TNFlow: Amortized Posterior Inference for Trans-Neptunian Object Surface Composition
Abstract
We present TNFlow, a transformer and normalizing flow architecture for inferring the surface composition of Trans-Neptunian Objects (TNOs) from their reflectance spectra. TNFlow is trained on synthetic spectra generated by the Shkuratov radiative transfer model to act as its inverse. TNFlow takes ∼0.7s to invert one spectrum on a single CPU core, returning a multimodal posterior over simplex-valid compositions and grain sizes. On synthetic spectra, the highest-weight mode achieves a mean total-variation distance of 0.149 from ground truth on the test split, and the model generalizes well to unseen combinations of known components. Qualitative tests on real JWST spectra show blindness or bias towards some materials. We suggest this could be attributed to either simulator fidelity or the training set.
Related
- MUSHY: Multimodal Flow-Based Amortized Bayesian Inference for Spectroscopic Data Fusion
- ENFR: Evidence Networks with Flow-Matched References for Amortized Simulation-Based Model Inference
- Misspecification in amortized X-ray spectral inference: a detection benchmark, a gain shift three schemes cannot see, and marginalizing it out
- Ganymede’s surface composition from 0.5 to 5.3 µm using combined VLT and JWST datasets and the resulting lessons learned
- Companion architectures of sub-Saturns: Distinct migration pathways across the Neptunian landscape
BibTeX
@misc{gaur2026tnflow,
title = {{TNFlow: Amortized Posterior Inference for Trans-Neptunian Object Surface Composition}},
author = {Agastya Gaur and Cristina M. Dalle Ore and Alessandra Ricca},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.04305},
eprint = {2609.04305},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.04305},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - TNFlow: Amortized Posterior Inference for Trans-Neptunian Object Surface Composition
AU - Agastya Gaur
AU - Cristina M. Dalle Ore
AU - Alessandra Ricca
PY - 2026
DA - 2026/09/03
PB - arXiv
DO - 10.48550/arxiv.2609.04305
UR - https://arxiv.org/abs/2609.04305
AB - We present TNFlow, a transformer and normalizing flow architecture for inferring the surface composition of Trans-Neptunian Objects (TNOs) from their reflectance spectra. TNFlow is trained on synthetic spectra generated by the Shkuratov radiative transfer model to act as its inverse. TNFlow takes ∼0.7s to invert one spectrum on a single CPU core, returning a multimodal posterior over simplex-valid compositions and grain sizes. On synthetic spectra, the highest-weight mode achieves a mean total-variation distance of 0.149 from ground truth on the test split, and the model generalizes well to unseen combinations of known components. Qualitative tests on real JWST spectra show blindness or bias towards some materials. We suggest this could be attributed to either simulator fidelity or the training set.
ER -