AMSC 600 /CMSC 760 Term Project Information

The topics we could have talked about this semester under the title "advanced linear numerical analysis" are quite diverse, and we can only cover a few in class. For your project, I want you to "dig into" one topic in depth -- either something we have talked about or something we don't talk about that you find interesting.

Deadlines and points:

  • By October 15, you should send me e-mail with the title and a short description of your project. There is no late penalty for this, but since each person must have a unique topic, it is in your best interest to submit this as soon as you can. I will let you know whether your topic is ok or not.
  • By noon on November 6 you should submit (by email) a bibliography of 5 sources and a brief summary (approx. 50 words per source) of why each is relevant to your topic. This part of the project is worth 25 points, and a late penalty will apply.
  • The final submission is due at 12 noon, Friday December 14. It is worth 175 points, and a late penalty will apply.
  • The late penalty is 15% penalty for submissions up to 24 hours late, 30% penalty for submissions 24--48 hours late, etc.
  • What to submit for the project:

    Your project must have these components:

  • A written discussion, in the form of at most 40 slides, prepared as if you were giving a 40-50 minute talk on your project to students in this class. (I will stop reading at the 40th slide.) The first slide gives only the title and your name.
  • A list of open problems.
  • A list of references, including html links.
  • Your project might also include
  • A class presentation. Let me know by October 30 if you want to do this.
  • A Matlab implementation of your ideas.
  • A brief paper discussing any original contributions you have made, however large or small.
  • How to submit: Submit your project by e-mail. The time stamp on the e-mail will determine whether the project is on time or late.

  • The slides, open questions, paper (optional), and references should be in plain text, html, pdf, or ps format. Please format the slides "4-up" (4 slides per side of paper) if possible.
    Microsoft-formatted documents (Word, Excel, Powerpoint, etc.) will not be accepted; submit a pdf or ps file for these.
  • Matlab programs should be in plain text, stored in files that can actually be run by Matlab.
  • Data files should be in .mat format or plain text.
  • The entire set of files should be bundled into a single file (in tar, zip, or gzip format) and attached to an e-mail.
  • After you send the email with the attachment, please send me email without an attachment, just telling me to look for your other message. (Sometimes the department spam filter chokes on attachments.)
  • I'll acknowledge your submission by e-mail after I have successfully extracted the files.

    Some questions that will be asked while evaluating a project:

  • Overall:
  • Does the project demonstrate understanding of the material?
  • Are the ideas presented clearly?
  • (For an "A") Does the project show evidence of independent thought?
  • The slide presentation and (optional) class presentation:
  • Is it clear and correct? Was it spell-checked?
  • Would other students in the class be able to understand it?
  • Is jargon explained, if it was not discussed in class?
  • Are the slides easy to read? Do they contain an appropriate amount of material?
  • Is all notation defined clearly?
  • Is all type at least 28pt?
  • The open questions:
  • Are the questions reasonable and interesting?
  • Do the questions show evidence of independent thought?
  • The references:
  • Are the references complete and appropriate?
  • Are the html links included (if available)?
  • The Matlab code:
  • Is it well designed and well documented?
  • Is it easy for me to run and to understand what it does?
  • Are the experiments well designed?
  • The paper:
  • Is it correct?
  • Is it clear and concise?
  • Is the work of interest?
  • Warning: The only failing grades I have given on term projects have resulted from lateness or plagiarism. If you use someone's ideas, cite the source. If you use a direct quote, use quotation marks and cite the source.

    How to get started: Some sample topics are listed below, but I welcome proposals of other topics, especially if they have relevance to your research interests.

    The Survival Guide gives journals and other sources of information that might lead you to ideas. If you have a topic that you find interesting but don't know whether it is a good idea, let's talk. And if you need help finding a topic, I'll be glad to talk to you about your interests.

    Throughout the semester: Let's talk about your project if you have questions, or if you have trouble finding appropriate references.

    A note on software:
    In general your grade will be higher if you use any available Matlab software rather than writing your own, since this will give you more time to be original.

    Projects chosen by other students this semester: Your project must be different from all of these, so either pick a different topic or check with me to make sure that your ideas are sufficiently different from what these students chose.

  • New! Implicitly Restarted Arnoldi and relation to linear system solving
  • New! Local Ensemble Transform Kalman Filter (LETKF) for data assimilation
  • New! Uses of the Lanczos method in finding eigenvalues of Hermitian matrices arising in adiabatic quantum computing
  • New! Saddlepoint problems
  • New! SVD for salient motion detection
  • Sample Topics: Choose one of these or propose a different one.

    Saddlepoint problems
    If we minimize a quadratic function subject to linear equality constraints, we obtain a linear system with a matrix of the form H = [A, B; B^T, 0], where A is symmetric and positive definite. The same problem often arises in solving pdes. The matrix H is symmetric but indefinite (both positive and negative eigenvalues) and iterative methods tend to be slow. There are many open questions about computing preconditioners.

    Implicitly Restarted Arnoldi and relation to linear system solving
    The implicitly restarted Arnoldi method is the method of choice for finding a few eigenvalues of a large sparse matrix. It is based on restarting the Krylov sequence using a vector (or set of vectors) derived from the previous sequence. Can the information it builds upon be used to restart the Krylov sequence for linear system solving?

    Field of values and applications
    The field of values of a matrix is the set containing the values x^H A x for all vectors x of norm 1. The concept has surprisingly practical applications; a recent issue of SIAM News discussed how a computation of the field of values of a matrix can be used to save fuel in space flight. Open questions include efficient algorithms for the computation of the field of values and new uses for it.

    Multilevel aggregation for Markov chain problems
    Multilevel algorithms have been applied to Markov chains to find the stationary vector (i.e., the eigenvector corresponding to the eigenvalue 1), but the theory is much less developed than that for so-called multigrid algorithms for differential equations. Can progress be made in the theory or in improving the heuristic efficiency?

    Image processing using linearization of total variation
    Image deblurring has typically been done using linear techniques, but a minimization problem called total variation is quite successful. Open questions include whether its cost can be reduced to be competitive in time with the linear methods.

    Cholesky-infinity and alternatives
    In applications such as function minimization by Newton's method, we form a Cholesky factorization of a matrix that should be positive definite but sometimes fails to be. Chol-inf is one algorithm for "fixing up" the matrix as the factorization is computed. Alternatives are also available, for example, those collected in a paper by Fang and O'Leary. Open questions include why Chol-inf works well and which method is best.

    CUR factorizations and accuracy guarantees
    CUR factorizations approximate a matrix A using C, containing a subset of the columns of A, R, containing a subset of the rows of A, and U, determined to minimize ||A - CUR ||. When the rows and columns are chosen by random sampling, there are some impressive theoretical bounds on how good the approximation is guaranteed to be (almost all of the time). Unfortunately, the dimensions of A need to be in the billions before these bounds apply. Open questions include how to reliably apply such methods to matrices with dimensions in the thousands.