Date of Award

12-2010

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Legacy Department

Computer Science

Advisor

Dean, Brian C

Committee Member

Jacobs , David

Committee Member

Madison , Wayne

Committee Member

Malloy , Brian

Abstract

In this dissertation we focus on different variations of the stable matching (marriage) problem, initially posed by Gale and Shapley in 1962. In this problem, preference lists are used to match n men with n women in such a way that no (man, woman) pair exists that would both prefer each other over their current partners. These two would be considered a blocking pair, preventing a matching from being considered stable. In our research, we study three different versions of this problem.
First, we consider batch testing of stable marriage solutions. Gusfield and Irving presented an open problem in their 1989 book The Stable Marriage Problem: Structure and Algorithms<\italic> on whether, given a reasonable amount of preprocessing time, stable matching solutions could be verified in less than O(n^2) time. We answer this question affirmatively, showing an algorithm that will verify k different matchings in O((m + kn) log^2 n) time.
Second, we show how the concept of an adaptive algorithm can be used to speed up running time in certain cases of the stable marriage problem where the disorder present in preference lists is limited. While a problem with identical lists can be solved in a trivial O(n) running time, we present an O(n+k) time algorithm where the women have identical preference lists, and the men have preference lists that differ in k positions from a set of identical lists. We also show a visualization program for better understanding the effects of changes in preference lists.
Finally, we look at preference list based matching as a heuristic for cost based matching problems. In theory, this method can lead to arbitrarily bad solutions, but through empirical testing on different types of random sources of data, we show how to obtain reasonable results in practice using methods for generating preference lists “asymmetrically” that account for long-term ramifications of short-term decisions. We also discuss several ways to measure the stability of a solution and how this might be used for bicriteria optimization approaches based on both cost and stability.

Share

COinS