Wednesday, December 20, 2023
MLIT Conference Hall, Online seminar via Webinar

MLIT General Laboratory seminar

Program of the seminar:

  1. Alexander Uzhinskiy
    Research on Machine Learning Solutions for Applied and Scientific Challenges Conducted at the Meshcheryakov Laboratory of Information Technologies

    The report presents findings from 2018 to 2023, focusing on two primary research areas: environmental monitoring utilizing Earth remote sensing data, and the application of contemporary automation and machine learning tools in agriculture. It will cover research directions, motivation, goals, objectives, attained results, published works, and future plans. Additionally, a brief overview of other areas within the speaker's sphere of activity will be provided.

  2. Danila Oleynik
    SPD Online Filter

    The expected event rate of the SPD experiment is about 3 MHz (pp collisions at 27 GeV and 1032 cm−2s−1 design luminosity). This is equivalent to a raw data rate of 20 GB/s or 200 PB/year, assuming a detector duty cycle of 0.3. The key challenge of SPD computing is the fact that no simple selection of physics events is possible at the hardware level, since the trigger decision would depend on the measurement of momentum and the vertex position, which requires tracking. Moreover, the free-running DAQ provides a continuous data stream that entails a sophisticated unscrambling prior to building individual events. That is the reason why any reliable hardware-based trigger system turns out to be overcomplicated, and the computing system has to cope with the full amount of data supplied by the DAQ system.
    The main goal of the SPD online filter facility is at least to decrease the data rate by a factor of 20, so that the annual growth of data, including simulated samples, stays within 10 PB.

  3. Artem Petrosyan
    Distributed Storage and Computing Environment for the SPD Experiment

    The SPD facility, being under construction at the NICA collider, will generate large data streams, which, after initial filtering, will need to be stored and processed. Given the expected amount of this data, both JINR resources and external resources provided by the collaboration participants are supposed to be used in organizing storage and processing. The report presents the status of work on the organization of the distributed storage and processing environment of SPD experiment data.

    Information on the seminar and the link to connect are available at Indico.