Uncertainty-aware deep learning for the Lyα forest: CNN-based absorber detection and characterization
Monthly Notices of the Royal Astronomical Society · 05.11.2026 · 553 (2)
DOI: 10.1093/mnras/stag2027 ↗Kurz gefasst
Fachartikel in der Zeitschrift Monthly Notices of the Royal Astronomical Society, online erschienen am 23.10.2026. Verfasst von Paryag Sharma und 4 weiteren Autor:innen. Die Arbeit nennt 70 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: University Grants Commission, Government of India.
Abstract (Originalsprache)
The Lyman-$\alpha$ forest is a powerful probe of the intergalactic medium and small-scale matter distribution, but deriving absorber properties traditionally requires computationally expensive Voigt-profile fitting. We present a convolutional neural network (CNN) that identifies and characterizes H i Ly$\alpha$ absorbers directly from quasar spectra. The model is trained on synthetic spectra generated from the IllustrisTNG simulation and fitted with the viper Voigt-profile fitting code to provide training labels. The network simultaneously predicts absorber presence, column density ($N_{\rm HI}$), Doppler parameter ($b_{\rm HI}$), and line centroid. On simulated spectra, the CNN reproduces the viper reference catalogue with an F1 score of $\sim 0.8$, with mean absolute errors of $\sim 0.18$ in $\log N_{\rm HI}$ and $\sim 0.10$ in $\log b_{\rm HI}$. It reproduces the H i column density distribution function (CDDF) and the $b_{\rm HI}$–$N_{\rm HI}$ relation obtained from viper, recovering similar CDDF slopes and a lower-envelope relation with an rms difference of only $0.36\, {\rm km\, s^{-1}}$. Applied to high-resolution UVES spectra, performance decreases to an F1 score of $\sim 0.5$, with mean absolute errors of $\sim 0.35$ in $\log N_{\rm HI}$ and $\sim 0.22$ in $\log b_{\rm HI}$. Latent-space analysis reveals a significant domain shift between the simulated and observational spectra, contributing to the reduced performance. Nevertheless, the CNN reproduces the observed CDDF and $b_{\rm HI}$–$N_{\rm HI}$ distributions obtained from the viper reference catalogue, yielding CDDF slopes consistent with viper and a lower-envelope rms difference of $3.20\, {\rm km\, s^{-1}}$ relative to viper. Monte Carlo dropout is implemented during inference to quantify predictive uncertainties. Together with its computational efficiency, the method provides a scalable and uncertainty-aware framework for Ly$\alpha$ forest analysis in upcoming spectroscopic surveys.
Förderung
- University Grants Commission
- Government of India
Zitieren
APA
Sharma, P., Khaire, V., Cheng, T., Chand, H., & Gaikwad, P. (2026). Uncertainty-aware deep learning for the Lyα forest: CNN-based absorber detection and characterization. Monthly Notices of the Royal Astronomical Society, 553(2). https://doi.org/10.1093/mnras/stag2027
Vancouver
Sharma P, Khaire V, Cheng T, Chand H, Gaikwad P. Uncertainty-aware deep learning for the Lyα forest: CNN-based absorber detection and characterization. Monthly Notices of the Royal Astronomical Society. 2026;553(2). doi:10.1093/mnras/stag2027
BibTeX
@article{Sharma2026,
author = {Sharma, Paryag and Khaire, Vikram and Cheng, Ting-Yun and Chand, Hum and Gaikwad, Prakash},
title = {Uncertainty-aware deep learning for the Lyα forest: CNN-based absorber detection and characterization},
journal = {Monthly Notices of the Royal Astronomical Society},
year = {2026},
volume = {553},
number = {2},
doi = {10.1093/mnras/stag2027}
}
EUVIMED