arXiv subjects

A. Jung

Publications and source records attributed to A. Jung.

At least 19 recordsLinked to original sources

arXiv2026

Trigger system for the Payload for Ultrahigh Energy Observations (PUEO) balloon-borne neutrino detector

The Payload for Ultrahigh Energy Observations (PUEO) is a NASA balloon-borne instrument for the detection of ultra-high energy (UHE) neutrinos with energies above $10^{17.5}~\textrm{eV}$ via either the Askaryan effect or geomagnetic emissions from an upward-going air shower. The main instrument trigger system for PUEO is a fully digital supersample rate beamformer based on 24 Xilinx Radio Frequency System-on-a-Chip (RFSoC) digitizers sampling 192 channels operating at $3~\textrm{GSa/s}$ and a system clock frequency of $375~\textrm{MHz}$. The trigger implements frequency band conditioning, dynamic radio-frequency interference (RFI) rejection, and matched filtering, with significant emphasis on optimization to reduce both the power and resource usage while maintaining sensitivity. The system implements 48 total synthetic antenna beams with up to 8 antennas each, covering a $\sim25^\circ$ range in zenith and $\sim60^\circ$ range in azimuth. Preflight testing demonstrated a trigger performance of a minimum signal-to-noise ratio (SNR) of $\sim1.5$ using simulated signals while consuming between $5-7~\textrm{W}$ in the trigger logic.

astro-ph.IM↗

arXiv2025

Federated Learning: From Theory to Practice

This book offers a hands-on introduction to building and understanding federated learning (FL) systems. FL enables multiple devices -- such as smartphones, sensors, or local computers -- to collaboratively train machine learning (ML) models, while keeping their data private and local. It is a powerful solution when data cannot or should not be centralized due to privacy, regulatory, or technical reasons. The book is designed for students, engineers, and researchers who want to learn how to design scalable, privacy preserving FL systems. Our main focus is on personalization: enabling each device to train its own model while still benefiting from collaboration with relevant devices. This is achieved by leveraging similarities between (the learning tasks associated with) devices that are encoded by the weighted edges (or links) of a federated learning network (FL network). The key idea is to represent real-world FL systems as networks of devices, where nodes correspond to device and edges represent communication links and data similarities between them. The training of personalized models for these devices can be naturally framed as a distributed optimization problem. This optimization problem is referred to as generalized total variation minimization (GTVMin) and ensures that devices with similar learning tasks learn similar model parameters. Our approach is both mathematically principled and practically motivated. While we introduce some advanced ideas from optimization theory and graph-based learning, we aim to keep the book accessible. Readers are guided through the core ideas step by step, with intuitive explanations.

cs.LG↗

arXiv2025

Cosmic Rays and the Askaryan Effect Reveal Subsurface Structure and Buried Ice on the Moon

We present the first full-wavelength numerical simulations of the electric field generated by cosmic ray impacts into the Moon. Billions of cosmic rays fall onto the Moon every year. Ultra-high energy cosmic ray impacts produce secondary particle cascades within the regolith and subsequent coherent, widebandwidth, linearly-polarized radio pulses by the Askaryan Effect. Observations of the cosmic ray particle shower radio emissions can reveal subsurface structure on the Moon and enable the broad and deep prospecting necessary to confirm or refute the existence of polar ice deposits. Our simulations show that the radio emissions and reflections could reveal ice layers as thin as 10 cm and buried under regolith as deep as 9 m. The Askaryan Effect presents a novel and untapped opportunity for characterizing buried lunar ice at unprecedented depths and spatial scales.

physics.space-ph↗

arXiv2024

Analysis of Total Variation Minimization for Clustered Federated Learning

A key challenge in federated learning applications is the statistical heterogeneity of local datasets. Clustered federated learning addresses this challenge by identifying clusters of local datasets that are approximately homogeneous. One recent approach to clustered federated learning is generalized total variation minimization (GTVMin). This approach requires a similarity graph which can be obtained by domain expertise or in a data-driven fashion via graph learning techniques. Under a widely applicable clustering assumption, we derive an upper bound the deviation between GTVMin solutions and their cluster-wise averages. This bound provides valuable insights into the effectiveness and robustness of GTVMin in addressing statistical heterogeneity within federated learning environments.

