A Strategy for Parallel Simulation of Declarative Object-Oriented Models of Generalized Physical Networks

For several years now, most of the growth of computing power has been made possible by exploiting parallel CPUs on the same chip; unfortunately, state-of-the-art software tools for the simulation of declarative, objectoriented models still generate single-threaded simulation code, showing an increasingly disappointing performance. This paper presents a simple strategy for the efficient computation of the right-hand-side of the ordinary differential equations resulting from the causalization of objectoriented models, which is often the computational bottleneck of the executable simulation code. It is shown how this strategy can be particularly effective in the case of generalized physical networks, i.e., system models built by the connection of components storing certain quantities and of components describing the flow of such quantities between them.