84个星系SPARC基准中重子旋转曲线重建的可复现双边界运动学校正
A Reproducible Two-Boundary Kinematic Correction for Baryonic Rotation-Curve Reconstruction in an 84-Galaxy SPARC Benchmark
浏览论文内容
中文总结 AI 辅助
本研究针对Flynn等人提出的omega运动学校正，基于84星系SPARC基准开展可复现验证，通过双边界系数校正将平均偏差降至30.15 km/s，明确了算法流程与结果，为天文数据变换提供可复现基准。
中文摘要 AI 辅助
我们对Flynn和Cannaliato（2025）提出的经验omega运动学校正进行了可复现的计算验证与失效分析。该算法设计极简：每个星系从最内侧和最外侧测得的旋转曲线点计算一个系数，应用于完整的观测径向轮廓，再与由SPARC气体、星系盘和核球分量组装的重子重建结果进行比较。我们保留了前人的固定84星系基准，并公布其确切成员，使得验证对象为变换本身，而非样本重选。在不将变换拟合到重子残差的情况下，主最大盘重建在固定84星系基准上将观测-重子平均偏差从51.82 km/s降至30.15 km/s。有界质光比优化进一步将描述性灵敏度拟合值降至25.45 km/s，而简单开普勒参考在我们之前的分析[15]中平均RMSE为74.20 km/s。从每个星系的84条记录重新计算显示，53个星系存在可分辨的质光优化收益（delta RMSE > 0.05 km/s），31个星系无可分辨变化。6个星系未超过开普勒参考；所有6个星系的Upsilon_max ≤ 0.111，而它们的omega值并不集中在样本高端。我们明确了完整的确定性工作流程、原生单位、端点不变量、不确定性传播及公式级回归检查，以防止分组和符号错误。完整的84星系面板集、总体误差分布及失效诊断均保留为可检查输出。该结果是可复现的天文数据变换基准，而非提出的力定律或暗物质、修正引力的替代方案。
英文摘要
We present a reproducible computational validation and failure analysis of the empirical omega kinematic correction introduced by Flynn and Cannaliato (2025). The algorithm is deliberately minimal: one coefficient per galaxy is calculated from the innermost and outermost measured rotation-curve points and applied to the full observed radial profile before comparison with a baryonic reconstruction assembled from SPARC gas, disk, and bulge components. We preserve the predecessor's frozen 84-galaxy benchmark and publish its exact membership so that the transformation, not sample re-selection, is the object of validation. Without fitting the transformation to the baryonic residual, the primary maximum-disk reconstruction reduces the mean observed-baryonic discrepancy from 51.82 to 30.15 km/s across the frozen 84-galaxy benchmark. Bounded mass-to-light-ratio optimization further reduces the descriptive sensitivity-fit value to 25.45 km/s, while the simple Keplerian reference has a mean RMSE of 74.20 km/s in our earlier analysis [15]. Recalculation from the 84 per-galaxy records shows a resolved mass-to-light optimization benefit (delta RMSE > 0.05 km/s) in 53 galaxies and no resolved change in 31. Six galaxies do not beat the Keplerian reference; all six occur at Upsilon_max <= 0.111, whereas their omega values are not concentrated at the high end of the sample. We specify the complete deterministic workflow, native units, endpoint invariants, uncertainty propagation, and formula-level regression checks required to prevent grouping and sign errors. The complete 84-galaxy panel set, population-level error distributions, and failure diagnostics are retained as inspectable outputs. The result is a reproducible astronomical data-transformation benchmark rather than a proposed force law or replacement for dark matter or modified gravity.
