Highlights of Aragats Research

А.Чилингарян; Физика высоких энергий в атмосфере Земли

(03.2021; ПРИРОДА - ЕЖЕМЕСЯЧНЫЙ ЕСТЕСТВЕННОНАУЧНЫЙ ЖУРНАЛ РОССИЙСКОЙ АКАДЕМИИ НАУК)

Национальная научная лаборатория имени А.Алиханяна (Ереванский физический институт) (Ереван, Армения)Новая методология физики высоких энергий в атмосфере характеризуется последовательным применением методов физики элементарных частиц и ядерной спектроскопии для выявления деталей работы ускорителей электронов, возникающих прямо над нашими головами в грозовых облаках. Совместно с потоками высокоэнергетических электронов и γ-лучей от лавины релятивистских убегающих электронов продолжительностью в несколько минут на исследовательской станции Арагац зарегистрированы также часовые изотропные потоки низкоэнергетических γ-лучей от изотопов 222Rn. Каждый вид космических лучей приносит свои особые признаки, позволяющие оценить структуру и силу атмосферного электрического поля.Эффект торможения мюонов, наблюдаемый одновременно с крупнейшим зарегистрированным увеличением потока электронов и γ-лучей на гореЛомницкий Штит, позволяет оценить максимальное значение атмосферного электрического поля. Измеряя энергетические спектры естественногоγ-излучения, мы обнаружили новый эффект циркуляции дочерних продуктов радона во время гроз. Сравнение энергетических спектров электронов и γ-лучей грозовых наземных превышений позволяет исследовать возникающие электрические структуры в атмосфере, которые ускоряют затравочные электроны космических лучей до ≈70 МэВ. Измеряя одновременно потоки нейтронов и γ-лучей нейтронным монитором и гибриднымдетектором частиц SEVAN, мы смогли доказать фотоядерное происхождение атмосферных нейтронов...читать далее

INTRODUCTION

Aragats research station (Figs. 1,3) of the Artem Alikhanyan National lab (Yerevan Physics Institute) with its unique equipment for measuring particle fluxes and atmospheric discharges is located on Mt. Aragats, at a distance of about 50 km from Yerevan. Aragats is an inactive, cone-shaped volcano with a diameter of about 40 km at its base. Its slopes are characterized by radial valleys that are deeply carved into the permeable volcanic rock. One of the world-first and largest high-mountain research stations was established on Mt. Aragats, at 3200 m elevation, 77 years ago in the middle of World War II in 1943. Since then, expeditions on Aragats continued uninterruptedly, in spite of insufficient funding, and electricity and fuel shortages during the recent history of Armenia. Currently, physicists of the Cosmic Ray Division of Yerevan Physics Institute with a reequipped and renewed facilities continue research in the field of galactic and solar cosmic rays, solar-terrestrial connections, atmospheric physics, and space weather. The main topic of research previously was the physics of the high-energy cosmic rays accelerated in our Galaxy and beyond (see review and references for early Aragats research in Chilingarian et al.,2009).

Figure 1. Aragats high-mountain station against the background of the biblical Mount Ararat, to which is less than 100

