Multi-Criteria Optimization in Answer Set Programming

Abstract

We elaborate upon new strategies and heuristics for solving multi-criteria optimization problems via Answer Set Programming (ASP). In particular, we conceive a new solving algorithm, based on conflictdriven learning, allowing for non-uniform descents during optimization. We apply these techniques to solve realistic Linux package configuration problems. To this end, we describe the Linux package configuration tool aspcud and compare its performance with systems pursuing alternative approaches.