1 Theory and Simulations of Materials, EPFL, Lausanne, Switzerland
2 PSI Centre for Scientific Computing, Theory, and Data, Paul Scherrer Institut, Villigen, Switzerland
3 Cavendish Laboratory, University of Cambridge, UK
EXTENDED ABSTRACT: The electronic structure of materials encodes their electronic, optical, magnetic and topological properties, and it is often discussed in terms of Bloch orbitals and band structures. Wannier functions provide an alternative, chemically intuitive representation; in particular, maximally localized Wannier functions - introduced close to 30 years ago - are widely used nowadays to calculate accurately and efficiently many properties of interest, from electron mobilities to BCS superconducting temperatures. Here, we first showcase two key algorithms that we recently developed (projectability disentanglement [1] and manifold remixing [2]) that allow to fully automate the calculation of Wannier functions that represent the most compact and most precise Hamiltonian describing the occupied and lowest unoccupied bands, and that can be mixed into desired target manifolds (e.g., the valence bands of insulators, or selected correlated bands). Starting from our MC3D database of experimentally known stoichiometric inorganics (https://mc3d.materialscloud.org), we construct ~2M Wannier functions for 21,737 materials and for a subset of 5032 that are band semiconductors or insulators. This allows us to identify high-performance thermoelectrics, discover candidate materials with large non-linear Hall response, and design heterostructures that support a two-dimensional electron gas at the interface. While these explorations are based on Kohn-Sham density-functional theory, I will highlight the next foundational step that delivers spectral functionals and spectral theories. Namely, I’ll discuss orbital-density-dependent functionals that describe the correct band structure and band gap [4], and functionals dependent on the spectral density that can also describe satellites and correlated-electron materials [5,6].
Keywords：computational materials discovery, Wannier functions, spectral functionals
Figure 1. 56 of the ~2M maximally localized Wannier functions constructed in this work.
References
[1] J. Qiao, G. Pizzi, and N. Marzari, Projectability disentanglement for accurate and automated electronic-structure Hamiltonians, npj Comput. Mater. 9, 208 (2023).
[2] J. Qiao, G. Pizzi, and N. Marzari, Automated mixing of maximally localized Wannier functions into target manifolds, npj Comput. Mater. 9, 206 (2023).
[3] J. Qiao, G. Pizzi, and N. Marzari, The electronic structure genome of inorganic crystals, in preparation (2025).
[4] E. Linscott, N. Colonna, R. De Gennaro, N. L. Nguyen, G. Borghi, A. Ferretti, I. Dabo, and N. Marzari, koopmans: an open-source package for accurately and efficiently predicting spectral properties with Koopmans functionals, Journal of Chemical Theory and Computation 19, 7097 (2023).
[5] T. Chiarotti, A. Ferretti, and N. Marzari, Energies and spectra of solids from the algorithmic inversion of dynamical Hubbard functionals, Physical Review Research 6, L032023 (2024).
[6] A. Ferretti and N. Marzari, Functional theory of the occupied spectral density, arXiv:2508.17245 (2025).
Nicola Marzari holds the Chair of Theory and Simulation of Materials at EPFL, where he also directs the National Centre on Computational Design and Discovery of Novel Materials MARVEL. He heads the Laboratory for Materials Simulations at the Paul Scherrer Institut and holds an Excellence Chair at the University of Bremen. He took up the post of Cavendish Professor of Physics at the University of Cambridge (UK) in 2025, transitioning there in full in 2026. Previously, he held the Toyota Chair for Materials Processing at the Massachusetts Institute of Technology and was the inaugural Statutory Chair of Materials Modelling at the University of Oxford (UK). He received a Laurea in Physics (1992) from the University of Trieste and a PhD in Physics (1996) from the University of Cambridge. His research is dedicated to the development and application of electronic-structure simulations to understand, predict, and design the properties and performance of novel materials and devices.