Searching for EeV photons with Telescope Array Surface Detector and neural networks.

J. Zvirzdin · 2025

Read Searching for EeV photons with Telescope Array Surface Detector and neural networks. by J. Zvirzdin (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-01638.txt.

Introduction

Zvirzdin searches for ultra-high-energy photons with the Telescope Array. Such photons can test astrophysics models and new physics. The array's surface detector has run for many years. Its data holds a vast set of air shower events.

A neural network sorts showers by their likely source particle. Photon showers differ in shape from proton showers. The sort yields updated caps on the diffuse photon flux. Tight caps rule out models that would flood the sky with such photons. They also bound exotic ideas about dark matter decay. Fourteen years of data give the caps weight, since rare photon showers need huge exposure before a null result means anything at all.

Catalog record

CREATORJ. Zvirzdin
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE41 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-01638

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Searching for EeV photons with Telescope Array Surface Detector and neural networks. J. Zvirzdin Abstract Ultra-high-energy photons play an important role in probing astrophysical models and beyond-Standard-Model scenarios. We report updated limits on the diffuse photon flux using Telescope Array’s Surface Detector data collected over 14 years of operation. Our method employs a neural network classifier to effectively distinguish between proton-induced and photon-induced events. The input data include both reconstructed composition-sensitive parameters and raw time-resolved signals registered by the Surface Detector stations. To mitigate biases from Monte Carlo simulations, we fine-tune the network with a subset of experimental data. The number of observed photon candidates is found to be consistent with the expected hadronic background, yielding upper limits on photon flux Φ γ ​ ( E γ > 10 19 ​ eV ) < 2.3 ⋅ 10 − 3 \Phi_{\gamma}(E_{\gamma}>10^{19}\text{eV})<2.3\cdot 10^{-3} , and Φ γ ​ ( E γ > 10 20 ​ eV ) < 3.0 ⋅ 10 − 4 \Phi_{\gamma}(E_{\gamma}>10^{20}\text{eV})<3.0\cdot 10^{-4} ( km 2 ⋅ sr ⋅ yr ) − 1 (\text{km}^{2}\cdot\text{sr}\cdot\text{yr})^{-1} . 1 Introduction Ultra-high-energy (UHE) photons of astrophysical origin, with energies above 10^{18} eV, are of interest for two main reasons. First, they can be produced via the Greisen-Zatsepin-Kuzmin (GZK) mechanism [ 44 , 83 , 82 , 68 , 40 , 47 , 41 ] in the interstellar medium. The resulting photon flux depends on the mass composition of cosmic rays, thus providing an independent tool for their analyses. Second, registering ultra-high-energy photons beyond the expected GZK flux level would provide strong evidence for models beyond the Standard Model of particle physics [ 26 , 23 , 35 ] , including, in particular, decaying dark matter [ 25 , 63 , 64 , 53 , 55 , 30 , 33 ] and violation of Lorentz symmetry [ 31 , 39 , 74 , 16 ] . Experimental searches for ultra-high-energy photons rely on complementary detection techniques, each with distinct advantages and limitations. Fluorescence Detectors measure fluorescence light from extensive air showers (EAS), allowing for the direct observation of the longitudinal shower development and the depth of shower maximum ( X_{max} ). This provides excellent discrimination between photon and hadron primaries, but their operation time is restricted to clear, moonless nights, resulting in a limited duty cycle that constrains the exposure accumulation for ultra-high-energy events. Conversely, scintillator-based Surface Detectors, measuring secondary particles at the ground level, operate continuously, offering the vast exposure required to probe low-flux regimes. However, they face challenges in distinguishing muon and electromagnetic components of EAS, whose ratio is the key for distinguishing hadron-induced and photon-induced EAS. To address this, experiments like Yakutsk [ 43 ] employed dedicated muon detectors to suppress the hadronic background. Similarly, the Pierre Auger Observatory has set stringent limits in the Southern Hemisphere by utilizing both hybrid observations [ 11 ] and water-Cherenkov Surface Detectors [ 15 ] , which possess different sensitivities to the muon component compared to