Search Landscape Analysis

Recently, I have been exploring the fitness landscapes of permutation optimization problems, problems where one must find an optimal ordering of some discrete set (e.g., scheduling problems). Such permutation problems fall into one of three broad categories, depending upon the structural characteristics of greatest influence on solution fitness (absolute positioning of elements, relative positioning of elements, or element precedences), although some problems may span more than one of these categories. Among other things, my research on permutation fitness landscapes has introduced a theoretical framework enabling exploring the effects of these structural properties on search performance. This includes the "Permutation in a Haystack" problem which enables specifying permutation fitness landscapes; as well as the "Calculus of Search Landscapes", a new tool for search landscape analysis that focuses on rates of change of fitness landscape topology, complementing other search landscape analysis tools such as fitness-distance correlation.

Selected Publications