18.336 Numerical Methods for Partial
Differential Equations
Spring 2008
Course Instructor: Jean-Christophe Nave ( jcnave@mit.edu - http://web.mit.edu/jcnave/www/
)
Tuesday / Thursday 11:00 –
12:30 in 2-136
Office Hours:
Wednesday 2-3pm
Course
Homepage: http://web.mit.edu/jcnave/www/courses/18.336.htm
TA: David Shirokoff (Office: 2-342) [shirokof (at) mit
(dot) edu]
TA Office
Hours: Tuesday 5-7pm
Course
Description:
Advanced introduction to applications and theory of
numerical methods for solution of partial differential equations, especially of
physically-arising partial differential equations, with emphasis on the
fundamental ideas underlying various methods. We will Start from fundamental
solutions of PDE and move toward numerical methods including finite difference
methods, spectral methods. Additional topics arising from the discretization
and solution of specific equations will be covered, i.e. stability,
consistency, convergence, Preconditioned Conjugate Gradient (PCG), Fast Fourier
Transforms (FFT), C and Matlab programming.
Prerequisites:
Some notion of programming using Matlab, C/C++, or
Fortran. + courses in Calculus, Linear Algebra, and Differential Equations.
Grading: 4 or 5 Homework sets (50%) and final project
(Including a final presentation) (50%)
Final
Projects:
I will suggest many final projects’ topics. However, I
strongly encourage students to come up with their own topic. The project can be
related to their own research, but I will NOT allow recycling or repackaging of
previous projects or research. The project has to be original and non-trivial.
List of
useful references:
Spectral Methods in Fluid Dynamics, Canuto, et al.
Computational Techniques for Fluid Dynamics (I and
II), C.A.J. Fletcher
Partial Differential Equations,
Numerical Recipes
Final Project Guidelines:
The write up should have the format (and professional look) of a journal
article. I strongly suggest the format of the journal of computational
physics (you should download and use their latex template).
Limit yourself to 20 pages not including
bibliographical references, appendices, and code.
At least the following section should be included:
-1- Background on the problem chosen, relevance with
respect to previous work done, and aim of the present work
-2- Governing equations
-3- Discretization / Numerical approach
-4- Stability / accuracy …
-5- Results
-6- Convergence study (grid refinement / time step refinement
…)
-7- Conclusion + future work
GOOD LUCK!!!
Homework:
1-
Tentatively due Thursday 2/21/08 or Tuesday 2/26/08:
Preliminary project proposal – 1 Page Max.
2-
Due Thursday 2/28/08: Project proposal: clear description of the problem to be solved. This
should contain a brief motivation, equations, expected approach (method to be
used, programming language), and bibliographical references.
3- PSET 1 delivered on Tuesday 3/4/08 due Tuesday 3/18/08
4- Midterm project Report due Thursday 4/10/08
5- PSET 2 delivered on Thursday 4/10/08 due Thursday 4/24/08
Lecture 1 – 2/05/08 We settled some administrative issues, went over the syllabus and the survey. I started the formal introduction to PDE’s with some definitions. Hand out 1: Common PDEs
Lecture 2 – 2/07/08 We covered well posedness, and derived solutions for the linear transport equation.
Lecture 3 – 2/12/08 We covered fundamental solutions to the Laplace’s and Poisson’s equations and showed mean-value properties, maximum principle, uniqueness …
Lecture 4 – 2/14/08 We covered fundamental solutions to the heat equation.
Lecture 5 – 2/21/08 Final project discussion
Lecture 6 – 2/26/08 Common Finite Difference approximation for the 1st, 2nd, and higher order derivatives and their analysis (order, efficiency, compactness …)
Lecture 7 – 2/28/08 Simple Boundary Value Problem
Lecture 8 – 3/04/08 Continued on consistency, stability, convergence
Lecture 9 – 3/06/08 Continued on consistency, stability, convergence
Lecture 10 – 3/11/08 Fourier analysis of PDE’s and started diffusion equation
Lecture 11 – 3/13/08 Von Neumann Stability analysis, stiffness of diffusion equation…
Lecture 12 – 3/18/08
Multi-dimensional problem, LOD, ADI…
Lecture 13 – 3/20/08 Elliptic systems
discretization, basic solution of the linear system (to be expanded in Lecture
15)
Lecture 14 – 4/01/08 Advection equation, Leap Frog, Lax-Friedrichs…
Lecture 15 – 4/03/08 Guest lecture by Benjamin Seibold on Conjugate Gradient methods and multigrid.
Lecture 16 – 4/08/08 Lax-Wendroff, Beam-Warming, method of characteristics, CFL condition