In: Proc. 8th Int. Workshop on Petri Net and Performance Models (PNPM'99), 8-10 October 1999, Zaragoza, Spain, pages 22-31. 1999.
Abstract: Kronecker-based approaches have been proposed for the solution of structured GSPNs with extremely large state spaces. Representing the transition rate matrix using Kronecker sums and products of smaller matrices virtually eliminates its storage requirements, but introduces various sources of overhead. We show how, by using a new data structure which we call matrix diagrams, we are able to greatly reduce or eliminate many of these overheads, resulting in a very efficient overall solution process.
Keywords: Kronecker representation, matrix diagrams, stochastic Petri nets.