Algebraic Characterization of Time-Varying Set Invariance

Abstract

The verification of temporal forward invariance for dynamical systems subject to time-varying constraints is a foundational problem in the analysis of safety-critical systems. Existing methodologies rely on restrictive geometric assumptions, such as the Linear Independence Constraint Qualification (LICQ), or suffer from exponential computational complexity when applied to time-varying polytopes. This paper establishes a rigorous, unified, and computationally tractable framework for temporal safety verification. First, utilizing tools from nonsmooth analysis, we prove that a Nagumo-type geometric inclusion evaluated strictly on the boundary of the augmented graph is both necessary and sufficient for temporal forward invariance. Second, we provide an algebraic characterization of this condition requiring only the weaker Abadie Constraint Qualification (ACQ), generalizing standard Control Barrier Function (CBF) boundary conditions to non-smooth, multi-constraint domains. Finally, for the practical class of Linear Time-Invariant (LTI) systems with time-varying polytopic constraints, we deploy Farkas' Lemma to eliminate exponential vertex enumeration. By introducing a Lipschitz-bounded time-discretization scheme, we reduce the infinite-dimensional continuous-time verification problem to a finite, highly tractable set of decoupled Linear Programs (LPs).