AI 中文总结
该研究验证DESI DR2的Lyα森林全形状分析，改进建模框架后发现BAO和AP参数稳定可靠，排除有偏差的fσ₈测量，提供更优的宽带AP测量。
AI 中文摘要
我们对暗能量光谱仪（DESI）数据发布2（DR2）莱曼α（Lyα）森林全形状分析进行验证。该分析结合三维Lyα森林自相关与类星体交叉相关，提取重子声学振荡（BAO）特征和宽带成团信号的信息，重点关注阿尔科克-帕钦斯基（AP）测量。与DESI DR1分析相比，DR2验证使用规模更大、更真实的模拟数据集，包括CoLoRe 2LPT和AbacusSummit Lyα森林模拟。建模框架通过对小尺度（<10 h⁻¹Mpc）的解析边缘化以及紫外背景涨落的影响得到改进。验证程序在盲解前完成，为目标宇宙学参数设定定量要求，通过数百次模拟实现评估。我们还通过对自相关和交叉相关的独立拟合、多个目录拆分，以及对模拟和盲态观测数据应用的大量分析与建模变体，进一步测试该分析。结果显示，BAO和AP参数满足所有验证要求，且在所有测试中保持稳定；相比之下，模拟研究揭示推断的增长速率参数fσ₈存在显著偏差，因此我们将该测量排除在最终分析之外。模拟、数据拆分及稳健性测试的一致性表明，DR2 Lyα全形状分析相较于以往Lyα森林研究，提供了可靠且大幅改进的宽带AP测量。
英文摘要
We present the validation of the Dark Energy Spectroscopic Instrument (DESI) Data Release 2 (DR2) Lyman-$α$ (Ly$α$) forest full-shape analysis. This analysis combines three-dimensional Ly$α$ forest auto-correlations and cross-correlations with quasars to extract information from both the baryon acoustic oscillation (BAO) feature and the broadband clustering signal, with primary emphasis on the Alcock-Paczynski (AP) measurement. Compared to the DESI DR1 analysis, the DR2 validation uses substantially larger and more realistic mock datasets, including CoLoRe 2LPT and AbacusSummit Ly$α$ forest simulations. The modeling framework is also improved through analytic marginalization over small scales ($<10$ $h^{-1}$Mpc) and the impact of ultraviolet background fluctuations. The validation program was completed prior to unblinding and defines quantitative requirements for the cosmological parameters of interest, which are evaluated using hundreds of mock realizations. We further test the analysis through independent fits to the auto- and cross-correlations, multiple catalog splits, and a broad suite of analysis and modeling variations applied to both mocks and blinded observational data. We find that the BAO and AP parameters satisfy all validation requirements and remain stable across all tests. In contrast, mock studies reveal a significant bias in the inferred growth-rate parameter $fσ_8$, leading us to exclude this measurement from the final analysis. The consistency across mocks, data splits, and robustness tests demonstrates that the DR2 Ly$α$ full-shape analysis provides a reliable and substantially improved broadband AP measurement over previous Ly$α$ forest studies.
