Math 221

Numerical Solution of

Partial Differential Equations

Part I

Class meets Tuesdays, Thursdays, 5:00 PM to 6:15 PM in Phillips 301

Office hours: M-Th, 4:00 PM to 5:00 PM, (MW in Phillips 307, TTh in Phillips 324 Computer Lab)

[Motivation] [Objectives] [Syllabus] [Grading] [Texts] [Assignments] [Computing] [Projects] [Final Examination]

Motivation

The physical world is described by a small set of laws. One of the most useful mathematical statements of these physical laws is in the form of partial differential equations (PDE's) describing local changes in physical parameters. Though it is usually straightforward to write down the particular PDE's that correspond to a given problem, finding a solution is much more difficult. For the vast majority of real-world problems approximate techniques must be used. One of the most productive approaches is to use a numerical approximation of the PDE's of interest. These techniques are used in studying chemical reactions, ecological models, astrophysical phenomena, aircraft design, financial models and in many other application domains.

Course Objectives

The development of numerical methods for PDE's and the ensuing theoretical development is within the province of applied mathematics. This course introduces the basic aspects of the discipline at the graduate level. At the end of the course the student should be able to understand the basic theory associated with solving each type of PDE encountered in practice, be adept at choosing methods appropriate for a specific application and be proficient in the computer solution of PDE's.

Syllabus

  • Basic equations of mathematical physics, other application domains

  • Minimal residual formulation

  • Numerical methods for ODE's

  • Linear advection diffusion equations

  • Parabolic equations

  • Hyperbolic equations

  • Equations of mixed type

  • Coupled PDE-ODE systems

Grading Policy

Grading shall be determined based on homework (HW, 35 points), supplementary reading (SR, 20 points), final project (FP, 15 points) and final examination (FE, 30 points). The number of points accumulated during classwork is mapped to a graduate grade as shown in the following table.

Clear Excellence

H

87-100 points

Entirely Satisfactory

P

70-87 points

Low Passing

L

50-70 points

Failed

F

0-49 points

During the course, 7 homework assignments shall be given. Each homework shall contain 5 required topics and a bonus topic, each worth 1 point for a maximum of 6 points. The bonus topics are meant to allow for makeups of previous deductions. Homework assignments are given every second Tuesday and due two weeks thereafter. The purpose of the homework assignments is to achieve familiarity with the course material.

Four reading assignments shall be given, once every 3 weeks in a staggered order among the students. Assignments are given to a small group of students attending the course, typically in pairs. Each assignment is graded with 5 points. The purpose of the reading assignments is to introduce students to the tasks involved in research work. Points are awarded to this end as follows:

  • 1 point for reading the assigned text;

  • 1 point for background research, i.e. placing the assigned text in the context of previous work;

  • 1 point for forward research, i.e. investigating the ramifications of the assigned text;

  • 1 point for working through some of the results from the assigned text;

  • 1 point for quality of presentation.

Reading assignments are presented to the class in the last 8 minutes of Thursday's class, starting on September 5. The purpose of the presentations is to instill the discipline of concise, informative talking in front of groups of colleagues. Of the 8 minutes, 5 minutes are allocated to presentation and 3 to questions and answers.

Revised grading policy

Reading assignments were an interesting experiment. However to do them right, much more time should be allocated and this would have been detrimental to coverage of a wide range of topics. The students having already presented reading assignments will receive the point credits as bonus points (2 points).

Also, homework assignments proved to be more time-consuming than first thought so only 5 assignments were completed.

The revised grading policy is:

(1) 50 points from homework (each homework is worth 10 points instead of the 5 originally intended)

(2) 30 points from final examination (final exam will have bonus questions for an extra 30 points)

(3) 20 points from project work

Points are translated to grades following the table above.

Course Texts

One of the tenets of graduate study is the ability to critically examine multiple sources in order to verify theories and to enhance understanding. Lecture notes for the course shall be provided online in Postscript and PDF formats. These are meant to document the progress of material presented in class. It is expected and required that the student supplement the lecture notes with reading from other sources.

Lecture notes

Lecture
Date

Topic

Notes

Lecture
Date

Topic

Notes

1, 8/20

Overview of PDE's

 .ps .pdf

  10/10

Lab: advection

Lab3

2, 8/22

ODE/PDE problems

 .ps .pdf

15, 10/15

Modified equations

 .ps .pdf

3, 8/27

Weighted residual

 .ps .pdf

  10/17

(Fall break - no class)

4, 8/29

ODE IVP

 .ps .pdf

16, 10/22

Non-linear hyperbolic equations

 .ps .pdf

5, 9/3

ODE num. methods

 .ps .pdf

17, 10/24

Hyperbolic systems

 .ps .pdf

6, 9/5

Euler method analysis

 .ps .pdf

18, 10/29

Finite volume

 .ps .pdf

7, 9/10

Stability, consistency

 .ps .pdf

19, 10/31

Splitting methods

 .ps .pdf

8, 9/12

Bnd. Loc., convergence

 .ps .pdf

  11/5

Lab: Spectral differentiation

 

9, 9/17

Fourier analysis

 .ps .pdf

20, 11/7

Spectral methods

 .ps .pdf

10, 9/19

1D Heat eq. FDM's

 .ps .pdf

21, 11/12

Compact finite differences

 .ps .pdf

11, 9/24

2D Heat eq.

