Combining AI and Physics-Based Simulations to Model Complex Real-World Systems

Machine Learning (ML) and physics-based simulations are two of the most promising approaches for understanding and predicting complex systems. ML excels at identifying patterns in data but requires large datasets to function effectively. Physics-based simulations, on the other hand, rely on detailed knowledge of the underlying physical phenomena but often struggle when this knowledge is incomplete.

This project aims to overcome these challenges by combining the strengths of both approaches. By integrating ML and physics-based simulations, we can tackle problems where we have data, but not enough, and where we understand the physics, but not entirely.

Many of the most important problems in contemporary science and engineering exist in this grey area, with partial data and incomplete physical knowledge.

One example is the modelling of human organs, which follow known physical laws but are also influenced by the activity of the nervous system—a factor that is currently difficult to replicate in multiphysics simulations. Another example is Data Assimilation, where incomplete and diverse information is integrated to provide new insights into a system’s behaviour.

As part of this project, you will develop novel techniques to couple ML and physics-based simulations and apply them to real-world problems. This interdisciplinary project offers the opportunity to gain expertise in both ML and multiphysics simulation while contributing to the theoretical advancements of both fields.

We are looking for motivated candidates with a First-Class degree or a strong Upper Second-Class degree (2:1) in engineering, physics, mathematics, or computer science, with a keen interest in computational modelling and machine learning.