Search

In this paper, a novel dataset is introduced, designed to assess student attention within in-person classroom settings. This dataset encompasses RGB camera data, featuring multiple cameras per student to capture both posture and facial expressions, in addition to smartwatch sensor data for each individual. This dataset allows machine learning algorithms to be trained to predict attention and correlate it with emotion. A comprehensive suite of attention and emotion labels for each student is provided, generated through self-reporting as well as evaluations by four different experts. Our dataset uniquely combines facial and environmental camera data, smartwatch metrics, and includes underrepresented ethnicities in similar datasets, all within in-the-wild, in-person settings, making it the most comprehensive dataset of its kind currently available. The dataset presented offers an extensive and diverse collection of data pertaining to student interactions across different educational contexts, augmented with additional metadata from other tools. This initiative addresses existing deficiencies by offering a valuable resource for the analysis of student attention and emotion in face-to-face lessons.

In this article the 2-adjunction that relates universal arrows and extensive monads is constructed explicitly. This 2-adjunction resembles the one that relates adjunctions and monads since the 2-category of universal arrows is isomorphic to the 2-category of adjunctions and the 2-category of extensive monads is isomorphic to the 2-category of monads. This article would be useful as a foundation for a theory relating pseudo adjunctions and pseudo monads for Gray-categories. On the other hand, it might function as an accesible tool for computer scientists on extensive monads.

The relationship between an active galactic nucleus (AGN) and its host galaxy is still far from being understood. Properties of the host galaxies of Seyfert nuclei, such as luminosity concentration, morphological type, metallicity, and age of the stellar population are expected to be related with nuclear activity -- either at the epoch of galaxy formation or in the present days via feeding of the central black-hole. In this paper we investigate whether stellar ages and metallicities are linked to the activity within the nucleus in a sample of AGN of various types. Our sample includes seven AGN, from Seyfert 1 to LINERs, observed with VLT/ISAAC and VLT/SINFONI. Based on an inverse method using a stellar library, we analyse H band infrared spectra, in a wavelength region devoid of emission lines, at a spectral resolution of about 3000, in the central few 100 pc. HST images are used to visualise the regions defined in each galaxy. For each galaxy, we give the results of the spectral synthesis, in particular the percentages of the stellar, power law and blackbody continua, and the percentages of various stellar types that account for the stellar lines. Out of the seven galaxies, three show strong and recent star formation in the inner 100 pc, while no star formation is detected in the three genuine Seyfert 2 galaxies. Beyond a radius of 100 pc, all show more or less recent star formation. Moreover we can conclude that the star formation history of the inner nucleus is highly heterogeneous., Comment: Accepted for publication in Astronomy & Astrophysics Typos and language editing corrected

This paper introduces the concept of domination in the context of colored graphs (where each color assigns a weight to the vertices of its class), termed up-color domination, where a vertex dominating another must be heavier than the other. That idea defines, on one hand, a new parameter measuring the size of minimal dominating sets satisfying specific constraints related to vertex colors. The paper proves that the optimization problem associated with that concept is an NP-complete problem, even for bipartite graphs with three colors. On the other hand, a weight-based variant, the up-color domination weight, is proposed, further establishing its computational hardness. The work also explores the relationship between up-color domination and classical domination and coloring concepts. Efficient algorithms for trees are developed that use their acyclic structure to achieve polynomial-time solutions., Comment: 23 pages, 14 figures

Federated Learning presents a nascent approach to machine learning, enabling collaborative model training across decentralized devices while safeguarding data privacy. However, its distributed nature renders it susceptible to adversarial attacks. Integrating blockchain technology with Federated Learning offers a promising avenue to enhance security and integrity. In this paper, we tackle the potential of blockchain in defending Federated Learning against adversarial attacks. First, we test Proof of Federated Learning, a well known consensus mechanism designed ad-hoc to federated contexts, as a defense mechanism demonstrating its efficacy against Byzantine and backdoor attacks when at least one miner remains uncompromised. Second, we propose Krum Federated Chain, a novel defense strategy combining Krum and Proof of Federated Learning, valid to defend against any configuration of Byzantine or backdoor attacks, even when all miners are compromised. Our experiments conducted on image classification datasets validate the effectiveness of our proposed approaches., Comment: Submitted to Neural Networks

