MTH208 NUMERICAL ANALYSIS THE SECOND SEMESTER 2021/2022 FINAL EXAMINATION
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
MTH208
THE SECOND SEMESTER 2021/2022 FINAL EXAMINATION
BACHELOR DEGREE - Year 3
TIME ALLOWED: 2.5 hours
NUMERICAL ANALYSIS
Question 1 (5 marks).
show that the sequence ①n = 32(1)π (n = 1, 2, . . .) converges quadratically as n y 必.
Question 2 (15 marks in total).
Consider the ixed point iteration scheme
pn+1 = g (pn ) .
(2-a) (5 marks). state the ixed point iteration theorem, i.e., the theory for ensuring convergence of the above scheme to ixed point p in the interval [a, b].
(2-b) (5 marks). Describe the upper bound for the absolute error |pn — a| .
(2-c) (5 marks). It is known that if g(①) e Ck ([a, b]) for k > 2 and g([a, b]) 军 [a, b]
g、(p) = . . . = g (k-1) (p) = 0,
then ①n converges to aixed point with kth order of convergence. Consider the following iteration for calculating a ixed point T1/3 :
①n+1 = a①n + b + c
Assuming that this iterative scheme converges as ①0 su伍ciently close to T1/3, determine the values of a,b,c such that the method has cubic convergence rate.
Question 3 (2O marks).
consider g = sin ①,
(3-a) (5 marks). Obtain the Lagrange interpolating polynomial from the data
sinO = O,
sin = ,
sin = ,
sin = 1
and use it to evaluate sin
(3-b) (5 marks). Find the error bound of the Lagrange interpolating polynomial at ① = . compare the error bound with the actual error at the point ① = .
(3-c) (1O marks). Let pn be the degree n Lagrange interpolating polynomial of cos(2①) on the uniformly spaced nodes ①0 , . . . , ①n on [O, 1] with ①j = jh, h = 1/n. prove that
0三(m)三(x)1 |cos(2①) — pn (①)| 一 O as n 一 钝.
Question 4 (1O marks).
(4-a) (5 marks). use the most accurate three-point endpoint formula to determine the missing value f、(1.4) in the following table (keep 4 signiicant digits):
① |
f (①) |
f、(①) |
1.1 |
9.O25 |
|
1.2 |
11.O2 |
|
1.3 |
13.46 |
|
1.4 |
16.44 |
|
(4-b) (5 marks). The data in part (4-a) were taken from the function f (①) = e2从. Compute the actual error at ① = 1.4 and the error bound using the error formula.
Question 5 (15 marks). Consider the deinite integral 11(3) ln(从2)d从.
(5-a) (5 marks). Approximate this integral using the composite simpson,s rule with 5 equal spaced nodes 从0 , 从1 , . . . , 从4 .
(5-b) (5 marks). Give a rigorous theoretical upper bound on the absolute error in your answer to (5-a).
(5-c) (5 marks). Based on the upper bound obtained in (5-b), determine theoretical values of n and h required to obtain an approximation accurate to within 10-4 using the composite simpson,s rule.
Question 6 (15 marks).
Given the initial-value problem
g、= g + t2 et , 1 三 t 三 2, g(1) = O,
with exact solution g(t) = t2 (et - e) :
(6-a) (5 marks). use Euler,s method with h = O.1 to approximate the values of g at t = 1.5 and t = 1.6.
(6-b) (5 marks). use the answers generated in part (6-a) and linear interpolation to approximate g(1.55).
(6-c) (5 marks). Based on the theoretical error estimate, compute the largest value of h to ensure that |gi - wi | 三 O.5 (the answer should be in the form x.xxxx).
Question 7 (10 marks).
Let
( 2 4
A = ' 4 10
( 6 16
6 )
16 '
28 )
be a given matrix in R3收3 .
(7-a) (5 marks). Find a lower triangular matrix L and a diagonal matrix D such that A = LDL」 .
(7-b) (5 marks). Compute an approximation to the largest eigenvalue of A by performing two steps of the power method with ①0 = [1, 1, 1]T . keep at least 4 decimal places in calculations.
Question 8 (1O marks).
( ,
consider the following linear system:〈
,
从1 + 从2 + 3从3 = 5
从1 + 3从2 + 从3 = 2
3从1 + 从2 + 从3 = —4
(8-a) (5 marks). Re-order the equations so that Jacobi iteration will converge to the exact solution as fast as possible. carry out 2 iterations with starting vector x(0) = (O, O, O). Explain why the iteration is convergent.
(8-b) (5 marks). The spectral radius of a matrix is di伍cult (expensive) to calculate. Fortunately, for any matrix M we have (by a theorem of linear algebra) that p(M) 三 IMI where p(M) is the spectral radius and I . I is any matrix norm. write your Jacobi iteration from (a) in the form x(n+1) = Tx(n) + b. calculate IT I“ and use this value to estimate the error after 2O Jacobi iterations.
2023-07-21