Course Overview for Stat/EE 520

Instructor

Don Percival, Applied Physics Laboratory (APL) and Department of Statistics

phone number: 543-1368 (office) or 543-1300 (APL operator)
e-mail address: dbp@apl.washington.edu
office hours: by appointment at APL (Henderson Hall).

Textbook

Spectral Analysis for Physical Applications: Multitaper and Conventional Univariate Techniques, by D.B. Percival and A.T. Walden, Cambridge University Press, 1993 (this should be on reserve in the Mathematics Research Library, 3rd floor of the Padelford Maze). Since one of the authors of this book is closely related to the instructor of this class, the instructor proposes to deal with charges of `conflict of interest' by blowing profits from the sale of this book (approximately $2.50 per book before taxes) on pastries, cakes and cookies (yum!) to be served during the presentation of the class projects.

Game Plan (Grades and All That)

Your course grade will be based on three parts: homework (35%), an in-class exam (30%) and a term project (35%). A set of homework problems will be due the first day that the class meets during weeks 2 to 9 of the quarter (i.e., Monday, unless this is a UW holiday, in which case the problems are due on Wednesday). Late homework will not be accepted unless you have made prior arrangements with me because of special circumstances. I will attempt to grade and return the problems the class period after I receive them so that you get quick feedback. Please feel free to contact me during the quarter if you are concerned about how you are doing. (The following should not need to be said, but, alas, recent experience forces me to say it. When I return your homework to you, I will supply you with a solution sheet. I have done this for a number of years, which means that it is possible to track down solutions to certain problems from a student who has previously taken the course. Don't do this!!! As far as learning the material is concerned, there is no value in merely copying a solution - and little in `peeking' at a solution - other than avoiding a poor grade on an assignment. Making use of past solutions sheets in any manner whatsoever is unacceptable: doing so makes a mockery of graduate school and is unfair to those graduate students (fortunately, the vast majority!) who approach homework as an opportunity for learning.)

The only exam for the class will be during the class period for Friday, March 7th. The exam will be closed book. No electronic devices of any nature are allowed, but you can bring in a single standard piece of paper (8.5 by 11 inches) with as much material as you can cram on both sides. I will give you more details about the exam toward the middle of the quarter.

For the term project, I want you to do either a critical review of the literature or - if you have a time series you wish to analyze - an analysis of a time series using the techniques discussed in this class. For the literature review, you should select a special topic related to spectral analysis, read the relevant papers in the literature and then critique them (please check with me about what you plan to do for the term project before you start it so that I can advise you whether or not it is appropriate). In either case, I want you to give an 8-10 minute presentation to the class about your project on Thursday, March 20th, from 8:30AM to 10:20AM (this is the time slot set aside for the final examination, but we'll be using it for the class presentations instead). In addition, during the final week of instruction (March 10th to 14th), I will arrange a 10 to 20 minute meeting with each of you so that we can talk about how your project is progressing. This will give me a chance to give you some comments about your project, which you should use in preparing your talk and a short written version of your project (approximately 5 pages in length, but no more than 10 pages total [PLEASE!!!]). These reports are due no later than 3PM on March 21st (Friday) and can be handed to the receptionist at APL (Henderson Hall).

Course Outline

We will be covering the following material:

introduction to spectral analysis (Chapter 1 of textbook; one lecture)
introduction to the theory of stationary random processes (Chapter 2; two lectures)
introduction to the Fourier analysis of deterministic sequences and functions (Chapter 3; four lectures)
theoretical foundations of spectral analysis, including the spectral representation theorem (Chapter 4; two lectures)
theory of linear filters (Chapter 5; one or two lectures)
nonparametric spectral analysis in theory and in practice (Chapter 6; about ten lectures)
multitaper spectral analysis (Chapter 7; four lectures)
parametric spectral analysis, including the `maximum entropy' method (Chapter 9; five or six lectures)
harmonic analysis (Chapter 10; three or four lectures)
if time permits, spectral analysis of multiple time series, including cross-spectral density and coherency (handouts; up to two lectures)

There is a home page on the Web for this class at http://www.stat.washington.edu/courses/stat520/winter08/ (but, if you looking at this on the Web, you have probably already figured that out!). Except for figures from the textbook, this Web page contains copies of all the overheads used during class lectures in three formats (LaTeX, PostScript and PDF). If you find it helpful, you can make copies of them and bring them to class. The Web site will also contain a list of assigned homework problems (ideas for other Web site items would be appreciated!).

Return to home page for Stat/EE 520.