{NRCL} - a model building approach to the {Bernays-Schönfinkel} fragment - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Conference Papers Year : 2015

{NRCL} - a model building approach to the {Bernays-Schönfinkel} fragment

Abstract

We combine constrained literals for model representation with key concepts from first-order superposition and propositional conflict-driven clause learning (CDCL) to create the new calculus Non-Redundant Clause Learning (NRCL) deciding the Bernays-Schönfinkel fragment. We use first-order literals constrained by disequalities between tuples of terms for compact model representation. From superposition, NRCL inherits the abstract redundancy criterion and the monotone model operator. CDCL adds the dynamic, conflict-driven search for a model. As a result, NRCL finds a false clause modulo the current model candidate effectively. It guides the derivation of a first-order ordered resolvent that is never redundant. Similar to 1UIP-learning in CDCL, the learned resolvent induces backtracking and, by blocking the previous conflict state via propagation, it enforces progress towards finding a model or a refutation. The non-redundancy result also implies that only finitely many clauses can be generated by NRCL on the Bernays-Schönfinkel fragment, which proves termination.

Dates and versions

hal-01247991 , version 1 (23-12-2015)

Identifiers

Cite

Gábor Alagi, Christoph Weidenbach. {NRCL} - a model building approach to the {Bernays-Schönfinkel} fragment. Frontiers of Combining Systems, 10th International Symposium (FroCos 2015), 2015, Wroclaw, Poland. pp.69-84, ⟨10.1007/978-3-319-24246-0_5⟩. ⟨hal-01247991⟩
80 View
0 Download

Altmetric

Share

Gmail Facebook X LinkedIn More