A Transferable Machine Learning Approach to Predict Optimized Orbitals for Electronic Structure Problems
Lucas van der Horst 1 , Maniraman Periyasamy 1 , Abhishek Y. Dubey 1 , Davide Bincoletto 2 , Jakob S. Kottmann 2 , and Daniel D. Scherer 1 · 2026
Read A Transferable Machine Learning Approach to Predict Optimized Orbitals for Electronic Structure Problems by Lucas van der Horst 1 , Maniraman Periyasamy 1 , Abhishek Y. Dubey 1 , Davide Bincoletto 2 , Jakob S. Kottmann 2 , and Daniel D. Scherer 1 (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-04174.txt.
Description
Lucas van der Horst and five co-authors train a graph neural network to predict molecular orbitals for electronic structure work. That replaces a costly classical step. Variational quantum eigensolver ansätze depend heavily on how well the molecular orbital basis captures electron correlation. Yet computing those coefficients classically is costly. It must be repeated for every molecular geometry. That bottleneck limits scaling across chemical space.
Their network predicts them from geometry and pairwise bonding. It was trained on small hydrogen systems across tens of thousands of shapes. The model transferred without retraining to larger unseen hydrogen chains, well beyond its training range in system size. Predicted orbitals yield energies close to the full classical result. Errors are larger on structured than on random cases. They can seed further quantum work.
Catalog record
| CREATOR | Lucas van der Horst 1 , Maniraman Periyasamy 1 , Abhishek Y. Dubey 1 , Davide Bincoletto 2 , Jakob S. Kottmann 2 , and Daniel D. Scherer 1 |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 52 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.04174