Empirical Analysis of ideal recombination on random decomposable problems

This document presents an empirical analysis of ideal crossover strategies within random additively decomposable problems (RADPs) using genetic algorithms (GAs). It explores population sizing, run duration, and the scaling behavior of selectorecombinative GAs while verifying the applicability of previously developed facetwise models based on adversarial test problems. Key findings indicate that GAs exhibit subquadratic scalability with problem size and differentiate between easy and hard problem instances based on signal variance characteristics.