Solution of EXAM #2 - Northern Illinois Universitydattab/math435.2009/SOLUTION_EXAM2.pdfSolution of...
Click here to load reader
Transcript of Solution of EXAM #2 - Northern Illinois Universitydattab/math435.2009/SOLUTION_EXAM2.pdfSolution of...
Solution of EXAM #2 1. (a) Derive the composite Trapezoidal rule.
So, the composite trapezoidal Rule is:
The error of composite trapezoidal Rule:
(b) Determine h so that the composite Trapezoidal rule gives the value of
21
0
xe dx−∫
With an accuracy of . 710ε −=
2(1 0) ( )12
h f η ε− ′′ <
2 22| ( ) | | ( ) | | ( 2 4 ) | 2x xf e x eη −′′ ′′= = − + ⋅ ≤
72 712 12 10 6 10
max{ ( )} 2h
fεη
−−⋅ ⋅
< = =′′
× , so 47.74597 10h −< × .
2. Approximate using Gaussian Quadrature with n=2. 21.5
1
xe dx−∫Sol: Changing of variables: 1 1[(1.5 1) (1 1.5)]
2 4x t= − + + = +
54
t
So, 14
dx dt=
22
51.5 1 ( )4 4
1 1 2 21 1
1 ( ) ( )4
txe dx e dt a f t a f t
− +−
−= = +∫ ∫
Where 25( )
4 41( )4
t
f t e− +
=
For 2n =
1 2 1a a= = ,
and the roots of the Legendre polynomial of degree 2 : 1,213
t = ±
So, 2 2
21 5 1 5( ) ( )1.5 4 44 3 4 3
1 1 2 21
1 1( ) ( )4 4
xe dx a f t a f t e e− + − +
− = + = +∫ .
(b) Construct a quadrature rule of the form:
1
0 1 21
1 1( ) ( ) (0) ( )2 2
f x dx A A f A f−
≈ − + +∫
Which is exact for all polynomial of degree 2≤ .
( ) 1f x =
1
0 111 2dx A A A
−= = + +∫ 2
( )f x x=
1
0 21
1 102 2
xdx A A−
= = − +∫
2( )f x x=
1 20 21
2 1 13 4 4
x dx A A−
= = +∫
3. Determine an upper bound for the global error at t=1 of Euler’s Method in
solving , fromy y′ = (0) 1y = 0t = , to 1t = , 0.5h = . (Note that the exact
solution is ( ) ty t e= ).
Sol: The global error ( )| | | ( ) | ( 12
iL t aEG i i
hME y t y e )L
−= − ≤ −
Where ( , )| | 1f t yLy
∂≥ =
∂| ( ) | | |t, t e M e′′ ≤ = , 0.5h = , 1 20.5, 1t t= = y =
( ) 0.5| | ( 1) ( 1) 1.167692 2
iL t aEG
hM eE e eL
− ⋅≤ − = − = .
4. (a) Given2 3
3( ) 12 6x xP x x= + + + , find a polynomial of degree 2 such that
is minimized. What is the minimum value?
2 ( )P x
3 21 1max | ( ) ( ) |
xP x P x
− ≤ ≤−
Sol: , 2 3 3 3( ) ( ) ( )P x P x a T x= −
3 33 3
3( ) 4 3 ( )4
T x x x T x x x= − ⇒ = − ,
So, 2 3
32 3 3 3
1 3 9( ) ( ) ( ) 1 ( ) 12 6 6 4 8 2
2x x xP x P x a T x x x x x= − = + + + − − = + + .
(b) Use Legendre polynomial of degree 2 to approximate 3( )f x x= on the [ 1 . ,1]−
Sol: , 0 0 1 1 2 2( ) ( ) ( ) ( )P x a x a x a x= Φ + Φ + Φ
Where 0 ( )xΦ , 1( )xΦ , 2 ( )xΦ are the legendre polynomial of degree 0, 1
and 2, respectively. And,
1
11 2
1
( ) ( )
( )
ii
i
x f x dxa
x dx−
−
Φ=
Φ
∫∫
.
0 ( ) 1xΦ = , 1( )x xΦ = , 22
1( )3
x xΦ = − ,
So,
1 3
10 1
1
x dxa
dx−
−
= ∫∫
,
1 3
11 1 2
1
x x dxa
x dx−
−
⋅= ∫∫
,
1 2 3
12 1 2 2
1
1( )31( )3
x x dxa
x dx
−
−
− ⋅=
−
∫
∫.
5. Bonus Problem
Find the best possible interpolating nodes for interpolation of ( ) xf x xe= in
using a polynomial of degree at most 2. [0,1.5]
Sol: The roots of the polynomial 33 ( ) 4 3T x x x= − are 1,2,3
30,2
x = ± .
Changing of nodes: 1.5 0 0 1.5 3 32 2 4i i ix x x
4− +
= + = + ,
So, 1 0.75x = , 2 1.39952x = , 3 0.10049x = are the best possible
interpolating nodes for interpolation of ( ) xf x xe= in using a
polynomial of degree at most 2.
[0,1.5]