Numerical Analysis Differential Equations

Numerical Analysis
Differential Equation
Hanyang University

Jong-Il Park

Department of Computer Science and Engineering, Hanyang University

Solving Differential Equation
Ordinary D.E.
Partial D.E.
Ordinary D.E.
Linear eg.
Nonlinear eg.

Initial value problem

Boundary value problem

) (
' '
t f y y = +
) (
' ' '
t f y y y = +
Usually no closed-form solution
linearization
numerical solution
0 ) 0 ( ) 0 ( , 0 .
' ' '
= = = + y y y y eg
3 ) 1 ( , 0 ) 0 ( , 10 5 4 .
' ' '
= = = + + y y y y y eg

Discretization in solving D.E.
Discretization

Errors in Numerical
Approach
Discretization error

Stability error
y
t
Grid Points
Exact sol.
d e D
y y e =
n d S
y y e =

sol. numerical :
sol. d discretize :
sol. exact :
n
d
e
y
y
y

Errors
Total error
S D
e e e + =
truncation round-off
D
e
S
e
0 increase
as t A 0
as t A 0
trade-off
t A
e
D
e
S
e

Local error & global error
Local error
The error at the given step if it is assumed that all the
previous results are all exact
Global error
The true, or accumulated, error

Useful concepts(I)
Useful concepts in discretization
Consistency

Order

Convergence

0 0 ) ( = A
D
e h t
5 5
2 2
) (
) (
h e h O
h e h O
D
D

0 h
e
y
y
t

Useful concepts(II)
stability
stable
unstable
y
t
Consistent
stable
Converge

Stability
Stability condition

0
'
) 0 ( , y y Ay y = =
eg.
Exact sol.
Euler method
At
e y y

=
0
0
1
1
) 1 (
) 1 (
y hA
y hA
hAy y y
n
n
n n n
+
=
=
=
=
For stability
A
h hA
2
0 1 | 1 | s < s
Amplification factor

n n n n
n n n
y h h y h y y
hf y y
) 1 ( 2 . 1 ) 2 . 1 (
1
1
= + =
+ =
+
+
h
h y
y h y y
hf y y
n
n n n
n n n
+
+
=
+ =
+ =
+ +
+ +
1
2 . 1
) 2 . 1 (
1 1
1 1
Explicit :
Implicit :
h increase
h large
h small
ye
y
t
y
t
explicit implicit
conditionally stable stable
y y y y y = = = + 2 . 1 2 . 0 ) 0 ( , 2 . 1
' '
eg.
= f
Implicit vs. Explicit Method

Modification to solve D.E.
Modified Differential Eq.
) ( .
'
t g Ay y eg = +
Diff.
eq.
Modified
D.E.
Discretization
Discretization by Euler method
) (
1 n n n n
Ay g h y y + =
+
<Consistency check>
= +
+ = + + +
+ + + =
+
' '
! 2
1
'
' ' 2
! 2
1
'
' ' 2
! 2
1
'
1
|
) ( | |
| |
hy g Ay y
Ay g h y y h hy y
y h hy y y
n n n
n n n n n n
n n n n
0 Let h
n n n
g Ay y = + |
' consistent ;
<Order>
) ( |
'
h O g Ay y
n n n
+ = +

Initial Value Problem: Concept

Initial value problem
Initial Value Problem

Simultaneous D.E.

High-order D.E.
0 0
'
) ( , ) , ( y t y y t f
dt
dy
y = = =
0 0 , 2 1
10 0 1 , 2 1 1
1
) ( ), , , , (
) ( ), , , , (
n n n n
n
n
y t y y y y t f
dt
dy
y t y y y y t f
dt
dy
= =
= =
) 1 (
0 0
) 1 ( '
0 0
'
0
) (
) 1 ( ' ' ' ) (
) ( , , ) ( ,
) , , , , , (
0

= = =
=
n n t
n n
y t y y t y y y
y y y y t f y

Well-posed condition


Taylor Series Method
) , (
'
y t f y =
n
n n
n
R h y h y h y y h t y + + + + + = +
) (
0 !
1
2 ' '
0 ! 2
1
'
0 0 0
) (
Truncation error
) (
0
h t y +
) (
0
t y
0
t h t +
0
h
h t t
y
n
h
R
n
n
n
+ s s
+
=
+
+
0 0
) 1 (
1
, ) (
)! 1 (
Taylor series method(I)

High order differentiation

Implementation

2 2
' ' '
' '
2
) ( ) ( ] [
))] ( , ( [
f f f f f f f f f
f f f
y
f f
t
f f f f f f
dt
d
y
f f f
dt
dy
y
f
t
f
t y t f
dt
d
y
yy y t y ty tt
y y ty tt y t
y t
+ + + + =
c
c
+
c
c
+ + = + =
+ =
c
c
+
c
c
= =
<Type 1>
Complicated computation
0
t
'
0
y
' '
0
y
' ' '
0
y
h
h
h
y
t
Less computation
accuracy
<Type 2>
h h h
0
t
1
t
2
t
3
t
'
0
y
' '
0
y
' ' '
0
y
'
1
y
' '
1
y
' ' '
1
y
'
2
y
' '
2
y
' ' '
2
y
....
y
t
More computation
accuracy
Requiring complicated
source codes
Taylor series method(II)

Euler method(I)
Euler Method
0
t
....
y
t
1
t
2
t
3
t
....
0
y
0
y
1
y
2
y
3
y
h
0 0
'
) ( ), , ( y t y y t f y = =
Talyor series expansion at t
o
1 0 0
2
0
' '
! 2
1
'
0 0 1
, ) ( t t h y h y y y s s + + =
0
0 0
) , (
f
y t f
=

