Solving Combinatorial Problems with Regular Local Search Algorithms

  • Authors:
  • Ramón Béjar;Felip Manyà

  • Affiliations:
  • -;-

  • Venue:
  • LPAR '99 Proceedings of the 6th International Conference on Logic Programming and Automated Reasoning
  • Year:
  • 1999

Quantified Score

Hi-index 0.01

Visualization

Abstract

In this paper we describe new local secirch algorithms for regular CNP formulcis and investigate their suitability for solving problems from the dom2uns of graph coloring and sports scheduling. First, we define suitable encodings for such problems in the logic of regular CNF formulas. Second, we describe Regular-GSAT and Regular-WSAT, as well as some varisuits, which are a natured generalization of two prominent local search algorithms - GSAT and WSAT - used to solve the prepositional satisfiability (SAT) problem in classical logic. Third, we report on experimented results that demonstrate that encoding graph coloring and sports scheduling problems as instances of the SAT problem in regular CNF formulas and then solving these instances with local search algorithms can outperform or compete with state-of-the-art approciches to solving hard combinatorial problems.