November 29, 2023 to December 1, 2023
Lamarr Institut, TU Dortmund University
Europe/Berlin timezone

Participant List

83 participants

The lists of participants grouped by the registration form they used to register for the event.

CS & Physics Meet-Up by Lamarr & B3D

Title First Name Last Name Affiliation Your Current Research Background
Abhinav Tyagi MPIfR Bonn
Mr Alexander Kier Astronomical Institute RUB Physics & Astronomy, Machine Learining & AI
Dr Amal Saadallah Lamarr Institute, TU Dortmund University Computer Science, Machine Learining & AI
Dr Ancla Müller AIRUB Physics & Astronomy, Radio Astronomy, Astrophysics
Prof. André Hinkenjann Hochschule Bonn-Rhein-Sieg, Institut für Visual Computing Computer Science, Machine Learining & AI, Big Data
Mr Andrei Kazantsev Max-Planck-Institut für Radioastronomie Radio Astronomy, Astrophysics
Mr Ankur Dev AIfA, University of Bonn Physics & Astronomy, Big Data, Radio Astronomy, Astrophysics
Aron Kordt TU Dortmund Physics & Astronomy, Computer Science
Biljana Mitreska TU Dortmund High-Energy-Physics
Dr Claudia Comito Forschungszentrum Jülich, Jülich Supercomputing Centre Physics & Astronomy, Computer Science, Machine Learining & AI, Big Data, Radio Astronomy
Mr Cyrus Walther TU Dortmund Machine Learining & AI, Astroparticle physics
Mr Dominik Baack TU Dortmund Physics & Astronomy, Computer Science
Prof. Dominik Bomans Astronomical Institute of the Ruhr University Bochum Machine Learining & AI, Big Data, Radio Astronomy, Astrophysics
Dr Dominik Elsaesser TU Dortmund
Mrs Ekaterina Moerova MPIfR
Mr Felix Geyer Astroparticle Physics - TU Dortmund University Physics & Astronomy, Machine Learining & AI, Radio Astronomy
Frank Bertoldi Bonn Universität Physics & Astronomy, Radio Astronomy, Astrophysics
Prof. Gennadiy Andriyenko Lamarr Institute / Fraunhofer Institute IAIS / City University London Computer Science, Machine Learining & AI, other (please add in the comments)
Mr Günther Heemann Astronomisches Institut RUB Physics & Astronomy, Machine Learining & AI
Prof. Hermann Heßling HTW Berlin Physics & Astronomy, Computer Science, Machine Learining & AI, Big Data, Radio Astronomy
Dr Holger Stiele JSC/FZJ Physics & Astronomy, Astrophysics
Dr Jens Buss Lamarr Institute, TU Dortmund University Physics & Astronomy, Computer Science, Machine Learining & AI, Astroparticle physics
Prof. Jens Teubner TU Dortmund • DBIS Group Computer Science, Data Bases
Mrs Jessica Koch Max-Planck-Institut für Radioastronomie
Dr Jinglan Zheng Bielefeld University Physics & Astronomy, Radio Astronomy
Prof. Johannes Albrecht TU Dortmund & Lamarr Physics & Astronomy, High-Energy-Physics
Mr Jonah Blank TU Dortmund High-Energy-Physics
Mr Julian Eßer Fraunhofer IML Machine Learining & AI
Dr Kai Polsterer HITS gGmbH Physics & Astronomy, Computer Science, Machine Learining & AI, Big Data, Data Bases, Radio Astronomy
Kamalpreet Kaur Argelander Institute for Astronomy, University of Bonn Physics & Astronomy, Machine Learining & AI, Radio Astronomy, Astrophysics
Ms Katharina Peters Lamarr Institute, TU Dortmund University
Dr Kaustuv Basu Universität Bonn Physics & Astronomy, Astrophysics
Dr Kevin Schmidt TU Dortmund University Physics & Astronomy, Computer Science, Machine Learining & AI, Radio Astronomy, Astrophysics