km in a straight line

Two surface arrays consisting of hundreds of plastic scintillators were measuring extensive air showers (EASs), the gigantic cascades of particles born in interactions of the primary high-energy proton or fully stripped nuclei with atoms of the terrestrial atmosphere. Since 2000, the CRD has been conducting constant monitoring of the cosmic ray flux at Aragats to study proton accelerators on the Sun and to alert about the dangerous consequences of solar flares. Fortunately, since the 90s, we have left a lot of elementary particle detectors from large ground-based installations that register extensive air showers (EAS), particle fluxes from interactions of protons and nuclei, accelerated to ultrahigh energies by galactic and extragalactic accelerators. After the completion of the EAS experiments on Aragats, research began on a new interesting topic - solar physics and space weather. Neutron monitors located at an altitude of 3200 and 2000 m, and a multitude of new particle detectors measuring the charged and neutral components of secondary cosmic rays, make Aragats one of the largest centers for the study of solar-terrestrial communications. During the 23rd solar cycle, many important solar energy events were measured, including the largest series of ground-level enhancements (GLE) and Forbush decreases in November 2003 (the so-called Halloween events, Chilingarian & Bostanjyan, 2010, see Fig. 2), and huge fluxes of the high-energy solar protons (Bostanjyan and Chilingarian, 2007, Chilingarian, 2009). As we can see in Fig. 2 ground level enhancements (GLE) and Forbush decreases (FD) are well pronounced at high latitudes (Oulu neutron monitor) comparing with middle and low altitudes (Aragats) due to in the first case registration of the huge amount of low energy secondaries from the solar protons, and in the second case, visa-verse suppression of the low energy cosmic ray flux. However, the geomagnetic storm (GMS) is better pronounced on the middle-low latitudes because due to lowering of the cutoff rigidity a huge amount of low energy secondaries from the galactic cosmic rays usually bended from Aragats location by the geomagnetic field.

Figure 2. Halloween events of 2003. By different colors we demonstrate galactic cosmic ray modulation effects by the violent solar activity (solar flares, interplanetary coronal mass ejections and shock waves).

The culmination of research in solar physics was the creation of the SEVAN (Space Weather Observing and Analysis Network) detector network, designed to improve basic space weather research and provide short-term predictions of space storms (Chilingarian et al., 2009, 2018). The SEVAN network consists of hybrid detectors that register charged and neutral components of secondary cosmic rays. The network registers varying fluxes of different types of secondary cosmic rays at different heights, longitudes and latitudes, thus becoming a powerful integrated system used to study the effects of solar activity. After an intense solar flare in January 2005, by which we determined the maximum energy of the solar proton accelerator, the activity of the sun gradually decreased, since then particles from solar flares have not reached Aragats. Only at high latitudes, where the geomagnetic field allows low-energy particles to reach the ground, have several terrestrial increases caused by solar flares been recorded. Accordingly, interest in the study of solar weather diminished, and, starting from 2008 CRD turns to investigations of the high-energy phenomena in the atmosphere. Existing and newly designed particle detectors and the unique geographical location of Aragats station allow observing in 10 years ≈ 500 particle bursts, which were called TGEs— thunderstorm ground enhancements. TGEs observed on Aragats are not only gamma rays, but also sizable enhancements of electrons, positrons, gamma rays and rarely also neutrons. Aragats physicists enlarge the possibilities for TGE research by coherent detection of the electrical and geomagnetic fields, radio emission from the atmospheric discharges, 222Rn progenies gamma radiation, lightning location, rain rate, temperature, relative humidity, and other meteorological parameters. The adopted multivariate approach of investigations allows connecting different fluxes, fields, and lightning occurrences and finally establishing a comprehensive model of the TGE and overall natural gamma radiation (NGR). We discover the new physical effect, namely, Radon circulation in the thunderous atmosphere, that enlarges NGR by several tens of percent.

The newly emerging field of HEPA comprises various physical processes extended to many cubic kilometers in thunderclouds and many hundred cubic kilometers in space. Our scientists on Armenia's Mt. Aragats have discovered mechanisms and characteristics of long-lasting particle multiplication and acceleration produced within thunderstorms and for the first time have measured the energy spectra of electrons and gamma rays of particle avalanches of atmospheric origin that reach the Earth's surface. As direct evidence of the RREA, for the first time was observed the fluorescence light emitted during the development of electron-gamma ray cascade in the atmosphere well correlated with registered by surface particle detectors the high-energy electron flux. Thus, first, we specify TGE phenomena by detecting simultaneous flux of high energy electrons and gamma rays, and neutrons; then we observe RREA by detecting particle showers coming from cloud (extensive cloud showers, next we prove the existence of the lower dipole, accelerated electrons downward; in the same year we perform simulations of the electron propagation in the strong atmospheric electric fields, proving origination of the runaway phenomena; and only in 2019 we present the comprehensive model of TGE and direct optical evidence of RREA origination in the thunderous atmosphere.

