Stochastic process algebras emerged about 15 years ago as system description techniques for performance modelling.  Originally they were focussed on the generation of continuous time Markov chains facilitating steady state and transient analysis numerically.  As with all state-based modelling techniques they suffer from problems of state space explosion.

Recently we have been exploring techniques to capture a representation of the system at the population level from the PEPA description, rather than capturing individual behaviours as happens in the CTMC semantics.  In this talk I wil explain the mapping to the population model, a set of non-linear ordinary differential equations, and illustrate it with a case study of an Internet worm infection.