Numerical Methods II, G63.2020/G22.2421

Spring 2005


First Day of Class: Monday January 24, 2005.
Time: Monday 5:00-7:00 p.m.
Location: WWH Room 101

Instructor: Anna-Karin Tornberg
Office: WWH 1119
Phone: 212 998 3299
email: tornberg (in the domain) cims.nyu.edu

Office hours: For shorter questions, drop by at any time. For longer consultations, please make an appointment.

Required text: A first course in the Numerical Analysis of Differential Equations by Arieh Iserles.
(Cambridge University Press, ISBN: 0-512-55655-4)


******* Final exam: The final exam will be an oral exam, in my office. Please see me to sign up for a slot, no later than May 2. Exams will be held on Wed May 4 (those of you who need early grades must take it then), and on Mon May 9 and Tue May 10. List of different topics that has been covered in the class , *******

Lectures: There are 13 lectures; each Monday from January 24 to May 2, with the exception of February 21 (President's Day) and March 14 (spring recess).
See the Syllabus below for a description of what material will be covered.

Assignments: There will be 7-8 homework assignments during the semester, due approximately every other week. They will be posted further down on this page as they are assigned. They will contain both theoretical problems as well as computer assignments. Computer assignments may be completed using Matlab or any other programming language of your preference. For advice concerning computing, see Professor Overton's Numerical Methods I home page.

Please hand your completed homework to me personally or leave them under my office door. Homework is due at midnight on the given date.

Class mailing list: If there are any important announcements, I will - in addition to posting them on this web page - also send them by email. To join this list, please send me an email.
Syllabus: This course will cover fundamental methods that are essential for numerical solution of differential equations. It is intended for students familiar with ODE and PDE and interested in numerical computing.

We will start by discussing interpolation by polynomials, including Hermite interpolation and cubic splines. This will be followed by a brief discussion of numerical quadrature (Gaussian quadrature was covered during the fall semester in Numerical methods I). Next, we will discuss numerical methods for the solution of nonlinear equations, primarily Newton's method.

After this, we will study the numerical approximation of ordinary differential equations. This will include methods such as the Adams' methods, backward differentiation (BDF) schemes and Runge-Kutta methods. We will discuss the concepts of convergence and stability. The definition of A-stability will be essential in the study of stiff ODEs. Finally, we will also devote some attention to the subject of error control and adaptive step size. This material considering the numerical discretization of ODEs is discussed in Chapters 1-5 of Iserles' book.

We will now turn to the approximation of a two-point boundary problem with finite differences, and from there to the discretization of the Poisson's equation. We will discuss Dirichlet and Neumann boundary conditions, and their implementation. We will study the order of accuracy of different approximations.The discretization will lead to a system of linear equations, and we will discuss the structure and some properties of this system matrix (Chapter 7). We will give an introduction to the finite element method. The weak formulation for the Poisson's equation and a few different choices of basis functions will be discussed (Chapter 8).

At this point, we should be ready to turn our attention to time dependent partial differential equations - both parabolic and hyperbolic. We will study finite difference discretizations of different equations and discuss the concepts of consistency, stability and convergence, introducing the Lax equivalence theorem and the CFL condition. We will discuss techniques for stability analysis. (Chapters 13-14).

In Numerical Methods I, several techniques for solving large linear system of equations where discussed. The system matrices that arise in our discretizations of PDEs are sparse, and we will add some specific coments regarding their solution. We will specifically study fast algorithms for solution of the discretized Poisson problem (Chapter 12) and, time permitting, discuss the fundamental ideas of multigrid techniques (Chapter 11).

Assignments
Assignment 1 , assigned Jan 24, due Feb 7
Assignment 2 , assigned Jan 31, due Feb 14
Assignment 3 , assigned Feb 14, due Feb 28
Assignment 4 , assigned Feb 28, due Wed March 23
Assignment 5 , assigned Mar 21, due Apr 4
Assignment 6 , assigned Apr 4, due Apr 18
Assignment 7 , assigned Apr 11, due Apr 25
Assignment 8 , assigned Apr 18, due May 2 - Extended until May 4 for those who do not need early grades.
NOTE! There are some typos in the version of Assignment 8 handed out in class. At several places, the time level is indicated as n+1, although it should be n. The version here should be correct.

Handouts
Construction of cubic spline , handed out Jan 31.
Explicit Runge-Kutta methods , handed out Feb 22.
Orthogonal polynomials and Gaussian quadrature , handed out Mar 21.