Inspired by the quantization of classical quantities and Rankin Selberg convolution, we study the anticommutator operation $\{\cdot, \cdot\}$, where $\{A,B\} = AB + BA$, applied to real symmetric random matrix ensembles including Gaussian orthogonal ensemble (GOE), the palindromic Toeplitz ensemble (PTE), the $k$-checkerboard ensemble, and the block $k$-circulant ensemble ($k$-BCE). Using combinatorial and topological techniques related to non-crossing and free matching properties of GOE and PTE, we obtain closed-form formulae for the moments of the limiting spectral distributions of $\{$GOE, GOE$\}$, $\{$PTE, PTE$\}$, $\{$GOE, PTE$\}$ and establish the corresponding limiting spectral distributions with generating functions and convolution. On the other hand, $\{$GOE, $k$-checkerboard$\}$ and $\{$$k$-checkerboard, $j$-checkerboard$\}$ exhibit entirely different spectral behavior than the other anticommutator ensembles: while the spectrum of $\{$GOE, $k$-checkerboard$\}$ consists of 1 bulk regime of size $\Theta(N)$ and 1 blip regime of size $\Theta(N^{3/2})$, the spectrum of $\{$$k$-checkerboard, $j$-checkerboard$\}$ consists of 1 bulk regime of size $\Theta(N)$, 2 intermediary blip regimes of size $\Theta(N^{3/2})$, and 1 largest blip regime of size $\Theta(N^2)$. In both cases, with the appropriate weight function, we are able to isolate the largest regime for other regime(s) and analyze its moments and convergence results via combinatorics. We end with numerical computation of lower even moments of $\{$GOE, $k$-BCE$\}$ and $\{$$k$-BCE, $k$-BCE$\}$ based on genus expansion and discussion on the challenge with analyzing the intermediary blip regimes of $\{$$k$-checkerboard, $j$-checkerboard$\}$., Comment: 48 pages, 12 figures

Multivariate Time Series Imputation (MTSI) is crucial for many applications, such as healthcare monitoring and traffic management, where incomplete data can compromise decision-making. Existing state-of-the-art methods, like Denoising Diffusion Probabilistic Models (DDPMs), achieve high imputation accuracy; however, they suffer from significant computational costs and are notably time-consuming due to their iterative nature. In this work, we propose CoSTI, an innovative adaptation of Consistency Models (CMs) for the MTSI domain. CoSTI employs Consistency Training to achieve comparable imputation quality to DDPMs while drastically reducing inference times, making it more suitable for real-time applications. We evaluate CoSTI across multiple datasets and missing data scenarios, demonstrating up to a 98% reduction in imputation time with performance on par with diffusion-based models. This work bridges the gap between efficiency and accuracy in generative imputation tasks, providing a scalable solution for handling missing data in critical spatio-temporal systems., Comment: 20 pages, 5 figures, 13 tables

This work studies several decoding algorithms for hyperbolic codes. We use some previous ideas to describe how to decode a hyperbolic code using the largest Reed-Muller code contained in it or using the smallest Reed-Muller code that contains it. A combination of these two algorithms is proposed when hyperbolic codes are defined by polynomials in two variables. Then, we compare hyperbolic codes and Cube codes (tensor product of Reed-Solomon codes) and propose decoding algorithms of hyperbolic codes based on their closest Cube codes. Finally, we adapt to hyperbolic codes the Geil and Matsumoto's generalization of Sudan's list decoding algorithm., Comment: arXiv admin note: substantial text overlap with arXiv:2107.12594

We combine the deepest X-ray survey from the Chandra Deep Field-South (CDF-S) `7-Ms' survey with the deepest mid-infrared (5.6$ \mu m$) image from the JWST/MIRI Deep Imaging Survey (MIDIS) in the Hubble Ultra-Deep Field (HUDF) to study the infrared counterparts and point-source emission of 31 X-ray sources with a median, intrinsic, rest-frame X-ray luminosity of $\log_{10}(L_{\rm Xc}^{\rm 0.5-7keV})$=42.04$\pm$0.22 erg $\rm s^{-1}$. The sample includes 24 AGN with a redshift range, as set by the X-ray detectability, of $z \simeq 0.5-3$. Through a multi-wavelength morphological decomposition, employing three separate classifications (visual, parametric and non-parametric) we separate (where present) the luminosity of the point-like AGN component from the remainder of the host-galaxy emission. The unprecedented mid-infrared sensitivity and imaging resolution of MIRI allows, in many cases, the direct characterisation of point-like (i.e. unresolved) components in the galaxies' emission. We establish a broad agreement between the three morphological classifications. At least 70% of the X-ray sources, including some classified as galaxies, show unresolved emission in the MIRI images, with the unresolved-to-total flux fraction at rest-frame 2$\mu m$ ranging from $\sim$0.2 to $\sim$0.9. At high X-ray luminosities ($\log_{10}(L_{\rm Xc}$)>43 erg $\rm s^{-1}$) we derive a consistent rest-frame near-infrared 2$ \mu m$ point-source luminosity to that derived for local AGN, whilst at lower X-ray luminosity we identify an excess in the 2$ \mu m$ emission compared to pre-JWST studies. We speculate this offset may be driven by a combination of Compton-thick AGN components and nuclear starburst, merger driven activity. Our observations highlight the complex nature of X-ray sources in the distant Universe and demonstrate the power of JWST/MIRI in quantifying their nuclear infrared emission. (Abridged), Comment: 21 pages, 7 figures. Submitted to A&A

