Machine Learning for Electronic Structure

Machine learning has become a powerful tool to replace or accelerate electronic-structure calculations, for example by predicting interatomic potentials with high accuracy and low cost. Increasingly, ML is also integrated into modern quantum chemistry and electronic-structure theory, enhancing algorithms that describe the quantum behavior of electrons in molecules and materials.

This special issue highlights three complementary themes:

Learning the ingredients of electronic-structure calculations: Contributions predicting electron densities, Hamiltonian elements, or other effective one- and two-body operators from atomic configurations, while exploiting symmetries and uncertainty quantification, are encouraged.

ML-accelerated many-body methods: Studies embedding ML models into wave-function theories—from coupled-cluster and multireference to variational and diffusion Monte Carlo—aiming to reduce scaling, extend reach, or improve accuracy.

Data-driven density functionals: Manuscripts developing and testing exchange–correlation functionals, kinetic-energy functionals, or orbital-free models with supervised, unsupervised, or reinforcement learning are welcome.