Scribd
Upload
76 views3 pages

Assignment01 PDF

This document is an exercise sheet for a numerical mathematics course. It contains 6 exercises on topics like matrix norms, the Gram-Schmidt algorithm, and the singular value decomposition. The exercises involve deriving formulas, proving properties, implementing algorithms in MATLAB, and analyzing computational complexity. Students are instructed to complete the exercises and submit their MATLAB code and results by a specified deadline.

Uploaded by

Hong Chul Nam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
76 views3 pages

Assignment01 PDF

This document is an exercise sheet for a numerical mathematics course. It contains 6 exercises on topics like matrix norms, the Gram-Schmidt algorithm, and the singular value decomposition. The exercises involve deriving formulas, proving properties, implementing algorithms in MATLAB, and analyzing computational complexity. Students are instructed to complete the exercises and submit their MATLAB code and results by a specified deadline.

Uploaded by

Hong Chul Nam
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

Prof. Ch.

Schwab Spring Semester 2019


F. Henrı́quez ETH Zürich
J. Opschoor Numerische Mathematik I D-MATH

Exercise Sheet 1

Exercise 1.1 Matrix Norms I


(1.1a) Let A ∈ Rn×n . Using the definition of the vector p-norm and the induced matrix p-norm,
derive the following explicit formulas for p ∈ {1, 2, ∞}:
n
X n
X p
kAk1 = max |aij |, kAk∞ = max |aij |, kAk2 = λmax (A> A),
j=1,...,n i=1,...,n
i=1 j=1

where λmax (A> A) denotes the maximum eigenvalue of A> A.


H INT: For kAk2 use the singular value decomposition of the matrix A.

(1.1b) Let the matrix A ∈ R2×2 be defined as


 
1 2
A= . (1.1.1)
−1 1/2

In view of the explicit formulas derived in subproblem (1.1a), compute the norm kAkp for p =
{1, 2, ∞}.

(1.1c) Write a M ATLAB function

function np = MatrixpNorm(A)

which takes as input a matrix A and returns a 3 × 1-vector np containing the p-norm of A for
p = 1, 2, ∞. Compare the numerical results with the ones obtained in subproblem (1.1b) for A
as in (1.1.4).
H INT: Use the M ATLAB built-in routines.

Exercise 1.2 Matrix Norms II


(1.2a) For A ∈ Rn×n let ρ(A) := max1≤i≤n |λi | be the spectral radius of A (i.e. the maximum
of the moduli of all eigenvalues of A). Is ρ a norm? Prove your claim.

(1.2b) Define submultiplicativity of a matrix norm k·kM . Moreover, define compatibility of a


matrix norm k·kM with vector norms k·kV , k·kW .
H INT: See [NMI, Sect. 0.7].

(1.2c) Show that the Frobenius norm k·kF on Cm×n is compatible with the vector norm k·k2 .

Exercise Sheet 1 Page 1 Numerische Mathematik I


m×n
(1.2d) Let A ∈ RP . First show that kQAkF = kAkF , if Q ∈ Rm×m is orthogonal. Then
2 p
show that kAkF = i=1 σi (A)2 , where p = min{m, n} and σi (A) is the ith singular value of A.

(1.2e) Consider the matrix  


2 3 −1
A := .
7 1 3
Compute kAkF and kAkp for p ∈ {1, 2, ∞}. What can you say about the convergence of the
series j∈N Aj ?
P

Exercise 1.3 Matrix Norm Equivalences


Prove the following norm equivalences:
p
(1.3a) For all A ∈ Cm×n , we have kAk2 6 kAkF 6 rank(A) · kAk2 .
√
(1.3b) For all A ∈ Cm×n , we have √1 · kAk∞ 6 kAk2 6 m · kAk∞ .
n
√
(1.3c) For all A ∈ Cm×n , we have √1 · kAk1 6 kAk2 6 n · kAk1 .
m

Exercise 1.4 Gram-Schmidt Algorithm


Consider the three vectors x1 , x2 , x3 ∈ R3 given by
     
1 3 0
x1 =  2  , x2 =  1 , x3 = 0.
 (1.4.1)
0 −1 1

(1.4a) Prove that x1 , x2 , x3 are linearly independent.

(1.4b) Use the Gram-Schmidt algorithm of the lecture notes to generate (by hand) an orthogonal
basis of R3 from the linearly independent vectors x1 , x2 , x3 given in (1.4.1).

Exercise 1.5 M ATLAB Implementation of the Gram-Schmidt Algorithm


(1.5a) Following Algorithm 0.55 of the lecture notes, implement a M ATLAB function

Q = gram schmidt algo(A),

to construct an orthonormal basis out of a set of k linearly independent vectors in Rd . Input is a


matrix A of size d × k whose columns correspond to k linearly independent vectors in Rd . Output
is a matrix Q, also of size d × k, whose columns correspond to k mutually orthogonal vector in
Rd of unitary length.

(1.5b) Test your implementation from subproblem (1.5a) using the vectors of Exercise 1.4.

(1.5c) Using the M ATLAB command rand, construct three vectors (i.e. k = 3 in Exercise
(1.5a)) with uniformly distributed entries in the interval [1, 10] of dimensions d = 102 , 103 , 104 , 105 ,
106 , 107 . Plot the total CPU time needed to construct the orthonormal basis using the M AT-
LAB function gram schmidt algo from subproblem (1.3a) (use the M ATLAB commands tic
toc to do so) in loglog scale. How does the CPU time behave with the dimension d of the vectors
taken into consideration?

Exercise Sheet 1 Page 2 Numerische Mathematik I


Exercise 1.6 Singular Value Decomposition and Matrix Norms
Let A be given by
 
5 3
A := .
0 −4

(1.6a) Find orthogonal matrices U and V in R2×2 and a diagonal Matrix Σ such that A =
UΣV> .

(1.6b) Compute the operator 2-norm, the Frobenius norm and the spectral radius of both A and
A−1 . For this, refer to [NMI, Prop 0.51] from the lecture notes.

Published on February 21, 2019.


To be submitted on March 5, 2019.
M ATLAB: Submit all file in the online system. Include the files that generate the
plots. Label all your plots. Include commands to run your functions. Comment
on your results.

Literatur
[NMI] Lecture Notes for the course “Numerische Mathematik I”.

Last modified on 20. Februar 2019

Exercise Sheet 1 Page 3 Numerische Mathematik I

You might also like