cs.LG↗

arXiv2023

Wind to start the washing machine? High-Resolution Wind Atlas for Finland

The current fossil fuel and climate crisis has led to an increased demand for renewable energy sources, such as wind power. In northern Europe, the efficient use of wind power is crucial for achieving carbon neutrality. To assess the potential of wind energy for private households in Finland, we have conducted a high spatiotemporal resolution analysis. Our main contribution is a wind power map of Finland that indicates the availability of wind power for given load profiles. As a representative example of power load, we consider the load profile of a household appliance. We compare this load profile against the wind power available nearby the weather stations of the Finnish meteorological institute.

math.NA↗

arXiv2023

Plug In and Learn: Federated Intelligence over a Smart Grid of Models

We present a model-agnostic federated learning method that mirrors the operation of a smart power grid: diverse local models, like energy prosumers, train independently on their own data while exchanging lightweight signals to coordinate with statistically similar peers. This coordination is governed by a graph-based regularizer that encourages connected models to produce similar predictions on a shared, public unlabeled dataset. The resulting method is a flexible instance of regularized empirical risk minimization and supports a wide variety of local models - both parametric and non-parametric - provided they can be trained via regularized loss minimization. Such training is readily supported by standard ML libraries including scikit-learn, Keras, and PyTorch.

cs.LG↗

arXiv2022

Solid State Detectors and Tracking for Snowmass

Tracking detectors are of vital importance for collider-based high energy physics (HEP) experiments. The primary purpose of tracking detectors is the precise reconstruction of charged particle trajectories and the reconstruction of secondary vertices. The performance requirements from the community posed by the future collider experiments require an evolution of tracking systems, necessitating the development of new techniques, materials and technologies in order to fully exploit their physics potential. In this article we summarize the discussions and conclusions of the 2022 Snowmass Instrumentation Frontier subgroup on Solid State and Tracking Detectors (Snowmass IF03).

physics.ins-det↗

arXiv2022

Explainable Empirical Risk Minimization

The successful application of machine learning (ML) methods becomes increasingly dependent on their interpretability or explainability. Designing explainable ML systems is instrumental to ensuring transparency of automated decision-making that targets humans. The explainability of ML methods is also an essential ingredient for trustworthy artificial intelligence. A key challenge in ensuring explainability is its dependence on the specific human user ("explainee"). The users of machine learning methods might have vastly different background knowledge about machine learning principles. One user might have a university degree in machine learning or related fields, while another user might have never received formal training in high-school mathematics. This paper applies information-theoretic concepts to develop a novel measure for the subjective explainability of the predictions delivered by a ML method. We construct this measure via the conditional entropy of predictions, given a user feedback. The user feedback might be obtained from user surveys or biophysical measurements. Our main contribution is the explainable empirical risk minimization (EERM) principle of learning a hypothesis that optimally balances between the subjective explainability and risk. The EERM principle is flexible and can be combined with arbitrary machine learning models. We present several practical implementations of EERM for linear models and decision trees. Numerical experiments demonstrate the application of EERM to detecting the use of inappropriate language on social media.

cs.LG↗

arXiv2022

flow-based clustering and spectral clustering: a comparison

We propose and study a novel graph clustering method for data with an intrinsic network structure. Similar to spectral clustering, we exploit an intrinsic network structure of data to construct Euclidean feature vectors. These feature vectors can then be fed into basic clustering methods such as k-means or Gaussian mixture model (GMM) based soft clustering. What sets our approach apart from spectral clustering is that we do not use the eigenvectors of a graph Laplacian to construct the feature vectors. Instead, we use the solutions of total variation minimization problems to construct feature vectors that reflect connectivity between data points. Our motivation is that the solutions of total variation minimization are piece-wise constant around a given set of seed nodes. These seed nodes can be obtained from domain knowledge or by simple heuristics that are based on the network structure of data. Our results indicate that our clustering methods can cope with certain graph structures that are challenging for spectral clustering methods.

