Compositional Invariant Generation via Linear Recurrence Analysis

This paper presents a new method for automatically generating numerical invariants for imperative programs. Given a program, our procedure computes a binary input/output relation on program states which over-approximates the behaviour of the program. It is compositional in the sense that it operates by decomposing the program into parts, computing an abstract meaning of each part, and then composing the meanings. Our method for approximating loop behaviour is based on first approximating the meaning of the loop body, extracting recurrence relations from that approximation, and then using the closed forms to approximate the loop. Our experiments demonstrate that on verification tasks, our method is competitive with leading invariant generation and verification tools.