Latent unified smooth Hamiltonians for excited state chemistry
Abstract
We describe a neural network architecture and training procedure designed to model electronic ground and excited states of arbitrary molecular systems. By indirectly learning a latent, implicit basis representation of the electronic-state Hamiltonian, the model offers a unified treatment of multiple electronic states, conical intersections, and non-adiabatic couplings. The formalism can be further extended to learn consistent latent representations of additional operators such as transition dipole moments, for example. To demonstrate the general capabilities of our architecture, we train and evaluate networks on two realistic photochemical systems, thymine and azobenzene. The resulting models accurately reproduce energies and oscillator strengths for the ground- and low-lying excited states relevant to the photochemistry of these systems. We highlight the performance of the trained networks by studying critical molecular geometries, including conical intersections and excited state minima. By construction, the proposed framework also recovers the emergence of Berry phase accumulation around conical intersections. By pairing key mathematical structure from quantum chemistry with the representation learning power of transformers, the presented architecture offers a qualitatively new path toward fast and accurate ground- and excited-state simulations.
Related
- From Peroxides to Qubits: Using Glow-Stick Chemistry to Introduce Excited States, Conical Intersections, and Quantum Computing to Undergraduates
- Unified adiabatic and diabatic excited-state description via the ensemble-variational quantum eigensolver
- Soft foam polyurethane recycling and foaming in reactive extrusion using partial aminolysis and a latent cross-linking chemistry
- Aromatic ring overlap affects excited-state properties of stacked nucleobases
- Active-Space Selection in Nonadiabatic Dynamics Simulations of Dense Manifolds of Excited States: When Local Diabatization Needs Great Care
BibTeX
@misc{juergens2026latent,
title = {{Latent unified smooth Hamiltonians for excited state chemistry}},
author = {David Juergens and Martin Stöhr and Andreas E. Hillers-Bendtsen and O. Jonathan Fajen and Todd J. Martínez},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.01871},
eprint = {2609.01871},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.01871},
}
RIS
TY - UNPB
TI - Latent unified smooth Hamiltonians for excited state chemistry
AU - David Juergens
AU - Martin Stöhr
AU - Andreas E. Hillers-Bendtsen
AU - O. Jonathan Fajen
AU - Todd J. Martínez
PY - 2026
DA - 2026/09/01
PB - arXiv
DO - 10.48550/arxiv.2609.01871
UR - https://arxiv.org/abs/2609.01871
AB - We describe a neural network architecture and training procedure designed to model electronic ground and excited states of arbitrary molecular systems. By indirectly learning a latent, implicit basis representation of the electronic-state Hamiltonian, the model offers a unified treatment of multiple electronic states, conical intersections, and non-adiabatic couplings. The formalism can be further extended to learn consistent latent representations of additional operators such as transition dipole moments, for example. To demonstrate the general capabilities of our architecture, we train and evaluate networks on two realistic photochemical systems, thymine and azobenzene. The resulting models accurately reproduce energies and oscillator strengths for the ground- and low-lying excited states relevant to the photochemistry of these systems. We highlight the performance of the trained networks by studying critical molecular geometries, including conical intersections and excited state minima. By construction, the proposed framework also recovers the emergence of Berry phase accumulation around conical intersections. By pairing key mathematical structure from quantum chemistry with the representation learning power of transformers, the presented architecture offers a qualitatively new path toward fast and accurate ground- and excited-state simulations.
ER -