Solving systems of rational equations through strategy iteration

  • Authors:
  • Thomas Martin Gawlitza;Helmut Seidl

  • Affiliations:
  • VERIMAG, France;Technische Universität München, Germany

  • Venue:
  • ACM Transactions on Programming Languages and Systems (TOPLAS)
  • Year:
  • 2011

Quantified Score

Hi-index 0.00

Visualization

Abstract

We present practical algorithms for computing exact least solutions of equation systems over the reals with addition, multiplication by positive constants, minimum and maximum. The algorithms are based on strategy iteration. Our algorithms can, for instance, be used for the analysis of recursive stochastic games. In the present article we apply our techniques for computing abstract least fixpoint semantics of affine programs over the relational template polyhedra domain. In particular, we thus obtain practical algorithms for computing abstract least fixpoint semantics over the abstract domains of intervals, zones, and octagons.