General announcement

Scope: Convex optimization problems arise in communication, system theory, VLSI, CAD, finance, inventory, network optimization, learning, computer vision, statistics, etc. Thanks to the recent advances in interior-point methods, various new solvers are now available and have made solving convex problems ever easier. Nevertheless, problems are often unrecognized as convex and, therefore, remain unsolved. This course is designed to be an exposure to convex optimization problems, their solution techniques and applications.

Topics covered: convex sets and functions, abstract convex optimization problems (QP, SOCP, SDP, etc.), duality, interior-point algorithms, and applications from

· CS, EE, and BioEng: statistical learning, image processing, computer vision, etc.
· Business: robust portfolio selection, etc.
· Statistics: robust regression, data fitting, etc.

Prerequisites: Linear algebra. Some knowledge in analysis, probability, and Matlab programming helps. No previous background in linear or nonlinear optimization is required.

Textbook: “Convex Optimization” by S.Boyd and L.Vendenberghe
Available both in print and online at www.stanford.edu/~boyd/cvxbook/

Course workload: approx. 5 homework sets, 1 take-home midterm exam (tentative), a few Matlab programming assignments, 1 group project

Credits: 3 credit hours

Instructor: Wotao Yin
Office Hour: 10:00-10:50 a.m. Tuesday and Thursday
Course webpage: www.caam.rice.edu/~wy1/CAAM554
Slides and homework: owlspace.rice.edu
Course flyer (made for Fall 2006) [link]