COS 323 - COMPUTING FOR THE PHYSICAL AND SOCIAL SCIENCES

Instructor: Ken Steiglitz, Room 421 CS Building, email ken@cs.princeton.edu, phone 258-4629, office hours TBA

Graduate Teaching Assistant:

Michael Schneider schneidr@cs.princeton.edu,

Undergraduate Assistants:

TBA

Undergraduate Course Designers:

Mike Carreno, Niki Kittur, J. Sheehan Maduraperuma (consultant).

COS 323 Computing for the Physical and Social Sciences

Principles of scientific computation, driven by current applications in biology, physics, economics, engineering, etc. Topics include: simulation, integration of differential equations, iterative optimization algorithms, stability and accuracy issues. Students will pursue projects in a variety of fields, writing their own computer programs and also using higher-level tools such as Mathematica.

QR Fall

Two lectures, one class. Prerequisites: COS 126 and MAT 104. K. Steiglitz

Don't panic! There will be regular assignments from various parts of this compendium, but we won't be covering all or even a large part of it. We'll also fill in details when it's too terse. But it has lots of good descriptions of the most well known and useful algorithms, plus code. You probably won't want to sell it when you're done with the course.

By the way, the entire book is online. We aren't going to get a license to download the software. You're going to use it as a guide, but roll your own short programs, which is much more instructive and satisfying.

Additional reading on applications will be handed out.

You should hand in for each assignment:

• A short description of your code including what it does, a summary of the control flow, and a summary of the major data structures. Your description could be the comments of a well-documented program without the code itself. It should make it clear what your model was and how you implemented it, without going into the details of pointers and indices.
• The results of your computations. Use graphs, numerical tables and/or statistical summaries, whatever you feel best communicates the behavior you observed in your experimentation.
• A short analysis of your data. What conclusions can you draw from your observations?
  
  • In the case of the population genetics simulation (Assignment 1), do your observations suggest new hypotheses that could be tested using another simulation, another mathematical model or a laboratory or field experiment? This can include qualitative statements (e.g. ``The data suggests one-legged robots are unstable.'') or quantitative statements (e.g. ``The robot's speed decreases as 4.6 x^2, where x is the slope of the hill it is climbing.'')
    
  • In the case of simulations of physical systems, what evidence is there that your results reflect reality? Which of the algorithms you used are likely to most accurately reflect real behavior? How should your model be changed to better reflect reality, and can you anticipate how that would affect the algorithms and computational resources you use.
  • In the case of any numerical computation, be sure to compare the different algorithms you used on the basis of speed, storage, accuracy, and ease of programming. Can you suggest guidelines about which algorithm to use under which circumstances in the context you studied?
  • Did any of the algorithms you used break? (That is, give nonsensical answers or none at all.) If you can, explain why the algorithms broke and what might be done to make them more robust.