View Image

GPAnalyser is a tool for fluid analysis of massively parallel systems. The core of GPA is based on the paper by Hayden et. al. showing how to derive systems of ordinary differential equations approximating moments of populations in models described in the Markovian process algebra PEPA.

GPA efficiently implements these techniques and extensions allowing calculation of accumulated rewards and computation of passage time probabilities. The techniques are slightly generalised and GPA supports a larger class of models. In addition to GPEPA, the input files can use a simple chemical-equations-like language. GPA also provides mechanisms to explore large parameter spaces, either by naive parameter sweeping or via calls to MATLAB's global optimisation toolbox.