Kostadin Cvejoski Fraunhofer IAIS Machine Learining & AI
Leonora Kardum Lamarr Institute, TU Dortmund Machine Learining & AI, Astroparticle physics, Astrophysics
Lucas Kock TU Dortmund Machine Learining & AI, Big Data
Dr Lukas Pfahler Lamarr Institute, TU Dortmund University Computer Science, Machine Learining & AI
Mr Maik Sowinski Forschungszentrum Jülich Physics & Astronomy, Astrophysics
Dr Maram Akila Fraunhofer IAIS Computer Science, Machine Learining & AI
Marcel Mielach Astronomisches Institut Ruhr-Universität Bochum Physics & Astronomy, Machine Learining & AI, Big Data
Mr Maurice Günder University of Bonn, Fraunhofer IAIS Physics & Astronomy, Computer Science, Machine Learining & AI, Astroparticle physics, Astrophysics
Prof. Michael Kramer Max-Planck-Institut fuer Radioastronomie Physics & Astronomy, Radio Astronomy, Astrophysics
Mirco Hünnefeld TU Dortmund Machine Learining & AI, Astroparticle physics
Dr Mirko Bunse Lamarr Institute, TU Dortmund University Computer Science, Machine Learining & AI
Mr Murad Elnagdi University of Bonn Computer Science, Machine Learining & AI
Prof. Natalia Andrienko Lamarr Institute / Fraunhofer Institute IAIS / City University London Computer Science, Machine Learining & AI, other (please add in the comments)
Mr Nils Wandel University of Bonn Computer Science, Machine Learining & AI
Mr Pascal Gutjahr TU Dortmund University Machine Learining & AI, Big Data, High-Energy-Physics, Astroparticle physics
Dr Ralf Antonius Timmermann AIfA, Uni Bonn Computer Science, Machine Learining & AI, Data Bases
Dr Ramesh Karuppusamy MPIfR Radio Astronomy
Mr Ramses Sanchez Lamarr Institute, University of Bonn Physics & Astronomy, Computer Science, Machine Learining & AI
Sadia Mahjabin Lamarr Institute, TU Dortmund University
Dr Sebastian Buschjäger Lamarr Institute, TU Dortmund University Computer Science, Machine Learining & AI
Mr Sebastian Konietzny TU Dortmund Computer Science, Machine Learining & AI
Ms Siba Mohsen TU Dortmund Computer Science, Machine Learining & AI
Dr Stefan Michaelis Lamarr-Institute Computer Science, Machine Learining & AI
Prof. Stefanie Walch-Gassner University of Cologne Physics & Astronomy, Astrophysics
Ms Subarna Chaki Argelander-Institut für Astronomie, University of Bonn Physics & Astronomy, Machine Learining & AI, Radio Astronomy, Astrophysics
Dr Thanh Liem Ngo University of Cologne Radio Astronomy, Astrophysics
Prof. Thomas Liebig Lamarr Institute Machine Learining & AI
Mr Thore Gerlach Fraunhofer IAIS Computer Science, Machine Learining & AI, other (please add in the comments)
Dr Tim Ruhe TU Dortmund Physics & Astronomy, Machine Learining & AI, Astroparticle physics
Dr Tim-Eric Rathjen University of Cologne
Dr Tobias Uelwer TU Dortmund Computer Science, Machine Learining & AI
Mr Toma Badescu Uni Bonn, Argelander Institut fur Astronomie Physics & Astronomy, Radio Astronomy, Astrophysics
Dr Vishnu Balakrishnan Max Planck Institute for Radio Astronomy, Bonn Physics & Astronomy, Machine Learining & AI, Big Data, Astrophysics
Prof. Wolfgang Rhode Physics Department, TU Dortmund University Physics & Astronomy, Machine Learining & AI, Astroparticle physics