Content generation and manipulation approaches based on deep learning methods have seen significant advancements, leading to an increased need for techniques to detect whether an image has been generated or edited. Another area of research focuses on the insertion and harmonization of objects within images. In this study, we explore the potential of using harmonization data in conjunction with a segmentation model to enhance the detection of edited image regions. These edits can be either manually crafted or generated using deep learning methods. Our findings demonstrate that this approach can effectively identify such edits. Existing forensic models often overlook the detection of harmonized objects in relation to the background, but our proposed Disharmony Network addresses this gap. By utilizing an aggregated dataset of harmonization techniques, our model outperforms existing forensic networks in identifying harmonized objects integrated into their backgrounds, and shows potential for detecting various forms of edits, including virtual try-on tasks.

This paper addresses the domain adaptation challenge for semantic segmentation in medical imaging. Despite the impressive performance of recent foundational segmentation models like SAM on natural images, they struggle with medical domain images. Beyond this, recent approaches that perform end-to-end fine-tuning of models are simply not computationally tractable. To address this, we propose a novel SAM adapter approach that minimizes the number of trainable parameters while achieving comparable performances to full fine-tuning. The proposed SAM adapter is strategically placed in the mask decoder, offering excellent and broad generalization capabilities and improved segmentation across both fully supervised and test-time domain adaptation tasks. Extensive validation on four datasets showcases the adapter's efficacy, outperforming existing methods while training less than 1% of SAM's total parameters., Comment: WACV25

In this paper, we study some existence and uniqueness results for systems of differential equations in which each of equations of the system involves a different Stieltjes derivative. Specifically, we show that this problems can only have one solution under the Osgood condition, or even, the Montel-Tonelli condition. We also explore some results guaranteeing the existence of solution under these conditions. Along the way, we obtain some interesting properties for the Lebesgue-Stieltjes integral associated to a finite sum of nondecreasing and left-continuous maps, as well as a characterization of the pseudometric topologies defined by this type of maps.

Retrieval-Augmented Generation (RAG) has become ubiquitous when deploying Large Language Models (LLMs), as it can address typical limitations such as generating hallucinated or outdated information. However, when building real-world RAG applications, practical issues arise. First, the retrieved information is generally domain-specific. Since it is computationally expensive to fine-tune LLMs, it is more feasible to fine-tune the retriever to improve the quality of the data included in the LLM input. Second, as more applications are deployed in the same real-world system, one cannot afford to deploy separate retrievers. Moreover, these RAG applications normally retrieve different kinds of data. Our solution is to instruction fine-tune a small retriever encoder on a variety of domain-specific tasks to allow us to deploy one encoder that can serve many use cases, thereby achieving low-cost, scalability, and speed. We show how this encoder generalizes to out-of-domain settings as well as to an unseen retrieval task on real-world enterprise use cases., Comment: 9 pages, 2 figures. Submitted to NAACL 2025 Industry Track

For the first time, we use relativistic mean-field (RMF) approximation with density-dependent couplings, adjusted by the DDME2 parameterization, to investigate the effects of dark matter on supernova remnants. We calculate the nuclear equation of state for nuclear and dark matter separately, under the thermodynamic conditions related to the evolution of supernova remnants. A mirrored model is adopted for dark matter, and its effect on remnant matter is studied using a two-fluid scenario. At each stage of the remnant evolution, we assume that dark and ordinary matter have the same entropy and lepton fraction, and a fixed proportion of dark matter mass fraction is added to the stellar matter to observe its effects on some microscopic and macroscopic properties of the star. We observe that dark matter in the remnant core reduces the remnant's maximum mass, radius, and tidal deformability. Moreover, dark matter heats the remnant matter and alters particle distributions, thereby decreasing its isospin asymmetry and increasing the sound speed through the matter., Comment: 15 pages, 6 figures

