Operating System Design (CS321)
My final project was a user level threads package. Here is a link
to my paper and the software I wrote.
Introduction to Numerical Methods (CS450)
My project was a liquid Argon simulation implemented pretty much straight out of
Frenkel and Smit.
This was the first of my classes taught by
Prof. Heath.
NA Seminar (CS491/591)
This is a regular seminar that the NA group has every semester, it is instructed
by
Prof. Heath.
The purpose of this seminar is to get us used to presenting in front of a group, as
well as presenting our research in a relatively informal setting.
Parallel Numerical Algorithms (CS454)
This class was also taught by
Prof. Heath.
The final project was a modification of the liquid argon
simulation, but to run in parallel. I implemented two methods,
first being a simple modification of the serial code to be
a redundant data method, the second is a more complex spatial
decomposition method. A link to the resulting writeup is below
Advanced Finite Elements (TAM574/CSE517)
For this class the project for everyone was to implement a 1d spacetime discontinuous
Galerkin FEM for linear elastodynamics (seems complicated!). This is where I discovered
both the difficulties of differential forms and discontinuous Galerkin methods. Here
is the resulting paper
Spectral Element Methods/High order FEM (CS598LO)
The instructor for this class was
Luke Olson.
For this project I investigated the Poisson Boltzmann problem and various difficulties
associated with it. I had a very hard time dealing with the discontinuous Galerkin formulation
used here. This was extraordinarily difficult for me from a software perspective, and also
very frustrating, I'm sure you'll get a flavor of this from reading the paper
Iterative and Multigrid Methods (CS550)
Once again
Luke Olson
was the instructor. For this project I considered geometric multigrid for adaptively refined meshes
(AMR).
In particular I looked at the multilevel adaptive technique (MLAT). This was the first MG algorithm
used for AMR meshes. It certainly has weaknesses. However, the largest obstacle
was implementing the data structures associated with AMR. Even the limited implementation done for
this project proved to be difficult (though I only had 2 weeks).
Last updated May 14, 2007