Posters

First Name Last Name Title Of The Poster Short Description Of The Topic
Abhinav Tyagi Pulsarnet - Accelerating the Discovery of Binary Pulsars through Machine Learning In the realm of radio astronomy, the discovery and analysis of binary pulsars present a unique opportunity to test the theories of gravity, particularly General Relativity, in the strong-field limit. PulsarNet introduces an advanced machine learning pipeline specifically tailored for this purpose. This pipeline uniquely processes the Fourier amplitude spectrum to identify binary pulsar candidates with increased efficiency and accuracy. It consists of two key components: a Support Vector Machine (SVM) classifier that effectively filters out non-signal segments in the Fourier domain and an attention-based neural network (NN) regressor that precisely predicts the Keplerian parameters of potential pulsars. PulsarNet stands out by significantly accelerating the discovery process compared to traditional template match-based methods, while achieving comparable sensitivity in the white-noise regime.
Claudia Comito Heat: accelerating massive data processing in Python Manipulating and processing massive data sets is challenging. In astrophysics as in the vast majority of research communities, the standard approach involves breaking up and analyzing data in smaller chunks, a process that is both inefficient and prone to errors. The problem is exacerbated on GPUs, because of the smaller available memory. Popular solutions to distribute NumPy/SciPy computations are based on task parallelism, introducing significant runtime overhead, complicating implementation, and often limiting GPU support to one vendor. This poster illustrates an alternative based on data parallelism instead. The open-source library Heat [1, 2] builds on PyTorch and mpi4py to simplify porting of NumPy/SciPy-based code to GPU (CUDA, ROCm, including multi-GPU, multi-node clusters). Under the hood, Heat distributes massive memory-intensive operations over multi-node resources via MPI communication. From a user's perspective, Heat can be used seamlessly in the Python array ecosystem. Supported features: - distributed (multi-GPU) I/O from shared memory - easy distribution of memory-intensive operations in existing code (e.g. matrix multiplication) - interoperability within the Python array ecosystem: Heat as a backend for your massive array manipulations, statistics, signal processing, machine learning... - transparent parallelism: prototype on your laptop, run the same code on HPC cluster. I'll also touch upon Heat's current implementation roadmap, and possible paths to collaboration. [1] https://github.com/helmholtz-analytics/heat [2] M. Götz et al., "HeAT – a Distributed and GPU-accelerated Tensor Framework for Data Analytics," 2020 IEEE International Conference on Big Data (Big Data), Atlanta, GA, USA, 2020, pp. 276-287, doi: 10.1109/BigData50022.2020.9378050.
Jean-Marco Alameddine High-energy lepton and photon propagation with the simulation framework PROPOSAL Current challenges in astroparticle physics like the muon puzzle in air shower physics or the upcoming launch of next-generation neutrino and gamma observatories require modern tools for the simulation of particle propagation, both from a technical and a physical standpoint. For those purposes, PROPOSAL is a simulation framework, developed at TU Dortmund University, that provides 3D Monte Carlo simulations of charged leptons and high-energy photons. PROPOSAL, which is usable in both C++ and Python, provides a high level of customizability, allowing the user to customize both the prop- agation environment and the underlying physical parametrizations, where up-to-date energy loss cross sections are available. In this contribution, we present PROPOSAL as a framework, as well as current applications where PROPOSAL is used. This includes the usage of PROPOSAL as an electromagnetic model for the shower simulation framework CORSIKA 8, as well as the usage of PROPOSAL for underground measurements of muon numbers, for example in the context of muography.
Jonah Blank Flavour Tagging at the LHCb experiment One of the prominent questions in particle physics is the apparent assymetry of matter and antimatter, which in the early universe led to an abundance of the former, resulting in our cosmos as we see it today. One important part in understanding this asymmetry is the so called Charge and Parity violation (CPV), which describes the different behaviour under the combination of charge conjugation and spatial reflection. To study this phenomenon, the decays of neutral B mesons, which oscillate between particle and antiparticle state are examined at the LHCb experiment. Particularly for those final states that can be reached from both, it is essential to know, in which flavour the B meson was produced. At LHCb, this information is obtained by (MachineLearning based) flavour tagging alogrithms, so called taggers. The existing taggers try to indentify individual tracks related to the production of the signal meson to provide this information on the initial state. When requiring a high tagging quality, using these classic approaches leads to low efficiencies, reducing the statistic power of the data samples. Furthermore starting from Run 3 the higher instantaneous luminosity at LHCb, leading to higher track multiplicities in the detected events, makes flavour tagging an increasingly challenging task. Therefore work is ongoing to develop an Inclusive Flavour Tagger, which aims to process the full event information exploiting more advanced machine learning techniques, to provide a better prediction of the initial B meson flavour.
Kaustuv Basu Deep Learning on the Cosmic Microwave Background Study of the Cosmic Microwave Background, the oldest light, offers various ways to learn about the intervening objects in the Universe. One such application is the determination of galaxy cluster masses, for which deep neural networks have already been applied successfully on simulated data. We aim to further such applications with more physics inputs and also train models on real data.
