Abstract:The predictions of hadronic interaction models for cosmic-ray induced air showers contain inherent uncertainties due to limitations of available accelerator data and theoretical understanding in the required energy and rapidity regime. Differences between models are typically evaluated in the range appropriate for cosmic-ray air shower arrays (10 15 -10 20 eV). However, accurate modelling of charged cosmic-ray measurements with ground based gamma-ray observatories is becoming more and more important.We assess … Show more
“…Secondly the network training is performed using simulated protons as the background events, however significant systematic uncertainties exist in the modelling of hadronic interactions in this energy range [35]. This behavioural uncertainty could result in a reduced performance when applying the trained networks to data due to incorrectly reproducing features within the air shower.…”
In this work, we present a new, high performance algorithm for background rejection in imaging atmospheric Cherenkov telescopes. We build on the already popular machine-learning techniques used in gamma-ray astronomy by the application of the latest techniques in machine learning, namely recurrent and convolutional neural networks, to the background rejection problem. Use of these machine-learning techniques addresses some of the key challenges encountered in the currently implemented algorithms and helps to significantly increase the background rejection performance between 100 GeV and 100 TeV energies. We apply these machine learning techniques to the H.E.S.S. telescope array, first testing their performance on simulated data and then applying the analysis to two well known gamma-ray sources. With real observational data we find significantly improved performance over the current standard methods, with a 20-25% reduction in the background rate when applying the recurrent neural network analysis. Importantly, we also find that the convolutional neural network results are strongly dependent on the sky brightness in the source region which has important implications for the future implementation of this method in Cherenkov telescope analyses.
“…Secondly the network training is performed using simulated protons as the background events, however significant systematic uncertainties exist in the modelling of hadronic interactions in this energy range [35]. This behavioural uncertainty could result in a reduced performance when applying the trained networks to data due to incorrectly reproducing features within the air shower.…”
In this work, we present a new, high performance algorithm for background rejection in imaging atmospheric Cherenkov telescopes. We build on the already popular machine-learning techniques used in gamma-ray astronomy by the application of the latest techniques in machine learning, namely recurrent and convolutional neural networks, to the background rejection problem. Use of these machine-learning techniques addresses some of the key challenges encountered in the currently implemented algorithms and helps to significantly increase the background rejection performance between 100 GeV and 100 TeV energies. We apply these machine learning techniques to the H.E.S.S. telescope array, first testing their performance on simulated data and then applying the analysis to two well known gamma-ray sources. With real observational data we find significantly improved performance over the current standard methods, with a 20-25% reduction in the background rate when applying the recurrent neural network analysis. Importantly, we also find that the convolutional neural network results are strongly dependent on the sky brightness in the source region which has important implications for the future implementation of this method in Cherenkov telescope analyses.
“…The main goal of this research is to understand the hadronic interaction models' deviations in the sub-TeV regime, where they have been seen to significantly disagree [2]. We chose an initial proton energy of 100 GeV (including the rest mass) to perform all simulations.…”
Section: Simulation Methodologymentioning
confidence: 99%
“…In a recent study [2], it was found that for cosmic-ray protons with an energy just above the typical switching energy, the properties of the simulated EASs have a strong dependency on the selection of the hadronic interaction model. With increasing energy of the incident proton (up to 100 TeV), the average differences in air shower properties between the models seemed to reduce.…”
In the sub-TeV regime, the most widely used hadronic interaction models disagree significantly in their predictions for post-first interaction and ground-level particle spectra from cosmic ray induced air showers. These differences generate an important source of systematic uncertainty in their experimental use. We investigate the nature and impact of model uncertainties through a simultaneous analysis of ground level particles and first interaction scenarios. We focus on air shower primaries with energies close to the transition between high and low energy hadronic interaction models, where the dissimilarities have been shown to be the largest and well within the range of accelerator measurements. Interaction models are shown to diverge as several shower scenarios are compared, reflecting intrinsic differences in the model theoretical frameworks. Finally, we discuss the importance of interactions in the energy regime where the switching between models occurs ($$<1$$
<
1
TeV) and the effect of the choice of model on the number of hadronic interactions within cosmic ray induced air showers of higher energies.
“…However a detailed study of these effects is beyond the scope of the work presented here. For a recent systematic study on the impact of the choice of hadronic interaction models on air shower simulation in the energy range of our interest we refer to [23]. The electromagnetic interactions are handled by the EGS4 model [24].…”
Section: Simulations and Atmospheric Propagationmentioning
Very-high-energy γ-ray astronomy based on the measurement of air shower particles at ground-level has only recently been established as a viable approach, complementing the well established air Cherenkov technique. This approach requires high (mountain) altitudes and very high surface coverage particle detectors. While in general the properties of air showers are well established for many decades, the extreme situation of ground-level detection of very small showers from low energy primaries has not yet been well characterised for the purposes of γ-ray astronomy. Here we attempt such a characterisation, with the aim of supporting the optimisation of next-generation γ-ray observatories based on this technique. We address all of the key ground level observables and provide parameterisations for use in detector optimisation for shower energies around 1 TeV. We emphasise two primary aspects: the need for large area detectors to effectively measure low-energy showers, and the importance of muon identification for the purpose of background rejection.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.