Description: The contours of an N-dimensional function exhibit
many forms ranging from "completely flat" to "totally chaotic". We wish to
develop simple tools and techniques for exploring these landscapes.
Number of students required: 1 or 2
Description: Particle Swarm Optimization is a technique for finding
good solutions to NP-Hard problems. This project consists in implementing such
an algorithm and evaluating the approach using standard test sets
to see how well they work. The programming should be done in C#.
Number of students required: 1 or 2
Description: Some recent investigations have suggested that certain
classes of NP hard problems based on data found in the real world (as opposed
to contrived artificial data used to test algorithms) have solutions whose
optimum values can be estimated by an examination of historical results. This
project is concerned with finding as many of these data sets as possible. No
programming is required - just lots of reading.
Number of students required: 1 or 2
Description: Paradiseo is a freely available package, written in C++, that performs optimization of complex
functions in many dimensions. The project will require a GUI that integrates the package into a visual system permitting
a user to optimize complex functions.
Number of students required: 1 or 2
Description: Available tabu algorithms are very effective in finding
pseudo-optimal solutions for optimization problems. An additional optimization
goal is to find optimal solutions to problems that have more than one function
to be optimized. This project consists in investigating ways of using
Tabu search to find such solutions.
Number of students required: 1 or 2
Description: The COMET programming language is a fairly new
language used for expressing constraints when modeling combinatorial
optimization problems. The project will require the downloading and loading
of the COMET system and then investigating what it can do.
Number of students required: 1 or 2