A universal machine learning model for the electronic density of states

TL;DR

This paper introduces PET-MAD-DOS, a universal transformer model capable of predicting electronic density of states across diverse materials, enabling efficient electronic structure analysis with minimal fine-tuning.

Contribution

The authors develop a universal machine learning model for electronic density of states using a transformer architecture trained on a large diverse dataset, capable of generalization and fine-tuning.

Findings

Accurately predicts DOS for diverse materials.

Achieves semi-quantitative agreement with system-specific models.

Fine-tuning improves performance with minimal data.

Abstract

In the last few years several ``universal'' interatomic potentials have appeared, using machine-learning approaches to predict energy and forces of atomic configurations with arbitrary composition and structure, with an accuracy often comparable with that of the electronic-structure calculations they are trained on. Here we demonstrate that these generally-applicable models can also be built to predict explicitly the electronic structure of materials and molecules. We focus on the electronic density of states (DOS), and develop PET-MAD-DOS, a rotationally unconstrained transformer model built on the Point Edge Transformer (PET) architecture, and trained on the Massive Atomistic Diversity (MAD) dataset. We demonstrate our model's predictive abilities on samples from diverse external datasets, showing also that the DOS can be further manipulated to obtain accurate band gap predictions. A fast evaluation of the DOS is especially useful in combination with molecular simulations probing matter in finite-temperature thermodynamic conditions. To assess the accuracy of PET-MAD-DOS in this context, we evaluate the ensemble-averaged DOS and the electronic heat capacity of three technologically relevant systems: lithium thiophosphate (LPS), gallium arsenide (GaAs), and a high entropy alloy (HEA). By comparing with bespoke models, trained exclusively on system-specific datasets, we show that our universal model achieves semi-quantitative agreement for all these tasks. Furthermore, we demonstrate that fine-tuning can be performed using a small fraction of the bespoke data, yielding models that are comparable to, and sometimes better than, fully-trained bespoke models.