Predicting the critical temperature of a superconductor is a two-part challenge: determining how atoms vibrate (phonons) and how those vibrations scatter electrons (coupling). Traditionally, both are calculated using Density Functional Theory (DFT), which is computationally expensive for complex crystals or large supercells. In this project, you will develop a hybrid workflow using Machine Learning Interatomic Potentials (MLIPs) to handle the vibrational part of the problem. This allows for fast structural relaxation and efficient calculation of the phonon spectrum. You will then use DFT to calculate the electronic coupling. This workflow combines the speed of machine learning with the proven accuracy of quantum mechanical calculations. If time permits, the project can be extended to utilize Machine-Learned Hamiltonians (ML-Ham) to predict electronic eigenvalues and deformation potentials directly, potentially bypassing the explicit DFT coupling step.
Prerequisite Knowledge:
Recommended Courses to take during the master programme: