Jane Hillston - Selected Publications#

1. A Compositional Approach to Performance Modelling. J. Hillston, Distinguished Dissertation Series, published by Cambridge University Press, June 1996. Also appeared as PhD thesis, Department of Computer Science, University of Edinburgh, 1994. CST-107-94.

[Hillston's thesis opened the field of quantitative formal methods and in particular the use of stochastic process algebras for performance modelling. It won the BCS/CPHC national Distinguished Dissertation competition and was published by Cambridge University Press. It has nearly 1700 citations and was subsequently reprinted in paperback in 2005.]

2. Product Form Solution for a Class of PEPA Models J. Hillston and N. Thomas, in Performance Evaluation 35, July 1999. pp. 171–192.

[Established a decomposed solution technique, based on the stochastic process algebra compositionality, which could be applied to the continuous time Markov chains underlying PEPA models. This was one of the key results leading to the award of the first Roger Needham award in 2004. The Roger Needham award is given annually to the UK-based computer scientist deemed to have achieved the most in the 10 years since PhD.]

3. Bio-PEPA: A Framework for the Modelling and Analysis of Biological Systems, F. Ciocchetta and J. Hillston, in Theoretical Computer Science 410 (33-34), pp. 3065-3084, 2009.

[This paper established a new process algebra designed to model intracellular processes and use in systems biology. The language has been taught in graduate courses at a number of institutions including the University of Edinburgh, University of Cambridge, University of Pisa, Ca' Foscari University Venice, The Stevens Institute and Tartu University. The language has also found use beyond systems biology to model ants, swarm robots and crowd dynamics.]

4. Fluid Flow Approximations of PEPA Models, J. Hillston, in the proceedings of the International Conference on Quantitative Evaluation of Systems (QEST) 2005. September 2005. pp. 33–42.

[This paper presents a systematic mapping from the PEPA high-level description of a continuous time Markov chain, to a set of ordinary differential equations which are much more efficient to solve and which can provide a good approximation of the behaviour in the case of large populations of agents in the PEPA model. It has been cited more than 320 times.]

5. Scalable Differential Analysis of Process Algebra Models, M. Tribastone, S. Gilmore and J. Hillston, in IEEE Transactions on Software Engineering, 38(1), 2012, pp. 205–219.

[This paper was the first to establish a formal semantics, in small-step operational style, establishing the link between a stochastic process algebra model and the mean-field approximation of the underlying continuous time Markov chain. This work stimulated a lot of interest as reflected by invited keynotes at CONCUR 2014 and PSI 2015 and Distinguished Lecture at Kings College London 2015.]

6. Continuous approximation of collective system behaviour: A tutorial, L. Bortolussi, J. Hillston, D. Latella and M. Massink, in Performance Evaluation 70(5), pp. 317–349, 2013.

[This tutorial gives a comprehensive overview of the use of continuous approximations (mean-field or fluid approximations) of population Markov chain models (such as are generated by stochastic process algebra models with large populations of agents). In addition of gathering a number of diverse results together and putting them into a common framework it also proves new results about the relationship between discrete and continuous time models via uniformisation.]

7. Model checking single agent behaviours by fluid approximation, L. Bortolussi and J. Hillston, in Information and Computation, 2015.

[This paper shows how the mean-field approach can be extended to stochastic verification using the Continuous Stochastic Logic. This allows properties of one agent within a large population to be checked very efficiently compared to statistical model checking, which would be the usual approach for models consisting of a large number of agents. The convergence of the results is proved and a model checking algorithm for time inhomogeneous Markov chains is established.]

8. Modelling and Analysis of Collective Adaptive Systems with CARMA and Its Tools, M. Loreti and J. Hillston, in “Formal Methods for the Quantitative Evaluation of Collective Adaptive Systems”, Volume 9700 of the series Lecture Notes in Computer Science,pp. 83–119, 2016.

[The CARMA modelling language developed during the QUANTICOL project greatly extends the expressiveness of previous stochastic process algebras. Agents have attributes in addition to process like behaviour and a variety of different communication mechanisms are supported. This chapter for a summer school gives a comprehensive review of the new language and the tools that have been developed to support it. The language has already been adopted by other researchers (Security group in Edinburgh, cloud computing group at Harbin Engineering University) leading to further publications. ]

9. Unbiased Bayesian Inference for Population Markov Jump Processes via Random Truncations, A. Georgoulas, J. Hillston and G. Sanguinetti, in Statistics and Computing 27(4), pp. 991-1002, 2017.

[This paper tackles the challenging problem of Bayesian inference over continuous time Markov processes. It introduces a class of pseudo-marginal sampling algorithms based on a random truncation method which enables a principled treatment of infinite state spaces with very good results. It has been downloaded 2500 times. ]

10. ProPPA: Probabilistic Programming for Stochastic Dynamical Systems, A. Georgoulas, J. Hillston and G. Sanguinetti, in ACM Transactions on Modelling and Computer Simulation (TOMACS) 28(1), 2018.

[This paper presents a formalism which tackles the problem of uncertainty in the parameters of a stochastic model, giving a formal framework for parameter estimation via machine learning. This blending of formal methods and machine learning is highly innovative. It provides a formalisation of uncertainty within quantitative models and a 'solution engine' approach for a wide-class of Bayesian learning problems, removing the need for a bespoke solution in every case. It has generated a lot of interest and invited talks at CMSB 2016, Simons Institute 2016, IFM 2017 and ATI Logic and Learning workshop 2018.]