=Paper=
{{Paper
|id=Vol-2679/short7
|storemode=property
|title=Educational and Outreach Resource for Astroparticle Physics
|pdfUrl=https://ceur-ws.org/Vol-2679/short7.pdf
|volume=Vol-2679
|authors=Yulia Kazarina,Vasiliy Khristyuk,Alexander Kryukov,Evgeny Postnikov,Vladimir Samoliga,Alexey Shigarov,Victoria Tokareva,Dmitriy Zhurov
}}
==Educational and Outreach Resource for Astroparticle Physics==
https://ceur-ws.org/Vol-2679/short7.pdf
Educational and Outreach Resource for
Astroparticle Physics
Yulia Kazarina1 , Vasiliy Khristyuk2 , Alexander Kryukov3 , Evgeny Postnikov3 ,
Vladimir Samoliga1 , Alexey Shigarov2 , Victoria Tokareva4 , and
Dmitriy Zhurov1
1
Applied Physics Institute of ISU,Irkutsk, Russia
2
Matrosov Institute for System Dynamics and Control Theory, Siberian Branch of
Russian Academy of Sciences, Irkutsk, Russia
3
Lomonosov Moscow State University, Skobeltsyn Institute of Nuclear Physics,
Moscow, Russia
4
Institute for Nuclear Physics, KIT, Karlsruhe, Germany
Abstract. The modern astrophysics is moving towards the enlarging of
experiments and combining the channels for detecting the highest en-
ergy processes in the Universe. To obtain reliable data, the experiments
should operate within several decades, which means that the data will
be obtained and analyzed by several generations of physicists. Thus, for
the stability of the experiments, it is necessary to properly maintain
not only the data life cycle, but also the human aspects, for example,
attracting, learning and continuity. To this end, an educational and out-
reach resource has been deployed in the framework of German-Russian
Astroparticle Data Life Cycle Initiative (GRADLCI).
Keywords: Astroparticle Physics, TAIGA observatory, Baikal-GVD neu-
trino telescope, astroparticle.online, Multi-messenger Astronomy, Deep
Learning, CNN, Gamma-Hadron Separation
1 Introduction
The only way to study high-energy processes occurring within and outside the
Milky Way is to detect the radiation and ultra-high-energy particles generated
by these processes. The flux of ultrahigh-energy cosmic rays, gamma rays and
neutrinos interacting with the atmosphere gives rise to cascades of secondary
particles. Reaching the ground, these cascades can cover areas of tens of km2 ,
moreover, with an increase in the energy of the initial particle, their flux drops
sharply, reaching one particle per year per thousand km2 . Thus, over the past few
years, experimental astrophysics of ultrahigh energies has been moving towards
the enlarging of experiments and combining the channels for detecting high-
energy processes named multi-messenger astronomy [1].
Copyright © 2020 for this paper by its authors. Use permitted under Creative
Commons License Attribution 4.0 International (CC BY 4.0).
The Baikal region is a unique place in Russia since two of the largest setups,
investigating three channels of multi-messenger astronomy, are deployed here:
the TAIGA gamma observatory [2], detecting cosmic rays and gamma rays, and
the Baikal-GVD deep underwater telescope, detecting neutrinos [3]. Already, the
flow of raw experimental data in these setups amounts to several terabytes per
day. With the expansion of existing and commissioning of new setups, the data
flow will grow many times over, which will lead to a slowdown in the rate of data
processing and a decrease in the efficiency of experiments in general. To avoid
such a scenario, it is necessary to pay great attention to planning the life cycle of
experimental data (from modeling to publishing data in the public domain and
archiving data), predicting the volume of data flow and assessing the prospects
for using new approaches to data processing to solve new physical problems.
Life cycle planning should address such issues as developing new approaches to
storing tasks, finding and setting physical tasks that can be solved within the
framework of this experiment, assessing the complexity and execution time of
tasks related to data analysis. Besides, real-time preliminary data analysis is
an important area for development. The presence of an online analysis system
will allow one to quickly respond to problems and improve data quality. Also,
online analysis will allow TAIGA and Baikal-GVD experiments to be prepared
for multi-messenger astronomy and interaction with other setups around the
world.
To meet this challenge the new Baikal Multimessenger Lab was established at
the Irkutsk State University (ISU) with the support of the Ministry of Education
and Science of the Russian Federation this year. The declared missions of the
Lab are:
– Creation of a common framework for experiments in the Baikal region (Baikal-
GVD and TAIGA);
– Integration of these setups into full-stack multi-messenger astronomy;
– Creation of a competitive school for astroparticle physics at ISU.
So, the important goal of the Lab is to attract more students to the astropar-
ticle physics and train highly qualified specialists in the field of data processing
and analysis for multi-messenger astronomy. The educational and outreach re-
source astroparticle.online contributes to the achievement of this goal.
This article is about the resource astroparticle.online [4], [5], its goals
and application for Baikal region experiments. In Section 2 there is the descrip-
tion of the resource on the whole and Section 3 is devoted to the interactive part
of the resource, developed for gamma-ray astronomy tasks using convolutional
neural networks (CNN).
2 Web Resource astroparticle.online
The deployment of the astroparticle.online resource (Fig. 1) was started in
2018 in the frame of the German-Russian Astroparticle Data Life Cycle Initia-
tive (GRADLCI) [6]. The resource is built on a free and open-source content
Fig. 1. Screenshot of the first page of the resource astroparticle.online.
management system WordPress. The servers of the platform are located at the
Matrosov Institute for System Dynamics and Control Theory. The main target
audience —- students, who are interested in astroparticle physics and would like
to collaborate with the Baikal region astroparticle physics experiments.
