Difference between revisions of "Automating Coinduction with Case Analysis"

From FSL
Jump to: navigation, search
Line 1: Line 1:
<pub id="goriac-lucanu-rosu-2010-icfem" template="PubDefaultWithAbstractAndTitle"> </pub>
<pubbib id="goriac-lucanu-rosu-2010-icfem" template="PubDefaultWithAbstractAndTitle"/>

Latest revision as of 18:52, 6 April 2016

Automating Coinduction with Case Analysis
Eugen-Ioan Goriac and Dorel Lucanu and Grigore Rosu
ICFEM'10, Springer, pp 220-236. 2010
Abstract. Coinduction is a major technique employed to prove behavioral properties of systems, such as behavioral equivalence. Its automation is highly desirable, despite the fact that most behavioral problems are $\Pi_2^0$-complete. Circular coinduction, which is at the core of the $\CIRC$ prover, automates coinduction by systematically deriving new goals and proving existing ones until, hopefully, all goals are proved. Motivated by practical examples, circular coinduction and $\CIRC$ have been recently extended with several features, such as special contexts, generalization and simplification. Unfortunately, none of these extensions eliminates the need for case analysis and, consequently, there are still many natural behavioral properties that $\CIRC$ cannot prove automatically. This paper presents an extension of circular coinduction with case analysis constructs and reasoning, as well as its implementation in $\CIRC$. To uniformly prove the soundness of this extension, as well as of past and future extensions of circular coinduction and $\CIRC$, this paper also proposes a general correct-extension technique based on equational interpolants.

Personal tools