Saved in:
Bibliographic Details
Main Author: Xiao, Jin
Format: Recurso digital
Language:
Published: Zenodo 2026
Subjects:
Online Access:https://doi.org/10.5281/zenodo.19155250
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866901892555079680
author Xiao, Jin
author_facet Xiao, Jin
contents <p>Data-driven methods can reproduce the accuracy of high-level quantum methods with reduced computational costs by learning features from low-level electronic structure calculations. We introduce the Deep Atomic Density-Based Tight-Binding (DeePaTB) framework, a novel machine-learning semiempirical quantum mechanical (MLSQM) framework built on our atomic density-based tight-binding (aTB) method. DeePaTB generates the electronic structure feature “eigenvalue of the local density matrix” computed using Amesp software as inputs to a deep model that predicts energies and related properties. Across diverse chemical systems, DeePaTB attains density functional theory (DFT)-level accuracy while retaining SQM computational effi ciency. This is applicable to closed-shell and open-shell species and systems under external electric fi elds, demonstrating the strong transferability of DeePaTB. Together, these results establish DeePaTB as a general, scalable approach that eff ectively balances accuracy and efficiency for large and complex quantum chemical computations.</p>
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_19155250
institution Zenodo
language
publishDate 2026
publisher Zenodo
record_format zenodo
spellingShingle DeePaTB: A Deep Learning-Powered Semi-Empirical Quantum Mechanical Method
Xiao, Jin
Machine learning
<p>Data-driven methods can reproduce the accuracy of high-level quantum methods with reduced computational costs by learning features from low-level electronic structure calculations. We introduce the Deep Atomic Density-Based Tight-Binding (DeePaTB) framework, a novel machine-learning semiempirical quantum mechanical (MLSQM) framework built on our atomic density-based tight-binding (aTB) method. DeePaTB generates the electronic structure feature “eigenvalue of the local density matrix” computed using Amesp software as inputs to a deep model that predicts energies and related properties. Across diverse chemical systems, DeePaTB attains density functional theory (DFT)-level accuracy while retaining SQM computational effi ciency. This is applicable to closed-shell and open-shell species and systems under external electric fi elds, demonstrating the strong transferability of DeePaTB. Together, these results establish DeePaTB as a general, scalable approach that eff ectively balances accuracy and efficiency for large and complex quantum chemical computations.</p>
title DeePaTB: A Deep Learning-Powered Semi-Empirical Quantum Mechanical Method
topic Machine learning
url https://doi.org/10.5281/zenodo.19155250