Design And Development Of Juno Offline Software System

The Jiangmen Neutrino Underground Observation(JUNO)is a major scientific research project in China,which is being built at a distance of 53 km from both the Yangjiang and Taishan nuclear power plants.JUNO is designed to determine the neutrino mass hierarchy and precisely measure multiple mixing parameters of neutrino oscillation as the primary physics goals,and is also capable to perform researches on atmospheric neutrinos,solar neutrinos,neutrinos from supernova burst,etc.The offline software system is a vital part of JUNO.It is developed for the offline data processing,including production of the Monte Carlo(MC)data,collection of the real data,data calibration and reconstruction,as well as providing tools for the physics analysis.The subject of this paper is the design and development of JUNO offline software,including the part of event data model(EDM),data management system,software validation system,as well as the research of parallel computing technologies for JUNO.The EDM plays a central role in the JUNO offline software.It describes the event data entities through all processing stages for both simulated and collected data,and provides persistency via the input/output(I/O)system.The special requirements of JUNO have challenged the design of the EDM.Firstly,the elec-tron anti-neutrino is detected via the inverse beta decay(IBD)reaction,requiring a correlation mechanism between the prompt-delayed signals produced by it.Sec-ondly,some special data analysis processes need event information from multiple offline data processing procedures,requiring the data correlation mechanism be-tween multiple processing procedures.Based on these special requirements,we designed the JUNO EDM,which not only satisfies the basic needs of various data processing procedures,but also provides flexible data correlation mechanisms.The data management services,including the memory management service and the data I/O services,are basic modules shared by the whole offline software.The memory management service is based on the DataBuffer provided by SNiPER,which manages the transient EDM objects in memory,and provides interfaces for analyzers to retrieve and store the event data.The data I/O services are responsible for the transportation of the transient EDM objects and the persistent event data.Due to the large scale of the JUNO detector,and the high trigger rate of backgrounds,the data I/O is one of the most serious performance bottlenecks of JUNO offline software.Therefore,the data I/O system is designed to support the lazy loading mechanism,in order to reduce the overhead caused by the data I/O.Currently,the parallelized prototype of the JUNO offline software is under development.The parallelization of the EDM and data management services is one of the most import,ant and difficult tasks.This paper also describes the design of the JUNO offline prototype,and the update of the EDM and data management modules,in order to support the parallel computing technology.The validation system of the offline software is a vital tool to endure the correctness of its physics results and to improve its quality.It can be divided into two levels.The first level is the unit test based on the software packages,which checks whether the software runs correctly and their execution performances,while the second level is the data production and validation,which check the offline software as a.integrated system.This paper introduces the design and implementation of JunoTest,which is applied as the indispensable tool to ensure the functionalities and performances of the JUNO offline software.A series of performance tests based on JunoTest are also performed,which show that the design of the EDM and data management systems satisfy the requirements of JUNO data processing.