Evaluating Las Vegas Algorithms - Pitfalls and Remedies
This work addresses the evaluation of stochastic search algorithms for combinatorial problems, which is incremental as it refines existing empirical methods rather than introducing a new paradigm.
The paper tackles the problem of evaluating Las Vegas Algorithms (LVAs) by proposing a novel methodology based on empirical run-time distributions, applied to Stochastic Local Search algorithms for SAT, highlighting pitfalls in existing methods and demonstrating benefits for performance comparison.
Stochastic search algorithms are among the most sucessful approaches for solving hard combinatorial problems. A large class of stochastic search approaches can be cast into the framework of Las Vegas Algorithms (LVAs). As the run-time behavior of LVAs is characterized by random variables, the detailed knowledge of run-time distributions provides important information for the analysis of these algorithms. In this paper we propose a novel methodology for evaluating the performance of LVAs, based on the identification of empirical run-time distributions. We exemplify our approach by applying it to Stochastic Local Search (SLS) algorithms for the satisfiability problem (SAT) in propositional logic. We point out pitfalls arising from the use of improper empirical methods and discuss the benefits of the proposed methodology for evaluating and comparing LVAs.