Manuel Dillenz and Jose Maria Castillo Robles at DTU Energy have leveraged LUMI to investigate electron transport through battery materials at the atomic level. Their approach integrates quantum mechanical simulations, molecular dynamics, and machine learning into a single analytical framework designed to evaluate charge transport characteristics in lithium manganese oxide batteries with both precision and computational efficiency.
The supercomputer's CPU and GPU capabilities proved critical to the endeavour, supplying the computational power needed to generate simulation datasets, construct machine learning models, and execute computationally intensive calculations. The LUMI user support team also played a key role, assisting the researchers in optimising their methodology for LUMI's AMD-based processing architecture.
Looking ahead, Dillenz and Castillo Robles intend to release their workflows and trained machine learning models to the scientific community as open resources. This move would allow researchers across the field to adapt and apply the methodology to diverse battery research applications beyond the specific material examined in this study.