Magnetic fields are ubiquitous across different physical systems of current interest; from the early Universe, compact astrophysical objects and heavy-ion collisions to condensed matter systems. A proper treatment of the effects produced by magnetic fields during the dynamical evolution of these systems, can help to understand observables that otherwise show a puzzling behavior. Furthermore, when these fields are comparable to or stronger than \Lambda_QCD, they serve as excellent probes to help elucidate the physics of strongly interacting matter under extreme conditions of temperature and density. In this work we provide a comprehensive review of recent developments on the description of QED and QCD systems where magnetic field driven effects are important. These include the modification of meson static properties such as masses and form factors, the chiral magnetic effect, the description of anomalous transport coefficients, superconductivity in extreme magnetic fields, the properties of neutron stars, the evolution of heavy-ion collisions, as well as effects on the QCD phase diagram. We describe recent theory and phenomenological developments using effective models as well as LQCD methods. The work represents a state-of-the-art review of the field, motivated by presentations and discussions during the "Workshop on Strongly Interacting Matter in Strong Electromagnetic Fields" that took place in the European Centre for Theoretical Studies in Nuclear Physics and Related Areas (ECT*) in the city of Trento, Italy, September 25-29, 2023., Comment: 325 pages-long review of recent topics of interest in the field of magnetic field effects on QED and QCD matter. To be susbmitted to PNPP

We present the dynamics of a single harmonically trapped ion interacting with a laser, considering a linear combination of two eigenstates of the system as the initial state. The conditions on the physical parameters that allow for the evolution of the system are discussed. We are able to obtain analytical results even though no approximations are realized and are able to show Rabi oscillations for such initial states., Comment: 8 pages, 5 figures

This paper presents a deep MIRI/JWST medium resolution spectroscopy (MRS) covering the rest-frame optical spectrum of the GN-z11 galaxy. The [OIII]5008 and H$\alpha$ emission lines are detected and spectroscopically resolved. The line profiles are well-modeled by a narrow Gaussian component with intrinsic FWHMs of 189$\pm$25 and 231$\pm$52 kms$^{-1}$, respectively. We do not find any evidence of a dominant broad H$\alpha$ emission line component tracing a Broad Line Region in a type 1 active galactic nuclei (AGN). However, a broad ($\sim$430-470 kms$^{-1}$) and weak ($<$ 20-30%) H$\alpha$ line component, tracing a minor AGN contribution in the optical, cannot be ruled out completely with the sensitivity of the present data. The physical and excitation properties of the ionized gas are consistent with a low-metallicity starburst forming stars at a rate of SFR(H$\alpha$)$=$24 $\pm$3$M_{\odot}$yr$^{-1}$. The electron temperature of the ionized gas is $T_{\mathrm{e}}$(O$^{++}$)$=$14000$\pm$2100K, while the direct-$T_{\mathrm{e}}$ gas-phase metallicity is 12+$\log$(O/H)$=$7.91$\pm$0.07 (Z=0.17$\pm$0.03Z$_{\odot}$). The optical line ratios locate GN-z11 in the starburst or AGN region but more consistent with those of local low-metallicity starbursts and high-$z$ luminous galaxies detected at redshifts similar to GN-z11. We conclude that the MRS optical spectrum of GN-z11 is consistent with that of a massive, compact, and low-metallicity starburst galaxy. Due to its high SFR and stellar mass surface densities, close to that of the densest stellar clusters, we speculate that GN-z11 could be undergoing a feedback-free, highly efficient starburst phase. Additional JWST data are needed to validate this scenario, and other recently proposed alternatives, to explain the existence of bright compact galaxies in the early Universe., Comment: accepted for publication in A&A (17 pages, 6 figures, 2 tables)

We study the effects of hyperons, delta baryons, and quark matter phase transitions on f-mode oscillations in neutron stars. Using the density-dependent relativistic mean-field model (DDME2) for the hadronic phase and the density-dependent quark mass (DDQM) model for the quark phase, we construct hadronic and hybrid equations of state (EoSs) consistent with astrophysical constraints. Including hyperons and delta baryons soften the EoS, reducing maximum masse, while phase transition to the quark matter further softens the EoS, decreasing the speed of sound and hence the maximum mass. f-mode frequencies, calculated using both the Cowling approximation and the general relativistic (GR) frameworks, reveal a significant overestimation by the Cowling method of about 10-30%, with discrepancies decreasing for more massive stars. We derive universal relations connecting the frequencies of the f-mode to the average density, compactness, and tidal deformability, finding significant deviations due to hyperons and delta baryons. Empirical relations for mass-scaled and radius-scaled frequencies are also provided, highlighting the importance of GR calculations for accurate modeling. These findings highlight the potential of gravitational wave asteroseismology to constrain neutron star EoSs and internal structure., Comment: 18 pages, 13 figures, 5 tables

