A Fast and Simple Parallel Algorithm for the Monotone Duality Problem

  • Authors:
  • Endre Boros;Kazuhisa Makino

  • Affiliations:
  • RUTCOR, Rutgers University, Piscataway 08854-8003;Department of Mathematical Informatics, University of Tokyo, Tokyo, Japan 113-8656

  • Venue:
  • ICALP '09 Proceedings of the 36th International Colloquium on Automata, Languages and Programming: Part I
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

We consider the monotone duality problem i.e., checking whether given monotone CNF φ and DNF *** are equivalent, which is a prominent open problem in NP-completeness. We construct a fast and simple parallel algorithms for the problem, that run in polylogarithmic time by using quasi-polynomially many processors. The algorithm exhibits better parallel time complexity of the existing algorithms of Elbassioni [11]. By using a different threshold of the degree parameter *** of φ in the algorithm, we also present a stronger bound on the number of processors for polylogarithmic-time parallel computation and improves over the previously best known bound on the sequential time complexity of the problem in the case when the magnitudes of |φ |, |*** | and n are different, e.g., |*** | = |φ | *** *** n for *** 1, where n denotes the number of variables. Furthermore, we show that, for several interesting well-known classes of monotone CNFs φ such as bounded degree, clause-size, and intersection-size, our parallel algorithm runs polylogarithmic time by using polynomially many processors.