Advancing Fusion Energy with AI: Developing Novel Compressed Representations for High-Dimensional Physics Data The University of Manchester in United Kingdom

This fully funded PhD project at The University of Manchester offers an exciting opportunity to advance fusion energy research by developing novel machine learning methods for high-dimensional physics data. The project addresses a critical challenge in the design of tokamak fusion reactors: predicting and modeling radiation-induced damage in structural materials. High-fidelity atomistic simulations generate vast, complex datasets describing the positions and evolution of millions of atoms under extreme conditions. However, the sheer volume and dimensionality of this data make it impractical for direct use in predictive models.

The core research goal is to create compressed, information-rich representations ("fingerprints") of atomic structures using advanced machine learning techniques. You will explore and develop new algorithms at the intersection of geometric deep learning, unsupervised representation learning, and time-series analysis. A major focus will be on graph neural network (GNN) approaches to capture long-range atomic neighborhoods and dynamic changes over time. The project provides access to a large, curated database of damaged atomic structures for training and validation, enabling you to design, test, and refine your models on real-world data.

Your work will directly contribute to the development of fast, efficient surrogate models for radiation damage, accelerating the in-silico design and qualification of new fusion reactor components. The project is part of the UKRI AI CDT in Decision Making for Complex Systems and is supported by the UK Atomic Energy Authority (UKAEA), offering a stipend of £31,000 for eligible students. The start date is September 2026, and the position is based at The University of Manchester, with industry collaboration.

Applicants should have a strong background in mathematics, computer science, or a related scientific field. Required application materials include academic transcripts, CV, a supporting statement, and contact details for two referees. An English language certificate is required if applicable. The university values diversity and inclusion, encourages applicants from all backgrounds, and offers flexible study options, including part-time study.

Apply by April 10, 2026, through the University of Manchester Application Portal under 'PhD in Artificial Intelligence'. For more information, visit the project page or contact the industry supervisor at [email protected].