The key activities of the Meshcheryakov Laboratory of Information Technologies (MLIT) are related to providing the Institute with network, computing, and information resources using the Multifunctional Information and Computing Complex (MICC) of JINR, as well as with mathematical and algorithmic support for a diverse range of research in high-energy physics, nuclear physics, condensed matter physics, and other areas.
The MICC plays a decisive role in studies requiring advanced computing power and data storage systems. Its uniqueness is ensured by the consolidation of all state-of-the-art information technologies from the network infrastructure with a bandwidth of up to 4х100 Gbps and the distributed data processing and storage system based on grid technologies and cloud computing to the hyperconverged computing infrastructure with liquid cooling for supercomputer applications (Govorun Supercomputer).
The JINR grid infrastructure includes one of the world’s seven Tier-1 grid sites for processing and long-term storage of data from the CMS experiment at the Large Hadron Collider (CERN), as well as a Tier-2 grid site that provides computing power and data storage systems for the LHC, NICA, and other large-scale international experiments, as well as support for users of JINR laboratories and Member States.
To process and store data from experiments underway at JINR, a heterogeneous computing environment integrating all MICC components was created on the DIRAC Platform. The platform’s main users are the experiments of the NICA Megascience Project and the Baikal-GVD Neutrino Experiment.
The JINR cloud infrastructure is a universal component of the MICC, designed to provide computing resources and host information and computing services. The major and most resource-intensive consumers of these resources are the Institute’s scientific projects, particularly the Baikal-GVD and JUNO Neutrino Experiments, as well as the experiments of the NICA Project. A distributed information and computing environment that integrates the cloud resources of the JINR Member States’ organizations was built as part of the cloud infrastructure.
One of the essential elements of the MICC is the data storage system. The primary storage systems used are dCache and the EOS for disk storage, as well as Enstore and the CTA for tape storage. The multi-layer data storage system ensures both operational and long-term data storage.
The MICC embraces the HybriLIT Heterogeneous Computing Platform – a multi-component system consisting of the Govorun Supercomputer, an educational testbed, network data storage systems, and a number of specialised testbeds. The platform is designed for carrying out high-performance and resource-intensive computations, training neural network models, as well as developing parallel applications and quantum algorithms. The supercomputer, named after N. N. Govorun, is the main resource for high-performance computing; it is a hyperconverged software-defined system with a multi-layer data processing and storage system built on unique liquid-cooled server solutions by the RSC company. The tasks solved on the supercomputer encompass calculations for the NICA experiments, quantum chromodynamics, research into the properties of superheavy element atoms, the radiation safety of JINR’s facilities, and radiation biology studies.
One of MLIT’s key activities has always been the provision of computational, algorithmic, and software support for the preparation and implementation of experimental and theoretical research conducted with JINR’s participation, as well as the design, development, and application of computational methods for modelling complex physical systems studied as part of the projects of the JINR Topical Plan. Mathematical methods and software, including those based on machine and deep learning algorithms, are developed for modelling physical processes and experimental facilities, processing and analysing experimental data in particle physics, nuclear physics, neutrino physics, radiobiology, etc. Particular attention is paid to the creation of systems for the distributed processing and analysis of experimental data, as well as information and computing platforms to support studies underway at JINR and other research centres.
The main direction in the modelling of complex physical systems, including the states of dense nuclear matter and quantum systems, is the development of methods, software packages, and the implementation of numerical studies based on the solution of corresponding systems of nonlinear, spatially multidimensional integral, integro-differential or partial differential equations with a large number of parameters, characterised by the presence of critical modes, bifurcations, and phase transitions with the integrated application of methods of computational physics, quantum information theory, and hybrid quantum-classical programming methods.
An important area of work is the development of the JINR Digital Ecosystem integrating existing and future services that support scientific, administrative, and social activities, as well as the maintaining the Institute’s engineering and IT infrastructure.
Special attention is paid to training highly qualified IT specialists capable of developing algorithmic and software solutions for megascience projects. The HybriLIT Educational and Testing Facility offers courses on parallel programming, machine learning, and deep learning. An annual information technology school is held for undergraduate and graduate students, aimed at engaging young professionals in solving problems in various scientific fields using high-performance and distributed computing, data analysis methods and algorithms, as well as modern information technologies.
MLIT JINR holds the “Distributed computing and Grid technologies in science and education” international conference, “Mathematical modelling and computational physics”, and a number of other major scientific conferences and events.
