Constraint reasoning based on interval arithmetic

The tolerance propagation approach

Eero Hyvönen

Research output: Contribution to journalArticleScientificpeer-review

122 Citations (Scopus)

Abstract

Interval constraint satisfaction (interval labeling) systems have traditionally been based on local Waltz filtering techniques that cannot in general determine global solutions. In contrast, this paper documents a related technique, tolerance propagation (TP), that generalizes the idea of numerical exact value propagation into interval propagation. In TP, consistency techniques based on the topology of the constraint net can be combined with techniques of interval arithmetic in a new fruitful way. In particular, by TP it is possible to determine global solutions for interval constraint satisfaction problems with arbitrary accuracy and without losing all attractions of simple local computations.
Original languageEnglish
Pages (from-to)71-112
JournalArtificial Intelligence
Volume58
DOIs
Publication statusPublished - 1992
MoE publication typeA1 Journal article-refereed

Fingerprint

tolerance
Constraint satisfaction problems
Labeling
Topology
Tolerance
Values
Constraint Satisfaction

Cite this

@article{13198b30c04f47db98964a9b15e591b9,
title = "Constraint reasoning based on interval arithmetic: The tolerance propagation approach",
abstract = "Interval constraint satisfaction (interval labeling) systems have traditionally been based on local Waltz filtering techniques that cannot in general determine global solutions. In contrast, this paper documents a related technique, tolerance propagation (TP), that generalizes the idea of numerical exact value propagation into interval propagation. In TP, consistency techniques based on the topology of the constraint net can be combined with techniques of interval arithmetic in a new fruitful way. In particular, by TP it is possible to determine global solutions for interval constraint satisfaction problems with arbitrary accuracy and without losing all attractions of simple local computations.",
author = "Eero Hyv{\"o}nen",
year = "1992",
doi = "10.1016/0004-3702(92)90005-I",
language = "English",
volume = "58",
pages = "71--112",
journal = "Artificial Intelligence",
issn = "0004-3702",
publisher = "Elsevier",

}

Constraint reasoning based on interval arithmetic : The tolerance propagation approach. / Hyvönen, Eero.

In: Artificial Intelligence, Vol. 58, 1992, p. 71-112.

Research output: Contribution to journalArticleScientificpeer-review

TY - JOUR

T1 - Constraint reasoning based on interval arithmetic

T2 - The tolerance propagation approach

AU - Hyvönen, Eero

PY - 1992

Y1 - 1992

N2 - Interval constraint satisfaction (interval labeling) systems have traditionally been based on local Waltz filtering techniques that cannot in general determine global solutions. In contrast, this paper documents a related technique, tolerance propagation (TP), that generalizes the idea of numerical exact value propagation into interval propagation. In TP, consistency techniques based on the topology of the constraint net can be combined with techniques of interval arithmetic in a new fruitful way. In particular, by TP it is possible to determine global solutions for interval constraint satisfaction problems with arbitrary accuracy and without losing all attractions of simple local computations.

AB - Interval constraint satisfaction (interval labeling) systems have traditionally been based on local Waltz filtering techniques that cannot in general determine global solutions. In contrast, this paper documents a related technique, tolerance propagation (TP), that generalizes the idea of numerical exact value propagation into interval propagation. In TP, consistency techniques based on the topology of the constraint net can be combined with techniques of interval arithmetic in a new fruitful way. In particular, by TP it is possible to determine global solutions for interval constraint satisfaction problems with arbitrary accuracy and without losing all attractions of simple local computations.

U2 - 10.1016/0004-3702(92)90005-I

DO - 10.1016/0004-3702(92)90005-I

M3 - Article

VL - 58

SP - 71

EP - 112

JO - Artificial Intelligence

JF - Artificial Intelligence

SN - 0004-3702

ER -