In the review paper we will describe the main discoveries of Aragats physicists of recent years.

Figure 3. Lake Kare-lich against the background of the southern peak of Mount Aragats. In the foreground is a magnet from the Alikhanov-Alikhanyan spectrometer; brought in 1945 by Artem Alikhanyan from Germany.

THE ENERGY SPECTRA AT “KNEE” ENERGIES: MAKET-ANI AND KASCADE RESULTS

High-energy processes in Universe are investigated with energetic primary particles (protons, fully stripped nuclei, gamma rays, and neutrinos) traveling millions of years and bringing to the solar system information on the most violent explosions from sites where they born. Neutral particles point on their sources, charged particles lost information about origin due to bending in galactic magnetic fields. However, by measuring the energy spectra of different nuclear groups we can acquire information on the mechanisms of particle acceleration. The secondary particles produced in hadronic collisions of primaries with air nuclei produce pions and kaons, which decay into muons and neutrinos, thus producing deep-penetrating components. The most intense component-electrons and photons-originates mainly from the fast decay of neutral pions into photons, which initiate electromagnetic showers. Thus, one primary particle generates millions of charged and neutral secondary particles that arrive on the earth's surface as Extensive air showers (EAS). Thousands of sensors distributed on areas up to several thousand square kilometer register EASs particles densities and energies. Integrating measured electron and muon densities over the area covered by the EAS leads to estimation of the electron and muon shower sizes, N_e and Nμ, which are related to the mass and energy of the primary hadron. In the 80-ths the construction of the ANI experiment (a joint initiative of YerPhI and Lebedev institute, Danilova et al., 1982) started at Aragats. ANI experiment aims to be the world's largest surface array measuring almost all species of secondary cosmic rays, their energies, and cross-correlations. In this concern, a new methodology for solving the inverse problem of cosmic rays was developed in YerPhI, see Fig. 4 (see for instance Chilingarian, 1982, 1989, 1990). From the distributions of EAS electrons, gamma rays, muons, and hadrons we should recover primary particle type and energy. Due to numerous random fluctuations and methodical errors, as well as, uncertainties in EAS simulation, this problem is extremely difficult to solve. We suggest, and then realize a scheme, that comprises multiple solving of direct problem of cosmic rays (simulations with known parameters of the primary hadron) and implementation of Bayesian and Neural Net models for final statistical inference on the type and energy of each particle, and as well on the "correct" model of the strong interaction at energies not achievable to modern colliders.

The implementation of this program first was done with the data from emulsion chambers (PAMIR experiment, Chilingarian et al., 1986), then for the Crab nebula observations by the WHIPPLE collaboration, then for the MAKET-AN|I surface array, and finally for the KASCADE experiment. It was the first successful attempt to exploit artificial intelligence (AI) and big data concepts for analyzing and making physical inferences from the largest astroparticle physics experiments. Proposals to use these techniques for LHC experiments also were submitted, but not realized (Chilingarian, 1989, Chilingarian and Zazyan, 1991). Outside the physics domain, this technique was successfully implemented for the genome analyses of colon cancer patients (Chilingarian et al., 2002). For the first time, not only one gene but the subclass of genes responsible for this disease were outline and identified by comparison samples of defective and sound cases. In 1997 the MAKET-ANI surface array in Armenia was launched in its full configuration with ~100 plastic scintillators of 1 m2 area each, viewed by photomultipliers. The efficiency of extensive air shower core selection from a surface of ~1000 m2 around the geometrical center of the array was above 95% for EASs generated by primary particles with energy ≥ 5x1014 eV. The compact array with well-calibrated detectors turned out to be very well suited for the energy and composition measurements at the "knee" of the cosmic ray spectrum. Using the non-parametric multivariate methodology of data analysis, the problem of event-by-event classification of EAS has been solved (Chilingarian & Zazyan 1991) using Bayesian and neural network techniques (Chilingarian et al., 2004, Chilingarian et al., 2007). In Fig. 5a we show the EAS classification procedure using 2 variables the electron size of EAS and the AGE parameter, related to the height of the first strong interaction of the primary hadron. The output of a feed-forward neural network used in the classification of simulated data was normalized to 0-1 interval 90 - the goal for the light nuclei, 1 - for the heavy). If we use 0.5 as a decision point for 2-way classification the misclassification of both types can be large; if we introduced the decision "not to make classification", the dashed area in Fig.5a, the purity of selection can be very high, of

