发表机构
Tata Institute of Fundamental Research; Université Paris-Saclay; Université Paris Cité; CEA; CNRS; New York University Abu Dhabi（塔塔基础研究院; 巴黎-萨克雷大学; 巴黎西岱大学; 法国原子能和替代能源委员会; 法国国家科学研究中心; 纽约大学阿布扎比分校）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
ROSE是一个模块化机器学习框架，用于从TESS数据自动测量红巨星的$\nu_{\max}$和$\Delta\nu$，通过合成谱训练和概率输出，在274,721颗恒星上实现了高效且可靠的星震学表征。
AI 中文摘要
来自Kepler和TESS的太空测光已为数十万颗红巨星提供了振荡谱，而PLATO和Roman银河系核球时域巡天将提供更多数据。我们提出了ROSE，一个用于红巨星自动星震学表征的模块化机器学习框架，并将其中的两个模块应用于最大振荡功率频率$\nu_{\max}$和大频率间隔$\Delta\nu$。每个模块都是一个神经网络，捆绑有其自身的预处理变换和输出分箱，因此每个参数都在最能有效展示它的表示中进行测量：$\nu_{\max}$来自功率密度谱，$\Delta\nu$来自以$\nu_{\max}$为中心的窗口上功率谱的功率谱。两者都仅基于一年基线生成的合成谱进行训练，并且都返回概率分布而非点估计。这使我们能够获取不对称不确定性和可靠性指数。针对留出的合成谱、训练分辨率下的Kepler谱以及TESS连续观测区巨星进行验证，我们在$\nu_{\max}$上获得了1.4%、3.3%和4.4%的稳健离散度，在$\Delta\nu$上获得了0.35%、0.61%和1.3%的稳健离散度，且$\nu_{\max}$的不确定性接近其名义覆盖率。该全自动流程应用于274,721颗至少被观测了十个扇区的TESS红巨星候选体，框架为56,224颗恒星返回了可靠的$\nu_{\max}$，为39,939颗恒星返回了可靠的$\Delta\nu$，每颗恒星耗时8毫秒。该样本遵循$\Delta\nu = 0.278 \nu_{\max}^{0.757}$，离散度为6.2%，红团簇相作为一个明显的过密区域突出显示，尽管两个模块都没有被提供演化标签或被要求生成演化标签。
英文摘要
Space-based photometry from Kepler and TESS has delivered oscillation spectra for hundreds of thousands of red giants, and PLATO and the Roman Galactic Bulge Time-Domain Survey will add more. We present ROSE, a modular machine-learning framework for automated asteroseismic characterisation of red giants, and apply two of its modules to the frequency of maximum oscillation power, $ν_{\rm max}$, and the large frequency separation, $Δν$. Each module is a neural network bundled with its own preprocessing transform and output binning, so that each parameter is measured in the representation that displays it most effectively: $ν_{\rm max}$ from the power-density spectrum, and $Δν$ from the power spectrum of the power spectrum taken over a window centred on $ν_{\rm max}$. Both are trained only on synthetic spectra generated at a one-year baseline and both return probability distributions rather than point estimates. This enables us to retrieve asymmetric uncertainty and a reliability index. Validating against held-out synthetic spectra, Kepler spectra at the training resolution and TESS continuous-viewing-zone giants, we obtain robust dispersions of 1.4, 3.3 and 4.4% in $ν_{\rm max}$ and 0.35, 0.61 and 1.3% in $Δν$, with the $ν_{\rm max}$ uncertainties close to their nominal coverage. The fully automated pipeline was applied to 274,721 TESS red-giant candidates observed for at least ten sectors, the framework returning reliable $ν_{\rm max}$ for 56,224 stars and reliable $Δν$ for 39,939, taking 8 ms per star. The sample follows $Δν= 0.278 ν_{\rm max}^{0.757}$ with a scatter of 6.2%, and the red-clump phase stands out as a distinct overdensity, although neither module is provided an evolutionary label or asked to produce one.
Comments22 pages, 4 figures, 3 tables
