My work sits between electronic-structure theory and machine learning.
Fine-tuning GRACE and ACE-type potentials on curated VASP MLFF datasets, then validating them against DFT energies, forces and phonon spectra. The goal is ab-initio accuracy at a cost that scales to systems classical DFT cannot reach.
Running and scaling these calculations on national and European supercomputers: Slurm job pipelines, containerised toolchains, and GPU workflows for both training and production molecular dynamics.
Quantum dynamics in symmetry-breaking states of correlated electrons: Antiferromagnetic phase
Deep learning frameworks for exploration of energy landscapes of perovskites
Transition metals: Matching microscopic quantum dynamics with macroscopic magnetic order
The quantum vortex states in extended Bose–Hubbard model: effects of lattice geometries, inter-particle interactions and spatial inhomogeneity