The resource has several sections: News on astronomy and astrophysics (up-
dated weekly), Science, Experiments, Projects dedicate to the Theory of mes-
sengers and astroparticle physics experiments and projects, respectively. These
sections aim to attract younger students and schoolchildren and contain text ma-
terials, also video materials that are borrowed from other sources with referring
to the source, quizzes for better assimilation of the material.
The section Online School, the largest one, is mostly original, partly based on
the new course in astrophysics launched at the ISU in 2019. It has the following
subsections:
– Data Analysis;
– Lections;
– Seminars;
– Labs;
– Popular Science;
– Tasks.
3 Client on CNN for Gamma/Hadron Separation
Ground-based gamma-ray astronomy studies very energetic radiation of galactic
and extragalactic origin by specially designed telescopes, the so-called Imaging
Air Cherenkov Telescopes (IACTs) [7]. With this technique, gamma-rays are ob-
served on the ground optically via the Cherenkov light emitted by air-showers of
secondary particles when a very-high-energy gamma-ray strikes the atmosphere.
Gamma-rays of such energies contribute only a fraction below one per million
to the flux of cosmic rays, most of which are protons [8]. Nevertheless, being
particles without electric charge they can be extrapolated back to their origin,
which makes them the best “messengers” of exotic and extreme processes and
physical conditions in the Universe. That is why particle identification (gamma-
ray discrimination against the cosmic-ray background) is an essential part of
data analysis for the IACT technique.
The subsection Data Analysis contains a prototype of the Astroparticle
CNN client developed for gamma-ray astronomy tasks as an interactive ser-
vice for students. This prototype provides access to an on-line analysis of the
gamma/hadron separation using convolutional neural networks developed as
part of the GRADLCI project [9]. The Monte Carlo events of the TAIGA-IACT
telescope are used as input for this prototype. The developed convolutional neu-
ral network gets the probability of the gamma reconstruction for each event
as a result. There is also a possibility to check your skills in gamma/hadron
separation using the telescope image.
A shower image is fitted as an ellipse. The ellipse is characterized by its axes
and has parameters: length, width, distance and the angular miss-alignment
of the major axis. In comparison with hadron showers gamma-ray ones have
more elliptic shape, less width and major axis pointed to the source. The neural
network underlying the prototype takes these features into account.
The interactive service contains:
– Guessing game on gamma/hadron separation (Fig. 2);
– Instruction how to define gamma-event using telescope image;
– Prepared datasets that are ready for downloading;
– Application for processing your own dataset;
– Tools:
• Script for data visualization + instruction;
• Script for converting data files to HDF5 format + instruction.
The back end of the service is deployed as a microservice in a Docker con-
tainer. It is written in Python and is based on the Django framework. SQLite
is used as the Database Management System, since it contains only informa-
tion about preloaded datasets. The database includes two tables: a dataset table
and an event table. The database structure is defined by the model described
using the Django framework. Tables are populated from HDF5 files with data
sets during system deployment. Also during deployment of the system the event
image files are generated.
4 Conclusion
The astroparticle.online resource is intended to become an education and
outreach instrument for the new Baikal Multimessenger Lab as well as to adver-
tise the experiments of the Baikal region such as the TAIGA gamma observatory
Fig. 2. Screenshot with guessing game on gamma/hadron separation.
and the Baikal-GVD neutrino telescope. We hope that this resource will attract
more students to the astroparticle physics and will allow us to train highly quali-
fied specialists in the field of data processing for astrophysical experiments. This
can solve a problem of great importance, namely, to prepare the above experi-
ments for multi-messenger astronomy and for the interaction with other setups
around the world.
The resource is filled with materials and tasks in astroparticle physics and
its content is regularly updated. The prototype of the Astroparticle CNN client
is developed. It is implemented in the resource astroparticle.online as an
interactive service for illustrating one of the most complicated challenges in
gamma-ray astronomy - the gamma/hadron separation.
Acknowledgements This work was supported by Russian Science Foundation
Grant 18-41-06003 (Section 2, 3), by the Helmholtz Society Grant HRSF-0027
and by the Russian Federation Ministry of Science and High Education (project.
FZZE-2020-0024). We are grateful to the members of the GRADLCI for the
informational support of our activity.
References
1. I. Bartos and M. Kowalski, Multimessenger Astronomy. 2399-2891, IOP Publishing,
2017.
2. N. Budnev et al., TAIGA - a hybrid array for high energy gamma astronomy and
cosmic ray physics, EPJ Web Conf., vol. 191, p. 01007, 2018.
3. A. D. Avrorin et al., The prototyping/early construction phase of the BAIKAL-GVD
project, Nucl. Instrum. Meth., vol. A742, pp. 82–88, 2014.
4. Y. Kazarina et al., Towards the Baikal Open Laboratory in Astroparticle Physics,
CEUR Workshop Proceedings, Vol. 2406, pp. 1-6, 2019.
5. Y. Kazarina et al., Application of HUBzero platform for the educational process in
astroparticle physics, CEUR Workshop Proceedings, Vol. 2267, pp. 553-557, 2018.
6. Bychkov, I. et al.Russian-German Astroparticle Data Life Cycle Initiative, Data
3(4), 56 (2018).
7. T. C. Weekes et al. Observation of TeV Gamma Rays from the Crab Nebula Using the
Atmospheric Cerenkov Imaging Technique, Astrophys. J., 342, pp.379–395, 1989.
8. E. Lorenz, R. Wagner Very-high energy gamma-ray astronomy, EPJ, H 37, pp.
459–513, 2012.
9. E.Postnikov et al. Gamma/Hadron Separation in Imaging Air Cherenkov Telescopes
Using Deep Learning Libraries TensorFlow and PyTorch, Journal of Physics: Con-
ference Series, Vol.1181, p. 012048, 2019.