Qualifying Examination Winter 2017

Similar documents
Preliminary Examination, Numerical Analysis, August 2016

(a) If A is a 3 by 4 matrix, what does this tell us about its nullspace? Solution: dim N(A) 1, since rank(a) 3. Ax =

Linear Algebra, part 2 Eigenvalues, eigenvectors and least squares solutions

Math 273, Final Exam Solutions

Second Order Elliptic PDE

Chapter 2 Formulas and Definitions:

Solutions Chapter 9. u. (c) u(t) = 1 e t + c 2 e 3 t! c 1 e t 3c 2 e 3 t. (v) (a) u(t) = c 1 e t cos 3t + c 2 e t sin 3t. (b) du

Finite difference method for elliptic problems: I

6.207/14.15: Networks Lectures 4, 5 & 6: Linear Dynamics, Markov Chains, Centralities

1 Inner Product and Orthogonality

Algebra I Fall 2007

Review problems for MA 54, Fall 2004.

Lecture 6 Positive Definite Matrices

1 Assignment 1: Nonlinear dynamics (due September

Maths for Signals and Systems Linear Algebra in Engineering

MATH 452. SAMPLE 3 SOLUTIONS May 3, (10 pts) Let f(x + iy) = u(x, y) + iv(x, y) be an analytic function. Show that u(x, y) is harmonic.

Lecture 15 Perron-Frobenius Theory

PH.D. PRELIMINARY EXAMINATION MATHEMATICS

1 T 1 = where 1 is the all-ones vector. For the upper bound, let v 1 be the eigenvector corresponding. u:(u,v) E v 1(u)

1 Adjacency matrix and eigenvalues

Analysis Comprehensive Exam, January 2011 Instructions: Do as many problems as you can. You should attempt to answer completely some questions in both

Mathematics Qualifying Exam Study Material

Q1 Q2 Q3 Q4 Tot Letr Xtra

Lecture 9: SVD, Low Rank Approximation

NATIONAL BOARD FOR HIGHER MATHEMATICS. M. A. and M.Sc. Scholarship Test. September 25, Time Allowed: 150 Minutes Maximum Marks: 30

Problem Set 1. Homeworks will graded based on content and clarity. Please show your work clearly for full credit.

Section 5.4 (Systems of Linear Differential Equation); 9.5 Eigenvalues and Eigenvectors, cont d

MIT Final Exam Solutions, Spring 2017

In particular, if A is a square matrix and λ is one of its eigenvalues, then we can find a non-zero column vector X with

Problem 1A. Find the volume of the solid given by x 2 + z 2 1, y 2 + z 2 1. (Hint: 1. Solution: The volume is 1. Problem 2A.

Discrete Math Class 19 ( )

PH.D. PRELIMINARY EXAMINATION MATHEMATICS

NOTES ON LINEAR ODES

Math 21b Final Exam Thursday, May 15, 2003 Solutions

November 18, 2013 ANALYTIC FUNCTIONAL CALCULUS

Qualifying Examination

Week Quadratic forms. Principal axes theorem. Text reference: this material corresponds to parts of sections 5.5, 8.2,

MATH 205C: STATIONARY PHASE LEMMA

Spectral Theorem for Self-adjoint Linear Operators

Lecture 8: Linear Algebra Background

The First Fundamental Form

Math 310 Introduction to Ordinary Differential Equations Final Examination August 9, Instructor: John Stockie

Final Exam May 4, 2016

Lecture 18. Ramanujan Graphs continued

Math 102, Winter Final Exam Review. Chapter 1. Matrices and Gaussian Elimination

you expect to encounter difficulties when trying to solve A x = b? 4. A composite quadrature rule has error associated with it in the following form

The University of Texas at Austin Department of Electrical and Computer Engineering. EE381V: Large Scale Learning Spring 2013.

Georgia Tech PHYS 6124 Mathematical Methods of Physics I

MATH 2250 Final Exam Solutions

Name: INSERT YOUR NAME HERE. Due to dropbox by 6pm PDT, Wednesday, December 14, 2011

Spectra of Adjacency and Laplacian Matrices

Extreme Values and Positive/ Negative Definite Matrix Conditions

Math 308 Practice Final Exam Page and vector y =

Fall 2017 Qualifier Exam: OPTIMIZATION. September 18, 2017

Math Final Exam.

Homework 1 Elena Davidson (B) (C) (D) (E) (F) (G) (H) (I)

Complex Dynamic Systems: Qualitative vs Quantitative analysis

Lecture 5: Random Walks and Markov Chain

Understand the existence and uniqueness theorems and what they tell you about solutions to initial value problems.