Power system oscillations are a significant concern for system operators, a problem that has grown due to the interconnection of inverter-based resources. To address this issue, various methods have been proposed to locate the sources of oscillations, which is essential for effective mitigation actions. A common characteristic of these methods is that they rely on phasor representation of oscillation phenomena. This paper takes a different approach by examining the actual voltage and current waveforms underlying the phasors. It is found that the presence of interharmonic components is both the necessary and sufficient condition for phasor oscillations. Moreover, the generation and propagation of interharmonic powers are identified as the true culprits behind power system oscillations and oscillatory instability. Based on these insights, two new methods are developed for locating oscillation sources: one for measurement-based monitoring applications and another for model-based system studies. These findings are validated through four field data-based and one simulation-based case studies., Comment: 13 pages and 27 figures. An earlier version was submitted to IEEE Trans. on Power System on Aug. 27, 2024 as TPWRS-01433-2024 (Review results unknown as of today). The current version is an improved version for record

In this work we constrain the value of $\sigma_8$ for the normal and self-accelerating branch of a DGP brane embedded in a five-dimensional Minkowski space-time. For that purpose we first constrain the model parameters $H_0$, $\Omega_{m0}$, $\Omega_{r0}$ and $M$ by means of the Pantheon+ catalog and a mock catalog of gravitational waves. Then, we solve numerically the equation for dark matter scalar perturbations using the dynamical scaling solution for the master equation and assuming that $p=4$ for the matter dominated era. Finally, we found that the evolution of matter density perturbations in both branches is different from the $\Lambda$CDM model and that the value of $\sigma_8=0.774\pm0.027$ for the normal branch and $\sigma_8=0.913\pm0.032$ for the self-accelerating branch., Comment: 14 pages, 5 figures

Multi-class semantic segmentation remains a cornerstone challenge in computer vision. Yet, dataset creation remains excessively demanding in time and effort, especially for specialized domains. Active Learning (AL) mitigates this challenge by selecting data points for annotation strategically. However, existing patch-based AL methods often overlook boundary pixels critical information, essential for accurate segmentation. We present OREAL, a novel patch-based AL method designed for multi-class semantic segmentation. OREAL enhances boundary detection by employing maximum aggregation of pixel-wise uncertainty scores. Additionally, we introduce one-vs-rest entropy, a novel uncertainty score function that computes class-wise uncertainties while achieving implicit class balancing during dataset creation. Comprehensive experiments across diverse datasets and model architectures validate our hypothesis., Comment: WACV 2025, 8 pages

This study investigates the radial oscillations of hybrid neutron stars, characterized by a composition of hadronic external layers and a quark matter core. Utilizing a density-dependent relativistic mean-field model that incorporates hyperons and baryons for describing hadronic matter, and a density-dependent quark model for quark matter, we analyze the ten lowest eigenfrequencies and their corresponding oscillation functions. Our focus lies on neutron stars with equations-of-state involving N, N + $\Delta$, N + H, and N + H + $\Delta$, featuring a phase transition to quark matter. Emphasizing the effects of a slow phase transition at the hadron-quark interface, we observe that the maximum mass is attained before the fundamental mode's frequency decreases for slow phase transitions. This observation implies the stability of stellar configurations with higher central densities than the maximum mass, called Slow Stable Hybrid Stars (SSHSs), even under small radial perturbations. The length of these SSHS branch depends upon the energy density jump between two phases and the stiffness of the quark EoS., Comment: Contribution for the conference proceedings of the "10th International Conference on Quarks and Nuclear Physics (QNP 2024)"

AI technologies are moving rapidly from research to production. With the popularity of Foundation Models (FMs) that generate text, images, and video, AI-based systems are increasing their complexity. Compared to traditional AI-based software, systems employing FMs, or GenAI-based systems, are more difficult to design due to their scale and versatility. This makes it necessary to document best practices, known as design patterns in software engineering, that can be used across GenAI applications. Our first contribution is to formalize two techniques, Task Decomposition and Retrieval-Augmented Generation (RAG), as design patterns for GenAI-based systems. We discuss their trade-offs in terms of software quality attributes and comment on alternative approaches. We recommend to AI practitioners to consider these techniques not only from a scientific perspective but also from the standpoint of desired engineering properties such as flexibility, maintainability, safety, and security. As a second contribution, we describe our industry experience applying Task Decomposition and RAG to build a complex real-world GenAI application for enterprise users: Workflow Generation. The task of generating workflows entails generating a specific plan using data from the system environment, taking as input a user requirement. As these two patterns affect the entire AI development cycle, we explain how they impacted the dataset creation, model training, model evaluation, and deployment phases., Comment: Under review; 12 pages, 8 figures

