Scribd
Upload
36 views

Final 2015

This document contains the final examination for a numerical methods course. It consists of 7 problems testing various numerical methods topics, including LU decomposition, bisection method, Lagrange polynomial interpolation, B-spline interpolation, least squares approximation, Simpson's rule for integration, Taylor series, and solving initial value problems. Students are asked to derive formulas, apply numerical methods, and explain concepts for full credit over the course topics.

Uploaded by

Sheikh Usman Ali
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
36 views

Final 2015

This document contains the final examination for a numerical methods course. It consists of 7 problems testing various numerical methods topics, including LU decomposition, bisection method, Lagrange polynomial interpolation, B-spline interpolation, least squares approximation, Simpson's rule for integration, Taylor series, and solving initial value problems. Students are asked to derive formulas, apply numerical methods, and explain concepts for full credit over the course topics.

Uploaded by

Sheikh Usman Ali
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

F INAL E XAMINATION

120202: ESM4A - N UMERICAL M ETHODS

Spring 2015

Prof. Dr. Lars Linsen


Jacobs University
May 21, 2015

Problem1: LU Decomposition.
 (7+2+6=15 points)
1 0 2
Let A =  0 4 2 .
2 2 14
(a) Give a constructive proof that an LU decomposition can be computed for any positive definite
matrix by deriving the matrices L and U in general form.

(b) Show that matrix A is positive definite.


(c) Apply the LU decomposition scheme derived in (a) to the given matrix A.

Problem 2: Non-linear Equations. (4+3+6+2=15 points)


Assume a function f ∈ C 0 [a, b], where r ∈ [a, b] for a root r of f . We want to apply the bisection method
to find r.
(a) Give the definition of stability of an iterative numerical method and show that the bisection method
is stable.

(b) Give the definition of consistency of an iterative numerical method and show that the bisection
method is consistent.
(c) Derive the convergence rate of the bisection method for [a, b] = [0, 1] and f (x) = x2 − 34 x + 81 .
(d) Can the bisection method be applied to the problem given in (c)? Explain.

Problem 3: Polynomial Interpolation. (3+4+5+3=15 points)


Given the interpolation problem that we want to interpolate the points p0 , . . . , pn at the knots u0 , . . . , un .
Lagrange interpolation follows the idea that the nodes are equal to the points such thatP the collocation
n
matrix becomes the identity matrix and the interpolating polynomial is given by p(u) = i=0 pi Lni (u).
(a) Using this idea of Lagrange interpolation, derive a closed-form expression of the Lagrange poly-
nomials Lni (u), i = 0, . . . , n. Explain all steps.
(b) Starting from the closed-form solution developed in (a), derive the recursive formula for Lagrange
polynomials (
Lk−1 (u) · u−uk for i = 0, . . . , k − 1
k
Li (u) = Pi k−1 k−1ui −uk u−uk
i=0 L i (u) · (1 − ui −uk ) for i=k
Hint: Use the partition-of-unity property of Lagrange polynomials (without proof).
(c) Starting from the recursive formula for Lagrange polynomials in (b), derive Aitken’s algorithm for
evaluating the interpolating polynomial p(u) for a fixed value u.

(d) Apply Aitken’s scheme derived in (c) to compute p(1) for the polynomial that interpolates the
points 0 and 1 at the knots 0 and 2, respectively.

1
Problem 4: Spline Interpolation. (4+5+4+2=15 points)
B-splines are splines of minimal support.
(a) Based on the property of minimal support, draw B-splines N00 (u), N01 (u), and N02 (u) over knot
sequence (0, 1, 2, 3). The drawings need to be qualitatively correct.
(b) Use the constructions in (a) to estimate N02 (0), N02 (1), N02 (2), and N02 (3). Explain how you obtain
the values.
Hint: No need to derive and evaluate the recursive formula for any u.
(c) Using (b),
Pnderive the collocation matrix for spline interpolation at knots 2, 3, . . . , n + 1 using spline
s(u) = i=0 ci Ni2 (u) defined over knot sequence 0, 1, . . . , n + 3.
Hint: If you did not solve (b), use placeholders for the function values of N02 (u).
(d) Is spline s(u) in (c) a natural spline? Explain your answer.

Problem 5: Least Squares Approximation and Gaussian Elimination. (4+2+7+2=15 points)


Given interpolation conditions p(0) = 0, p(1) = 2, and p(2) = 1. Assume that p(u) is a linear polynomial
given with respect to the monomial basis.
(a) Derive the collocation matrix for this interpolation problem and the respective linear equation
system to determine p(u).
Hint: This leads to an overdetermined system of linear equations.
(b) Derive the error that is minimized when solving the overdetermined system of (a) in the least-
squares sense.
(c) Use Gaussian elimination with scaled partial pivoting to solve the normal equations (i.e., when
solving the overdetermined system in the least-squares sense). Provide the scaled partial pivots
and explain your steps.
(d) Compute the error according to (b) for the solution estimated in (c).

Problem 6: Integration. (5+6+4=15 points)


R1
The composite Simpson rule for computing a definite integral I = 0 f (x)dx is given by
 n n

2 2 −1
h X X
I = f (0) + f (1) + 4 f ((2i − 1)h) + 2 f (2ih)
3 i=1 i=1

1
using an even number of intervals n and step size h = n. Its asymptotic error is O(h4 ).
(a) Derive a recursive scheme from the composite Simpson rule assuming that n is a power of 2.
(b) Use Richardson extrapolation to the recursive scheme in (a) to derive an estimate for I with an
improved asymptotic error.
(c) Apply your estimate in (b) to compute I for f (x) = x2 using n = 4 intervals.

Problem 7: Taylor Series and Differential Equations. (3+4+8=15 points)


P∞ f (k) (x) k
(a) A Taylor series expansion developed at x is given by f (x + h) = k=0 k! h .
Derive the
truncated Taylor series expansion and the respective error term when truncating the Taylor series
for f (x) = log2 (2x) developed at x = 1 after the nth term.
(b) For which values of h does the Taylor series in (a) represent the function? Prove your claim.
(c) Consider an initial value problem given by the initial condition x(0) = 1 and the ordinary differential
equation x0 (t) = log2 (2x(t)). Derive the second-order Taylor series method for solving initial value
problems and apply one iteration step to the given problem for step size h = 0.1.

Notes:
• No books, notes, or calculators may be used.
• Duration: 115 minutes.
• Grades in percentages will be computed out of 90 points.

You might also like