Applied/Numerical Analysis Qualifying Exam

Lecture: Local Spectral Methods (1 of 4)

Final exam (practice 1) UCLA: Math 32B, Spring 2018

Linear algebra for computational statistics

5 Linear Algebra and Inverse Problem

MTH 5102 Linear Algebra Practice Final Exam April 26, 2016

OR MSc Maths Revision Course

Today: eigenvalue sensitivity, eigenvalue algorithms Reminder: midterm starts today

Math 273 (51) - Final

Lecture 13 Spectral Graph Algorithms

ODE, part 2. Dynamical systems, differential equations

Lecture 7: Positive Semidefinite Matrices

NUMERICAL SOLUTION OF GENERAL SINGULAR PERTURBATION BOUNDARY VALUE PROBLEMS BASED ON ADAPTIVE CUBIC SPLINE

MATH 1553 SAMPLE FINAL EXAM, SPRING 2018

Chapter 18. Remarks on partial differential equations

LINEAR ALGEBRA QUESTION BANK

McGill University Department of Mathematics and Statistics. Ph.D. preliminary examination, PART A. PURE AND APPLIED MATHEMATICS Paper BETA

Linear Algebra Review

CSCE 750 Final Exam Answer Key Wednesday December 7, 2005

SPRING 2006 PRELIMINARY EXAMINATION SOLUTIONS

Linear algebra and applications to graphs Part 1

Final Exam Practice 3, May 8, 2018 Math 21b, Spring Name:

( sin(t), cos(t), 1) dt =

MA22S3 Summary Sheet: Ordinary Differential Equations

Multivariable Calculus

University of Colorado Denver Department of Mathematical and Statistical Sciences Applied Linear Algebra Ph.D. Preliminary Exam January 22, 2016

Machine Learning for Data Science (CS4786) Lecture 11

Assignment #10: Diagonalization of Symmetric Matrices, Quadratic Forms, Optimization, Singular Value Decomposition. Name:

Total 170. Name. Final Examination M340L-CS

A converse to the Ambrosetti-Prodi theorem

Problem Set 9 Due: In class Tuesday, Nov. 27 Late papers will be accepted until 12:00 on Thursday (at the beginning of class).

Physical Metaphors for Graphs

Rambo s Math GRE Practice Test. Charles Rambo

5 Quiver Representations

QUALIFYING EXAMINATION Harvard University Department of Mathematics Tuesday September 21, 2004 (Day 1)

EE363 homework 7 solutions

0.1 Naive formulation of PageRank

Preliminary/Qualifying Exam in Numerical Analysis (Math 502a) Spring 2012

B 1 = {B(x, r) x = (x 1, x 2 ) H, 0 < r < x 2 }. (a) Show that B = B 1 B 2 is a basis for a topology on X.

On the concentration of eigenvalues of random symmetric matrices

LECTURE NOTES ELEMENTARY NUMERICAL METHODS. Eusebius Doedel

Transcription:

Qualifying Examination Winter 2017 Examination Committee: Anne Greenbaum, Hong Qian, Eric Shea-Brown Day 1, Tuesday, December 12, 9:30-12:30, LEW 208 You have three hours to complete this exam. Work all problems. Start each problem on a new page. You must show all steps and prove all claims, unless you rely on a standard theorem or result, in which case you should refer to it. You may use a computer to check your work, but all derivations must be your own. You are not allowed to access the internet or any application file existing on your computer. Any application file you use must be created from scratch. 1. An undirected graph consists of N nodes and a set of edges that connect given nodes. The adjacency matrix for the graph, A, has entries A ij = 1 if there is an edge between nodes i and j and A ij = 0 otherwise. Note that A ij = A ji, because we say if there is an edge between nodes i and j, then there is also an edge between nodes j and i. We do not allow self-connections, so that diagonal entries A ii are always 0. The degree of a node is the number of edges that connect to it. A d-regular random graph is a graph of which every node has degree d, for some integer d N, but the edges are otherwise in random positions. For the adjacency matrix A of a d-regular, undirected random graph with N nodes, answer the following questions. (a) Can A ever have negative eigenvalues? If yes, give an example, if not, explain why not. (b) Can A ever have imaginary eigenvalues? If yes, give an example, if not, explain why not. (c) Identify a vector v and value λ that is an eigenvector-eigenvalue pair for the adjacency matrix of every d-regular undirected random graph with N nodes? (d) Give a lower bound on the eigenvalues of the adjacency matrix for any d-regular undirected random graph. 2. Consider difference equations of the form u n+2 + a 1 u n+1 + a 0 u n = kbf(u n+1 ), for the initial value problem u (t) = f(u(t)), where k = t n+1 t n is the timestep. 1

(a) Determine the coefficients a 0, a 1, and b that give the highest order local truncation error and say what that order is. (b) Is the resulting method convergent? Explain why or why not. (c) Determine which, if any, of the points 1, ±i, ± 1 i lie in the region of absolute 2 stability. 3. Let f(z) = z + 1 2 and let γ be the unit circle oriented counterclockwise. (a) Show that f is not holomorphic on any domain that contains γ. (b) Find a function g that is holomorphic on some domain containing γ and such that g(z) = f(z) on γ. [Hint: Write f(z) = (z + 1)( z + 1).] (c) Use g to evaluate γ z + 1 2 dz. 4. Find the appropriate boundary value y(0, ɛ) such that the two-point problem { ɛy (x) + y (x) + y(x) = 0 y(0, ɛ) as determined, y(1) = 1, has a smooth solution with an asymptotic power series expansion y(x, ɛ) a j (x)ɛ j, j=0 that is uniformly valid for x [0, 1]. 2

