Techniques for Efficient Lazy-Grounding ASP Solving

Research output: Chapter in Book/Report/Conference proceedingConference contributionProfessional

Researchers

  • Lorenz Leutgeb
  • Antonius Weinzierl

Research units

  • Vienna University of Technology

Abstract

Answer-Set Programming (ASP) is a well-known and expressive logic programming paradigm based on efficient solvers. State-of-theart ASP solvers require the ASP program to be variable-free, they thus ground the program upfront at the cost of a potential exponential explosion of the space required. Lazy-grounding, where solving and grounding are interleaved, circumvents this grounding bottleneck, but the resulting solvers lack many important search techniques and optimizations. The recently introduced ASP solver Alpha combines lazy-grounding with conflict-driven nogood learning (CDNL), a core technique of efficient ASP solving. This work presents how techniques for efficient propagation can be lifted to the lazy-grounding setting. The Alpha solver and its components are presented and detailed benchmarks comparing Alpha to other ASP solvers demonstrate the feasibility of this approach.

Details

Original languageEnglish
Title of host publicationDeclare 2017 - Conference on Declarative Programming
EditorsDietmar Seipel, Michael Hanus, Salvador Abreu
Publication statusPublished - 14 Sep 2017
MoE publication typeD3 Professional conference proceedings
EventInternational Conference on Applications of Declarative Programming and Knowledge Management - Würzburg, Germany
Duration: 19 Sep 201721 Sep 2017
Conference number: 21
http://declare17.de/inap/

Conference

ConferenceInternational Conference on Applications of Declarative Programming and Knowledge Management
Abbreviated titleINAP
CountryGermany
CityWürzburg
Period19/09/201721/09/2017
Internet address

ID: 16807829