course on the price of lower efficiency.

Figure 4. The overall scheme of the statistical inference for the inverse problem solving by multiple solutions of the direct problem with varying parameters.

More than a million EASs detected in 1999-2004 have been carefully examined and used for the estimation of energy spectra of light and heavy nuclei (Chilingarian et al, 2004). In Fig. 5b we show the energy spectra of light and heavy nuclei obtained by the methodology shown in Fig. 5a. The evidence from the recovered energy spectra can be summarizedas follows (Vardanyan et al., 1989, Chilingarian et al., 1999):

Theestimated energy spectrum ofthe light mass groupof nuclei shows a very sharp knee: Dg ^≈ 0.9, compared to^≈0.3 for the all-particleenergy spectra.
In contrast, the energy spectrum of the heavy mass group of cosmic rays shows no break in the energy interval of 10¹⁵ - 2x10¹⁶ eV.
The MAKET-ANI results on the rigidity-dependent position of the knee, points towards SuperNovae remnants (SNRs) as a most favorable source of galactic cosmic rays and Fermi-type acceleration as the mechanism of hadron acceleration. Further results from the KASCADE array and the AGILE and Fermi orbital gamma ray observatories, as well as several theoretical extensions of Fermi mechanism, confirms MAKET-ANI results.

Figure 5. Left the output of the Neural Network trained to distinguish "light" and "heavy" nuclei (from Chilingarian e t al., 2004); Right: Energy spectra oflight and heavy nucleiobtained by neural classificationand energy estimation. TheEAS characteristics used areshower size and shape(age parameter).

The KASCADE experiment (Antony et al., 2003, 2005) that combines the observation of the electromagnetic and muon components with measurements of the hadronic component, operates in 1996-2003. In Fig. 6a we demonstrate the 3-way classification of the simulated EAS reaching the KASCADE detector's sensors. The Electron and Muon sizes of showers were calculated according to the experimental procedures. Because KASCADE measures additional important parameters (number of muons) it was possible to divide all hadron masses not in 2-way as for MAKET-ANI detector, but perform 3-way classification (Fig. 6a), adding the intermediate nuclei class (Oxygen, green).

However, the accuracy of the classification (errors of the first and second type) are smaller for the MAKET-ANI, than for KASCADE, see Table 1. The EASs measured by MAKET-ANI on the height of 3200 m is very near to the cascade maximum and contain more information about the primary than the already degraded cascade measured 3.2 km lower at sea level by the KASCADE detector. The 2-dimensional distributions of light and heavy classes by muon and electron sizes are shown in Fig. 6b. The proton originated events (red spots) as expected are grouped in the left upper corner (enhanced electron size and suppressed muon size) and Iron nuclei originated EASs are grouped lower (more muons, fewer electrons). By squares, triangles, and open circles we show different possible versions of light and heavy nuclei selection. The recovered energy spectra of 3 types of nuclei are shown in Fig.7. There is overall good agreement with the MAKET-ANI data. Thus, the KASCADE data confirms the MAKET-ANI inference on the origin of the "knee". Further observation of the high-energy gamma ray flux from the super-novae remnants performed by orbiting gamma ray observatories and imaging Cherenkov telescopes also confirms physical inference obtained by MAKET and KASCADE surface arrays.

Figure 6. The energy spectra of the 3 species of primary hadrons recovered from the KASCADE observations by the neural classification methods.

