论文标题

通过大调查数据,机器学习和宇宙学模拟解码星光

Decoding Starlight with Big Survey Data, Machine Learning, and Cosmological Simulations

论文作者

Blancato, Kirsten

论文摘要

恒星和恒星的集合编码了恒星物理和星系进化的丰富签名。由于其环境和内在性质都影响了特性,恒星保留了有关天体物理现象的信息,而这些现象本来是不可直接观察的。在时间域中,可以使用恒星的观察到的亮度变异性来研究在恒星表面和恒星内部发生的物理过程。在银河尺度上,恒星的性质,包括化学丰度和恒星年龄,是银河系来源的多维记录。以银河系和轨道特性,这为我们的银河形成以来的发展详细介绍了细节。延伸到本地群体之外,未解决的恒星种群的属性使我们能够研究宇宙中星系的多样性。 通过检查恒星的特性,以及它们如何在一系列的空间和时间尺度上变化,该论文将恒星中的信息与银河形成和进化中的全局过程联系起来。我们开发了新的方法来确定我们直接观察到的变异性,包括旋转和表面重力在内的恒星特性。我们为银河系的化学丰富历史提供了新的见解,并追踪了不同的恒星爆炸,以捕捉数十亿年的进化。我们通过研究构成星系的最初恒星质量分布的差异来提高对恒星和星系如何联系的知识和理解。在建立这些知识时,我们强调了数据和理论之间的当前紧张局势。通过综合数值模拟,大型观察数据集和机器学习技术,这项工作使有价值的方法论贡献从当前和未来的恒星观察结果的各种集合中最大化见解。

Stars, and collections of stars, encode rich signatures of stellar physics and galaxy evolution. With properties influenced by both their environment and intrinsic nature, stars retain information about astrophysical phenomena that are not otherwise directly observable. In the time-domain, the observed brightness variability of a star can be used to investigate physical processes occurring at the stellar surface and in the stellar interior. On a galactic scale, the properties of stars, including chemical abundances and stellar ages, serve as a multi-dimensional record of the origin of the galaxy. In the Milky Way, together with orbital properties, this informs the details of the evolution of our Galaxy since its formation. Extending beyond the Local Group, the attributes of unresolved stellar populations allow us to study the diversity of galaxies in the Universe. By examining the properties of stars, and how they vary across a range of spatial and temporal scales, this Dissertation connects the information residing within stars to global processes in galactic formation and evolution. We develop new approaches to determine stellar properties, including rotation and surface gravity, from the variability that we directly observe. We offer new insight into the chemical enrichment history of the Milky Way, tracing different stellar explosions that capture billions of years of evolution. We advance knowledge and understanding of how stars and galaxies are linked, by examining differences in the initial stellar mass distributions comprising galaxies, as they form. In building up this knowledge, we highlight current tensions between data and theory. By synthesizing numerical simulations, large observational data sets, and machine learning techniques, this work makes valuable methodological contributions to maximize insights from diverse ensembles of current and future stellar observations.

扫码加入交流群

加入微信交流群

微信交流群二维码

扫码加入学术交流群,获取更多资源