This report presents a formal approach to reducing sequences in PEPA components. Performing the described amalgamation procedure we may remove, from the underlying Markov chain, those states for which local detailed balance equations cannot be formulated. This transformation may lead to a simpler model with product form solution. Some classes of reduced models preserve those performance measures which we are interested in and, moreover, the steady state solution is much easier computationally. Keywords PEPA, decomposability, reversible Markov chains, insensitive stochastic processes
Jane Hillston, J. Tomasik