PEACE: Covariant learning of nonadiabatic manifolds with parity-resolved Hamiltonians
Abstract
Nonadiabatic molecular dynamics provides mechanistic insight into light-driven processes and informs the design of molecules and materials for solar energy conversion, photocatalysis and photo switching. Accurately describing these processes requires a representation that respects electronic symmetry and consistently relates energies to interstate couplings. Here we introduce PEACE, which combines a parity-equivariant latent Hamiltonian with a learned electronic connection. Controlled ablations reveal the complementary roles of symmetry-allowed state mixing and electronic-frame variation in reproducing crossing structures and relaxation dynamics. PEACE closely reproduces excited-state population dynamics from first-principles simulations, while its extension to spin-orbit coupling enables simulations of intersystem crossing. These results demonstrate that a more complete incorporation of the underlying physics into learned electronic representations leads to more accurate predictions of nonadiabatic dynamics.
Related
- Active-Space Selection in Nonadiabatic Dynamics Simulations of Dense Manifolds of Excited States: When Local Diabatization Needs Great Care
- Nonadiabatic excited-state dynamics with quantum Monte Carlo-trained machine learning: azomethane as a stringent test
- Machine Learning Hamiltonians with Polarizations for Heterogeneous Systems
- Chemistry-Informed Transfer and Interaction-Resolved Learning Towards Data-Efficient Materials Exploration
- WISP: Learning spin-resolved electron localization from atomic environments
BibTeX
@misc{gao2026peace,
title = {{PEACE: Covariant learning of nonadiabatic manifolds with parity-resolved Hamiltonians}},
author = {Rongzhi Gao and Shuguang Chen and Yang Zhou and GuanHua Chen and Ziyang Hu and ChiYung Yam},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2610.09576},
eprint = {2610.09576},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2610.09576},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - UNPB
TI - PEACE: Covariant learning of nonadiabatic manifolds with parity-resolved Hamiltonians
AU - Rongzhi Gao
AU - Shuguang Chen
AU - Yang Zhou
AU - GuanHua Chen
AU - Ziyang Hu
AU - ChiYung Yam
PY - 2026
DA - 2026/10/07
PB - arXiv
DO - 10.48550/arxiv.2610.09576
UR - https://arxiv.org/abs/2610.09576
AB - Nonadiabatic molecular dynamics provides mechanistic insight into light-driven processes and informs the design of molecules and materials for solar energy conversion, photocatalysis and photo switching. Accurately describing these processes requires a representation that respects electronic symmetry and consistently relates energies to interstate couplings. Here we introduce PEACE, which combines a parity-equivariant latent Hamiltonian with a learned electronic connection. Controlled ablations reveal the complementary roles of symmetry-allowed state mixing and electronic-frame variation in reproducing crossing structures and relaxation dynamics. PEACE closely reproduces excited-state population dynamics from first-principles simulations, while its extension to spin-orbit coupling enables simulations of intersystem crossing. These results demonstrate that a more complete incorporation of the underlying physics into learned electronic representations leads to more accurate predictions of nonadiabatic dynamics.
ER -