Modelling and analysis of large data sets for LUX and LZ experiments

谢菲尔德大学的粒子物理学和粒子天体物理学小组已经参与暗物质的搜索超过20年。该博士项目的主管正在领导由STFC资助的LUX-ZEPLIN（LZ）建设项目英国部分的英国数据中心和模拟工作包。该项目将建造和运行世界上最灵敏的暗物质实验，使用7吨液态氙靶来探测WIMP。预计LZ每年的运行将存储约1 PB的数据，LZ协作目前正在开发蒙特卡罗建模和数据分析软件工具。2017年成功完成了第一次模拟数据挑战赛，模拟数据处理链。计划于2018年和2019年进行进一步的模拟数据挑战，涉及具有探测器参数的数据库，完全实现的探测器响应和完整的事件重建，为2020年的数据采集做准备。谢菲尔德小组负责大量与事件发生器、蒙特卡罗模拟和分析软件开发相关的任务。英国还将托管基于GridPP资源的两个数据中心之一，包括谢菲尔德HEP集群。除了参与LZ实验，谢菲尔德小组还在为LZ的前身--LUX实验进行模拟和数据分析。探测器的运行已于两年前停止，但仍需要分析大量数据集，特别是为帮助设计着陆区实验而进行的不同来源的各种校准运行的数据。这一分析应辅之以探测器效应的完整蒙特卡罗分析。博士生将在LUX实验（1-2年）的数据模拟和分析中发挥关键作用。他们还将协助设计和实施未来着陆区数据建模和分析软件（1-3年），并在着陆区实验开始校准和科学运行时（3-4年）参与数据分析。预计学生将在LZ探测器将运行的Sanford地下研究设施中花费6-12个月的LTA。这将在博士项目的第3年进行，届时将进行主要探测器调试、操作准备、工程和校准运行。他还将看到和分析来自LZ实验的第一批数据，特别强调使用重合事件验证背景模型。对LUX和LZ探测器数据集的大规模模拟和分析将使学生有机会学习这些实验中使用的软件工具和计算机基础设施。学生预计将在LBNL（美国伯克利实验室）度过6个月，该实验室将托管美国数据中心，或与大型数据集处理（费米实验室或BNL）相关的其他设施。LBNL拥有国家能源研究计算中心的超级计算设施，其中包括一个基于Cray XC 40的强大的高性能计算系统科里，该系统将与英国GridPP资源一起用于LZ数据分析和模拟。学生将学习实验与美国数据中心之间以及美国和英国两个数据中心之间的数据传输所使用的软件和程序。他们将参与开发数据管理和控制软件。

The Particle Physics and Particle Astrophysics group at the University of Sheffield has been involved in dark matter searches for more than 20 years. The supervisor of this PhD project is leading the Work Package on the UK Data Centre and Simulations within the UK part of the LUX-ZEPLIN (LZ) construction project, funded by STFC. This is the project to build and operate the world most sensitive dark matter experiment with 7 tonnes of liquid xenon target to detect WIMPs. LZ is expected to store about 1 PB of data per year of operation and the LZ Collaboration is currently developing software tools for Monte Carlo modelling and data analysis. The first mock data challenge with simulated data processing chain has been successfully completed in 2017. Further mock data challenges involving databases with detector parameters, fully implemented detector response and full event reconstruction are scheduled for 2018 and 2019 in preparation for data taking in 2020. The Sheffield group is responsible for a large number of tasks linked to event generators, Monte Carlo simulations and development of analysis software. UK will also host one of the two data centres based on GridPP resources including Sheffield HEP cluster. In addition to the involvement in the LZ experiment, the Sheffield group is working on simulations and data analysis for the predecessor of LZ - the LUX experiment. The detector operation has stopped 2 years ago but large data sets, in particular data from various calibration runs with different sources, carried out to help designing the LZ experiment, still need to be analysed. This analysis should be complemented by the full Monte Carlo of the detector effects. The PhD student will play key role in the simulation and analysis of data from the LUX experiment (years 1-2). They are also expected to contribute to the design and implementation of software for modelling and analysis of future LZ data (years 1-3) and take part in the data analysis when the LZ experiment will start calibration and science runs (years 3-4). The student is expected to spend 6-12 months on LTA at the Sanford Underground Research Facility where the LZ detector will be operating. This will happen in year 3 of the PhD project when the main detector commissioning, preparation to operations, engineering and calibration runs will take place. He will also see and analyse the first data coming from the LZ experiment with a special emphasis on validating the background model using coincident events. Large-scale simulations and analysis of data sets from LUX and LZ detectors will give the student an opportunity to learn software tools and computer infrastructure used in these experiments. The student is expected to spend 6 months at LBNL (Berkeley Lab, USA) that will host the US data centre, or in some other facility linked to large data set processing (Fermilab or BNL). LBNL hosts the National Energy Research Computing Center's supercomputing facility that includes a powerful High Performance Computing system Cori based on Cray XC40 that will be used for LZ data analysis and simulations together with the UK GridPP resources. The student will learn the software and procedures used in data transfer between the experiment and the US data centre, as well as between the two data centres in the US and the UK. They will participate in the developing the software for data management and control.