ARCH-COMP21 Category Report: Stochastic Models

Abstract

This report presents the results of a friendly competition for formal verification and policy synthesis of stochastic models. It also introduces new benchmarks within this category, and recommends next steps for this category towards next year’s edition of the competition. The friendly competition took place as part of the workshop Applied Verification for Continuous and Hybrid Systems (ARCH) in Spring/Summer 2021.

Temporal Logic Control of Nonlinear Stochastic Systems Using a Piecewise-Affine Abstraction

“…These hyper-parameters include among others the diameter of the discretization grid [35], confidence values for sampling-based computations [46], and the number of basis functions [25]. A comparison between the current tools handling correct-by-design synthesis of stochastic systems applied to a set of standard benchmarks can be found in the ARCH competition reports [2,1,3].…”

ARCH-COMP22 Category Report: Stochastic Models

“…In (1), both random variables follow an exponential distribution; the time delay in location ℓ 0 can hence be modeled by one random variable exp(0.1 + 0.08). This is not possible in (2), as X 1 follows an exponential distribution and X 2 a folded normal distribution.…”

“…The Heated Tank benchmark stems from the safety literature; there it is a well-known example of a Piecewise Deterministic Markov Process (PDMP) [29]. This made the Heated Tank benchmark a logical candidate for inclusion in the set of ARCH stochastic models [1,2,3,4].…”

“…In [4,2,3], Heated Tank version 4.0 has been evaluated by the tools modes and HYPEG and by the framework SDCPN&IPS. In [1], Heated Tank version 4.III has been evaluated by the tools FIGARO and PyCATSHOO, and by the framework SDCPN&IPS.…”

“…graphical Petri Net model of version 4.0[2]. The details of these exyensions have been specified in subsection 5.4 of[1]. In 2021 and 2022, Heated Tank version 4.III has been evaluated by FIGARO, PyCATSHOO, SDCPN&MC, SDCPN&IPS, SDCPN&IS, and HYPEG.…”