TwinSpecNet: Extending APOGEE's chemical reach to low-S/N spectra via empirical paired learning
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917447114686464 |
|---|---|
| author | Sun, Weijia Chiappini, Cristina Nepal, Samir |
| author_facet | Sun, Weijia Chiappini, Cristina Nepal, Samir |
| contents | Large spectroscopic surveys rely on automated pipelines to deliver homogeneous stellar labels, but a substantial fraction of observations are at low signal-to-noise ratio (S/N), where label estimates become imprecise or are omitted. In APOGEE, these low-S/N spectra visits sample faint and distant populations -- the bulge, outer halo, and satellite systems -- yet still encode recoverable chemical information. We present TwinSpecNet (TSN), a paired-learning framework that exploits APOGEE's multi-visit observing strategy: by training on empirical low-/high-S/N spectral twins of the same stars, TSN learns to suppress stochastic noise while preserving the ASPCAP label scale. TSN employs a Vision Transformer encoder with dual objectives: reconstructing high-S/N flux from low-S/N visits and predicting stellar parameters and abundances with calibrated uncertainties. TSN reduces label scatter relative to visit-level ASPCAP for S/N<60 visits. TSN reproduces the ASPCAP scale with residual scatters of $σ$< 19 K in $T_{\mathrm{eff}}$, $σ\sim$0.06 dex in $\log g$, and $σ\sim$0.03 dex in Fe/H. TSN tightens intra-cluster abundance dispersions, recovers cleaner chemical sequences in inner-disk and bulge and satellite samples, and improves C/N-based age precision for APOKASC giants from 1.70 to 1.59 Gyr. By learning survey-specific noise patterns from repeated observations, TSN demonstrates how empirical paired learning can extend the chemical reach of existing spectroscopic data, providing a template applicable to other multi-visit surveys. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2604_26491 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | TwinSpecNet: Extending APOGEE's chemical reach to low-S/N spectra via empirical paired learning Sun, Weijia Chiappini, Cristina Nepal, Samir Astrophysics of Galaxies Large spectroscopic surveys rely on automated pipelines to deliver homogeneous stellar labels, but a substantial fraction of observations are at low signal-to-noise ratio (S/N), where label estimates become imprecise or are omitted. In APOGEE, these low-S/N spectra visits sample faint and distant populations -- the bulge, outer halo, and satellite systems -- yet still encode recoverable chemical information. We present TwinSpecNet (TSN), a paired-learning framework that exploits APOGEE's multi-visit observing strategy: by training on empirical low-/high-S/N spectral twins of the same stars, TSN learns to suppress stochastic noise while preserving the ASPCAP label scale. TSN employs a Vision Transformer encoder with dual objectives: reconstructing high-S/N flux from low-S/N visits and predicting stellar parameters and abundances with calibrated uncertainties. TSN reduces label scatter relative to visit-level ASPCAP for S/N<60 visits. TSN reproduces the ASPCAP scale with residual scatters of $σ$< 19 K in $T_{\mathrm{eff}}$, $σ\sim$0.06 dex in $\log g$, and $σ\sim$0.03 dex in Fe/H. TSN tightens intra-cluster abundance dispersions, recovers cleaner chemical sequences in inner-disk and bulge and satellite samples, and improves C/N-based age precision for APOKASC giants from 1.70 to 1.59 Gyr. By learning survey-specific noise patterns from repeated observations, TSN demonstrates how empirical paired learning can extend the chemical reach of existing spectroscopic data, providing a template applicable to other multi-visit surveys. |
| title | TwinSpecNet: Extending APOGEE's chemical reach to low-S/N spectra via empirical paired learning |
| topic | Astrophysics of Galaxies |
| url | https://arxiv.org/abs/2604.26491 |