Algebraic Reasoning for (Un)Solvable Loops (Invited Talk)

Abstract

Loop invariants describe valid program properties that hold before and after every loop iteration. As such, loop invariants are the workhorses in formalizing loop semantics and automating the formal analysis and verification of programs with loops. While automatically synthesizing loop invariants is, in general, an uncomputable problem, when considering only single-path loops with linear updates (linear loops), the strongest polynomial invariant is in fact computable [Michael Karr, 1976; Markus Müller-Olm and Helmut Seidl, 2004; Laura Kovács, 2008; Ehud Hrushovski et al., 2018]. Yet, already for loops with "only" polynomial updates, computing the strongest invariant has been an open challenge since 2004 [Markus Müller-Olm and Helmut Seidl, 2004]. In this invited talk, we first present computability results on polynomial invariant synthesis for restricted polynomial loops, called solvable loops [Rodríguez-Carbonell and Kapur, 2004]. Key to solvable loops is that one can automatically compute invariants from closed-form solutions of algebraic recurrence equations that model the loop behaviour [Laura Kovács, 2008; Andreas Humenberger et al., 2017]. We also establish a technique for invariant synthesis for classes of loops that are not solvable, termed unsolvable loops [Daneshvar Amrollahi et al., 2022]. We next study the limits of computability in deriving the (strongest) polynomial invariants for arbitrary polynomial loops. We prove that computing the strongest polynomial invariant of arbitrary, single-path polynomial loops is very hard [Julian Müllner, 2023] - namely, it is at least as hard as the Skolem problem [Graham Everest et al., 2003; Terrence Tao, 2008], a prominent algebraic problem in the theory of linear recurrences. Going beyond single-path loops, we show that the strongest polynomial invariant is uncomputable already for multi-path polynomial loops with arbitrary quadratic polynomial updates