Maik Sowinski Identification and Classification of Young Star Clusters The GAIA DR3 presents scientists with new opportunities to investigate star clusters due to the increased number of light sources (~1.6 billion) and improved precision of the data. A natural ansatz is now to attempt to create a comprehensive catalogue of star clusters using Machine Learning tools. The specific tools to be used are clustering algorithms, such as DBSCAN, HDBSCAN, and OPTICS. These divide a number of (unlabeled) data points in a metric space (e.g., 3D Euclidean position space or 6D position- velocity space) into mathematical clusters based on their spatial proximity in a process called cluster analysis (or simply clustering). Ideally, a mathematical cluster found by such an algorithm shall correspond to a physical star cluster. To make this happen reliably is the very aim.
Mirko Bunse Unfolding from a Computer Science perspective Unfolding corrects measurements of distributions if these measurements are affected by distortions and limited acceptance of the detector. We show that Computer Science knows unfolding under a different term, as "quantification learning" or as "class prior estimation". Through this connection, we use advancements made in Computer Science, both concerning the theoretical understanding of the problem and concerning novel methods, to unfold measurements of distributions with greater success.
Mirko Bunse How to learn from real IACT data Imaging atmospheric Cherenkov telescopes (IACTs) typically require simulations to obtain labeled training data for reconstruction tasks. For the specific task of gamma hadron classification, we show that no simulations are needed if the direction of each event is employed as a so-called "noisy label". Machine learning research on the theory of class-conditional label noise provides us with formal proofs that accurate classifiers can indeed be learnt from labels of this kind. Our experiments with data from the FACT telescope demonstrate that this approach can outperform state-of-the-art classification pipelines.
Nils Wandel Approaching PDEs with Physics-Driven Deep Learning Partial Differential Equations (PDEs) play an important role in describing continuous physical systems such as fluids, waves, cloth and many more. Thus, by solving these equations, one can simulate for example fluids or garment in computer graphics or analyse lift and drag coefficients in engineering applications. However, in most scenarios, analytic solutions of PDEs are not available and traditional numerical solutions are computationally expensive. In contrast, recent deep-learning based methods promise great gains in efficiency by infering solutions in a single forward pass through a neural network. Here, we want to present our physics-driven approach to train neural surrogate models for PDEs without precomputed ground truth data. This way, we achieve fast, stable and differentiable simulations that can be used for example in interactive real-time applications or in inverse problems.
Pascal Gutjahr The Relevance of Muon Deflections for Neutrino Telescopes Large-scale neutrino telescopes have the primary objective to detect and characterize neutrino sources in the universe. These experiments rely on the detection of charged leptons produced in the interaction of neutrinos with nuclei. Angular resolutions are estimated to be better than 1 degree, which is achieved by the reconstruction of muons. This angular resolution is a measure of the accuracy with which the direction of incoming neutrinos can be determined. Since muons can traverse distances of several kilometers through media, the original muon direction can differ from the muon direction inside the detector due to deflections by stochastic interactions and multiple scattering. In this contribution, a recently published study of muon deflections based on the simulation tool PROPOSAL is presented. Muons with various energies are propagated through different media over several distances. Data-Monte-Carlo comparisons as well as comparisons to the simulation tools MUSIC and Geant4 are performed. Finally, the impact of muon deflections on large-scale neutrino telescopes is discussed.
Sascha Mücke Efficient Light Source Placement using Quantum Computing NP-hard problems regularly come up in video games, with interesting connections to real-world problems. In the game Minecraft, players place torches on the ground to light up dark areas. Placing them in a way that minimizes the total number of torches to save resources is far from trivial. We use Quantum Computing to approach this problem. To this end, we derive a QUBO formulation of the torch placement problem, which we uncover to be very similar to another NP-hard problem. We employ a solution strategy that involves learning Lagrangian weights in an iterative process, adding to the ever growing toolbox of QUBO formulations. Finally, we perform experiments on real quantum hardware using real game data to demonstrate that our approach yields good torch placements.
Sebastian Buschjäger Predictive Maintanance TBA
Tim-Eric Rathjen Optical emission-line diagnostics of the simulated interstellar medium in different environments​ ​We present the first results from applying optical emission-line diagnostics to high-resolution magneto-hydrodynamics (MHD) simulations of the multi-phase interstellar medium (ISM) via post-processing. We model the ISM in different galactic environments by increasing the initial gas surface density, 𝚺, in a range from 10 to 100 M⦿ / pc^2. Star formation is self-regulated through stellar feedback, including supernovae but also early feedback in the form of stellar winds and ionising radiation of the massive stars. We couple our simulations to the photo-ionisation code cloudy and predict the optical line emission of the ISM. We find that the overall ionisation parameter of a system increases with increasing gas surface density, leading to a shift in the line-ratio diagnostics diagram (BPT diagram) along the star-forming sequence. However, for strong starburst systems at higher gas surface densities, up to 30% of the total optical emission originates from outside of H II regions, moving the BPT signal towards the regime of shock excited emission.​
Vukan Jevtic Flavour Tagging at the LHCb Experiment Presenting together with Jonah Blank