 .ps .pdf

22, 11/14

Finite element methods

 .ps .pdf

  9/26

Lab: Fourier space

 Lab2

 11/19

Lab: 2D Navier-Stokes equations

 

12, 10/1

Linear advection: characteristics

 .ps .pdf

23, 11/21

Finite element formulations

 .ps .pdf

13, 10/3

Basic advection methods

 .ps .pdf

24, 11/26

Ritz formulation example

 .ps .pdf

14, 10/8

Advection schemes: stability

 .ps .pdf

35, 12/3

Project presentations

 

General bibliography

The following texts serve as general background material. Students are encouraged to actively read from these texts material related to the current lecture. Homework and final examination questions from this material should be expected.

Finite Difference Schemes and Partial Differential Equations, John Strikwerda

Numerical Methods for Conservation Laws, Randall LeVeque

Finite Difference Schemes for Computational Fluid Dynamics, Phillip Colella & Gerry Puckett

Suggested reading for specific topics

Material specific to a certain topic shall be listed here as the course progresses. Generally the books or articles will be placed on reserve at the Mathematics Library in Phillips for the time period indicated

P.G. Garabedian, Partial Differential Equations, Chapter 1, The Method of Power Series. (17 pp.). On reserve: Aug. 22 - Aug. 29. Questions to ponder: (1) Consider the relevance and applicability of the methods used in the Cauchy-Kowalewski theorem to practical computations; (2) Try to find a biography of Sophia Kowalewski (also transcribed as Sonja Kovalevski) and place the Cauchy-Kowalewski theorem in historical context.

G. Dahlquist, Numerical Methods, Prentice-Hall, 1974, Sections 7.1 (Difference operators), 7.2 (Richardson extrapolation), Chapter 8 (Differential equations)

Assignments

Homework assignments

Date issued

Date due

Topic

Homework

Solution

8/20

8/29

Finite differences

.ps .pdf

(too easy!)

9/3

9/17

ODE's

.ps .pdf

.html .nb

9/17

10/1

Heat equation

.ps .pdf

.ps .pdf
Makefile
dif1D.f90
dif1Dg.f90
dif1D.m
dif1Dcn.m

10/1

10/15

Hyperbolic equations

.ps .pdf

.ps .pdf

10/15

10/29

Non-linear hyperbolic eqs.

.ps .pdf

.ps .pdf

10/29

11/12

.ps .pdf

11/12

11/26

.ps .pdf

.ps = Postscript File, .pdf = Portable Document File, .html = Web page, .nb = Mathematica notebook .m = Matlab m file .f90=Fortran 90 source file

Supplementary reading assignments

  1. R. Courant, K. O. Friedrichs and H. Lewy, "Ueber die partiellen Differenzengleichungen der mathematischen Physik," Mathematische Annalen 100 (1928), 32-74. Translated as: "On the partial difference equations of mathematical physics," IBM Journal of Resarch and Development 11 (1967), 215-234. Students: Sarah Batey, Joohee Lee, Xiaoyu Zheng. To be presented to class on September 5.

  2. Dahlquist, G. "A Special Stability Problem for Linear Multistep Methods". BIT, 3:27 (1963). I've checked extensively, but it seems that this classic is not available in the UNC libraries or at Duke. The basic material from the article is well covered in the book: Numerical Methods, Dahlquist, G. and Björck, Äke, sections 8.5.3, 8.5.4, Prentice-Hall, 1974. Students: Alison Hall, Joohee Li, Omar Tolbert. To be presented to class on September 12. Note: checking that the original article or some reprint is not available took some time, so the assignment was posted later than I would have liked. This is compensated by the fact that the material is closely related to the course work and is quite concise.

 

Computational work

It is essential that students acquire a basic familiarity with the process of developing and using scientific software. The homework assignments shall contain gradually more difficult problems that are intended to be solved using programs written by the students. The problems in the first homework assignments are small enough that they can solved using Matlab. Later problems are more computationally intensive and require writing programs in a compiler language such as C or Fortran. Students are encouraged to observe advanced techniques by attending the Scientific Computing Club (SCC) that shall meet during the Fall 2002 semester.

Fourier solution of common PDE's (09/26/02)

Here is a Matlab m-file script that computes solutions to an advection, heat, dispersive equation using Fourier analysis. The procedures closely mirror the analytical framework presented in the lecture notes.

Lab2

Basic advection equation schemes (10/10/02)

Here is a Matlab m-file script that applies the upwind, Lax-Friedrichs and Lax-Wendroff schemes to the constant velocity advection equation.

Lab3

Spectral differencing (11/05/02)

A small Matlab m-file showing how to compute derivatives spectrally.

Lab4

2D Navier-Stokes equations (11/14/02)

A Matlab m-file to solve the 2D Navier-Stokes equations in the vorticity - stream function formulation.

Lab5

 

Final project topics

Individual final project topics shall be listed here on November 5. Students must submit their choice of topic by November 12. Projects are due and shall be presented to class on December 3.

Final examination preparation

Questions similar to those students can expect on the final examination. .ps .pdf

Some solutions: .ps .pdf

Some more solutions: .ps .pdf

Final examination questions and solutions: .ps .pdf