Python (Weighted) Signal Temporal Logic Library

A comprehensive Python library for constructing, evaluating, and optimizing Weighted Signal Temporal Logic (WSTL) formulas.

Features

• Weighted Robustness: Quantitative semantics with weights. If weights are set to 1, reduces to standard STL robustness.
• Dual Backends with Unified API:
  • switch backends with a single call; all formula classes work identically across backends
  • Backends:
    • numpy (default): numpy robustness computation via wstlpy
    • torch: differentiable robustness via wstltorch, compatible with gradient-based optimization and cuda