Table 1. The 2 types of errors committed during the classification of simulated events for MAKET-ANI and KASCADE experiments arrangement. For the MAKET-ANI experiment electron size and AGE parameters were used; for the KASCADE experiment - the electron and muon sizes of the EAS.

	MAKET	KASCADE
	light heavy	light heavy
light heavy	0.720 0.280 0.240 0.760	0.688 0.312 0.338 0.662

The comparison of MAKET-ANI and enlarged KASCADE-GRANDE arrays are shown in Fig. 8.

Figure 7.

OPENING NEW WINDOW TO UNIVERSE WITH IMAGING AIR CHERENKOV TELESCOPES (IACT)

Recently the GeV-TeV gamma-ray astronomy landscape has witnessed the blossom of several brilliant results. These span from seminal results on previous astrophysical targets to the discovery of new classes of TeV sources. The large collection area, the low energy threshold, and fast reaction of the modern IAST systems, did contribute to unveil the new TeV component of GRBs and to set the stage for the association of the gamma-ray and neutrino emission from blazars (MAGIC collaboration, 2019a, 2019b, H.E.S.S. collaboration, 2019). The first GRB detected by the MAGIC telescopes, known as GRB 190114C, reveals for the first time the highest-energy photons measured from these objects. This ground-breaking achievement by MAGIC provides critical new insight for understanding the physical processes at work in GRBs, which are still mysterious. On January 14th, 2019, a GRB was discovered independently by two space satellites: the Neil Gehrels Swift Observatory and the Fermi Gamma-ray Space Telescope. Within 22 seconds, its coordinates in the sky were distributed as an electronic alert to astronomers worldwide, including the MAGIC Collaboration, which operates two 17m diameter Cherenkov telescopes located in La Palma, Spain, see Fig. 9.

Figure 8. Twin MAGIC-Telescopes (Major Atmospheric Gamma-Ray Imaging Cherenkov Telescopes) at La Palma, Canarias

An automatic system processes in real-time the GRB alerts from satellite instruments and makes the MAGIC telescopes point rapidly to the sky position of the GRB in just 50 seconds after the beginning of the GRB. GeV-TeV emission from a set of blazars did allow us to provide an unprecedented study of the cosmological extragalactic background light. IASTs did also overcome the barrier of 100 TeV maximum energy, looking out into the realm of the galactic cosmic rays' sources, the Pevatrons. The new multi-messenger era does pose new challenges, on which the IACT will play its part. The present generation of IACTs is successfully operating for two decades, paving the path to the Cherenkov Telescope Array (CTA) observatory, presently under construction. However, the IAST era started 30 years ago when gamma rays from the CRAB nebula were detected by the WHIPPLE collaboration (Weekes et al., 1989). The largest challenge in proving IACTs was the rejection of hadron-initiated air showers and the selection of the gamma-initiated showers coming from the source under question (Aharonian, et al., 1989). The typical signal-to-notice ratio for the first IACTs (WHIPPLE, HEGRA) due to the extreme outnumbering of gamma-ray induced air showers by those from cosmic- ray was 10-5 and less. Traditionally, for rejecting this huge background, so-called Hillas Parameters (Hillas, 1985), that parameterize the image obtained from the PM pattern are used, see Fig.10. Images from hadronic showers appear to be both, longer and wider, than those from gamma rays.

Figure 9. Hadron and gamma images captured by the mirror dish of IAST. Definition of the image parameters, so-called, Hillas parameters.

