Automated AFGL quantum number assignment for CO2 isotopologues using a graph neural network
Abstract
Accurate quantum number assignment for calculated molecular energy levels is a critical bottleneck in generating line broadening parameters for comprehensive line lists for radiative transfer applications. We present an automated pipeline for assigning Air Force Geophysics Laboratory (AFGL) quantum numbers to CO2 calculated rovibrational states lying below 15,000cm−1 across all 12 stable isotopologues. A GraphSAGE graph neural network is trained transductively on empirical (MARVEL) energy levels, exploiting inter-isotopologue perturbation chains and intra-isotopologue rotational ladder edges to propagate assignment information to unlabelled calculated states. Physical uniqueness is enforced locally by a Hungarian algorithm solver operating within groups of states sharing the same polyad, rotational quantum number, and parity. A five-generation bootstrap loop iteratively promotes high-confidence predictions into the training set, expanding coverage without additional labelling effort. The pipeline assigns 224,650 previously unlabelled states over 12 isotopologues, accounting for 10.7% of all available states (including MARVEL-derived levels), with coverage now increased to 97.4% below 5000cm−1. The pipeline includes a novel decision tree method for converting AFGL to Herzberg notation in asymmetric isotopologues, while the architecture and Hungarian uniqueness enforcement are applicable beyond CO2, any molecular system with a conserved polyad-like quantum number and a large body of unlabelled computed states is a natural target for this approach, suggesting a pathway toward automated quantum number annotation for the next generation of large-scale computed line lists.
Versions
Related
- A Tutorial on Automated Mass Spectral Analysis Using NucleoSpec for Compositional Assignment of Nucleic Acid-Silver Complexes and Nanoclusters.
- Ab initio phasing based on topological restraints: automated determination of the space group and the number of molecules in the unit cell
- Layer-number-controlled Fe-tea polyphenol metal-phenolic network-coated PLCL/gelatin electrospun membranes for active preservation of yellow sweet cherries
- Correlation between normal character and Raman intensities of isotopologues of carbon dioxide
- Mono-substituted isotopologues of HC₃N: characterization of the resonance systems for the ¹³C-and ¹⁵N-species
BibTeX
@misc{barnfield2026automated,
title = {{Automated AFGL quantum number assignment for CO$_2$ isotopologues using a graph neural network}},
author = {Marco G. Barnfield and Sergei N. Yurchenko and Jonathan Tennyson},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.13803},
eprint = {2609.13803},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.13803},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Automated AFGL quantum number assignment for CO_2 isotopologues using a graph neural network
AU - Marco G. Barnfield
AU - Sergei N. Yurchenko
AU - Jonathan Tennyson
PY - 2026
DA - 2026/09/12
PB - arXiv
DO - 10.48550/arxiv.2609.13803
UR - https://arxiv.org/abs/2609.13803
AB - Accurate quantum number assignment for calculated molecular energy levels is a critical bottleneck in generating line broadening parameters for comprehensive line lists for radiative transfer applications. We present an automated pipeline for assigning Air Force Geophysics Laboratory (AFGL) quantum numbers to CO2 calculated rovibrational states lying below 15,000cm^−1 across all 12 stable isotopologues. A GraphSAGE graph neural network is trained transductively on empirical (MARVEL) energy levels, exploiting inter-isotopologue perturbation chains and intra-isotopologue rotational ladder edges to propagate assignment information to unlabelled calculated states. Physical uniqueness is enforced locally by a Hungarian algorithm solver operating within groups of states sharing the same polyad, rotational quantum number, and parity. A five-generation bootstrap loop iteratively promotes high-confidence predictions into the training set, expanding coverage without additional labelling effort. The pipeline assigns 224,650 previously unlabelled states over 12 isotopologues, accounting for 10.7% of all available states (including MARVEL-derived levels), with coverage now increased to 97.4% below 5000cm^−1. The pipeline includes a novel decision tree method for converting AFGL to Herzberg notation in asymmetric isotopologues, while the architecture and Hungarian uniqueness enforcement are applicable beyond CO2, any molecular system with a conserved polyad-like quantum number and a large body of unlabelled computed states is a natural target for this approach, suggesting a pathway toward automated quantum number annotation for the next generation of large-scale computed line lists.
ER -