The recently launched James Webb Space Telescope (JWST) is opening new observing windows on the distant universe. Among JWST's instruments, the Mid Infrared Instrument (MIRI) offers the unique capability of imaging observations at wavelengths $\lambda > 5\mu$m. This enables unique access to the rest frame near infra-red (NIR, $\lambda \ge 1$\mum) emission from galaxies at redshifts $z>4$ and the visual ($\lambda \gtrsim 5000$\AA) rest frame for $z>9$. We here report on the guaranteed time observations (GTO) from the MIRI European Consortium, of the Hubble Ultra Deep Field (HUDF), forming the MIRI Deep Imaging Survey (MIDIS), consisting of an on source integration time of $\sim41$ hours in the MIRI/F560W (5.6 $\mu$m) filter. To our knowledge, this constitutes the longest single filter exposure obtained with JWST of an extragalactic field as yet., Comment: submitted to A&A on July 30, 2024

We give an explicit presentation of a family of finite-dimensional pointed Hopf algebras over an algebraically closed field of characteristic zero that constitute all liftings of Nichols algebras of diagonal Cartan type $B_{3}$ over a finite abelian group.

Context. Spider pulsars are millisecond pulsars in tight binary systems, in which a low-mass companion star is heated and ablated by the pulsar wind. Their observations allow one to study stellar evolution with formation of millisecond pulsars and physics of superdense matter in neutron stars. However, spiders are rare due to difficulties of their discovery using typical radio search techniques. The Fermi $\gamma$-ray source 4FGL J1544.2$-$2554 was recently proposed as a pulsar candidate, and its likely X-ray and optical counterparts with the galactic coordinates $l\approx344.\!\!^\circ76$, $b\approx22.\!\!^\circ59$ and the magnitude $G\approx20.6$ were found using the eROSITA and Gaia surveys. Aims. Our goals are to study whether the source is a new spider pulsar and to estimate its fundamental parameters. Methods. We performed the first optical time-series multi-band photometry of the object. We used the Lomb-Scargle periodogram to search for its brightness periodicity and fitted its light curves with a model of direct heating of the binary companion by the pulsar wind. Results. The source shows a strong brightness variability with a period of $\approx$ 2.724 h and an amplitude of $\gtrsim$ 2.5 mag, and its light curves have a single broad peak per period. These features are typical for spider pulsars. The curves are well fitted by the direct heating model, resulting in an orbit inclination of the presumed spider system of $\approx 83^\circ$, a companion mass of $\approx 0.1$ M$_\odot$, its ''day-side'' and ''night-side'' temperatures of $\approx 7200$ K and $\approx 3000$ K, a Roche-lobe filling factor of $\approx 0.65$ and a distance of $\approx 2.1$ kpc. Conclusions. Our findings suggest that 4FGL J1544.2$-$2554 is a spider pulsar. This encourages searches for the pulsar millisecond pulsations in the radio and $\gamma$-rays to confirm its nature., Comment: 6 pages, accepted for publication in Astronomy & Astrophysics

Here, we present the outcomes from the second Large Language Model (LLM) Hackathon for Applications in Materials Science and Chemistry, which engaged participants across global hybrid locations, resulting in 34 team submissions. The submissions spanned seven key application areas and demonstrated the diverse utility of LLMs for applications in (1) molecular and material property prediction; (2) molecular and material design; (3) automation and novel interfaces; (4) scientific communication and education; (5) research data management and automation; (6) hypothesis generation and evaluation; and (7) knowledge extraction and reasoning from scientific literature. Each team submission is presented in a summary table with links to the code and as brief papers in the appendix. Beyond team results, we discuss the hackathon event and its hybrid format, which included physical hubs in Toronto, Montreal, San Francisco, Berlin, Lausanne, and Tokyo, alongside a global online hub to enable local and virtual collaboration. Overall, the event highlighted significant improvements in LLM capabilities since the previous year's hackathon, suggesting continued expansion of LLMs for applications in materials science and chemistry research. These outcomes demonstrate the dual utility of LLMs as both multipurpose models for diverse machine learning tasks and platforms for rapid prototyping custom applications in scientific research., Comment: Updating author information, the submission remains largely unchanged. 98 pages total

The reliability of supervised classifiers is severely hampered by their limitations in dealing with unexpected inputs, leading to great interest in out-of-distribution (OOD) detection. Recently, OOD detectors trained on synthetic outliers, especially those generated by large diffusion models, have shown promising results in defining robust OOD decision boundaries. Building on this progress, we present NCIS, which enhances the quality of synthetic outliers by operating directly in the diffusion's model embedding space rather than combining disjoint models as in previous work and by modeling class-conditional manifolds with a conditional volume-preserving network for more expressive characterization of the training distribution. We demonstrate that these improvements yield new state-of-the-art OOD detection results on standard ImageNet100 and CIFAR100 benchmarks and provide insights into the importance of data pre-processing and other key design choices. We make our code available at \url{https://github.com/LarsDoorenbos/NCIS}.

