Xiang & Rix Subgiant Catalog

• The paper outlines the construction of a subgiant catalog using Gaia astrometry and LAMOST DR7 spectroscopy with isochrone fitting to deliver precise ages.
• It employs a consistent age methodology anchored by key parameters including Tₑff, [Fe/H], and [α/Fe] with median fractional uncertainties around 5–10%.
• The catalog supports Galactic archaeology and cosmological inference, validated via wide binaries and rigorous quality cuts.

Searching arXiv for the Xiang & Rix stellar sample and its uses. The Xiang & Rix sample is a catalog of Milky Way subgiants constructed from Gaia astrometry plus LAMOST DR7 spectroscopy, with stellar ages inferred from spectroscopically anchored isochrone fitting. In the literature summarized here, it is used both as a precision age catalog for Galactic archaeology and as a large old-star sample for cosmological inference. Its defining observables are $T_{\rm eff}$, $M_K$, $[\mathrm{Fe/H}]$, and $[\alpha/\mathrm{Fe}]$, together with Gaia+2MASS photometry, and later work treats its full age likelihoods as the basis for reconstructing the oldest tail of the Galactic age distribution (Shariat et al., 9 Oct 2025, Banik et al., 1 Jul 2026).

1. Definition and scope

The sample is explicitly a subgiant catalog, not a giant-star catalog. In the wide-binary validation study, Xiang & Rix (2022) is described as a catalog of subgiants constructed from Gaia astrometry plus LAMOST DR7 spectroscopy, with the spectroscopy supplying the key chemical and atmospheric information $T_{\rm eff}$, $[\mathrm{Fe/H}]$, and $[\alpha/\mathrm{Fe}]$. The parent sample was selected via an HR-diagram cut intended to isolate subgiants, and the stars included in that validation are all subgiants (Shariat et al., 9 Oct 2025).

In the cosmological application, the Xiang & Rix sample is the starting point for an old-star analysis based on 247,103 stars after removing one duplicate entry. Those stars are described as lying in the subgiant region of the $(T_{\rm eff}, M_K)$ plane, where ages are much more diagnostically constrained than on the unevolved main sequence. After additional quality and chemistry-based cuts tailored to the oldest-star problem, the nominal analysis retains 155,600 stars within 5 kpc (Banik et al., 1 Jul 2026).

2. Observables, isochrones, and age inference

The age methodology, as summarized in the validation literature, is an isochrone fit to the observables

$$T_{\rm eff},\quad M_K,\quad [\mathrm{Fe/H}],\quad [\alpha/\mathrm{Fe}],$$

together with Gaia+2MASS photometry. Xiang & Rix fit these observables to a grid of Yonsei–Yale ($Y^2$) isochrones, so the published ages are isochrone ages anchored by spectroscopic chemistry and atmospheric parameters, with luminosity information from the $M_K$0-band absolute magnitude and Gaia distance information. The same literature emphasizes that accurate chemistry, especially $M_K$1, is critical: even a $M_K$2 offset of 0.20 dex can shift inferred subgiant isochrone ages by $M_K$3 Gyr (Shariat et al., 9 Oct 2025).

A later use of the sample exploits the fact that the published stellar age likelihoods extend up to 20 Gyr. In that work, each star is associated with a full per-star age likelihood $M_K$4 on a 140-knot grid from 0.1 to 20 Gyr, with ages still understood as coming from the YY-based Xiang & Rix pipeline rather than being recomputed from scratch. This weak truncation at high age is a central reason the sample is useful for oldest-star and cosmic-age inference (Banik et al., 1 Jul 2026).

3. Sample construction and quality control

Several quality cuts from Xiang & Rix are repeatedly singled out because they govern the age precision actually attained. Atmospheric parameters were derived only for LAMOST DR7 spectra with SNR $M_K$5, and the catalog was restricted to $M_K$6 K, where the spectral fits were deemed robust. The sample excluded stars with $M_K$7 mag to avoid contamination by He-burning horizontal-branch stars, removed stars whose spectro-photometric distances and Gaia parallax distances disagreed by more than $M_K$8, and removed unresolved binaries flagged by Gaia diagnostics. The result is described as a relatively clean subgiant sample with minimized contamination from misclassified evolutionary states and unresolved companions (Shariat et al., 9 Oct 2025).

The cosmological reuse of the sample adds a second layer of filtering. It requires

$M_K$9

which effectively restricts the analysis to stars within 5 kpc, and also requires

$[\mathrm{Fe/H}]$0

to reduce contamination from partially resolved binaries. Because the raw catalog still shows an implausible population of stars with very large ages and very small age uncertainties, that work further imposes chemistry-based cuts in age–metallicity and age–$[\mathrm{Fe/H}]$1 space: $[\mathrm{Fe/H}]$2

$[\mathrm{Fe/H}]$3

together with an external consistency check against Gaia DR3 FLAME ages (Banik et al., 1 Jul 2026).

4. External validation with wide binaries

The most direct empirical audit of the Xiang & Rix sample uses wide binaries, whose two components are assumed to be coeval and generally co-chemical. The validation starts from a Gaia wide-binary catalog, cross-matches each age catalog with a $[\mathrm{Fe/H}]$4 positional tolerance, and retains systems in which both components of the same wide binary appear in the same age catalog and both have ages. For Xiang & Rix, this yields 12 matched binaries initially, i.e. 24 Xiang & Rix subgiant stars, although one pair is later treated as an outlier and marked “not used in statistical analysis,” leaving 11 binaries in the main assessment (Shariat et al., 9 Oct 2025).

