TLTk: A Toolbox for Parallel Robustness Computation of Temporal Logic Specifications

Joseph Cralley, Ourania Spantidi, Bardh Hoxha, Georgios Fainekos

Abstract: This paper presents the Temporal Logic Toolkit (TLTk), a modular falsification tool for signal temporal logic specifications developed in Python and C. At the core of the tool, an algorithm for robustness computation is utilized that supports multi-threaded CPU/GPU computation. The tool enables memory-efficient, parallel, robustness computation of system traces. In addition, the python implementation enables the addition and modification of temporal operators for application-specific scenarios. The performance of the tool is evaluated against state-of-the-art robustness computation engines DP-TaLiRo and Breach on a number of benchmark problems.