Wednesday, May 19, 2021
MLIT Conference Hall, Online seminar via Webex
Daria Priakhina

Data center simulation and data analysis on the use of data components

A program that enables the synthesis of modeling and monitoring processes for the development of systems for storing and processing large data arrays is developed. It has been applied to simulate a number of systems.
Currently, a new software tool for modeling is being developed. It is derived from the approach based on the representation of information processes as byte streams and the use of probability distributions of significant data acquisition processes; in particular, the probabilities of losses of incoming information for different configurations of the data center equipment are to be defined. The first results of modeling centers for processing and storing data of the BM@N experiment of the NICA project for the next run scheduled for 2021 are obtained.
In addition, work that is aimed at data mining on the use of the HybriLIT platform to enhance the efficiency of using high-performance computing platforms has started.
Alexander Gusev

Finite element method for solving boundary value problems for systems of ordinary differential equations

We have developed an algorithmic approach to the construction of computational schemes of the finite element method of high order of accuracy for solving boundary value problems for systems of ordinary differential equations of the second order in the strong coupled channels method, namely, scattering problems with both uncoupled and non-uncoupled asymptotic solutions and a different number of open channels in left and right asymptotic regions, as well as problems for bound and metastable states. Using the developed approach and its software implementation, calculations of the spectrum of vibrational-rotational bound and metastable states of beryllium dimer (JQS&RT(2021)) and the cross-section of sub-barrier reactions of the fusion of heavy ions (PRC(2021)) are performed.
Igor Pelevanyuk

Integration of geographically distributed heterogeneous resources based on the DIRAC Interware

The DIRAC Interware platform enables the integration of distributed heterogeneous computing resources and storage systems into a unified system. Since 2009, it has been developed as a versatile open source tool. DIRAC encompasses a number of services related to load management, data management, a web interface, accounting of consumed resources, authorization and authentication, workload management and some others. Four interfaces are presented to the user to choose from: web interface, command line interface, python programming interface and REST interface. Today, the unified DIRAC-based environment, which comprises both computing resources and data storage systems, is used for centralized Monte-Carlo generation of the MPD experiment, for the study of the SARS-CoV-2 virus within the Folding@Home project and for Monte-Carlo generation of the Baikal-GVD experiment. An approach to evaluate the performance of heterogeneous computing resources on the basis of the results of the automatic launch of the DB12 benchmark is developed.

More information on the seminar and the link to connect via Webex are available at Indico.