Your Project
Overview: The goal is for you to
get
practical experience in numerical analysis of continuum problems by
working on a project of direct interest to you. As you are all
graduate students, the obvious advice is to pick a project in
consultation with your advisor; see the "subject" comment below.
You will submit a one page Proposal
on Friday 3/3; a short Version
1.0
on Friday 4/1; and a final and longer Version 2.0 on Wednesday 5/11;
see comments on length and format below.
Subject of Your Project:
You should pick a continuum
mathematical model like
- a
(single or small system of) partial differential equation(s), or
- a
(single or small system of) delay/stochastic differential equations, or
- a (function space) variational inequality.
You must identify which one of
these it is! Most ordinary differential equation problems
lead to numerical analysis rather unlike the content of the course, and
thus are not recommended.
You will do a numerical analysis of a particular problem for your
model. The problem could be a
- a boundary value problem, or
- an initial and boundary value problem, or it could be
- an inverse problem, or
- a stability analysis (of the model).
<>You must identify, early in your
project, which one of these it is!
>
Content of Your Project:
You will do a numerical analysis, with computed example, for this
model. In particular, three things are required:
- numerical analysis
- practical computation
- discussion of scientific/engineering meaning (including
conclusions about the role/meaning of your particular computation)
Note the first part: You must make a serious attempt to understand and
explain the efficiency, stability,
and accuracy (i.e. truncation or approximation error, and
possibly convergence) of your numerical method(s).
On the other hand, I do not expect you to derive the equations of the
model. That is the subject for other courses. Instead,
careful statement of the equations of the model, explanation of the
symbols, and careful references suffice. You should state the type of the
problem (i.e. "..an initial/boundary problem for a nonlinear evolution
equation of mixed parabolic and elliptic type..." or "... an inverse
problem for a linear, elliptic boundary value problem..." or "... a
nonlinear stationary variational inequality ...") as precisely as
possible.
I strongly recommend that you do the practical computation in MATLAB, and I do not expect you to
produce a production quality code. Rather, your goal should be a
functional and readable prototype. (Something you can build
upon!) Please use MATLAB's
built-in linear algebra, and don't write your own! You don't need
to use MATLAB tricks, but you may find that learning to vectorize and
avoid loops will help you understand.
You are encouraged to look for methods other than finite differences
for your practical computation. In particular, consider
- finite difference (and finite volume),
- finite element, and
- spectral
methods. These methods differ both in implementation and in their
analysis.
In what follows I state what you
will turn in at various stages, but some of the work is mine,
too. I will be giving you
feedback on each part. You can and should also come to me and
talk about your project at any stage.
Length and Format of Proposal (due
3/3/05): The proposal should be one page and should include at
least three references. (If you find it hard to pack it on to one
page, then think of this as practice for extracting money from funding
sources; their length requirements can be much more painful.)
Briefly state the continuum model, the scientific/engineering context,
the type of the problem (see above), and the numerical method you
propose to apply. Then state what forms of analysis of the method
you think are feasible; you need not include that analysis yet.
At least one of your references should address the numerical analysis
of your chosen (or closely related) model, problem, and method.
Length and Format of Project Version
1.0 (due 4/1/05): The point here is to get you
to do a little bit of everything that
will go into the final product. Version 1.0
should be a very good skeleton on which to build Version 2.0. It
needs to be readable but it need not be polished.
The length should be between 5 and 15 pages total.
Please include the following 7 section headings (meaning in
parentheses):
- Introduction (including
scientific/engineering context and sketch of the numerical analysis you
do, especially the high points)
- Continuum Model and Problem
(here should go the PDE, etc, and clear specification of the particular
problem including boundary conditions and parameters; include one or
two exact solutions if known)
- Numerical Method (state
the method, discuss the reasons for this choice, discuss
implementability and efficiency; excerpts of the code may go here)
- Analysis (truncation- or
approximation- error issues, stability, convergence)
- Results or Conclusion
- References
- Appendix/Appendices (put
the actual codes here, tangential analysis or computation, failed
attempts if significant)
The Results/Conclusion part should be rather short in version 1.0
because you are likely not able to draw definitive conclusions yet.
Length and Format of Project Version
2.0 (due 5/11/05): Now the point is to fill out
the skeleton, of course. You will undoubtedly need to alter your
earlier plans, so don't worry if version 2.0 has different
computational or analytical details than version 1.0.
Version 2.0 is the final version, and I encourage you to
make it a little bit clean-and-neat. Please use ordinary 8.5 by
11 paper and staple it in the upper left corner. Please do not
put it in a folder, etc. (I will have a stack of these things to
grade and I don't need yours to take up extra space.)
The length should be between 15 and 30 pages total and
should include exactly the same sections as above.
2/20/05 ELB