Interpolants for Linear Arithmetic in SMT

  • Authors:
  • Christopher Lynch;Yuefeng Tang

  • Affiliations:
  • Department of Mathematics and Computer Science, Clarkson University, Potsdam, USA 13676;Department of Mathematics and Computer Science, Clarkson University, Potsdam, USA 13676

  • Venue:
  • ATVA '08 Proceedings of the 6th International Symposium on Automated Technology for Verification and Analysis
  • Year:
  • 2008

Quantified Score

Hi-index 0.00

Visualization

Abstract

The linear arithmetic solver in Yices was specifically designed for SMT provers, providing fast support for operations like adding and deleting constraints. We give a procedure for developing interpolants based on the results of the Yices arithmetic solver. For inequalities over real numbers, the interpolant is computed directly from the one contradictory equation and associated bounds. For integer inequalities, a formula is computed from the contradictory equation, the bounds, and the Gomory cuts. The formula is not exactly an interpolant because it may contain local variables. But local variables only arise from Gomory cuts, so there will not be many local variables, and the formula should thereby be useful for applications like predicate abstraction. For integer equalities, we designed a new procedure. It accepts equations and congruence equations, and returns an interpolant. We have implemented our method and give experimental results.