Discussion groups

First Name Last Name Affiliation Topic To Discuss Short Description Of The Context Potential Time Slots:
Andrei Kazantsev Max-Planck-Institut für Radioastronomie Deep Learning for real-time classification of astronomical radio signals Continuous developments in systems for recording radio astronomical signals lead to a natural increase in the volume of data acquired. This increase, in turn, exacerbates the challenges associated with the storage and processing of this information. A potential solution to this challenge is the development of systems underpinned by deep machine learning. These systems, at the stage of signal acquisition, could classify the data and retain only those portions that contain significant scientific information for further analysis. However, the challenge resides in ensuring that these systems not only exhibit high accuracy across a vast array of signals but also maintain exceptional sensitivity to avoid overlooking weak signals that still have significant scientific value. This presentation will discuss efforts to implement such a model, utilizing radio pulsar data from the Effelsberg telescope as a case study. Wednesday 16:00 - 18:00
Dr. Vishnu Balakrishnan Max Planck Institute for Radio Astronomy, Bonn Leveraging Machine Learning for Next-Generation Radio Pulsar Searches: Challenges and Opportunities The discovery of new radio pulsars has significant implications for both Gravitational and Condensed Matter Physics. With the emergence of large datasets on the order of petabytes requiring quasi-real-time analysis, computational methods, particularly machine learning, have been increasingly important. This session will begin with a review of the progress made in applying machine learning techniques to radio pulsar searches over the last decade. Following this, we will delve into the most recent advancements and the challenges that lie ahead. The aim is to foster a robust dialogue and establish collaborations between researchers in computer science and astronomy, working toward the next frontier in radio pulsar discovery. Thursday 11:30 - 12:30, Thursday 16:00 - 17:00, Friday 11:30 - 12:30