0% found this document useful (0 votes)
11 views

Numerical Methods

Solution

Uploaded by

xinyueya
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
11 views

Numerical Methods

Solution

Uploaded by

xinyueya
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 8

NUMERICAL METHODS

HOMEWORK 10

• Due date: Wednesday, November 20, 2024 at 11:59PM on Can-


vas/Gradescope
• The assignment will be graded out of 20 points. Point values of each problem
are labeled.
• Remember: You are allowed to discuss the assignment with your fellow stu-
dents or consult other references as long as the solution write up is your own
and you provide citation of who you discussed with or what reference you
used.
Exercise 1 (4 points). We want to compute
Z 1
I(f ) = f (x)e x dx.
0

Determine weights w0 , w1 and points x0 , x1 for which the quadrature rule Q(f ) =
w0 f (x0 ) + w1 f (x1 ) is exact for polynomials of degree three or less, i.e.
Solve this problem by writing equations corresponding to Q(xi ) = I(xi ) for i =
0, 1, 2, 3 and then solving R 1 the equations.
Hint: Use the formula 0 x e x dx and try to solve the system by changing variables
n

to
x0 = x̄ , x0 = x̄ + , w0 = w̄ ⌘, w1 = w̄ + ⌘.
R1 n x
Also, the formula 0 x e dx = n! might be useful.
Exercise 2 (3 points). We now consider the inner product
Z 1
(f, g) = f (x)g(x)e x dx.
0

The exercise finds a quadratic polynomial q that is orthogonal to polynomials of


degree 1.
a) (1 points) Write q(x) = x2 + a1 x + a0 and solve two linear equations for a0 , a1
given by (q, 1) = 0 and (q, x) = 0.
b) (2 points) Beginning with the basis {1, x, x2 } for polynomials of degree 2 and
q0 (x) = 1, use the Gram Schmidt orthogonalization process to compute q1 of
degree 1 orthogonal to q0 and then q2 of degree 2 orthogonal to q0 , q1 .
Exercise 3 (2 points). Verify that the roots of the polynomial from the previous
exercise are the points found in Exercise 1, and that the weights are
Z 1
wi = `i (x)e x dx
0

where `i is the Lagrange polynomial corresponding to the points {x0 , x1 }.


1
2 NUMERICAL METHODS HOMEWORK 10

Exercise 4 (4 points). We again consider integrals of the form


Z 1
I(f ) = f (x)x 1/2 dx
0
1/2
We consider Gauss quadrature rule with this weight function w(x) = x and inner
product Z 1
1/2
(p, q) = p(x)q(x)x dx.
0
a) (2 points) Let q(x) = x2 + a1 x + a0 and determine a0 , a1 such that (q, 1) = 0 and
(q, x) = 0.
b) (2 points) What are the nodes x0 , x1 and weights w0 , w1 such that the quadrature
rule
Q(f ) = w0 f (x0 ) + w1 f (x1 )
is exact for polynomials of degree less than or equal to 3, i.e. I(p) = Q(p) for
deg(p)  3?
Exercise 5 (4 points). Consider the integral that we want to compute
Z 1
I(f ) = f (x)(1 x2 ) 1/2 dx
1
Consider the quadrature rule
n
X
Q(f ) = wi f (xi )
i=0

where xi are the roots of the Chebyshev polynomial Tn+1 (x) = cos((n + 1) arccos(x))
and the weights are the typically Gauss quadrature weights wi = I(`i ). This rule is
exact for polynomials of degree 2n + 1.
Derive an estimate for the quantity
E(f ) = |Q(f ) I(f )|
using polynomial interpolation theory.
Hint: You may want to use the inequality for a function g : [ 1, 1] ! R
Z 1
|g(x)|(1 x2 ) 1/2 dx  ⇡ max |g(x)|.
1 1x1

Another true statement (but you should show this) is for a continuous function
g : [ 1, 1] ! R, we have
n
X
|Q(g)| = wi g(xi )  ⇡ max |g(x)|.
1x1
i=0
NUMERICAL METHODS HOMEWORK 10

Exercise 6 (3 points). We again consider integrals of the form


Z 1
I(f ) = f (x)dx
1
We consider Gauss quadrature rule with this weight function w(x) = 1 and inner
product Z 1
(p, q) = p(x)q(x)dx.
1
a) (1 point) Let q(x) = 5x3 3x. Show that (q, x) = (q, x2 ) = 0. Notice that
(q, 1) = 0 because q is an odd function. R1
b) (2 points) Let xi be the roots of q and let the weights be wi = 1 `i (x)dx. We
consider the quadrature rule
5 ⇣ p ⌘ 8 5 ⇣p ⌘
Q(f ) = w0 f (x0 ) + w1 f (x1 ) + w2 f (x2 ) = f 3/5 + f (0) + f 3/5 .
9 9 9
Let f (x) = x5 7x4 . Compute Q(f ) and then state the error Q(f ) I(f ) using
theorems from class.
1 Q Xi Iexi 5 xie xd x no Xo wix

For i o w.tw Se dx 1

For 1 1 wox.tw x for e dx 1

For i 2 WoXo w xp for X'exdx 2

For i 3 Woxo w xp ff Xe dx 6
Thing in Xo x̅ 8 x x̅ 8 Wo w̅ n w int n

w̅ 1
x̅ 8 w̅ n x̅ sicity 1
x̅ sit w̅ no ix stains 2
x̅ 8 w̅ n x̅ 8pct 7 6 4
Xo 2 52 X 2 52 W
24 W
2 a q 1 1 1 4 a ao e dx
2 a 1 do o 2 a do 0

q X f a ao xe dx
3 9 2 ao 1 6 29 an o

91 90 2 91 4
291 a 6 no 2

q X X2 4 2

b Go X 1

q 1 7 X co 1 where co 1
X 1

q x X C 9 x c 8 X where C
1 1 4
X2 41 1 2
2
co
X 4 2

3 nots X 4 2 0 X 1 42 2 a
Xo 2 T X 2 52 same as EX 1

weights l

wo
flex e
1ft F
w f b ex e same as Ex 1
4 a
q 1 St x ax ao x dx
32
Ex 290 72
Ja
a 2 no 0
1 2
q X Sf x aix ao x x dx

Ix Ea a

I F a no 0

5 alt 2 do
Fania
1
q x X

x
b Xo X are roots of q

IF x I x 3 its
e wo
flex x ax 1

lie we flax x ax 1
5 Ecf D f Icf 1
1 O f p 0 p Icf p I p

0 f p Icf p since Q p 1 p

Since Ʃ Wi I 1 T 101971 ƩWig Xi


Ʃ Wig xi max gx Wi Imax g x

let g f p then Q f p 1 a max f p x

Given Stigenice xitraxsi.fif i i


Icf p f fix px 1 x2 dx
S if x p x 1 25 dx
p1 11
T.im Y1fx
1 oct p Icf p 2 fix p x
Ecf 19
3 3
6 a
q x 51115 dx
3
X5 11 1 1 1 1 1 0

q 5115 3 3 7X dx
Ex Ix É É ÉÉ 0

4
b Q f 0 7

É C F 7 1 E1 to E 7 1751
4
2 5 7 FF

Q f 1 f o since off is exact for degree up to 5

You might also like