With a growing number of facilities able to monitor the entire sky and produce light curves with a cadence of days, in recent years there has been an increased rate of detection of sources whose variability deviates from standard behavior, revealing a variety of exotic nuclear transients. The aim of the present study is to disentangle the nature of the transient AT 2021hdr, whose optical light curve used to be consistent with a classic Seyfert 1 nucleus, which was also confirmed by its optical spectrum and high-energy properties. From late 2021, AT 2021hdr started to present sudden brightening episodes in the form of oscillating peaks in the Zwicky Transient Facility (ZTF) alert stream, and the same shape is observed in X-rays and UV from Swift data. The oscillations occur every about 60-90 days with amplitudes of around 0.2 mag in the g and r bands. Very Long Baseline Array (VLBA) observations show no radio emission at milliarcseconds scale. It is argued that these findings are inconsistent with a standard tidal disruption event (TDE), a binary supermassive black hole (BSMBH), or a changing-look active galactic nucleus (AGN); neither does this object resemble previous observed AGN flares, and disk or jet instabilities are an unlikely scenario. Here, we propose that the behavior of AT 2021hdr might be due to the tidal disruption of a gas cloud by a BSMBH. In this scenario, we estimate that the putative binary has a separation of about 0.83 mpc and would merge in about 70000 years. This galaxy is located at 9 kpc from a companion galaxy, and in this work we report this merger for the first time. The oscillations are not related to the companion galaxy., Comment: 11 pages, 5 figures, 2 tables, plus appendices. Accepted for publication in Astronomy & Astrophysics. NASA press release available at https://science.nasa.gov/missions/swift/nasas-swift-studies-gas-churning-monster-black-holes/ doi: 10.1051/0004-6361/202451305

We analyze the proton$\text{-}$box contribution to the hadronic light$\text{-}$by$\text{-}$light part of the muon's anomalous magnetic moment, which is the first reported baryonic contribution to this piece. We follow the quark$\text{-}$loop analysis, incorporating the relevant data$\text{-}$driven and lattice proton form factors. Although the heavy mass expansion would yield a contribution of $\mathcal{O}(10^{-10})$, the damping of the form factors in the regions where the kernel peaks, explains our finding $a_{\mu}^{\rm{p-box}}=1.82 (7)\times 10^{-12}$, two orders of magnitude smaller than the forthcoming uncertainty on the $a_{\mu}$ measurement and on its Standard Model prediction., Comment: 18 pages, 4 figures

This paper reports on a comprehensive study of an ultra-relativistic positron beam deflection by an oriented quasi-mosaic crystal. The analysis was carried out for the positron energy of 530 MeV incident on the quasi-mosaic bent Si(111) crystal. This particular case was chosen because it has recently been studied experimentally at the Mainz Microtron MAMI. The results of the relativistic molecular dynamics simulations were compared with the experimental observations and a good agreement was found. The presence of planar channeling, de-channeling, volume reflection and volume capture processes in the angular distribution of deflected positrons for different beam-crystal alignments has been studied. Predictions have been made for certain crystal orientations for which experimental data are lacking.

The Dice Similarity Coefficient (DSC) is the current de facto standard to determine agreement between a reference segmentation and one generated by manual / auto-contouring approaches. This metric is useful for non-spatially important images; however, radiation therapy requires consideration of nearby Organs-at-Risk (OARs) and their radiosensitivity which are currently unaccounted for with the traditional DSC. In this work, we introduce the OAR-DSC which accounts for nearby OARs and their radiosensitivity when computing the DSC. We illustrate the importance of this through cases where two proposed contours have similar DSC, but lower OAR-DSC when one contour expands closer to the surrounding OARs. This work is important because the OAR-DSC may be used by deep learning auto-contouring algorithms in a radiation therapy specific loss function, thereby progressing on the current disregard for the importance of these differences on the final radiation dose plan generation, delivery, and risks of patient toxicity., Comment: Proceedings of the XX-th international conference on the use of computers in radiation therapy (ICCR)

