Three linearization techniques for multivariate polynomials in static analysis using convex polyhedra
نویسندگان
چکیده
We present three linearization methods to over-approximate non-linear multivariate polynomials with convex polyhedra. The first one is based on the substitution of some variables by intervals. The principle of the second linearization technique is to express polynomials in the Bernstein basis and deduce a polyhedron from the Bernstein coefficients. The last method is based on Handelman’s theorem and consists in using products of constraints of a starting polyhedron to over-approximate a polynomial. As a part of the VERASCO project, the goal is to prove such methods with the proof assistant Coq.
منابع مشابه
A Linearization Technique for Multivariate Polynomials Using Convex Polyhedra Based on Handelman-Krivine's Theorem
We present a new linearization method to over-approximate non-linear multivariate polynomials with convex polyhedra. It is based on Handelman-Krivine’s theorem and consists in using products of constraints of a polyhedron to over-approximate a polynomial on this polyhedron. We implemented it together with two other linearization methods that we will not detail in this paper, but that we shall u...
متن کاملPolyhedral Approximation of Multivariate Polynomials Using Handelman's Theorem
Convex polyhedra are commonly used in the static analysis of programs to represent over-approximations of sets of reachable states of numerical program variables. When the analyzed programs contain nonlinear instructions, they do not directly map to standard polyhedral operations: some kind of linearization is needed. Convex polyhedra are also used in satisfiability modulo theory solvers which ...
متن کاملRefinement to Certify Abstract Interpretations, Illustrated on Linearization for Polyhedra
Our concern is the modular development of a certified static analyzer in Coq: we extend a certified abstract domain of convex polyhedra with a linearization procedure approximating polynomial expressions. In order to help such a development, we propose a proof framework, embedded in Coq, that implements a refinement calculus. It allows to hide for proofs several low-level aspects of the computa...
متن کاملBernstein's polynomials for convex functions and related results
In this paper we establish several polynomials similar to Bernstein's polynomials and several refinements of Hermite-Hadamard inequality for convex functions.
متن کاملThree Rings of Polyhedral Simple Functions
We survey three ways to multiply elements of the additive subgroup of the group of real-valued functions on R(d) which is generated by the indicator functions of polyhedra. In the resulting commutative rings, identities often correspond to useful techniques of decomposition of polyhedra. We are led immediately to various interesting topics, including Ehrhart polynomials, mixed volumes, Gram's r...
متن کامل