By placing cuts on the image width and length, the signal-to-noise ratio was significantly enlarged (Chilingarian & Cawley, 1990, 1991a and 1991b). A program package, called, "Analysis and Nonparametric Inference (ANI) was under development from 1985 and implemented on different platforms in Lebedev institute, CERN, Forshungszentrum, Karlsruhe, and other centers and was widely used for data analysis from large experiments in high energy astrophysics (Chilingarian, 1985). ANI package was used both for research of the background rejection methods and for the "big" experimental data analysis. The comparison of different background rejection methods is shown in Table 1, where DIFF = N_on^*-N_off^* is the estimate of the signal, DIFF/ N_off^* is the estimate of the signal-to-noise ratio, N_off^*/ N_off is the estimate of background suppression by the technique used. Where N_on.Is the number of gamma ray candidates measured during a scan when telescope axes follow the source of high-energy gamma rays in question; N_off - the same duration scan pointed to the sky where there are no high-energy sources (background). The asterisks mean that the numbers were obtained after implying corresponding cuts.

is an overall measure of the "quality" of background rejection.

Thus, the Neural network provides the best result enlarging the quality of background suppression from 4.8 on raw data to 35.8 and simultaneously keeping 70% of the signal estimated by the raw data difference of ON and OFF scans. For the comparison of different background rejection methods see also (Bock et al., 2004).

Table 2. Comparison of the different background rejection methods for the WHIPPLE detection of CRAB, 1988-1989

	N^*_on	N^*_off	s	DIFF	DIFF/N^*_off	N^*_off/ N_off
Raw Azwidth Wedgecut¹ Supercut² Neural3 4::5::1	506255 14622 6017 4452 6278	501408 11389 3381 1766 2858	4,8 20.4 27.2 34.3 35.8	4847 3233 2636 2686 3420	0.01 0.28 0.78 1.52 1.20	0.0227 0.0067 0.0035 0.0057

¹Chilingarian, A.A. and M.F. Cawley. Application of multivariate analysis to atmospheric Cherenkov imaging data from the Crab nebula. Proc. 22 ICRC, 1, 460-463, Dublin, 1991.
²Punch, M., C.W. Akerlof, M.F. Cawley, et. al. Supercuts: an improved method of selecting gamma-rays. Proc. 22nd Internal. Cosmic Ray Conf., Dublin, 1, 464-467, 1991

³Chilingaryan A. A., Neural classification technique for background rejection in high energy physics experiments, Neorocomputing, 6, 497, 1994.

However, the question remains how to find an optimal set of parameters to reject the background and not too strict suppress the efficiency of event selection. Physicists from CRD solve this problem by introducing a new method directly optimizing the nonlinear multidimensional shape of the best gamma-cluster (Chilingarian, 1994, Chilingarian and Cawley, 1994, Chilingarian, 1995), Fig.11. In this method, we do not use the a-priory information on the shape of the signal domain. This shape is very difficult to simulate due to a variety of very high-energy gamma ray sources and numerous random and methodical errors. Using only experimental information from the scans when telescope axes were pointed on the source in question and open space, we avoid additional uncertainties connected with using nonreliable a-priory information. Thus, we use only experimental information: the source + background and pure background. In the upper row of Fig. 10, we show the model of signal (nonlinear shape) overlaid on the background and uniformly distributed random background. In the middle row we show one of the steps of working of iterative algorithm optimizing the signal domain shape, and in the bottom row - the final shape that covers signal domain and minimally includes background. 2 algorithms were used (left and right sides of Fig. 10, see details in Chilingarian, 1997).

Figure 10. Implementation on a toy problem the new type of NN - Mapping Networks (MP). MP maximizes the signal significance, directly optimizing the shape of the Gamma Cluster.

REFERENCES:

Aharonian F.A., Chilingarian A.A., Pluasheshnikov A.et.al., (1991) A multidimensional analysis of the Cherenkov images of air showers induced by very high energy y-rays and protons1991, NIM A302, 522; YerPhI preprint 1171 (48)-79, 1989.

T.Antoni, A. Chilingarian, et. al.,for the KASCADE collaboration, The Cosmic-Ray Experiment KASCADE, Nuclear Instruments & Methods, A513, (490-510), 2003.

T. Antoni, W.D. Apel, A. Chilingarian, et al., KASCADE Measurements of Elemental Groups of Cosmic Rays: Results and Open Problems, Astroparticle Physics, 24, (1-25), 2005.