2017 Winter Qual Day 2 Wednesday, December 13, 9:30-12:30, LEW 208 You have three hours to complete this exam. Work all problems. Start each problem on a new page. You must show all steps and prove all claims, unless you rely on a standard theorem or result, in which case you should refer to it. You may use a computer to check your work, but all derivations must be your own. You are not allowed to access the internet or any application file existing on your computer. Any application file you use must be created from scratch. 1. Let A be an n by n real symmetric matrix with eigenvalues λ 1... λ n. Let x be any real n-vector with x 2 = 1. Show that the Rayleigh quotient r(x) = x T Ax satisfies λ 1 r(x) λ n. Show that, by varying x while keeping x 2 = 1, the Rayleigh quotient can take on every value in the interval [λ 1, λ n ]. 2. Consider a general description of a system of N nonlinearly coupled units, as given by du i N dt = u i + w ij g(u j ), 1 i N. (1) j=1 Here, u i is the activity of the i th unit, and the matrix W gives the connection weights among these units; in particular, its element w ij is the connection weight between unit j and unit i. Finally, g( ) is a smooth monotonically increasing function that describes how the strength of interaction between units depends on their activities. A system is said to be bi-directional and full coupling if w ij = w ji, and w ij 0 for i j. For such a system, answer the following questions. (a) Is there a system of this form such with an equilibrium that is asymptotically stable: that is, any initial condition that begins near the equilibrium will tend back to the equilibrium as time tends to? If yes, give an example, numerically or analytically or both; if no, explain why not. (b) Consider the energy function H(u) = 1 w ij V i (u)v j (u) + 2 ij i Vi 0 g 1 (V )dv where u = (u 1,..., u N ) T, V i = g(u i ). Compute the time evolution of the energy, d dt H(u(t)), along solutions u(t) to the ODE above. Show that d dt H(u(t)) < 0 in the case where you have bi-directional and full coupling. In this case, is there a system of this form that has a nonconstant periodic solution? If so, give an example; if not, state why not. 3

3. Consider linear partial differential equation with boundary values: ( ) u(x, t) 2 u u(x, t) = ɛ + b(x) + c(x)u(x, t); c(x) < 0. (2a) t x 2 x x [ 1, 1], u( 1, t) = u 1, u(1, t) = u 1, (2b) in which b(x), c(x) are smooth functions of x [ 1, 1]. (a) Show that if twice differentiable function u(x, 0) > 0, then u(x, t) 0 for t 0. Similarly, if u(x, 0) < 0, then u(x, t) < 0 for t 0. (b) Now consider the stationary solution to (2a): ( ) 2 u ɛ + b(x) u(x) + c(x)u(x) = 0 (3) x 2 x with the boundary conditions in (2b). Because of the ɛ, in addition to the posibility of boundary layers at x = ±1, there could also be internal layers. Where should an internal layer be located if it has one? Why? 4. Consider the linear ODE system u (t) = Au(t), with an arbitrary initial vector u(0). (a) Show that for a general real, constant matrix A, the necessary and sufficient condition for u(t) 2 to decrease monotonically (for initial vector u(0)) is that the eigenvalues of A + A T be negative. (b) Give an example to show that in general it is not sufficient to have the eigenvalues of A in the left half-plane; that is, write down a matrix A whose eigenvalues are all in the left half-plane but for which the 2-norm of the solution to u = Au with some initial vector u(0) does not decay monotonically. 4

2017 Winter Qual Day 3 Thursday, December 14, LEW 208 You have 2 hours to complete this exam. DO 1 OF THE FOLLOWING 2 PROB- LEMS. You must show all steps and prove all claims, unless you rely on a standard theorem or result, in which case you should refer to it. You are allowed to use a computer for any part of any problem, but you are not allowed to access the internet or any application file existing on your computer. Any application file you use must be created from scratch. We recommend you hand in all your work, even if it did not produce any results. 1. Let A = {a ij } be a real n by n matrix. (a) Suppose A F A + ti F t R, ( n 1/2. where F denotes the Frobenius norm: A F = i,j=1 ij) a2 Show that the trace of A is 0. If you don t see how to do this, write a code to test the result numerically. Perhaps you can get some insight from your numerical experiments. (b) Suppose A 2 A + ti 2 t R, where 2 is the spectral norm; i.e., the largest singular value. Show that the left and right singular vectors of A corresponding to the largest singular value are orthogonal to each other. If you don t see how to do this, write a code to test the result numerically. Perhaps you can get some insight from your numerical experiments. 2. Consider singularly perturbed ordinary differential equation ( ) 2 u ɛ + bx u(x) + cu(x) = 0, x 2 x where b, c are two constants, with two-point boundary conditions u(0) = u 0, u(1) = u 1. Please describe the asymptotic behavior of the solution u(x, ɛ) as ɛ 0. 5