Numerical Analysis Math446/OR481 - DL1

Mid-terms Exam Fall 2022

INSTRUCTIONS
• This is a Take-home exam and is worth 15% of your total grade.
• This exam consists of three pages including this cover page.

• This examination is open book, open notes, open mind, open computer, open Web. However, you
cannot talk to anyone about the exam, nor look at anyone else’s exam, nor allow anyone to see your
exam.
• There are nine problems in this exam divided into three parts. Part-A has five problems each worth
15 points. Part-B has three problems each worth 20 points. Part-C has one problem worth 30 points.

• Write (or type) your solutions clearly and show complete work for full credit. Please try to start every
problem on a new sheet. Also, write your name and GMU ID-number on each page.
• To obtain full credit, the completed solutions (via scanned soft copy) must be uploaded to blackboard
no later than midnight on October 10, 2022 (Monday). Any delay will result in loss of points.

• Remember! You are not allowed to collaborate on this exam. Any evidence of cheating or group work
will automatically result in a failing grade and will result in disciplinary measures consistent with the
University’s academic dishonesty policy.
• Please read the statement in quotes below. Complete the information in this coverpage and submit
it along with your solutions. Note that “inappropriate assistance” is assistance from (or to) anyone
other than the instructor Dr. Padmanabhan Seshaiyer. By signing below, you are stating that you
have read and understand that you will not receive or give any inappropriate assistance on this exam.

“I have neither received nor given inappropriate assistance on this examination.”

Full Name:

GMU ID Number:

Signature:

1
 Part A - Show complete work for full credit.

1. (15 points) If a marathoner can run 26.2 miles in 2.2 hours, then prove that there was at least two
points in the race where he was running exactly at 11 miles per hour. Clearly give details of the
mathematical facts you need to come to the conclusion.
2. (15 points) Show that |ea − eb | ≤ |a − b| for all a, b ≤ 0.

3. (15 points) Consider solving the quadratic equation x2 + 100x + 1 = 0 for its roots. Using a four digit
decimal arithmetic with chopping find the two roots to within 0.01% accuracy in relative error. (You
may assume the exact roots are x1 = −0.010001 and x2 = −99.989999).
4. (15 points) Using any of the methods for solving one-variable equations, find the value of x that
produces the closest point on the graph of y = x2 to the point (1, 0) to within an accuracy of 10−4 .

5. (15 points) Consider solving the nonlinear system given by:

x2 − 10x + y 2 + 8 = 0
2
xy + x − 10y + 8 = 0

using Newton’s Method for systems. Write a MATLAB program to perform 4 iterations of the Newton’s
Method with the initial vector (0.5, 0.5). Attach your code and the output of each iteration.

Part B - Show complete work for full credit.

ln(x + 1) − ln(x)
6. (20 points) Let f (x) = .
2
(a) Use three-digit decimal arithmetic with rounding to evaluate f (1000).
(b) Rewrite f (x) in a form that avoids the loss of significant digits and evaluate f (x) for x = 1000
once again.
(c) The exact answer may be assumed to be f (1000) = 0.4998 × 10−3 . Compare the relative errors
for the answers obtained in (a) and (b).
7. (20 points) Consider applying Newton’s method to calculate the reciprocal of a number R.
(a) Establish the iterative scheme: xk+1 = xk (2 − Rxk ).
(b) Show that this iterative scheme yields quadratic convergence.
8. (20 points) Suppose f (x) is continuous in [x1 , x2 ] and that x1 and x2 are in [a, b].
(a) Suppose that p and q are positive constants. Show that there is a number c between x1 and x2
pf (x1 ) + qf (x2 )
such that: f (c) = .
p+q
(b) Give an example to show that the result in part (a) does not necessarily hold when p and q have
opposite signs with p 6= −q.

2
 Part C - Show complete work for full credit. Attach your code and output.

9. (30 points) Consider the fixed-point method to solve for the positive real root of the polynomial
20x3 + x2 − 36 = 0 using the iteration function
r
36
g(x) =
20x + 1
that satisfies x = g(x).

(A) Find the positive real root of 20x3 + x2 − 36 = 0. Show your work. (You may use any approach
to find the zeros of the polynomial.)

(B) Evaluate the approximate root by using the fixed-point method algorithm with an initial guess
x0 = 1 that is accurate upto 10−6 . Using the first 10 iterations of the approximate solution
x1 , x2 , . . . , x10 , study the convergence of the method employed.

(C) Define the ∇xi = xi+1 − xi for i ≥ 0 and ∆xi = xi+2 − 2xi+1 + xi for i ≥ 0. Write a MATLAB
program to implement the following steps:
(a) Using the fixed-point algorithm with x0 = 1, determine x1 = g(x0 ) and x2 = g(x1 ). Use the
three values x0 , x1 , x2 to find x̂ as follows:

(∇x0 )2
x̂ = x0 −
∆x0
Call S1 = x̂ for this first approximation.

(b) Using the fixed-point algorithm with x0 = x̂, and repeat the previous step. That is, determine
x1 = g(x0 ) and x2 = g(x1 ). Use the three values x0 , x1 , x2 to find x̂ as follows:

(∇x0 )2
x̂ = x0 −
∆x0
Call S2 = x̂ for this second approximation.
(c) Perform 10 iterations of this process to obtain S1 , S2 , . . . , S10 and use this sequence to study
the convergence of method employed here.
(D) Compare the convergence in steps (B) and (C).