Search papers, labs, and topics across Lattice.
This paper examines the distinct computational challenges and accuracy needs in electronic and nuclear density functional theory (DFT) that can be addressed through machine learning (ML) and artificial intelligence (AI). It contrasts the efficacy of deep neural networks for Kohn-Sham (KS) DFT with the potential of symbolic regression in constructing kinetic energy functionals (KEFs) for both electronic and nuclear DFT. The findings suggest that while deep learning is effective for KS DFT, symbolic regression may offer conceptual advantages that have yet to be fully exploited in the context of DFT applications.
Symbolic regression could unlock new efficiencies in kinetic energy functional construction, challenging the dominance of deep neural networks in DFT applications.
We overview similarities and, importantly, differences in computational bottlenecks and accuracy requirements that can be addressed with machine learning (ML) and artificial intelligence (AI) techniques in electronic and nuclear DFT. From these follow different promising methodological and algorithmic choices depending on whether one machine learns the exchange correlation (XC) functional, the kinetic energy functional (KEF), the density or the basis functions. In particular, while the popular deep neural networks remain a potent choice in the context of KS DFT, we highlight their disadvantages when building KEFs and highlight conceptual advantages - yet to be fully realized - of symbolic regression for both electronic and nuclear DFT. We point out promising approaches that can be carried from the more extensively investigated ML-enhanced electronic DFT to nuclear DFT.
