PRoTECT is an open-source software tool, with a user-friendly GUI, for the purposes of formal verification using polynomial-type barrier certificates for safety specifications. PRoTECT can handle discrete-time stochastic systems, continuous-time stochastic systems, discrete-time deterministic systems and continuous-time deterministic systems.

Synthesizing and validating barrier certificates for unknown dynamical systems with latent states and polynomial dynamics using Bayesian inference

Tools for verifying safety of control systems that may experience timing uncertainty

A robust framework for formal verification of multi-agent systems powered by LLMs, ensuring safe and reliable decision-making. Integrates machine learning with formal methods to validate agent behaviors, safety constraints, and real-world simulations in critical domains like autonomous driving and robotics.

Code for paper "Bridging Neural ODE and ResNet: A Formal Error Bound for Safety Verification"

Security verifier of zero-knowledge proofs using Gröbner bases built on top of the open-source Circom compiler.

AI agents that spend real money. Autonomous department heads with on-chain budgets, safety verification, and USDC settlement on Monad via x402.