Higher-precision groundness analysis

Michael Codish, Samir Genaim, Harald Søndergaard, Peter J. Stuckey

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

3 Scopus citations

Abstract

Groundness analysis of logic programs using Pos-based abstract interpretation is one of the clear success stories of the last decade in the area of logic program analysis. In this work we identify two problems with the Pos domain, the multiplicity and sign problems, that arise independently in groundness and uniqueness analysis. We describe how these problems can be solved using an analysis based on a domain Size for inferring term size relations. However this solution has its own shortcomings because it involves a widening operator which leads to a loss of Pos information. Inspired by Pos, Size and the LSign domain for abstract linear arithmetic constraints we introduce a new domain LPos, and show how it can be used for groundness and uniqueness analysis. The idea is to use the sign information of LSign to improve the widening of Size so that it does not lose Pos information. We prove that the resulting analyses using LPos are uniformly more precise than those using Pos.

Original languageEnglish
Title of host publicationLogic Programming - 17th International Conference, ICLP 2001, Proceedings
EditorsPhilippe Codognet
PublisherSpringer Verlag
Pages135-149
Number of pages15
ISBN (Electronic)9783540429357
DOIs
StatePublished - 1 Jan 2001
Event17th International Conference on Logic Programming, ICLP 2001 - Paphos, Cyprus
Duration: 26 Nov 20011 Dec 2001

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume2237
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference17th International Conference on Logic Programming, ICLP 2001
Country/TerritoryCyprus
CityPaphos
Period26/11/011/12/01

Fingerprint

Dive into the research topics of 'Higher-precision groundness analysis'. Together they form a unique fingerprint.

Cite this