发表机构
The University of Texas at Austin; Scuperta, Intelligent Embedded Systems Department; Cosmic Frontier Center, The University of Texas at Austin; Kapteyn Astronomical Institute, University of Groningen; Centre de Recherche Astrophysique de Lyon UMR5574; University of Geneva; Institute for Astronomy, University of Edinburgh; Williams College（德克萨斯大学奥斯汀分校; Scuperta智能嵌入式系统部门; 德克萨斯大学奥斯汀分校宇宙前沿中心; 格罗宁根大学卡普坦天文研究所; 里昂天体物理研究中心UMR5574; 日内瓦大学; 爱丁堡大学天文研究所; 威廉姆斯学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本研究利用约10^9M⊙虚拟星系的辐射流体动力学模拟，结合LzLCS+样本的58个星系观测数据，通过Si II和C II吸收线恢复电离光子逃逸与星系标度关系，为高红移星系光谱分析提供了可靠方法。
AI 中文摘要
本研究利用一个约10^9M⊙质量的虚拟星系的辐射流体动力学模拟，对来自LzLCS+样本的58个星系的堆叠HST/COS光谱中观测到的Si II和C II线轮廓展开研究。我们将三个恒星质量区间（M★ ≤ 10^8M⊙、10^8–10^9M⊙、M★ ≥ 10^9M⊙）及三种堆叠方法（均值、中位数、加权平均）的堆叠结果，与包含22500条 mock 光谱的数据库进行对比，探究该模拟能否准确复刻真实气体特征、揭示星系属性的清晰趋势，并间接估算电离光子逃逸分数（f_esc）。我们发现，在所有质量区间内，模拟轮廓与观测轮廓均呈现合理吻合（χ² < 1）。值得注意的是，从最佳拟合 mock 轮廓中提取等效宽度（EW）和f_R等线属性，为直接经验趋势提供了可靠替代方案，尤其适用于噪声常使结果出现偏差的低信噪比（S/N）区间。尽管基于单一虚拟天体推导的模拟星际介质（LIS）特征，却展现出与恒星质量（M★）、恒星形成率（SFR）及f_esc的清晰相关性，与已确立的经验标度关系相符。我们还发现，匹配度最高的 mock 光谱主要源自模拟中对应紫外光度峰值和强烈星暴阶段的时间步长，表明这些活跃期产生了恒星形成星系中观测到的星际介质多样性。最后，基于模拟的估算值（f_esc^virtual）再现了f_esc的质量依赖趋势，且与生成堆叠结果的平均f_esc高度一致。这种基于模拟的框架为解释高红移星系的光谱观测（包括推断f_esc及表征物理标度关系）提供了可行方法，适用于再电离纪元的星系研究。
英文摘要
In this work, we use a radiation-hydrodynamic simulation of a single $\sim10^9M_{\odot}$ virtual galaxy to study Si II and C II line profiles seen in stacked HST/COS spectra of 58 galaxies from the LzLCS+ sample. We compare stacks across three mass bins ($M_{\star} \leq 10^8M_{\odot}$, $10^8$-$10^9M_{\odot}$, and $M_{\star} \geq 10^9M_{\odot}$) and three stacking methods (mean, median, and weighted average) to a library of 22,500 mock spectra. We investigate whether the simulation can accurately mimic real gas features, reveal clear trends with galaxy properties, and provide indirect estimates of the ionizing escape fraction ($f_{\rm esc}$). We find reasonable agreement between simulated and observed profiles ($χ^2 < 1$) across all mass regimes. Notably, extracting line properties such as EW and $R_f$ from best-fit mock profiles provides a robust alternative to direct empirical trends, particularly in the low-S/N regime where noise frequently biases results. The simulation-based LIS features, although derived from a single virtual object, exhibit clear correlations with $M_{\star}$, SFR, and $f_{\rm esc}$, mirroring established empirical scaling relations. We find that the best-matching mock spectra predominantly originate from simulation time steps corresponding to peak UV luminosity and intense starburst phases, suggesting that these active periods generate the ISM diversity observed in star-forming galaxies. Finally, simulation-based estimates ($f_{\rm esc}^{\rm virtual}$) reproduce the observed mass-dependent trends in $f_{\rm esc}$ and are in close agreement with the average $f_{\rm esc}$ of the generated stacks. This simulation-based framework establishes a relevant methodology for interpreting spectroscopic observations, including inferring $f_{\rm esc}$ and characterizing physical scaling relations, in high-redshift galaxies from the Epoch of Reionization.
Comments20 pages, 8 figures
