Machine learning electronic structure methods

Henkes T, Dral PO, Müller C (2026)

Publication Type: Book chapter / Article in edited volumes

Publication year: 2026

Edited Volumes: Handbook of Electronic Structure Theory

Series: Methods and Applications

Pages Range: 289-327

ISBN: 9780443265969

DOI: 10.1016/B978-0-443-26596-9.00009-0

Abstract

Machine learning (ML) revolutionized how we do science, most strikingly reflected by two Nobel prizes in natural sciences (physics and chemistry) awarded in the same year (2024) for ML-related research. Electronic structure (ES) methods and their predictions are no exception: ML revamps them by speeding them up and also, often, improving their accuracy. It is hard to find an area related to ES still untouched by ML. In this chapter, we cram together this vast and exciting field of ML ES methods to provide at least a starting point for the readers. We provide a cursory overview of the field and go into details in specific, key areas such as the mathematical background of a selection of the machine learning force fields (MLFFs). This chapter is the reflection of the authors' interests and is surely not aiming at the full coverage. The topics covered are MLFFs, including their underlying ML algorithms and descriptors, active learning for construction of data, universal models, methods for excited-state simulations, learning from multiple levels of theory (Δ-learning, transfer learning, etc.), overview of databases, universal models, ML-improved density functional theory, and others briefly mentioned. As is seen from the above overview, this chapter almost exclusively focuses on supervised learning but, of course, other ML approaches are useful and applied in ML ES too. In any case, in this rapidly developing field, this chapter might be outdated the moment it is printed; hence, after reading it and getting an initial overview, you might want to check out newer, online, resources.

Authors with CRIS profile

Carolin Müller Juniorprofessur für die Theorie elektronisch angeregter Zustände

Involved external institutions

University of Luxembourg LU Luxembourg (LU) Xiamen University CN China (CN)

How to cite

APA:

Henkes, T., Dral, P.O., & Müller, C. (2026). Machine learning electronic structure methods. In Handbook of Electronic Structure Theory. (pp. 289-327).

MLA:

Henkes, Tobias, Pavlo O. Dral, and Carolin Müller. "Machine learning electronic structure methods." Handbook of Electronic Structure Theory. 2026. 289-327.

BibTeX: Download