R.K. Bock, A. Chilingarian, et. al., Methods for Multidimensional Event Classification: a Case Study Using Images from a Cherenkov Gamma-Ray Telescope, Nuclear Instruments and Methods (NIM) in Physics Research A516, (511-528), 2004.

N.K. Bostanjyan, A. Chilingarian et al., On the production of highest energy solar protons at 20 January 2005, J. Adv. Space Res. 39, (1456-1459), 2007.

A.Chilingarian, Analysis of Data Interpretation Methods for ANI Experiment, Problems of Nuclear Science and Technology, techniques of physics experiment series, 3/12/1982.

Chilingarian, S. Galfayan, М. Zazyan, A. Dunaevsky, The Multivariate Analysis of Data Obtained in Experiments with Roentgen-Emulsion Chambers and in EAS Experiments, Preprint of Physics Institute of USSR Academy of Science, № 332, 1986

A.Chilingarian, Statistical Conclusions under Nonparametric A Priori Information, Computer Physics Communication, Vol.54, (381-390), 1989.

A.Chilingarian, Dimensionality Analysis of Multiparticle Production at High Energies, Preprint YerPhi -1213(90), 1989

Chilingarian A. Statistical Method of Elementary Particle Mass Determination Via Indirect Measurements Using Simulation Data, Preprint YerPhi-1221(7), 1990

Chilingarian, G. Zazyan, The Analysis of Multiparticle Production at High Energies with Renie Dimensions, Computer Physics Communication 56, 1991.

Chilingarian A. A., Statistical Decisions under Nonparametric a priori information, Computer Physics Comunication, 54, 381, 1989.

A.Chilingaryan, N. Gevorgyan, A. Vardanyan, D. Jones and A. Szabo, Multivariate Approach for Selecting Sets of Differentially Expressed Genes, Math. Biosciences, Vol.176(1), (59-69), Elsevier Science Inc. PII: S0025-5564(01)00105-5, 2002.

A.Chilingarian, G.Zazyan, On the possibility of investigation of the mass composition and energy spectra of primary cosmic ray (PCR) in the energy range from 10¹⁵ to 10¹⁷ eV using EAS data, Nuovo Cimento C, Serie 1 (ISSN 0390-5551), vol. 14 C, Nov.-Dec. 1991, p. 555-568, http://adsabs.harvard.edu/abs/1991NCimC..14..555C

A.Chilingarian , G. Gharagyozyan , G. Hovsepyan , S. Ghazaryan , L. Melkumyan, and A. Vardanyan, Light and Heavy Cosmic Ray Mass Group Energy Spectra as Measured by the MAKET-ANI Detector, Astrophysical Journal, 603:L29-L32, 2004, http://iopscience.iop.org/1538-4357/603/1/L29/fulltext/

A.Chilingarian , G. Gharagyozyan , G. Hovsepyan , S. Ghazaryan , L. Melkumyan , A. Vardanyan, E.Mamidjanyan, V.Romakhin, and S. Sokhoyan Study of extensive air showers and primary energy spectra by MAKET-ANI detector on Mount Aragats, Astroparticle Physics, Volume 28, Issue 1, September 2007, Pages 58-71 http://www.sciencedirect.com/science/article/pii/S0927650507000552

Vardanyan A., A.Chilingarian , M.Roth for the KASCADE collaboration, in Proceedings of the workshop ANI 99, Nor-Amberd, 1999, Preprint FZK 672, p.23.

Chilingarian A., Gharagyozyan G., et.al., (1999), Proc. Of the ANI 99 Workshop Edited by Chilingarian , Haungs, Rebel, Zazyan, Nor Amberd, Armenia. Forschungszentrum, Kahrlsrue, preprint #6472

Chilingarian A.,1985, Analysis and Nonparametric Inference in High Energy Physics and Astroparticle Physics, 1997. Program Package ANI, (User Manual) http://crd.yerphi.am/ANI_User_Guide_Introduction Chilingarian, A.A. and M.F. Cawley, Multivariate analysis of Crab Nebula data, Note to Whipple collaboration from July 5, 1990.

