x) = f ( x) , e.g. x 3 , sin x
a a
If f is even, then a
f ( x)dx = 2 f ( x)dx
0

a
If f is odd, then
a
f ( x)dx = 0

odd * odd=even
even * even =even
odd * even =odd

1 Edited by Prof. Yung-Jung Hsu

 (iii) Cosine and sine series:
(a) If f is an even function on (-P,P), then its Fourier expansion:
1 p 2 p
p p p 0
a0 = f ( x ) dx = f ( x)dx

1 p n n

p
an = f ( x ) cos dx = 2 f ( x) cos xdx
p p p 0 p
1 p n
bn =
p p
f ( x) sin
p
xdx = 0

a0 n
f ( x) = + an cos x Fourier cosine series
2 n=1 p
(b) If f is an odd function on (-P,P), then its Fourier expansion:

n
f ( x) = bn sin x Fourier sine series
n =1 p
2 p n
, where bn =
p 0
f ( x) sin
p
xdx

Example: Expand f ( x) = 1, < x < 0

f ( x) = 1,0 < x < in a Fourier series
Note that f (x) is odd, and the interval ( p, p ) = ( , )
Since f (x) is odd at ( , ) , f (x) can be simply expanded in sine series.

n
f ( x) = bn sin x
n =1 p
2 p n 2 2 1 (1) n
p 0 0
bn = f ( x ) sin dx = f ( x ) sin nxdx = *
p n

(2) Generality in Fourier expansion

In general, f (x) defined on (a, a + 2 p), a R , can also be expanded in a
Fourier series.
Suppose that f (x) is a function defined on (0,2 p ) , and Fourier series is
given by:
a0 n n
f ( x) = + [an cos x + bn sin x]
2 n=1 p p
1 2p
p 0
, where a0 = f ( x)dx

1 2p n
an =
p 0
f ( x) cos
p
xdx

2 Edited by Prof. Yung-Jung Hsu

 1 2p n
bn =
p 0
f ( x) sin
p
xdx

(3) Half-range expansion

odd even function,
Fourier series
If f ( x) = 1, 0 x < , and f (x) is an odd function
,then f (x) Fourier series sine series

n
, f ( x) = bn sin x
n =1 p
If f ( x) = 1, 0 x < , and f (x) is an even function
,then f (x) Fourier series cosine series
a0 n
+ an cos
, f ( x) = x
2 n =1 p
(0, ) (0, ) (0,2 p)
, Fourier series expansion.

(II) Partial Differential Equation

PDE :
1.
2. Laplace
3.

(1) (Separation of variable method)

*: PDE+ B. C. (2 )
: =0 (homogeneous)
= or (non- homogeneous)
Example1: Give T ( x, y ) , a function of x & y :
Steady-state, solid and no heat generation 2T = 0

2T 2T
+ =0
x 2 y 2

3 Edited by Prof. Yung-Jung Hsu

 B.C. : x = 0, T = 0 y = 0, T = 0
x = L, T = 0 y = b, T = f ( x)
By separation of variable method:
2T 2T
Let T ( x, y ) = X ( x) * Y ( y ) + =0
x 2 y 2
d2 X d 2Y 1 d 2 X 1 d 2Y
Y ( y) + X ( x ) = 0 + =0
dx 2 dy 2 X dx 2 Y dy 2
1 d 2 X 1 d 2Y
2
= 2
= 2 , :Eigen value constant
X dx Y dy

d2 X
From
1 : + 2 X = 0 (2nd order homo ODE)
dx 2

Let X = e r , r 2 + 2 = 0 , r 2 = i , X = A cos x + B sin x

BC1 : AY ( y ) = 0 , A = 0

BC2 : B sin L = 0 ,Q B 0 , sin L = 0

n
= , n = 1,2,3... (eigen value)
L
n
X = B sin( x) (eigen function)
L

d 2Y
From 2 :
2
2 Y = 0 , r 2 2 = 0 , r 2 = , Y = Ce y + De y ( BC)
dy

e y + e y e y e y
cosh y = , sinh y =
2 2

Y = C cosh y + D sinh y ( BC)

n n
Y = C cosh( y ) + D sinh( y)
L L
n
BC3 : 0 = C cosh 0 + D sinh 0 = C ,Y = D sinh( y)
L

4 Edited by Prof. Yung-Jung Hsu


n n
T ( x, y ) = X ( x) * Y ( y ) = B sin( x) * D sinh( y)
n =1 L L

n n
= En sin( x) * sinh( y)
n =1 L L

nb nx
BC4 : f ( x) = En sinh( ) * sinh( )
n =1 L L
By the definition of Fourier sin-series expansion, we have

n nx
L
L
En sinh b = f ( x) sin( )dx
L 20 L

nx
L
L
20 f ( x) sin(
L
)dx
En =
nb
sinh( )
L
nx
L
L

nx ny 20 f ( x) sin(
L
)dx
T ( x, y ) = En sin( ) * sinh( ) , where En =
L L n
n =1
sinh b
L