cs.LG↗

arXiv2022

Dual-Readout Calorimetry for Future Experiments Probing Fundamental Physics

In this White Paper for the 2021 Snowmass process, we detail the status and prospects for dual-readout calorimetry. While all calorimeters allow estimation of energy depositions in their active material, dual-readout calorimeters aim to provide additional information on the light produced in the sensitive media via, for example, wavelength and polarization, and/or a precision timing measurements, allowing an estimation of the shower-by-shower particle content. Utilizing this knowledge of the shower particle content may allow unprecedented energy resolution for hadronic particles and jets and new types of particle flow algorithms. We also discuss the impact continued development of this kind of calorimetry could have on precision on Higgs boson property measurements at future colliders.

physics.ins-det↗

arXiv2020

Extreme Universe Space Observatory on a Super Pressure Balloon 1 calibration: from the laboratory to the desert

The Extreme Universe Space Observatory on a Super Pressure Balloon 1 (EUSO-SPB1) instrument was launched out of Wanaka, New Zealand, by NASA in April, 2017 as a mission of opportunity. The detector was developed as part of the Joint Experimental Missions for the Extreme Universe Space Observatory (JEM-EUSO) program toward a space-based ultra-high energy cosmic ray (UHECR) telescope with the main objective to make the first observation of UHECRs via the fluorescence technique from suborbital space. The EUSO-SPB1 instrument is a refractive telescope consisting of two 1m$^2$ Fresnel lenses with a high-speed UV camera at the focal plane. The camera has 2304 individual pixels capable of single photoelectron counting with a time resolution of 2.5$μ$s. A detailed performance study including calibration was done on ground. We separately evaluated the properties of the Photo Detector Module (PDM) and the optical system in the laboratory. An end-to-end test of the instrument was performed during a field campaign in the West Desert in Utah, USA at the Telescope Array (TA) site in September 2016. The campaign lasted for 8 nights. In this article we present the results of the preflight laboratory and field tests. Based on the tests performed in the field, it was determined that EUSO-SPB1 has a field of view of 11.1$^\circ$ and an absolute photo-detection efficiency of 10%. We also measured the light flux necessary to obtain a 50% trigger efficiency using laser beams. These measurements were crucial for us to perform an accurate post flight event rate calculation to validate our cosmic ray search. Laser beams were also used to estimated the reconstruction angular resolution. Finally, we performed a flat field measurement in flight configuration at the launch site prior to the launch providing a uniformity of the focal surface better than 6%.

astro-ph.IM↗

arXiv2020

Federated Learning From Big Data Over Networks

This paper formulates and studies a novel algorithm for federated learning from large collections of local datasets. This algorithm capitalizes on an intrinsic network structure that relates the local datasets via an undirected "empirical" graph. We model such big data over networks using a networked linear regression model. Each local dataset has individual regression weights. The weights of close-knit sub-collections of local datasets are enforced to deviate only little. This lends naturally to a network Lasso problem which we solve using a primal-dual method. We obtain a distributed federated learning algorithm via a message passing implementation of this primal-dual method. We provide a detailed analysis of the statistical and computational properties of the resulting federated learning algorithm.

cs.LG↗

arXiv2018

Analysis of Network Lasso for Semi-Supervised Regression

We apply network Lasso to semi-supervised regression problems involving network structured data. This approach lends quite naturally to highly scalable learning algorithms in the form of message passing over an empirical graph which represents the network structure of the data. By using a simple non-parametric regression model, which is motivated by a clustering hypothesis, we provide an analysis of the estimation error incurred by network Lasso. This analysis reveals conditions on the the network structure and the available training data which guarantee network Lasso to be accurate. Remarkably, the accuracy of network Lasso is related to the existence of sufficiently large network flows over the empirical graph. Thus, our analysis reveals a connection between network Lasso and maximum flow problems.

