Atomistic molecular dynamics (MD) simulations, in principle, allow calculating materials properties at different thermodynamic conditions. However, first-principles based MD simulations are computationally demanding, and reaching the time scales required for thermodynamic equilibration is often not achievable in practice.
We are looking for a summer student to investigate whether machine-learning can be used to extract knowledge from short non-equilibrium MD simulations that allows predicting equilibrium properties. A concrete example of such a thermodynamic quantity is the heat capacity, which is of critical importance for the computational prediction of phase diagrams.
As a summer student, you will learn the basics of atomistic MD simulations and machine-learning techniques such as Gaussian Process regression and tree-based models.
Direct Supervisor: Alexander Urban
Position dates: 6/1/2020 - 8/31/2020
Hours per Week: 35 Hours
Qualifications: Experience in Python programming is required, and basic knowledge of thermodynamics and/or statistical mechanics would be useful.
Eligibility: SEAS only