Team MExICo

Overall Objectives
Scientific Foundations
Application Domains
New Results
Contracts and Grants with Industry
Other Grants and Activities

Section: New Results


Concurrent Recursive Programs

We introduced a new automata class, called concurrent visibly pushdown automata , to model recursive programs that communicate via shared memory. While most previous approaches to those kind of systems restrict to the analysis of an already existing program, our new model allows one to tackle the synthesis problem : the automatic generation of programs from global specifications. Our results extend classical theorems from the theory of Mazurkiewciz traces to the recursive setting [10] .

Local Safety and Local Liveness for Distributed Systems

Volker Diekert and Paul Gastin have defined local versions of the classical safety and liveness properties. By local we mean that we are interested in the local views of processes and not in global configurations or snapshots. The advantage is to obtain reasonable complexity, i.e., PSPACE as for sequential systems. They charachterized local safety by local closure and local liveness by local density. Restricting to first-order definable properties, they proved a decomposition theorem in the spirit of the separation theorem for linear temporal logic. They also characterized local safety and local liveness by means of canonical local temporal logic formulae.


Refinements were obtained in the analysis of unfoldings for the purposes of diagnosis. Stefan Haar's study of the reveals relation in occurrence nets has led to novel criteria on observability in labeled Petri nets, presented in [19] .

Automata theory

We made a detailed re-investigation of the emptiness problem for Büchi automata, which is at the heart of LTL model checking [18] . More precisely, we studied on-the-fly algorithms of the type that are used in the acclaimed Spin tool. We compared existing and new algorithms experimentally suite and proposed improvements. Compared with the algorithm implemented in Spin, our best algorithm is faster by about 33 per cent on average. The paper was named among the best six at MEMICS and invited for a journal publication.


Logo Inria