2u u
Example2: Heat equation: k 2 = (u : temperature)
x t
BC1: x = 0, u = 0 BC2: x = L, u = 0 BC3: t = 0, u = x( L x)

Let u ( x, t ) = X ( x) * T (t ) PDE

2 X T 1 2 X 1 T
k = X = 0 = = 2
x 2 t X x 2
kT t
d2 X
+ 2 X = 0
1 x = 0, u = 0
dx 2
x = L, u = 0
d 2T
+ k2 T = 0
2 t = 0, u = x( L x)
dt

1 : r 2 + 2 = 0 , r 2 = i , X = A cos x + B sin x
From

BC1 : A cos ( x) * T (t ) = 0 , A = 0

BC2 : B sin L * T (t ) = 0 ,Q B 0 , sin L = 0

5 Edited by Prof. Yung-Jung Hsu

 n
= , n = 1,2,3...
L
n
X = B sin( x)
L
2 n 2 2 kt
2 : r 2 + k2 = 0 , r = k2 , T = Ce k t = C exp(
From )
L2

n 2 2 kt n
u = X ( x) * T (t ) = C exp( 2
) * B sin( x)
n =1 L L

n 2 2 kt n
= Dn exp( 2
) sin( x)
n =1 L L

n
BC3 : u = Dn sin( x) = x( L x)
n =1 L
By the definition of Fourier sin-series expansion, we have

nx
L
2
Dn =
L0 x( L x) sin(
L
)dx

n 2 2 kt nx nx
L
2
u = Dn exp( 2
) sin( ) , where Dn = x( L x) sin( )dx
n =1 L L L0 L

(2) Laplace transform for PDE

Assume T ( x, t ) , 0 x b , t 0 , L[T ( x, t )] = T ( x, s)
T
1. L[ ] = sT ( x, s ) T ( x,0)
t
2T T
2. L[ ] = s 2 T ( x, s ) sT ( x,0) Tt ( x,0) , where Tt =
t 2
t
T d
3. L[ ] = T ( x, s )
x dx
2T d2
4. L[ ] = T ( x, s )
x 2 dx

T T
Example: +x = 0 , T ( x,0) = 0 , T (0, t ) = 4t
x t
Take Laplace on each part of PDE
dT
+ x[ sT T ( x,0)] = 0 ,
dx
dT
T ( x,0) = 0 + xsT = 0 (1st order linear ODE)
dx

6 Edited by Prof. Yung-Jung Hsu

 1 4
T ( x, s ) = c * exp( sx 2 ) , T (0, s ) = L[T (0, t )] = L[4t ] = 2
2 s
4 4 1
c= 2
, T ( x, s ) = 2 * exp( sx 2 )
s s 2
4 1
PDE T ( x, t ) = L1[T ( x, s )] = L1[ 2
* exp( sx 2 )]
s 2
1 2 1
= 4(t x )u (t x 2 ) (By 2nd-transition theorem)
2 2
u 2 u
Heat equation: k =
x 2 t
u 2 2u
Wave equation: 2 =
x 2 t 2
2u 2u
Laplace equation: + =0
x 2 y 2

(3) (combination of variable method)

* homo. ODE + non-homo. BC
Example: Given Q( x, t ) a function of x & t
Q 2Q
= 2 BC: t = 0, Q = 0 x = 0, Q = 1 x = , Q = 0
t x
By combination of variable method
Let = axbt c , Q( x, t ) = Q( )
Q dQ c ' c
= * = Q = caxbt c 1 = c
t d t t t t
Q dQ b '
= * = Q
x d x x
2Q Q bQ ' b 2
= ( )= + 2 (Q' )'
x 2
x x x 2
x
Q 2Q
ODE: = 2
t x
c ' bQ ' b 2 x2
Q = [ + 2 (Q' )' ]
t x 2
x
2
cx '
Q = [bQ ' + b 2 (Q' )' ] = [bQ ' + b 2Q'+b 2Q '' ]
t
x2
( : b : c = 2 : (1) b 2 b = 0 b = 1 )
t

7 Edited by Prof. Yung-Jung Hsu

 1
( ) 2 Q ' [Q ' + Q'+Q '' ]
2 a

Q '' + Q ' = 0 (2nd-order, ODE, homo,)
2a 2
du
Q ' = u , Q '' =
d
du
+ 2 u = 0 (1st-order ODE)
d 2a
2
ln u + = ln c1
4a 2
1 1
=1 a = ( error function )
4a
2
4
x
= ax bt c =
4t
ln u + 2 = ln c1
2 dQ
u = c1e =
d

Q = c1 e d + c2
2

BC: = 0, Q = 1 & = , Q = 0
BC1 c2 = 1
1
BC2 c1 =
e
2
d
0

e
2
d
Q = 1 0

= 1 erf ( ) = erfc( )
e
2
d
0

8 Edited by Prof. Yung-Jung Hsu