arXiv subjects

Tim Lukas Holch

Publications and source records attributed to Tim Lukas Holch.

10 recordsLinked to original sources

arXiv2026

An ALMA view of the Jet-Arc CO clouds toward the TeV $\gamma$-ray source HESS J1023-575 and Westerlund 2; Evidence for the footprints of microquasar jets, the very powerful cosmic-ray accelerator in the Galactic disk

The TeV $\gamma$-ray source HESS J1023-575 (HESSJ 1023 hereafter) is one of the brightest H.E.S.S. sources near the young massive cluster Westerlund 2. HESS J1023 shows a remarkable positional alignment with the Jet and Arc CO clouds on its eastern and western sides over 170 pc length. We have carried out sub-pc scale observations of the CO clouds with ALMA and have discovered that the clouds consist of numerous thin filamentary features of $\sim$0.5 pc width and 10--20 pc length at distance of 7.5 kpc, which are well aligned with the Jet-Arc axis. Based on the magneto-hydrodynamical model of microquasar jets launched from {the center of the $\gamma$-ray source} HESS J1023-575, we present an interpretation that the thin filamentary clouds are the footprints of the microquasar jets on the HI gas. The model also explains the dissimilar Jet vs. Arc clouds in terms of HI density difference on each side. By using the density of the CO and HI gas and the $\gamma$-ray luminosity, we have calculated the cosmic ray proton energy $W_{\rm p}$ to be 7$\times$10$^{48}$ erg under the hadronic scheme, which is ten times larger than those derived in the TeV $\gamma$-ray SNRs RX J1713.7-3946 and RX J0852.0-4622. It is likely that HESS~J1023 has been active over 1-10 Myr, which is significantly longer than the duration of cosmic ray acceleration of the SNRs. HESS~J1023 is therefore an outstanding source of cosmic rays equivalent to at least 1000 SNRs, and is possibly the most powerful CR accelerator in the Galactic disk. A high energy compact source in HESS~J1023, which is likely a Myr-old black hole or neutron star, remains veiled due to heavy extinction.

astro-ph.HE↗

arXiv2025

Convolution and Graph-based Deep Learning Approaches for Gamma/Hadron Separation in Imaging Atmospheric Cherenkov Telescopes

The identification of $γ$-rays from the predominant hadronic-background is a key aspect in their ground-based detection using Imaging Atmospheric Cherenkov Telescopes (IACTs). While current methods are limited in their ability to exploit correlations in complex data, deep learning-based models offer a promising alternative by directly leveraging image-level information. However, several challenges involving the robustness and applicability of such models remain. Designing model architectures with inductive biases relevant for the task can help mitigate the problem. Three such deep learning-based models are proposed, trained, and evaluated on simulated data: (1) a hybrid convolutional and graph neural network model (CNN-GNN) using both image and graph data; (2) an enhanced CNN-GNN variant that incorporates additional reconstructed information within the graph construction; and (3) a graph neural network (GNN) model using image moments serving as a baseline. The new combined convolution and graph-based approach demonstrates improved performance over traditional methods, and the inclusion of reconstructed information offers further potential in generalization capabilities on real observational data.

astro-ph.HE↗

arXiv2025

Agent-based code generation for the Gammapy framework

Software code generation using Large Language Models (LLMs) is one of the most successful applications of modern artificial intelligence. Foundational models are very effective for popular frameworks that benefit from documentation, examples, and strong community support. In contrast, specialized scientific libraries often lack these resources and may expose unstable APIs under active development, making it difficult for models trained on limited or outdated data. We address these issues for the Gammapy library by developing an agent capable of writing, executing, and validating code in a controlled environment. We present a minimal web demo and an accompanying benchmarking suite. This contribution summarizes the design, reports our current status, and outlines next steps.

cs.SE↗

arXiv2023

Validating Monte Carlo simulations for an analysis chain in H.E.S.S

Imaging Air Cherenkov Telescopes (IACTs) detect very high energetic (VHE) gamma rays. They observe the Cherenkov light emitted in electromagnetic shower cascades that gamma rays induce in the atmosphere. A precise reconstruction of the primary photon energy and the source flux depends heavily on accurate Monte Carlo (MC) simulations of the shower propagation and the detector response, and therefore also on adequate assumptions about the atmosphere at the site and time of a measurement. Here, we present the results of an extensive validation of the MC simulations for an analysis chain of the H.E.S.S. experiment with special focus on the recently installed FlashCam camera on the large 28 m telescope. One goal of this work was to create a flexible and easy-to-use framework to facilitate the detailed validation of MC simulations also for past and future phases of the H.E.S.S. experiment. Guided by the underlying physics, the detector simulation and the atmospheric transmission profiles were gradually improved until low level parameters such as cosmic ray (CR) trigger rates matched within a few percent between simulations and observational data. This led to instrument response functions (IRFs) with which the analysis of current H.E.S.S. data can ultimately be carried out within percent accuracy, substantially improving earlier simulations.

astro-ph.IM↗

arXiv2021

Probing the hadronic nature of the gamma-ray emission associated with Westerlund 2

Star-forming regions have been proposed as potential Galactic cosmic-ray accelerators for decades. Cosmic ray acceleration can be probed through observations of gamma-rays produced in inelastic proton-proton collisions, at GeV and TeV energies. We analyze more than 11 years of Fermi-LAT data from the direction of Westerlund 2, one of the most massive and best-studied star-forming regions in our Galaxy. The spectral and morphological characteristics of the LAT source agree with the ones in the TeV regime (HESS J1023-575), allowing the description of the gamma-ray source from a few hundreds of MeV to a few tens of TeVs. We will present the results and discuss the implications of the identification with the stellar cluster and the radiation mechanism involved.