Euler method(II)
Error
Eg. y =-2x
3
+12x
2
-20x+8.5, y(0)=1

Euler method(III)
1
2
2 ' '
! 2
1
1
, ) (
) (
+
+
s s + + =
+ + =
n n n n n
n n n n
t t h O h f y
h y h f y y
=
+
+ = + + + + =
1
0
1 0 1 1 2 0 1 0
) ( ) ( ) ( ) (
n
i
i i n n n
y y y y y y y y y y y
) ( ) ( ) (
2
1
) (
2
1
) (
_
' '
0
2
_
' '
0
2 ' '
1
0
2
1
h O h y t t
h y
h
t t
h y e
n
n
i
n
i
t
= =
=
=
Generalizing the relationship

Eulers approx.
truncation error
Error Analysis
Accumulated truncation error
h
t t
n
t t y y
n
n
n
i
i n
0
0
1
0
' '
1
' '
, ) ( ) (
=
s s =

=

; 1
st
order

Eg. Euler method

Modified Euler method: Heuns method
Modified Eulers Method
Why a modification?
error
modify
n
t
n
t
1 + n
t
C
n
y
1 +
P
n
y
1 +
P
n
f
1 +
n
f
2
1
P
n n
f f
+
+
1 + n
t
Predictor

Average slope

Corrector
n n
P
n
hf y y + =
+1
2
) , ( ) , (
2
1 1
'
1
'
'
P
n n n n n n
y t f y t f y y
y
+ + +
+
=
+
=
)] , ( ) , ( [
2
1 1
'
1 + + +
+ + = + =
n n n n n n
C
n
y t f y t f
h
y y h y y

Heuns method with iteration
Iteration
significant
improvement

Error analysis
Error Analysis
Taylor series

Total error

) ( ] [
) ( ) (
) (
3 '
1
'
2
' ' ' 3
! 3
1
' '
1
2
2
1
'
' ' ' 3
! 3
1
' ' 2
2
1
'
1
h O y y y
y h h O
h
y y
h hy y
y h y h hy y y
n n
h
n
n n
n n
n n n n
+ + + =
+
)
`
+ + =
+ + + =
+
+
+
truncation
3
rd
order
) (
2
h O
; 2
nd
order method
Significant improvement over Eulers method!

Eg. Euler vs. Modified Euler
e Euler Modified
Euler Method
improvement

Runge-Kutta method
Runge-Kutta Method
Simple computation

very accurate

The idea

code source Easy . , , no
' ' '
y y
) , , (
1
h y t h y y
n n n n
| + =
+
where
) , (
) , (
) , (
1 1
1 1 2
1
2 2 1 1

+ + =
+ + =
=
+ + + =
n n n n n
n n
n n
n n
y t f k
y t f k
y t f k
k a k a k a
| o
| o
|

Second-order Runge-Kutta method
Second-order Runge-Kutta method
) , (
) , (
) (
1 1 2
1
2 2 1 1 1
| o + + =
=
+ + =
+
n n
n n
n n
y t f k
y t f k
k a k a h y y

Taylor series expansion
( ) ( )
n
n
y
f
n
t
f
n
h
n y t
h
n n n
R f k
y f f f f h y y
+ + + =
+ + + + =
c
c
c
c
+
1 1 2
' ' '
! 3 ! 2 1
) (
) ( ) ( ) (
3 2
| o

n n y t n n n
hR f f ha f a a h y y + + + + + =
+
) ( ) )( (
1 1 2 2 1 1
| o

Equating and
) , (
2
,
2
, 1
1 2 1 2 2 1 n n
y t f
h
a
h
a a a = = = + | o

Modified Euler - revisited
) (
2 1 2 1
k k y y
h
n n
+ + =
+
) , (
1 n n
y t f k =
) , (
1 2
hk y h t f k
n n
+ + =
Modified Euler method
set
1 1 1 1
, ,
2
1
hk h a = = = | o
2
1
2
= a
Modified Euler method is a kind
of 2
nd
-order Runge-Kutta method.
P1
P2
) , (
1 n n
y t f k =
) , (
1 2
hk y h t f k
n n
+ + =
1 + n
y

Other 2nd order Runge-Kutta methods
Midpoint method

Ralstons method

1 1 1 2 1
2
,
2
, 1 , 0 k
h h
a a = = = = | o
))) , ( , ( (
2 2 1 n n
h
n
h
n n n
y t f y t f h y y + + + =
+
) 3 (
2 1
4
1
k k y y
h
n n
+ + =
+
) , (
1 n n
y t f k =
) , (
1
3
2
3
2
2
hk y h t f k
n n
+ + =

Comparison: 2
nd
order R-K method

Comparison: 2
nd
order R-K method
Eg. y =-2x
3
+12x
2
-20x+8.5, y(0)=1


Fourth-order Runge-Kutta
Taylor series expansion to 4-th order
accurate
short, straight, easy to use
) , (
) , (
) , (
) , (
} ) ( 2 {
3 4
2 2 2 3
1 2 2 2
1
4 3 2 1 6
1
1
hk y h t f k
k y t f k
k y t f k
y t f k
k k k k h y y
n n
h
n
h
n
h
n
h
n
n n
n n
+ + =
+ + =
+ + =
=
+ + + + =
+
) (
1
P f
n
t
2
h
n
t + h t
n
+
P1
P2
P3
P4
) (
3
P f
) (
2
P f
) (
1
P f
significant improvement over
modified Eulers method
4-th order Runge-Kutta methods

Runge-Kutta method

Eg. 4-th order R-K method
Significant
improvement

Discussion
Better!

Comparison
(5
th
order)

Numerical Analysis Differential Equations

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Numerical Analysis Differential Equations

Загружено:

Авторское право:

Доступные форматы

Numerical Analysis

Taylor series method(I)

Generalizing the relationship

Вам также может понравиться