Monte Carlo algorithms are widely used in the nuclear industry for a range of applications, including criticality calculations, shielding problems and reactor design. Many of these algorithms are based on alternating between simulating the underlying radiation transport process and population control methods in a similar fashion to sequential Monte Carlo (SMC) methods.

The broad aim of this project is to interpret these industrial algorithms in an SMC framework, and using the SMC formalism to improve their performance. An essential first step will be deriving appropriate Feynman–Kac flows to describe industrial simulation algorithms. Potential applications of such a result include

1) The use of adaptive resampling methods to guide the alternation of simulation and population control steps.

2) Implementation of techniques for variance estimation from SMC output.

3) Development of convergence diagnostics based on guidelines used in MCMC methods.

There is a wide range of industrial algorithms available to study, but the so-called superhistory powering method [1] is envisaged as a good starting point.

The project will involve a mix of mathematical analysis and programming and is being delivered via an PhD studentship supported by an EPSRC Industrial CASE award, and Jacobs Clean Energy. The project team are Emma Horton, Andreas Kyprianou, Paul Smith and Usman Laden.