Coordination Level Modeling and Analysis of Parallel Programs using Petri Nets

In the last fifteen years, the high performance computing (HPC) community has claimed for parallel programming environments that reconciles generality, higher level of abstraction, portability, and efficiency for distributed-memory parallel computing platforms. The Hash component model appears as an alternative for addressing HPC community claims for fitting these requirements. This paper presents foundations that will enable a parallel programming environment based on the Hash model to address the problems of "debugging", performance evaluation and verification of formal properties of parallel program by means of a powerful, simple, and widely adopted formalism: Petri nets.