Mining the LSST data stream for distant supernovae

挖掘遥远超新星的 LSST 数据流

• 批准号: 2039272
• 金额: --
• 依托单位: Lancaster University
• 依托单位国家: 英国
• 项目类别: Studentship
• 财政年份: 2017
• 资助国家: 英国
• 起止时间: 2017 至 无数据
• 项目状态: 已结题
• 来源: https://gtr.ukri.org/projects?ref=studentship-2039272
• 关键词: Mining; LSST; data; stream; distant

The LSST (Large Synoptic Survey Telescope) will be unique in its ability to image the observable southern sky every few nights. This will generate a vast dataset of transient objects, resulting in 100,000 transient alerts every night. Among these among will be of order 50,000 Type Ia supernovae per year, which can be used as cosmological distance indicators for studying dark energy. Picking out these SNe based on their photometry is a massive data challenge that has been recognised internationally (LSSTC and LSST:UK have recently funded a series of workshops on this issue).The PhD project will start with work on these selection algorithms, using simulations as input. The work will make use of machine learning algorithms that are currently being developed within the LSST:UK consortium (of which IH is a member). The focus will be to assess the impact of spectroscopy (i.e. redshift and transient classification information for a training sample) on the selection algorithm. Such spectroscopy is expected to be provided by the 4MOST fibre-fed multi-object spectrograph instrument, a project that Lancaster has recently joined.The second phase of the project will be preparing for 4MOST's survey. This involves designing the software interface between LSST transient detection and placement of 4MOST's spectroscopic fibres. Sophisticated algorithms will be developed to visualise and merge priorities from the transient target list with those from 4MOST's other surveys, and to optimally configure the approximately 2000 fibres per 4MOST observation (totaling over 50 million fibre positions must be defined over 4MOST's 5-year survey).

大型综合巡天望远镜（LSST）的独特之处在于它每隔几个晚上就能对可观测的南方天空进行成像。这将产生一个巨大的瞬态对象数据集，每晚产生10万个瞬态警报。其中每年约有5万颗Ia型超新星，可以作为研究暗能量的宇宙学距离指标。根据光度法来挑选这些超新星是一项巨大的数据挑战，这在国际上已经得到了认可（LSSTC和LSST:UK最近资助了一系列关于这个问题的研讨会）。博士项目将从研究这些选择算法开始，使用模拟作为输入。这项工作将利用LSST:UK联盟（IH是其成员之一）目前正在开发的机器学习算法。重点是评估光谱学（即训练样本的红移和瞬态分类信息）对选择算法的影响。这种光谱预计将由4MOST纤维馈送多目标光谱仪提供，这是兰开斯特大学最近加入的一个项目。该项目的第二阶段将是为4MOST的调查做准备。这包括设计LSST瞬态检测和4MOST光谱光纤放置之间的软件接口。将开发复杂的算法，将瞬态目标列表中的优先级与4MOST的其他调查中的优先级进行可视化和合并，并优化配置每次4MOST观测的约2000个光纤（4MOST的5年调查必须确定总计超过5000万个光纤位置）。

项目成果

4MOST Consortium Survey 10: The Time-Domain Extragalactic Survey (TiDES)

• DOI: 10.18727/0722-6691/5129
• 发表时间: 2019-03
• 期刊: arXiv: Instrumentation and Methods for Astrophysics
• 作者: E. Swann;Mark Sullivan;J. Carrick;S. Hoenig;I. Hook;R. Kotak;K. Maguire;R. McMahon;R. Nichol;S. Smartt