[PAST EVENT] Honors Thesis Defense - Isabel Agostino
Title: Approximating Star-Discrepancy with a Genetic Algorithm
Abstract: The star-discrepancy is a measure of uniformity commonly used in quasi-Monte Carlo methods. However, its computation is NP-hard, making study of star-discrepancy for high dimensions and large point sets nearly impossible. Many approximation techniques have been researched over the years, so a from nothing development of a genetic algorithm was implemented and studied. Choice of the various metrics and mechanisms for the genetic algorithm are discussed before implementation and analysis of results. The approximation results are then compared to an established genetic algorithm for approximating star-discrepancy and issues with our algorithm discussed.