The test statistic is the normalized age residual

$[\mathrm{Fe/H}]$5

for which realistic independent uncertainties imply $[\mathrm{Fe/H}]$6 and $[\mathrm{Fe/H}]$7. For the full matched Xiang & Rix sample, the study finds $[\mathrm{Fe/H}]$8, but this is dominated by a single strong outlier. That pair has ages $[\mathrm{Fe/H}]$9 Gyr and $[\alpha/\mathrm{Fe}]$0 Gyr, together with a $[\alpha/\mathrm{Fe}]$1 discrepancy in $[\alpha/\mathrm{Fe}]$2; the system is nevertheless described as a highly credible wide binary, with chance alignment probability $[\alpha/\mathrm{Fe}]$3, projected separation $[\alpha/\mathrm{Fe}]$4 au, and distance $[\alpha/\mathrm{Fe}]$5 pc. After removing that pair, the scatter drops to $[\alpha/\mathrm{Fe}]$6 (Shariat et al., 9 Oct 2025).

This leads to the central assessment of the sample: subgiant ages based on spectroscopic metallicities from Xiang & Rix are generally consistent within their reported uncertainties, implying that fractional uncertainties of $[\alpha/\mathrm{Fe}]$7 are realistically achievable. The same study gives a concrete catalog-level figure: the median fractional age uncertainty is $[\alpha/\mathrm{Fe}]$8. Among the three benchmark age catalogs tested there, the Xiang & Rix subgiant ages are the only ones judged consistent with their quoted precisions after removal of the single abundance-discrepant outlier (Shariat et al., 9 Oct 2025).

5. Oldest-star selection and cosmological reuse

The Xiang & Rix sample has also been repurposed for a lower-bound estimate on the age of the Universe. In that setting, the sample’s size and its age grid up to 20 Gyr are treated as crucial, because the catalog is not hard-truncated at the standard cosmological age. The authors do not recompute Xiang & Rix ages; instead they reinterpret the published per-star likelihoods $[\alpha/\mathrm{Fe}]$9 as observational likelihoods and reconstruct the latent age distribution $T_{\rm eff}$0 of the selected sample (Banik et al., 1 Jul 2026).

A further filter is imposed through comparison to Gaia DR3 FLAME. Because published FLAME ages were truncated at 13.5 Gyr, missing FLAME ages are reconstructed from the age–mass relation

$T_{\rm eff}$1

and the empirical YY–FLAME trend is fitted as

$T_{\rm eff}$2

Only stars within a $T_{\rm eff}$3 orthogonal band around this relation are retained in the nominal cosmology analysis (Banik et al., 1 Jul 2026).

The main reconstruction uses MCMC over a discretized latent age distribution, with posterior

$T_{\rm eff}$4

From that reconstruction, the inferred age of the oldest star is

$T_{\rm eff}$5

Varying the quality cuts changes this to as low as

$T_{\rm eff}$6

for a much lower age-dependent metallicity ceiling, or as high as

$T_{\rm eff}$7

for a much higher one (Banik et al., 1 Jul 2026).

6. Interpretation, strengths, and limitations

The principal strength of the Xiang & Rix sample is that it is a carefully quality-controlled subgiant regime in which spectroscopy supplies $T_{\rm eff}$8 and $T_{\rm eff}$9, exactly the quantities identified as decisive for precise isochrone ages. This is the basis for the claim that these are among the most trustworthy presently available field-star isochrone ages for evolved stars, at least in the differential sense tested by wide binaries (Shariat et al., 9 Oct 2025).

At the same time, the literature is explicit about what the sample does not establish. Wide binaries mainly test relative consistency, not the absolute age scale. Because both components have nearly identical chemistry and similar $[\mathrm{Fe/H}]$0, parameter-dependent systematics tend to cancel, and any global age-scale bias would not appear in a coevality test. The wide-binary constraints are therefore described as a lower limit on the true total age uncertainties, and the strongest claim is validation of reported relative or statistical uncertainties rather than proof that absolute ages are unbiased at the same $[\mathrm{Fe/H}]$1 level (Shariat et al., 9 Oct 2025).

The main limitations are also sharply defined. First, the direct wide-binary validation rests on only 11 statistically used binaries after excluding one outlier. Second, the problematic pair with a $[\mathrm{Fe/H}]$2 $[\mathrm{Fe/H}]$3 discrepancy illustrates a likely failure mode: age precision is only as good as abundance precision and abundance systematics. Third, the cosmological application depends materially on age–chemistry cuts and on the interpretation of very old, metal-poor, $[\mathrm{Fe/H}]$4-enhanced subgiants; in that context the age-dependent metallicity ceiling is the dominant sample-definition systematic (Shariat et al., 9 Oct 2025, Banik et al., 1 Jul 2026).

Within those limits, the Xiang & Rix sample occupies a distinctive position in recent stellar-population work: it is simultaneously a large spectroscopically anchored subgiant catalog, an externally tested benchmark for field-star age precision, and a chemically filtered old-star sample from which the oldest Galactic age tail can be reconstructed.