Transferability in machine-learned density functionals

Stefan Vuckovic, Heng Zhao, Elias Polak

The development of density functional approximations stands at a crossroads: while machine-learned functionals show potential to surpass their human-designed counterparts, their transferability to unseen chemistry lags behind. I will address several aspects of transferability of density functionals, focusing on how to define it [1] and how to solve the transferability issues in machine-learned functionals [2].

1. H. Zhao, T. Gould, S. Vuckovic, Phys. Chem. Chem. Phys. 2024, (accepted), 10.1039/D4CP00878B.
2. S. Vuckovic, T. Gould, B. Chang, S. Dale, ChemRXiv 2024, 10.26434/chemrxiv-2023-5075x-v2.