Sunteți pe pagina 1din 2

MA 210 - Introduction to Numerical Analysis

End-Semester Examination (Spring 2007 - 2008)


Department of Mathematics, IIT Bombay
Max. Marks: 40 • Time: 2:30 to 5:30 PM • Date: 22/04/2008

Instructions:

1. Write your Name and Roll Number clearly on your answer book as well as every supplement you may
use. A penalty of 2 marks will be imposed in case any one of these is not written.
2. Number the pages of your answer book and make a question-page index on the front page. A penalty
of 1 mark will be imposed in case the index is incomplete.
3. The answer to each question should start on a new page. If the answer for a question is split into two
parts and written in two different places, the first part alone will be corrected.
4. Calculators are needed and only scientific calculators are allowed. Any kind of programing device is not
allowed.
5. Formulas used need not be proved but needs to be stated clearly.
6. There are 5 questions asked. Each question carries 8 marks. Answer all the questions.

1. (i) Find the condition number for


f (x) = bx ,
where b is a positive constant. (2 Marks)

(ii) State Lagrange interpolation formula. Find a polynomial P (x) of degree ≤ 2


such that
P (−1) = −1, P (1.5) = 2.4, P (2) = 4.2
using Lagrange interpolation formula. Show that this polynomial representa-
tion is unique. (3 Marks)

(iii) Write an algorithm for composite Simpson’s rule to approximate the integral
Z b
I= f (x)dx. (3 Marks)
a

2. (i) Find the value of c in the equation cos x = c such that the nth iteration of the
Newton-Raphson’s method in solving this equation takes the form xn = (−1)n
for n = 0, 1, · · · . (3 Marks)

(ii) State the secant method for solving a non-linear equation f (x) = 0. (2 Marks)

(iii) Find the root of the equation x2 −2x−1 = 0 using secant method with x0 = 2.6
and x1 = 2.5 by performing the iteration till
|xn+1 − xn | < 0.001.
Round the intermediate calculations in each step (including the final solution)
to five digits after the decimal point. (3 Marks)
3. (i) Write an algorithm for Gauss-Seidel method to solve an n × n system of linear
equations (without using matrix notations). (3 Marks)

(ii) Let A be an n × n non-singular matrix. Suppose the vector b of the system


Ax = b contains error. Obtain the estimate of the relative error in the corre-
sponding solution vectors. (5 Marks)

4. Consider the initial value problem


dy
= x2 + y 2; y(0) = 1. (∗)
dx
(i) State the Runge-Kutta method of order 4 for solving the initial value problem
(*). (2 Marks)

(ii) Use the Runge-Kutta method of order 4 to approximate the values of the so-
lution y of the problem (*) at x = 0.1, 0, 2 and 0.3, using h = 0.1. Round the
intermediate calculations in each step to five digits after the decimal point,
and round the final results of each step to four such places. (6 Marks)

5. Show that the finite difference method


 n
ui+1 +un

n+1 i−1
ui − 2 uni+1 − uni−1
+A =0
∆t 2∆x
for solving the linear advection equation with sound speed A is conditionally
stable. (8 Marks)

S-ar putea să vă placă și