Automatic Generation of Polynomial Loop Invariants: Algebraic Foundations

  • Authors:

  • Enric Rodríguez-Carbonell;Deepak Kapur

  • Affiliations:

  • Technical University of Catalonia, Barcelona, Spain;University of New Mexico, Albuquerque, NM

  • Venue:
  • ISSAC '04 Proceedings of the 2004 international symposium on Symbolic and algebraic computation
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents the algebraic foundation for an approach for generating polynomial loop invariants in imperative programs. It is first shown that the set of polynomials serving as loop invariants has the algebraic structure of an ideal. Using this connection, a procedure for finding loop invariants is given in terms of operations on ideals, for which Grobner basis constructions can be employed. Most importantly, it is proved that if the assignment statements in a loop are solvable (in particular, affine) mappings with positive eigenvalues, then the procedure terminates in at most 2m+1 iterations, where m is the number of variables in the loop. The proof is done by showing that the irreducible subvarieties of the variety associated with the polynomial ideal approximating the invariant polynomial ideal of the loop either stay the same or increase their dimension in every iteration. This yields a correct and complete algorithm for inferring conjunctions of polynomial equations as invariants. The method has been implemented in Maple using the Groebner package. The implementation has been used to automatically discover nontrivial invariants for several examples to illustrate the power of the techniques.