即将开展的发射线光谱巡天（如 Euclid 和 Roman Space Telescope）将受到星际污染源带来的系统效应影响：这些星系由于其发射光谱中的谱线混淆，导致其红移和与我们之间的距离被错误计算。尤其棘手的是涉及两条发射波长接近的谱线之间混淆的污染源（例如被误认为 \oiii 的 H$β$ 发射体），因为它们与目标星系在空间上高度相关。它们在观测星系目录的 3D 分布中引入了一种特定模式，这会改变星系相关函数中 BAO 峰值的位置，并使使用该样本进行的任何宇宙学分析产生偏差。在此，我们提出了一种预测星系目录中污染源比例的新方法，使用图神经网络（GNNs）学习污染源比例的后验分布，同时对宇宙学和星系偏差进行边缘化处理。该方法使用以晕作为星系代理的模拟进行开发。GNN 能够利用通常在宇宙学分析中不考虑的小尺度信息，推断出污染源比例后验分布的均值和标准差。将大尺度信息作为全局属性注入图中，提高了 GNN 在对宇宙学进行边缘化处理时的性能。
英文摘要：
Upcoming emission-line spectroscopic surveys, such as Euclid and the Roman Space Telescope, will be affected by systematic effects due to the presence of interlopers: galaxies whose redshift and distance from us are miscalculated due to line confusion in their emission spectra. Particularly pernicious are interlopers involving the confusion between two lines with close emitted wavelengths, like H$β$ emitters confused as \oiii, since those are strongly spatially correlated with the target galaxies. They introduce a particular pattern in the 3D distribution of the observed galaxy catalog that can shift the position of the BAO peak in the galaxy correlation function and bias any cosmological analysis performed with that sample. Here we present a novel method to predict the fraction of interlopers in a galaxy catalog, using Graph Neural Networks (GNNs) to learn the posterior distribution of the interloper fraction while marginalizing over cosmology and galaxy bias. The method is developed using simulations with halos acting as a proxy for galaxies. The GNN can infer the mean and standard deviation of the posterior distribution of interloper fraction using small-scale information that is usually not considered in cosmological analyses. The injection of large-scale information into the graph as a global attribute improves the performance of the GNN when marginalizing over cosmology.
Interloper bias in future large-scale structure surveys
下一代分光巡天将利用发射线星系作为示踪物，绘制可观测宇宙的大尺度结构。虽然每一次巡天都将用一条特定的发射线来绘制天空，但交错的发射线可以伪装成巡天在不同红移处的预期发射线。没有被移除的星系的干涉线会污染功率谱测量，混合各种红移的相关性并稀释真实信号。我们评估了功率谱污染的可能性，发现一个比0.2%更差的干扰者分数可能会使未来调查的功率谱测量偏差超过10%的统计误差，同时也会使功率谱推断产生偏差。我们还构建了一个预测宇宙学参数偏差的形式主义，证明了0.15%-0.3%的闯入者分数可能会使增长率偏差超过10%的误差，这可能会影响对即将到来的重力调查的约束。我们使用COSMOS模拟目录（CMC），与发射线重新缩放，以更好地再现最近的数据，预测潜在的闯入者分数的主焦点摄谱仪（PFS）和宽视场红外巡天望远镜（WFIRST）。我们发现，二次线识别，或确认星系红移找到相关的发射线，可以删除PFS的闯入者。对于WFIRST，我们使用CMC来预测WFIRST H $α$调查可以达到0.2%的目标，但需要灵敏的光学和近红外测光。对于WFIRST [OIII]调查，预测的闯入者比例达到百分之几，必须估计它们的影响并在统计上消除（例如，使用深度训练样本）。（节录）
Filtering Interlopers with Photometry and Diagnostic Features: A Machine Learning Framework Validated with CSST Slitless Spectroscopy
Shanghai Jiao Tong University（上海交通大学）；State Key Laboratory of Dark Matter Physics, Shanghai Jiao Tong University（上海交通大学暗物质物理全国重点实验室）；Department of Astronomy, School of Physics and Astronomy, Shanghai Jiao Tong University（上海交通大学物理与天文学院天文学系）；Key Laboratory for Particle Astrophysics and Cosmology (MOE)/Shanghai Key Laboratory for Particle Physics and Cosmology, Shanghai Jiao Tong University（上海交通大学粒子天体物理与宇宙学教育部重点实验室/上海粒子物理与宇宙学重点实验室）
A novel analysis of contamination in Lyman-break galaxy samples at $\boldsymbol{z\sim6-8}$: spatial correlation with intermediate-redshift galaxies at $\boldsymbol{z\sim1.3-2}$
Spectroscopic confirmation of z~7 LBGs: probing the earliest galaxies and the epoch of reionization
我们在ESO/VLT上展示了FORS 2超深光谱活动的最终结果，以确认从我们的VLT/Hawk-I成像调查中选择的三个独立领域的z~7“z-波段脱落”候选者。我们特别报道了在NTT深场中新发现的两个红移约为6.7的星系：这两个星系都显示出Ly-alpha发射线，其静止坐标系EW为15-20 A量级，光度为2-4 X 10^{42} erg/s。我们还提出了一个样本的i-dropout星系的超深观测的结果，从中我们设置了一个坚实的上限的分数的闯入者。在观察到的20个z-dropout中，我们确认了5个星系在6.7 < z < 7.1。这系统地低于根据较低红移观测得出的预期：特别是有一个显着缺乏对象与中间Ly-alpha EW（20和55 A之间）。我们的结论是LBG中Ly-alpha发射的分数从z~3到z~6不断增加的趋势很可能从z~6到z~7反转。解释观测到的LAE分数在辍学人口与再电离之间的快速变化需要一个快速演变的中性部分的氢在宇宙中。假设宇宙在z=6时完全电离，并采用Dijkstra et al.（2011）的半解析模型，我们发现我们的数据要求中性氢分数在时间Δ z ~ 1内发生Δ chi_{HI} ~ 0.6量级的变化，前提是逃逸分数在相同的红移间隔内没有显著增加。
Detections of Faint Lyman-alpha Emitters at z = 5.7: Galaxy Building Blocks and Engines of Reionization
We report results of a unprecedentedly deep, blind search for Ly-alpha emitters (LAEs) at z = 5.7 using IMACS, the Inamori-Magellan Areal Camera & Spectrograph, with the goal of identifying missing sources of reionization that could also be basic building blocks for today's L* galaxies. We describe how improvements in wide field imaging with the Baade telescope, upgrades to IMACS, and the accumulation of ~20 hours of integration per field in excellent seeing led to the detection of single-emission-line sources as faint as F ~ 2 x 10^{-18} ergs s^{-1} cm^{-2}, a sensitivity 5 times deeper than …
