COntext-free model checking for Recursive PrObabilistic pRogrAms

Información del proyecto

CORPORA

Identificador del acuerdo de subvención: 101107303

DOI

10.3030/101107303

Proyecto cerrado

Fecha de la firma de la CE 2 Marzo 2023

Fecha de inicio 1 Septiembre 2023

Fecha de finalización 31 Agosto 2025

Financiado con arreglo a

• Marie Skłodowska-Curie Actions (MSCA)

Coste total

Sin datos

Aportación de la UE

€ 183 600,96

Coordinado por

TECHNISCHE UNIVERSITAET WIEN
Austria

Periodic Reporting for period 1 - CORPORA (COntext-free model checking for Recursive PrObabilistic pRogrAms)

Período documentado: 2023-09-01 hasta 2025-08-31

Resumen del contexto y de los objetivos generales del proyecto

The CORPORA project tackles the challenge of verifying the reliability and safety of software systems that rely on advanced probabilistic reasoning. Probabilistic programs are increasingly used to model and implement complex decision-making under uncertainty, from autonomous systems and cybersecurity protocols to cognitive science models. However, existing verification techniques are ill-equipped to analyse programs that incorporate nested conditioning and other intricate stochastic phenomena, which are essential to accurately represent real-world processes.
CORPORA has developed cutting-edge verification techniques and tools that enable scalable, mathematically rigorous analysis of probabilistic programs with recursive and nested structures. By integrating probabilistic model checking with expressive temporal logics based on context-free languages, the project delivers a novel theoretical and algorithmic foundation for reasoning about stochastic systems at a level of precision previously unattainable. The resulting open-source tools allow researchers and practitioners to detect design flaws, prove safety and correctness properties, and optimize decision-making systems.

Trabajo realizado desde el comienzo del proyecto hasta el final del período abarcado por el informe y los principales resultados hasta la fecha

Within the CORPORA project, we advanced the verification of probabilistic programs by demonstrating, for the first time, how to use a highly expressive temporal logic capable of describing recursive aspects of probabilistic phenomena. This makes it possible to specify and check properties of systems where events are nested or depend on previous decisions, which is particularly valuable for modeling the theory of mind in multi-agent systems—situations where agents reason about the beliefs and intentions of others. We developed new algorithms and implemented them in POPACheck, a prototype tool that automates the verification of such complex programs. This achievement proves the feasibility of verifying context-free properties in probabilistic settings and provides a solid foundation for future scalable tools for reasoning about uncertainty in critical applications.

Avances que van más allá del estado de la técnica e impacto potencial esperado (incluida la repercusión socioeconómica y las implicaciones sociales más amplias del proyecto hasta la fecha)

The results of CORPORA go beyond the state of the art by delivering the first verification framework capable of analysing recursive probabilistic programs against context-free temporal properties, addressing a fundamental gap in formal verification research. By enabling rigorous reasoning about nested stochastic behaviours, CORPORA opens new possibilities for advancing trustworthy AI, autonomous systems, and cognitive modelling. To maximize impact, the project followed rigorous open-access practices, ensuring that all theoretical results, algorithms, and publications are freely accessible to the research community. The POPACheck prototype was designed with clarity, robustness, and reproducibility in mind, providing a solid foundation for future research and tool development. These choices ensure that CORPORA’s contributions are easy to adopt, extend, and build upon, supporting long-term innovation in formal verification and contributing to Europe’s leadership in safe and explainable computing.

logo-final.png