Stellar Age Compression Reshapes Interpretations of the Milky Way Thick-Disk Formation History

Technical Summary: Stellar Age Compression Reshapes Interpretations of the Milky Way

Problem Statement
The formation timescale of the Milky Way's thick disk remains a central debate in Galactic archaeology. Previous studies, relying primarily on spectroscopically inferred stellar ages (e.g., via astroNN), have suggested a rapid assembly history characterized by a short formation timescale (~1 Gyr), bursty enrichment, and an early-dominated buildup. These conclusions are largely drawn from the Age–Metallicity Relation (AMR), which exhibits a steep negative slope in spectroscopic data. However, stellar ages are not directly observable; they are derived through inference pipelines sensitive to observational quality (signal-to-noise ratio, parallax precision). There is a growing concern that these inference methods may introduce a systematic "compression" bias, where ages are shrunk toward the sample mean, potentially distorting population statistics and leading to erroneous physical interpretations of the Galaxy's early evolution.

Methodology
To isolate the impact of age definition from sample selection effects, the authors employ a rigorous comparative framework using a single, identical stellar sample drawn from APOGEE spectroscopic data and Gaia parallaxes.

• Independent Age Scales: The study compares two distinct age scales for the same stars: spectroscopic inferred ages (astroNN) and asteroseismic ages (APOKASC-3). The latter serves as an independent external anchor with a different systematic error budget (based on oscillation frequencies rather than spectral fitting).
• Coarsened Exact Matching (CEM): To ensure a fair comparison, the authors apply CEM to match stars across a five-dimensional physical covariate space ($T_{\text{eff}}$, $\log g$, [Fe/H], [$\alpha$/Fe], distance, and Galactic coordinates). Age information is explicitly excluded from the matching process, ensuring that differences in diagnostics arise solely from the age definition, not sample composition.
• Transport Inversion Experiments: The authors distinguish between additive noise and compressive transformations. They model the relationship between inferred and seismic ages using a linear transport operator ($a_{\text{infer}} = \alpha + \lambda \cdot a_{\text{seismo}} + \epsilon$). Theoretical constraints establish that additive noise can only broaden age distributions, whereas a compressive map ($\lambda < 1$) narrows them.
• Diagnostics: Key formation-history metrics are quantified, including the AMR slope, formation timescale ($\Delta t$), peak formation age, and cumulative formation fraction (CFF). Statistical significance is assessed via bootstrap resampling ($N=10^4$).

Key Results
The analysis reveals that switching from spectroscopic inferred ages to seismic anchoring systematically alters the interpretation of thick-disk formation:

1. AMR Flattening: The AMR slope decreases significantly from $-3.29 \pm 0.08$ Gyr dex$^{-1}$ (inferred) to $-1.86 \pm 0.15$ Gyr dex$^{-1}$ (seismic), a difference of $\Delta a = +1.43$ ($z \approx 8.4$). This indicates a much slower chemical enrichment rate than previously inferred.
2. Extended Assembly: The formation timescale widens from 3.04 Gyr to 3.55 Gyr, and the peak formation age shifts from 9.1 Gyr to 6.0 Gyr. This contradicts the "rapid burst" scenario, suggesting a more extended assembly history.
3. Compression Evidence: The bias between inferred and seismic ages is not random noise but a structured function of observational quality. In low-quality regimes, inferred ages compress toward the mean. Transport inversion experiments confirm that only a compressive transformation ($\lambda \approx 0.85$) can simultaneously reproduce the narrower age distribution, steeper AMR, and rapid-formation-like observables seen in spectroscopic data. Additive noise models fail to replicate this pattern.
4. Metallicity Dependence: The compression coefficient $\lambda$ is lowest (strongest compression, $\lambda \approx 0.79$) at the metal-poor end and approaches unity ($\lambda \approx 0.94$) at the metal-rich end, consistent with the expectation that metal-poor thick-disk stars are more susceptible to observational quality issues.

Significance and Claims
The paper argues that the "rapid formation" signature of the Milky Way's thick disk—traditionally derived from spectroscopic ages—may be an artifact of observationally driven age compression rather than an intrinsic bursty formation history.

• Definition Sensitivity: The core finding is that statistical interpretations of Galactic formation history depend sensitively on the stellar age definition itself. Changing the age scale while holding the stellar sample constant is sufficient to invert the physical narrative from "rapid, early-dominated assembly" to "extended, continuous formation."
• Mechanism Identification: The study demonstrates that a compressive transport map is sufficient to generate rapid-formation-friendly observables without requiring an intrinsically bursty history.
• Caution for Future Surveys: The authors caution that future large-scale Galactic surveys (e.g., SDSS-V, 4MOST, WEAVE) must explicitly model age-definition-dependent compression effects. Merely reducing statistical uncertainties without addressing systematic compression may systematically amplify rapid-formation features in the inferred history.

The paper concludes that while seismic ages are not absolute truth, the systematic differences between independent age scales are robust enough to challenge the prevailing rapid-assembly paradigm, necessitating a re-evaluation of the Milky Way's early evolutionary timeline.