We have established one interface for communicating LISP-STAT processes, based on a message-passing communication model, and demonstrated the use of this to accelerate a simple coarse-grained computation. We are currently generalising this interface, initially to support compiled Lisp systems.
Our prototype implementation will be used to provide support for two statistical applications: the first is the computationally-intensive Monte Carlo approach to the analysis of large sparse social mobility tables proposed by Smith and McDonald [4]; the second is a modified approach to a Markov Chain Monte Carlo (Gibbs Sampler) technique for the analysis of contingency tables. We believe that both applications will gain significant benefit from the distributed LISP-STAT environment.