NATIONAL UNIVERSITY OF SINGAPORE

FACULTY OF SCIENCE

SEMESTER 1 EXAMINATION 2010-2011

MA2213 Numerical Analysis I

November 2010 — Time allowed : 2 hours

INSTRUCTIONS TO CANDIDATES

1. This is a closed book examination. Each student is allowed to bring one piece
of A4-sized two-sided help sheet into the examination room.

2. This examination paper contains a total of SIX (6) questions and comprises
FOUR (4) printed pages.

3. Answer ALL questions in Section A. Each question in Section A carries 20

marks.

4. Answer not more than TWO (2) questions from Section B. Each question in
Section B carries 20 marks.

5. Candidates may use calculators. However, they should lay out systematically
the various steps in the calculations.
 PAGE 2 MA2213

SECTION A

Answer all the questions in this section. Section A carries a total of 60 marks.

Question 1. [ 20 marks ]

(a) Solve the following system using Gaussian elimination with partial pivoting and
four digits rounding arithmetic.

0.729x + 0.81y + 0.9z = 0.6867

x + y + z = 0.8338
1.331x + 1.21y + 1.1z = 1.

(b) Let  
2.25 −3 4.5
A =  −3 5 −10 
.


4.5 −10 34
Find a factorization of the form A = LLT where L is lower triangular with
positive diagonal entries. Explain why if there is no such a factorization.

Question 2. [ 20 marks ]

(a) Find the Hermite interpolating polynomial for the following data using Newton’s
divided-difference formula.
x 0 1 3
f (x) 0 1 -1
f 0 (x) 1 0 0
(b) Let i0 , i1 , · · · , in be a rearrangement of the integers 0, 1, · · · , n. Suppose x0 , x1 , · · · , xn
are distinct points. Show that

f [xi0 , xi1 , · · · , xin ] = f [x0 , x1 , · · · , xn ],

where f [xi0 , xi1 , · · · , xin ] and f [x0 , x1 , · · · , xn ] are nth divided differences.

. . .–3–
 PAGE 3 MA2213

Question 3. [ 20 marks ]
(a) Determine constants a, b, c that will produce a quadrature formula
Z 1
xf (x)dx = af (−c) + bf (c)
−1
that has the highest degree of precision. Determine the degree of precision.
(b) Determine the four missing entries in the following table for the approximation
of the integral Z 1
4
dx
0 1 + x2
with Romberg integration. (Keep 6 decimal digits.)
n h Rn,1 Rn,2 Rn,3 Rn,4
1 1 3.000000
2 1/2 R2,1 R2,2
3 1/4 R3,1 3.141568 3.142117
4 1/8 3.138988 3.141592 3.141594 R4,4

SECTION B
Answer not more than two questions from this section. Section B carries a total of
40 marks.

Question 4. [ 20 marks ]
(a) Determine the discrete least squares trigonometric polynomial approximation
S6 (x) in the form
5
a0 X
S6 (x) = + a6 cos 6x + (ak cos kx + bk sin kx),
2 k=1

for f (x) = cos 7x + sin 7x on the interval [−π, π] with 16 points (xj , yj ), where
xj = −π + 8j π for j = 0, 1, · · · , 15.

(b) Let {(xj , yj )}2m−1

j=0 be 2m points where xj = −π + mj π, yj = f (xj ) for j =
0, 1, · · · , 2m − 1. Suppose an interpolatory trigonometric polynomial P (x)
of degree m and a least squares trigonometric polynomial Q(x) of degree
n < m are obtained to approximate f (x) with the given 2m points .
Rπ
The integral −π f (x)dx can then be approximated byR integrating P R(x) or Q(x).
π π
Which do you think will give a better approximation, −π P (x)dx or −π Q(x)dx?
What happens when m → ∞? Justify your answer.

. . .–4–
 PAGE 4 MA2213

Question 5. [ 20 marks ]
Let x0 , x1 , · · · , xn be distinct real points, and consider the following interpolation
problem. Choose a function
n
ck ekx
X
P (x) =
k=0
such that P (xi ) = yi , for i = 0, 1, · · · , n with the given data {yi } .
Show that this problem can be reduced to an ordinary polynomial interpolation and
then use Lagrange interpolation polynomial to find a general formula for P (x).

Question 6. [ 20 marks ]
The set of Legendre polynomials {P0 (x), P1 (x), · · · , Pn (x), · · · , } is orthogonal on
[−1, 1] with respect to the weight function w(x) = 1 where Pk (x) is a polynomial of
degree k. That is, for any i 6= j, one has
Z 1
Pi (x)Pj (x)dx = 0.
−1

Moreover, any Legendre polynomial Pk (x) has k distinct roots, lying in the interval
(−1, 1). Let x1 , x2 , · · · , xn be the roots of the nth LegendreR polynomial Pn (x). Define
1
the following quadrature formula to approximate integral −1 f (x)dx,
n
X
G(f ) = ci f (xi )
i=1

where n
Z 1 Y x − xj
ci = dx for i = 1, 2, · · · , n.
−1 j=1,j6=i xi − xj

(i) Show that the quadrature formula is exact for any polynomial h(x) of degree
less than n.
(ii) Show further that the quadrature formula is exact for any polynomial h(x)
of degree at least n but less than 2n using the fact that h(x) can be written
as h(x) = q(x)Pn (x) + r(x) where Pn (x) is the nth Legendre polynomial and
q(x), r(x) are polynomials of degree less than n.
(iii) Find a quadrature formula with degree of precision 5 given the Legendre poly-
nomial
3
P3 (x) = x3 − x.
5

END OF PAPER