PhD Studentship: Advanced Monte Carlo for Rapid Atomistic Simulations: from Complex Materials to Protein Folding

Many important scientific problems involve rare transitions between stable states. Examples include the different ways that a protein can fold and the different solid crystal structures of complex materials. This leads, however, to very slow simulations, which makes modelling a real challenge. This project will develop advanced simulation algorithms with highly accelerated dynamics, providing routes to materials discovery for green energy and the connections between disease and protein folding.

Background:

This project is based on event-chain Monte Carlo [1]. This rejection-free Monte Carlo algorithm advances particles along ballistic-style trajectories, without the constraints of physical dynamics. This leads to significant freedom when choosing the particle dynamics, with certain choices recently shown to circumvent analogous challenges in a foundational model [2]. The recent development of rejection-free approaches to tempering [3] creates further opportunities to accelerate exploration of different crystal structures (in solid materials) and folding configurations (in proteins). This project will develop and unify these two key techniques in the context of complex materials and protein folding. We also aim to develop AI frameworks to characterise the power of these techniques.

Project outcomes:

This project brings together expertise from the Warwick Centre for Predictive Modelling with world-leading Monte Carlo experts in Warwick Statistics. Its principal outcome will be an advanced computational technique for simulating so-called multimodal systems such as complex materials with different crystalline phases and protein folding – a foundational challenge with broad relevance across computational science and engineering. The development of this novel technique will lead to new software, which will be released open-source on GitHub with rigorous documentation. We anticipate publishing 2-3 peer-reviewed journal articles. The student will also present their work at conferences and workshops and we hope to discuss our outputs with experimentalists.

The skills acquired during this project will position you for careers in AI research, computational materials science, data science, national laboratories, the tech industry or academic research.

References:
[1] PRE 80, 056704 (2009)
[2] PRE 99, 043301 (2019)
[3] J. R. Stat. Soc. B: Stat. Methodol. 84, 321 (2022)

Informal enquiries to michael.faulkner@warwick.ac.uk are welcome.

Scholarship:

The award will cover the UK tuition fee level, plus a tax-free stipend, currently £21,805, paid at the prevailing UKRI rate for 3.5 years of full-time study. International candidates are welcome to apply, but must be able to cover the difference in the fee levels.

Eligibility:

Applicants should have a first or a strong 2:1 honours degree in physics, maths, statistics or similar. An MSc or similar is preferred. Experience with numerical programming in statistical physics and/or Bayesian computation is desirable.

How to apply:

Candidates should submit an expression of interest by sending a CV and supporting statement outlining their skills and interests in this research area via the above 'Apply' button. If this initial application is successful, we will invite you to submit a formal application.

Advert information