基于深度学习管道的TGSS ADR1巨型射电源候选体目录
A Catalog of Giant Radio Source Candidates from TGSS ADR1 Using a Deep Learning--based Pipeline
浏览论文内容
中文总结 AI 辅助
该研究基于TGSS ADR1数据，用含射电源检测模型与似然比方法的深度学习管道，构建含5595个GRS候选体的目录，分析其物理性质分布，发现高功率异常值挑战自相似演化模型，还识别出286个与星系团关联的候选体。
中文摘要 AI 辅助
我们提出了一个包含5595个巨型射电源（GRS）候选体的目录，其定义为最大投影线性尺寸（LLS）超过0.7 Mpc。其中4566个若被确认将是新发现。这些候选体通过对TIFR GMRT天空巡天的第一替代数据发布（TGSS ADR1）进行系统自动搜索识别，所使用的是一种基于深度学习的管道，该管道整合了射电源检测模型与似然比方法。在全部样本中，LLS可达约4.5 Mpc，红移范围为z=0.056至至少z=2.385。该目录包含4210个巨型射电星系、1030个巨型射电类星体，剩余355个候选体目前未分类。我们系统分析了它们的关键物理性质分布，包括谱指数、弯曲角（BA）、射电功率（P₁₅₀）以及r波段绝对星等。结果显示，GRS候选体以直型为主（弯曲角中位数为11.0°），且遵循“L型”的BA-LLS分布。射电功率与LLS（P₁₅₀-D）图显示，在P₁₅₀<10²⁸ W Hz⁻¹处存在聚集，而LLS>1 Mpc的稀有高功率异常值（>10²⁸.⁵ W Hz⁻¹）对自相似演化模型提出挑战。此外，还发现286个GRS候选体与已编目星系团相关联。
英文摘要
We present a catalog of 5,595 giant radio source (GRS) candidates, defined by a largest (projected) linear size (LLS) exceeding 0.7 Mpc. Of these, 4,566 would be new discoveries if confirmed. These candidates were identified through a systematic and automated search of the first Alternative Data Release of the TIFR GMRT Sky Survey (TGSS ADR1) using a deep learning--based pipeline that incorporates a radio sources detection model and the likelihood ratio method. Across the full sample, the LLS reaches up to $\sim$4.5~Mpc, with redshifts ranging from $z=0.056$ to at least $z=2.385$. The catalog comprises 4,210 giant radio galaxies and 1,030 giant radio quasars, with the remaining 355 candidates currently unclassified. We systematically analyze the distributions of their key physical properties, including spectral index, bending angle (BA), radio power ($P_{150}$), and $r$-band absolute magnitude. Our results show that GRS candidates are predominantly straight (a median BA of $11.0^\circ$) and follow an ``L-shaped" BA--LLS distribution. The radio power vs. LLS ($P_{150}$--$D$) diagram reveals a concentration at $P_{150} < 10^{28}$ W Hz$^{-1}$, while rare high-power outliers ($>10^{28.5}$ W Hz$^{-1}$) with LLS $>$ 1 Mpc challenge self-similar evolutionary models. Additionally, 286 GRS candidates are found to be associated with cataloged galaxy clusters.
