Machine Learning for Probabilistic Modelling of Non-equilibrium Time Series Beyond the Markovian Paradigm

About the Project

The overarching aim of this project is to find synergies between methods and ideas of modern machine learning and of statistical mechanics for the study of stochastic dynamics with application to the analysis of time series. In particular, the project will examine and develop methods that go beyond the Markovian paradigm. It will consider a range of time series data, focusing on those that show challenging properties of uncertainty, irregularity and mixed-modality. It will examine a range of models and techniques that go beyond Markovian approaches, including state-space models, tensor networks, and machine learning frameworks such as recurrent neural networks and transformers. Models and datasets will be studied and benchmarked in key tasks relating to both prediction/forecasting and anomaly detection. Comparison with known analytic methods and established Markov models will be made wherever possible. Expected outcomes include a unified non-Markovian framework for time series analysis, a suite of relevant datasets, and large-scale statistical studies comparing different methods. The successful candidate will be jointly supervised by: Dr Edward Gillman (https://www.nottingham.ac.uk/physics/people/edward.gillman) and Professor Juan P. Garrahan (https://www.nottingham.ac.uk/physics/people/juan.garrahan)

Entry requirements

Open to UK nationals only. Expected starting date October 2025. We are seeking candidates with:

• Relevant subject matter experience at required level (e.g. 2.1 or above undergraduate degree in physics, mathematics or computer science)
• Willingness to adapt and work across different disciplines
• Ability to work independently and cooperatively
• Commitment to inclusivity, responsible research and innovation

How to apply

Applications should be submitted by following the steps outlined on the page: https://www.nottingham.ac.uk/physics/studywithus/postgraduate/howtoapply.aspx

In the “Research Proposal Section” of the online application simply state that you are applying to the open position on “Machine Learning for Probabilistic Modelling” with Dr Edward Gillman and Professor Juan P. Garrahan as supervisors.

Funding Notes

Full tuition fee waiver pa (Home Students only) and stipend at above UKRI rates pa (currently at £20,780 for 2025/26 academic year, increasing in line with inflation). Research training and support grant (RTSG) of £3000 per year. Funding is available for 4 years.