stat.ML↗

arXiv2017

The EUSO@TurLab Project

The TurLab facility is a laboratory, equipped with a 5 m diameter and 1 m depth rotating tank, located in the Physics Department of the University of Turin. The tank has been built mainly to study problems where system rotation plays a key role in the fluid behaviour such as in atmospheric and oceanic flows at different scales. The tank can be filled with different fluids of variable density, which enables studies in layered conditions such as sea waves. The tank can be also used to simulate the terrestrial surface with the optical characteristics of different environments such as snow, grass, ocean, land with soil, stones etc., fogs and clouds. As it is located in an extremely dark place, the light intensity can be controlled artificially. Such capabilities of the TurLab facility are applied to perform experiments related to the observation of Extreme Energy Cosmic Rays (EECRs) from space using the fluorescence technique, as in the case of the JEM-EUSO mission, where the diffuse night brightness and artificial light sources can vary significantly in time and space inside the Field of View (FoV) of the telescope. Here we will report the currently ongoing activity at the TurLab facility in the framework of the JEM-EUSO mission (EUSO@TurLab).

astro-ph.IM↗

arXiv2017

Interfilament Resistance at 77 K in Striated HTS Coated Conductors

Striating HTS coated conductor (CC) tapes into narrow filaments offers the possibility of reducing the tapes' magnetization losses without unreasonably decreasing their current-carrying capability. However, realizing well-separated striations presents technological challenges, especially if the number of filaments is large and/or if a thick layer of metallic stabilizer is present. In these situations, the filaments can be easily coupled and their effectiveness to reduce magnetization losses is strongly diminished or eliminated. While the onset of coupling is well visible from magnetization loss measurements, the actual path of the coupling current is unknown. In this contribution we present a systematic study of the transverse resistance in HTS CC samples in order to get a deeper understanding of those paths. The measured samples differ in terms of manufacturer (SuperPower and SuperOx), and presence and thickness of stabilizer material. In addition, oxidation is used as a means to increase the resistance between the filaments in non-stabilized samples. The results are interpreted with a chain network model. This work provides useful insights on the factors determining the transverse resistance in striated HTS CCs, thus indicating ways to improve the effectiveness of the striation process for AC loss reduction.

cond-mat.supr-con↗

arXiv2015

HERAFitter, Open Source QCD Fit Project

HERAFitter is an open-source package that provides a framework for the determination of the parton distribution functions (PDFs) of the proton and for many different kinds of analyses in Quantum Chromodynamics (QCD). It encodes results from a wide range of experimental measurements in lepton-proton deep inelastic scattering and proton-proton (proton-antiproton) collisions at hadron colliders. These are complemented with a variety of theoretical options for calculating PDF-dependent cross section predictions corresponding to the measurements. The framework covers a large number of the existing methods and schemes used for PDF determination. The data and theoretical predictions are brought together through numerous methodological options for carrying out PDF fits and plotting tools to help visualise the results. While primarily based on the approach of collinear factorisation, HERAFitter also provides facilities for fits of dipole models and transverse-momentum dependent PDFs. The package can be used to study the impact of new precise measurements from hadron colliders. This paper describes the general structure of HERAFitter and its wide choice of options.

hep-ph↗

arXiv2015

QCD analysis of $W$- and $Z$-boson production at Tevatron

Recent measurements of the $W$-boson charge asymmetry and of the $Z$-boson production cross sections, performed at the Tevatron collider in Run II by the D0 and CDF collaborations, are studied using the HERAFitter framework to assess their impact on the proton parton distribution functions (PDFs). The Tevatron measurements, together with deep-inelastic scattering data from HERA, are included in a QCD analysis performed at next-to-leading order, and compared to the predictions obtained using other PDF sets from different groups. Good agreement between measurements and theoretical predictions is observed. The Tevatron data provide significant constraints on the $d$-valence quark distribution.

hep-ph↗

arXiv2014

Planning the Future of U.S. Particle Physics (Snowmass 2013): Chapter 8: Instrumentation Frontier

These reports present the results of the 2013 Community Summer Study of the APS Division of Particles and Fields ("Snowmass 2013") on the future program of particle physics in the U.S. Chapter 8, on the Instrumentation Frontier, discusses the instrumentation needs of future experiments in the Energy, Intensity, and Cosmic Frontiers, promising new technologies for particle physics research, and issues of gathering resources for long-term research in this area.

hep-ex↗