Machine learning allows accurate electronic structure calculations to be made at large scales in material modeling
The electronic structure of matter is a key factor in both fundamental and applied research. Examples include drug design, energy storage, and the arrangement of electrons within atoms. The lack of a simulation method that is both high-fidelity and scalable across time and lengthscales has been a major roadblock to the development of these technologies.
Researchers at the Helmholtz-Zentrum Dresden-Rossendorf in Gorlitz (HZDR), Germany, and Sandia National Laboratories, Albuquerque (New Mexico), U.S.A., have developed a machine-learning-based simulation technique that surpasses traditional electronic structure simulation methods.
The Materials Learning Algorithms software stack (MALA) allows access to previously inaccessible length scales. The journal npj Computational Materials published the work.
Source:
https://phys.org/news/2023-07-machine-enables-accurate-electronic-large.html