astro-ph.HE↗

arXiv2021

The upgraded Data Acquisition System of the H.E.S.S. telescope array

The High Energy Stereoscopic System (H.E.S.S.) is an array of five Imaging Atmospheric Cherenkov Telescopes located in the Khomas Highland of Namibia. H.E.S.S. observes gamma rays above tens of GeV by detecting the Cherenkov light that is produced when Very High Energy gamma rays interact with the Earth's atmosphere. The H.E.S.S. Data Acquisition System (DAQ) coordinates the nightly telescope operations, ensuring that the various components communicate properly and behave as intended. It also provides the interface between the telescopes and the people on shift who guide the operations. The DAQ comprises both the hardware and software, and since the beginning of H.E.S.S., both elements have been continuously adapted to improve the data-taking capabilities of the array and push the limits of what H.E.S.S. is capable of. Most recently, this includes the upgrade of the entire computing cluster hosting the DAQ software, and the accommodation of a new camera on the large 28m H.E.S.S. telescope. We discuss the performance of the upgraded DAQ and the lessons learned from these activities.

astro-ph.IM↗

arXiv2021

Probing the hadronic nature of the gamma-ray emission associated with Westerlund 2

Star-forming regions have been proposed as potential Galactic cosmic-ray accelerators for decades. Cosmic-ray acceleration can be probed through observations of gamma-rays produced in inelastic proton-proton collisions, at GeV and TeV energies. In this paper, we analyze more than 11 years of Fermi-LAT data from the direction of Westerlund 2, one of the most massive and best-studied star-forming regions in our Galaxy. In particular, we investigate the characteristics of the bright pulsar PSR J1023-5746 that dominates the gamma-ray emission below a few GeV at the position of Westerlund 2, and the underlying extended source FGES J1023.3-5747. The analysis results in a clear identification of FGES J1023.3-5747 as the GeV counterpart of the TeV source HESS J1023-575, through its morphological and spectral properties. This identification provides new clues about the origin of the HESS J1023-575 gamma-ray emission, favouring a hadronic origin of the emission, powered by Westerlund 2, rather than a leptonic origin related to either the pulsar wind nebula associated with PSR J1023-5746 or the cluster itself. This result indirectly supports the hypothesis that star-forming regions can contribute to the cosmic-ray sea observed in our Galaxy

astro-ph.HE↗

arXiv2019

HexagDLy - Processing hexagonally sampled data with CNNs in PyTorch

HexagDLy is a Python-library extending the PyTorch deep learning framework with convolution and pooling operations on hexagonal grids. It aims to ease the access to convolutional neural networks for applications that rely on hexagonally sampled data as, for example, commonly found in ground-based astroparticle physics experiments.

cs.CV↗

arXiv2018

Application of Deep Learning methods to analysis of Imaging Atmospheric Cherenkov Telescopes data

Ground based gamma-ray observations with Imaging Atmospheric Cherenkov Telescopes (IACTs) play a significant role in the discovery of very high energy (E > 100 GeV) gamma-ray emitters. The analysis of IACT data demands a highly efficient background rejection technique, as well as methods to accurately determine the energy of the recorded gamma-ray and the position of its source in the sky. We present results for background rejection and signal direction reconstruction from first studies of a novel data analysis scheme for IACT measurements. The new analysis is based on a set of Convolutional Neural Networks (CNNs) applied to images from the four H.E.S.S. phase-I telescopes. As the H.E.S.S. cameras pixels are arranged in a hexagonal array, we demonstrate two ways to use such image data to train CNNs: by resampling the images to a square grid and by applying modified convolution kernels that conserve the hexagonal grid properties. The networks were trained on sets of Monte-Carlo simulated events and tested on both simulations and measured data from the H.E.S.S. array. A comparison between the CNN analysis to current state-of-the-art algorithms reveals a clear improvement in background rejection performance. When applied to H.E.S.S. observation data, the CNN direction reconstruction performs at a similar level as traditional methods. These results serve as a proof-of-concept for the application of CNNs to the analysis of events recorded by IACTs.

astro-ph.IM↗

arXiv2017

Probing Convolutional Neural Networks for Event Reconstruction in γ-Ray Astronomy with Cherenkov Telescopes

A dramatic progress in the field of computer vision has been made in recent years by applying deep learning techniques. State-of-the-art performance in image recognition is thereby reached with Convolutional Neural Networks (CNNs). CNNs are a powerful class of artificial neural networks, characterized by requiring fewer connections and free parameters than traditional neural networks and exploiting spatial symmetries in the input data. Moreover, CNNs have the ability to automatically extract general characteristic features from data sets and create abstract data representations which can perform very robust predictions. This suggests that experiments using Cherenkov telescopes could harness these powerful machine learning algorithms to improve the analysis of particle-induced air-showers, where the properties of primary shower particles are reconstructed from shower images recorded by the telescopes. In this work, we present initial results of a CNN-based analysis for background rejection and shower reconstruction, utilizing simulation data from the H.E.S.S. experiment. We concentrate on supervised training methods and outline the influence of image sampling on the performance of the CNN-model predictions.

astro-ph.IM↗