Anton Beiersdorfer | Searcharxiv

arXiv2026

Adaptive Slater Koster Parameters: Crossing Oxidation States with Density Functional Tight Binding

We propose to adapt the confined pseudo-atomic orbitals underpinning the precalculated Slater-Koster (SK) interaction tables in Density Functional Tight Binding (DFTB) to local atomic environments. We demonstrate significant improvement in electronic structure and energetics in the application to a partially oxidized Ni surface and Li insertion into graphite, where we assign optimal SK parameters to metal atoms in different oxidation states. Further analysis reveals the smoothness of the SK integrals across the varying oxidation states. Exploiting this, we introduce a site-resolved machine-learning scheme for fully adaptive DFTB. Using atomic descriptors and simple regression architectures already established in the context of machine-learning interatomic potentials, our scheme achieves 95% band-structure accuracy across all Ni-O binary compositions in the Materials Project.

cond-mat.mtrl-sci↗

arXiv2024

Broad Yet Narrow: Super-resolution techniques to simulate electronic spectra of large molecular systems

An accurate treatment of electronic spectra in large systems with a technique such as time dependent density functional theory (TDDFT) is computationally challenging. Due to the Nyquist sampling theorem, direct real time simulations must be prohibitively long in order to recover a suitably sharp resolution in frequency space. Super-resolution techniques such as compressed sensing and MUSIC assume only a small number of excitations contribute to the spectrum, which fails in large molecular systems where the number of excitations is typically very large. We present a new approach that combines exact short time dynamics with approximate frequency space methods to capture large narrow features embedded in a dense manifold of smaller nearby peaks. We show that our approach can accurately capture narrow features and broad quasi-continuum of states at the same time - even when the features overlap in frequency. Our approach is able reduce the required simulation time by a factor of 20-40 with respect to standard Fourier analysis and shows promise for the accurate whole-spectrum prediction of large molecules and materials.

cond-mat.mtrl-sci↗