Chilingarian, A.A. and M.F. Cawley (1991). Application of multivariate analysis to atmospheric Cherenkov imaging data from the Crab nebula. Proc. 22 ICRC, 1, 460-463, Dublin, 1991a.

Chilingarian, A.A., Classification of the gamma and proton images with aid of mathematical models of Nural Networks, Proc. 22ICRC, 1, 483, Dublin, 1991b.

Chilingarian A. A.: (1994) Neural classification technique for background rejection in high energy physics experiments. Neorocomputing, 6; 497.

Chilingarian, A.A. and M.F. Cawley , Optimizing the non-linear gamma-ray domain in {VHE} gamma-ray astronomy using neural-network classifier Proc. 24 ICRC, 3, p. 742, Rome, 1994.

Chilingarian A. A.(1995), Detection of weak signals against background using neural network classifiers. Pattern Recognition Letters, 16; 333.

Chilingarian, A., The Non-linear Signal Domain Selection using a New Quality Function in Neural Net Training, Nuclear Instruments and Methods (NIM), 389A, (242), 1997.

Chilingarian, R. Mirzoyan and M. Zazyan, Cosmic Ray research in Armenia, Advances in Space Research 44 (2009) 1183-119.

A.Chilingarian and N. Bostanjyan, On the relation of the Forbush decreases detected by ASEC monitors during the 23rd solar activity cycle with ICME parameters, Advances in Space Research, 45, Issue 5, 2010, 614-621.

A.Chilingarian, Statistical study of the detection of solar protons of highest energies at 20 January 2005, Advances in Space Research 43 (2009) 702-707

A.Chilingarian, G. Hovsepyan, K. Arakelyan, S. Chilingaryan, V. Danielyan, K. Avakyan, A. Yeghikyan, A. Reymers, S. Tserunyan, Space Environmental Viewing and Analysis Network (SEVAN), Earth, Moon and Planets: Vol.104, Issue 1, (195), 2009.

A.Chilingarian, V. Babayan, T. Karapetyan, et al., The SEVAN Worldwide network of particle detectors: 10 years of operation, Advances in Space Research 61 (2018) 2680-2696.

Т. Danilova, A. Dunaevsky, A. Erlikin, A. Chilingarian, et al., Project of Hadron Interaction in the Energy Ranges103 -105 TeV Experiment (ANI Experiment), Proceedings of Academy of Science of Armenian Soviet Republic, physics series, Vol. 17, issue 3-4, 1982.

Hillas, A. M. 1985, Cerenkov Light Images of EAS Produced by Primary Gamma Rays and by Nuclei, Proceedings from the 19th International Cosmic Ray Conference, Volume 3 (OG Sessions), p.445

MAGIC collaboration, Teraelectronvolt emission from the γ-ray burst GRB 190114

Nature Vol 575, 455-458 (2019)

MAGIC collaboration, Observation of inverse Compton emission from a long γ-ray burst

Nature Vol 575, 459-463 (2019)

H.E.S.S. collaboration, A very-high-energy component deep in the γ-ray burst afterglow

Nature Vol 575, 464-467 (2019)

Weekes, T. C., Cawley, M., Fegan, D., et al. 1989, Observation of TeV Gamma Rays from the Crab Nebula Using the Atmospheric Cerenkov Imaging Technique, The Astrophysical Journal, 342, 379

ON THE PRODUCTION OF HIGHEST ENERGYSOLARPROTONS AT 20 JANUARY 2005

SHAPE OF THUNDERSTORM GROUND ENHANCEMENT AND RADON PROGENY RADIATION

HIGH-ENERGY PHYSICS IN ATMOSPHERE: IN SITU MEASUREMENTS OF THE RUNAWAY BREAKDOWN (RB) AND THUNDERSTORM GROUND ENHANCEMENTS (TGES) ON ARAGATS MOUNTAIN