At SPD collaboration meeting On 12–16 May 2025, the IX SPD (Spin Physics Detector) collaboration meeting took place in Armenia. At A. Alikhanyan National Science Laboratory (AANL), scientists discussed the progress of work on creating the experimental facility and exchanged experiences and results. On the first day of the meeting, MLIT Senior Researcher, SPD Software Coordinator Danila Oleynik spoke at the plenary session about the development of the experiment’s computing infrastructure. The MLIT team is actively elaborating a primary data processing system (SPD Online Filter). Significant progress in developing a middleware that enables multi-stage, high-throughput data processing has been made. In December 2024, the load testing of the distributed data processing system was conducted on the basis of the JINR MLIT Multifunctional Information and Computing Complex (MICC) and Petersburg Nuclear Physics Institute (PNPI) of the National Research Center “Kurchatov Institute”. Since March of this year, the system has been put into operation, and over 400 million events have been simulated for the experiment, almost the entire volume of events has been reconstructed, and thus the total number of processed events is close to a billion (excluding events processed during load testing). The development of the common Sampo software platform for the SPD experiment application software, which will replace the existing SPDRoot platform, is underway. The setup of a dedicated data storage for the SPD experiment was completed. The storage system is implemented using EOS technology. With a raw storage capacity of 7.2 PB (5.3 PB usable), it will meet the experiment’s needs for the next 3–4 years. The speaker highlighted the need to develop systems for monitoring the computing infrastructure and data processing to ensure stability and efficiency in operations. MLIT Senior Researcher, SPD Deputy Software Coordinator Artem Petrosyan presented an overview of the offline data processing system for the SPD experiment. Work is being performed in the following directions: storage and processing systems, migration from X.509 certificates to JWT, development of a processing control panel, support of the data management system. At present, the transfer from the institute-wide storage system eos.jinr.ru to an installation recently allocated for the collaboration spd-eos.jinr.ru is being carried out. PNPI also participates in storing the experiment’s data, and automatic data replication is configured between the centers. The JINR CICC, a complex of information and computing resources for the Institute’s local and international projects, and the computing farm at PNPI are primarily responsible for task processing. Recently, colleagues from Samara University have purchased equipment on which the required software has been installed during the winter and spring of this year, and tests on sending tasks have already been conducted. For the more convenient management of data processing, a web application, i.e., a processing control panel, was created. Over the past few months, within the distributed data processing environment of the SPD experiment, nearly a billion physical events have been processed as part of Monte Carlo simulation, and more than 200 TB of data have been written to storage. A project to develop a unified service for the consolidated monitoring of the infrastructure and services was launched. The transfer from the old VOMS system to the SPD Identity and Access Management Service (IAM), deployed and supported at MLIT JINR, was completed. This made it possible to make the IAM service the only access point to the experiment’s computing infrastructure. All communication between users and services was transferred from X.509 standard certificates to JSON Web Token technology. Now users do not need to obtain certificates to work with collaboration services, it is enough to authenticate in SPD IAM. For the past six months since the last collaboration meeting, the system itself has been connected to JINR SSO, and thus the experiment’s computing infrastructure and intermediate services are fully integrated into the perimeter of JINR information security. MLIT laboratory assistant Nikita Monakov delivered for the first time at the collaboration meeting a talk on the status of a project to develop the data processing management system. Its user-friendly web interface makes it easy to send requests to the workload management system. The system is handed over to users. Since the spring of this year, task description has been handled by the representative of the physical groups of the collaboration responsible for data processing. The users of the system have created over 20 simulation requests and produced almost a billion physical events and more than 200 TB of data. The system is integrated with CRIC, PanDA, Rucio, and allows minimizing the human factor in creating tasks for the functioning of the distributed computing environment, as well as automating a multitude of operations that previously required manual processing. The system is developing. MLIT trainee researcher Alexey Konak presented the status of work on the Rucio data management system. The system performs its major functions perfectly. Statistics on the data stored under the system’s management were considered. Since the previous meeting, the system has been updated to the latest stable version, modules specific to the SPD collaboration have been elaborated for registering group accounts, integrating with the CRIC information system, setting up internal monitoring panels, and exporting information about free and occupied space on available storage systems managed by Rucio to the PanDA system. The development status and plans for the enhancement of the SPD Online Filter middleware, a hardware and software complex for the high-throughput processing of primary data, was presented by MLIT trainee researcher Nikita Greben. The talk described the main components of the software, namely, the data management system, the workflow management system, and the workload management system. MLIT laboratory assistant Polina Korshunova delivered a report on the data management system for SPD Online Filter. The tasks of the data management system and its components, as well as the implemented functionality of each of the services: dsm-register, dsm-manager, dsm-inspector, were reviewed. The implementation of the stage of integration of the data management system with the workload management system and the workflow management system was considered. In addition, Dzmitry Budkouski from VBLHEP JINR demonstrated the results of the development of the methodology, algorithms and software for investigating the cluster production of hadrons in the SPD experiment. The work was performed jointly with MLIT represented by the Laboratory’s Director Sergei Shmatov and Yerevan Physics Institute represented by Armen Tumasyan. The possibility of observing particle clusters and defining the kinematics of partons of a hard process by their parameters was tested. Two types of processes were considered: qg –> qγ and all QCD processes simulated by the Pythia and Herwig generators. The particles were clustered using the anti-k_{T} algorithm implemented in the FastJet package. As a result, it was shown that using a decision tree it was possible to reconstruct the parton transverse momentum. A comparison of the kinematic properties of the particles predicted by the Herwig and Pythia generators for the SPD experiment energy was given. Throughout the week, the SPD collaboration participants discussed in detail the current state of work on the major subsystems of the facility, electronics, and software. Particular attention was paid to the consideration of the physics research program of the experiment’s first stage. Over 160 scientists from all over the world took part in the IX SPD collaboration meeting in a mixed format. The program of the event embraced more than 60 talks devoted to the investigation of the spin structure of polarized particles in collisions at high luminosity, the creation of the experimental facility, and the development of the experiment’s data processing system. Read also on JINR's website.
