Machine-learning-based molecular modelling of nanoscale geological processes
Describing geological processes on the nanoscale is challenging, partly because current molecular dynamics force fields for geologically relevant materials work well only within a moderate range of conditions. In this project, we aim at developing a systematic machine-learning based method for tailoring force fields to geologically relevant simulation conditions, to study the emergent properties robustly.
Geological processes shape the earth, erect mountains and shatter the crust during earthquakes. Many geological processes ultimately involve details at the nanoscale. For example, in reactions where rocks expand in the presence of water, atomic scale details may determine whether water can make it to the reaction site, or whether the reaction is shut off.
In this project, we will use machine learning to tailor models to specific nanoscale geological processes. We will traverse the traditionally infeasible parameter spaces of interatomic force fields and adapt them to the problem at hand. For example, if we want to investigate the properties of a water film between silica surfaces at extreme pressures, we train the potential to describe just that, using potentially both empirical and more fundamental calculated properties.
You will learn to use machine learning for optimizing interatomic force fields. You will also learn to run high-throughput atomistic simulations to incrementally improve simulations and analyses using various machine learning techniques. These skills are useful both in an academic and an industrial setting.
The project may include collaborations with the Center for Advanced Computing and Simulations at University of Southern California.
- MSc in physics or a mathematically focused MSc in geosciences.
- Candidates with documented experience in statistical or computational physics, scientific programming, molecular dynamics simulations, and experience from machine learning will be prioritized.
Call 1: Project start autumn 2021
This project is in call 1, starting autumn 2021. Read about how to apply