Object-oriented data analysis framework for neutron scattering experiments

摘要

Materials and Life Science Facility (MLF) of Japan Proton Accelerator Research Complex (J-PARC) is one of the facilities that provided the highest intensity pulsed neutron and muon beams. The MLF computing environment design group organizes the computing environments of MLF and instruments. It is important that the computing environment is provided by the facility side, because meta-data formats, the analysis functions and also data analysis strategy should be shared among many instruments in MLF. The C++ class library, named Manyo-lib, is a framework software for developing data reduction and analysis softwares. The framework is composed of the class library for data reduction and analysis operators, network distributed data processing modules and data containers. The class library is wrapped by the Python interface created by SWIG. All classes of the framework can be called from Python language, and Manyo-lib will be cooperated with the data acquisition and data-visualization components through the MLF-platform, a user interface unified in MLF, which is working on Python language. Raw data in the event-data format obtained by data acquisition systems will be converted into histogram format data on Manyo-lib in high performance, and data reductions and analysis are performed with user-application software developed based on Manyo-lib. We enforce standardization of data containers with Manyo-lib, and many additional fundamental data containers in Manyo-lib have been designed and developed. Experimental and analysis data in the data containers can be converted into NeXus file. Manyo-lib is the standard framework for developing analysis software in MLF, and prototypes of data-analysis softwares for each instrument are being developed by the instrument teams.