scintillators. Early constraints were also established by the AGASA experiment [ 75 ] . None of the experiments reported an excess of photon-induced events above the expected hadronic background, producing photon-like events due to fluctuations of EAS evolution. In this paper, we search for ultra-high-energy photons using Telescope Array (TA) experiment data. TA is a large-scale cosmic ray experiment located in Utah, USA, 39°17’ 49” N 112°54’ 31” W, at approximately 1.4 km above sea level [ 18 , 80 , 81 ] . It aims to study ultra-high-energy cosmic rays (UHECR) using its Surface Detector (SD), Fluorescence Detector (FD), and Low Energy (TALE) extension. Operating since 2008, TA has accumulated a large number of EAS events, allowing for thorough studies of cosmic ray properties. Unlike FD, SD operates under almost all weather conditions, but provides less informative data. To compensate for this, we employ a neural network classifier that is capable of leveraging complex patterns in data to effectively discern between photon- and proton-induced events. In the astrophysical field, neural networks have been successfully applied in TA [ 56 , 50 , 54 ] , Pierre Auger Observatory [ 2 , 12 , 42 ] , KASCADE-Grande [ 65 ] , and neutrino telescopes [ 49 , 48 , 34 , 76 , 57 , 58 ] . These examples illustrate the potential of using machine learning in our analysis using TA SD data. Our method incorporates a blind optimization of the neural network classification threshold to establish the most stringent constraint on the photon flux. To mitigate discrepancies between Monte Carlo simulations and the experimental data, the technique includes a procedure for fine-tuning the neural networks using a relevant subset of the experimental data. This allows us to achieve robust and efficient separation between hadron-induced and photon-induced EAS. The TA SD array, depicted in Figure 2 , comprises 507 above-ground scintillation stations arranged in a square grid covering an area of approximately 700 km 2 . Each SD station consists of two layers of plastic scintillator, which register signals with a time resolution of 20 ns [ 18 ] . All stations are calibrated in real time using atmospheric muons, with the signal measured in conventional units called a "minimal ionizing particles" (MIPs). For our analysis, we retain 128 consecutive time bins from each layer of triggered SD stations, which we refer to as waveforms. An example of a waveform registered by a single SD station is presented in Figure 2 . Figure 1 : Schematic representation of the arrangement of the TA SD and FD stations, [ 18 ] . Figure 2 : An example of processed and calibrated waveforms registered by a TA SD station. Blue and orange lines depict signals from the upper and lower scintillator layers, respectively. The sensitivity threshold of the TA SD for photon-induced events is approximately 10^{18.1} eV (see Figure 3 ). While this allows us to derive an integral limit for E>10^{18} eV, the very limited exposure between 10^{18} and 10^{18.5} eV implies that such a limit would be dominated almost entirely by higher-energy photons, and hence could be misleading. We therefore present results only for E>10^{18.5} eV, where the TA SD has sufficient direct sensitivity. 2 Method We start by outlining the general scheme of the analysis and stating our main assumptions. The method comprises four main steps, which we describe below. 1. Data selection and preparation. We perform Monte Carlo (MC) simulations of 14 years of TA SD data, modeling air showers initiated by protons and photons. Heavy nuclei such as iron produce showers that develop earlier, exhibit denser cores, and higher muon-to-electromagnetic ratios than proton-induced cascades [ 73 , 72 , 29 ] . This makes them easier to distinguish from photon-induced air showers, yielding protons as the main background source to photons signal. 2. Neural network training. We train a neural network classifier to distinguish between proton- and photon-induced EAS events using a MC-simulated data set. For each event the neural network outputs a number \xi\in[0;1] representing its confidence that a given event is proton-like (closer to 0) or photon-like (closer to 1). Monte Carlo simulations might have slight differences from experimental data that are irrelevant for standard algorithmic reconstruction but are utilized by the

Read the complete text