Imaging atmospheric Cherenkov telescopes (IACTs) are used to observe very high-energy photons from the ground. Gamma rays are indirectly detected through the Cherenkov light emitted by the air showers they induce. The new generation of experiments, in particular the Cherenkov Telescope Array Observatory (CTAO), sets ambitious goals for discoveries of new gamma-ray sources and precise measurements of the already discovered ones. To achieve these goals, both hardware and data analysis must employ cutting-edge techniques. This also applies to the LST-1, the first IACT built for the CTAO, which is currently taking data on the Canary island of La Palma. This paper introduces a new event reconstruction technique for IACT data, aiming to improve the image reconstruction quality and the discrimination between the signal and the background from misidentified hadrons and electrons. The technique models the development of the extensive air shower signal, recorded as a waveform per pixel, seen by CTAO telescopes' cameras. Model parameters are subsequently passed to random forest regressors and classifiers to extract information on the primary particle. The new reconstruction was applied to simulated data and to data from observations of the Crab Nebula performed by the LST-1. The event reconstruction method presented here shows promising performance improvements. The angular and energy resolution, and the sensitivity, are improved by 10 to 20% over most of the energy range. At low energy, improvements reach up to 22%, 47%, and 50%, respectively. A future extension of the method to stereoscopic analysis for telescope arrays will be the next important step., Comment: Accepted in A&A

The Workflows Community Summit gathered 111 participants from 18 countries to discuss emerging trends and challenges in scientific workflows, focusing on six key areas: time-sensitive workflows, AI-HPC convergence, multi-facility workflows, heterogeneous HPC environments, user experience, and FAIR computational workflows. The integration of AI and exascale computing has revolutionized scientific workflows, enabling higher-fidelity models and complex, time-sensitive processes, while introducing challenges in managing heterogeneous environments and multi-facility data dependencies. The rise of large language models is driving computational demands to zettaflop scales, necessitating modular, adaptable systems and cloud-service models to optimize resource utilization and ensure reproducibility. Multi-facility workflows present challenges in data movement, curation, and overcoming institutional silos, while diverse hardware architectures require integrating workflow considerations into early system design and developing standardized resource management tools. The summit emphasized improving user experience in workflow systems and ensuring FAIR workflows to enhance collaboration and accelerate scientific discovery. Key recommendations include developing standardized metrics for time-sensitive workflows, creating frameworks for cloud-HPC integration, implementing distributed-by-design workflow modeling, establishing multi-facility authentication protocols, and accelerating AI integration in HPC workflow management. The summit also called for comprehensive workflow benchmarks, workflow-specific UX principles, and a FAIR workflow maturity model, highlighting the need for continued collaboration in addressing the complex challenges posed by the convergence of AI, HPC, and multi-facility research environments.

Large Language Models~(LLMs) have demonstrated capabilities across various applications but face challenges such as hallucination, limited reasoning abilities, and factual inconsistencies, especially when tackling complex, domain-specific tasks like question answering~(QA). While Knowledge Graphs~(KGs) have been shown to help mitigate these issues, research on the integration of LLMs with background KGs remains limited. In particular, user accessibility and the flexibility of the underlying KG have not been thoroughly explored. We introduce AGENTiGraph (Adaptive Generative ENgine for Task-based Interaction and Graphical Representation), a platform for knowledge management through natural language interaction. It integrates knowledge extraction, integration, and real-time visualization. AGENTiGraph employs a multi-agent architecture to dynamically interpret user intents, manage tasks, and integrate new knowledge, ensuring adaptability to evolving user requirements and data contexts. Our approach demonstrates superior performance in knowledge graph interactions, particularly for complex domain-specific tasks. Experimental results on a dataset of 3,500 test cases show AGENTiGraph significantly outperforms state-of-the-art zero-shot baselines, achieving 95.12\% accuracy in task classification and 90.45\% success rate in task execution. User studies corroborate its effectiveness in real-world scenarios. To showcase versatility, we extended AGENTiGraph to legislation and healthcare domains, constructing specialized KGs capable of answering complex queries in legal and medical contexts., Comment: 30 pages, 7 figures; Submitted to COLING 2025 System Demonstrations Track

We analysed ALMA FIR (1.3 mm) dust continuum and CO emission of 12 starburst galaxies at $z\sim 2.1-3.6$, selected for their extreme brightness in the rest-UV with $M_{\rm UV} = -23.4$ to $-24.7$. We also analysed VLT HAWK-I $H$- and $K_{\rm s}$-band images. The galaxies are characterised by negligible dust attenuations with blue UV spectral slopes ($-2.62$ to $-1.84$), very young stellar populations of $\sim 10$ Myr, and powerful starbursts with a high mean specific star formation rate of $\rm 112~Gyr^{-1}$, placing them $\sim 1.5$~dex above the main sequence at similar redshifts and stellar masses ($M_{\rm stars} \sim (1.5-4.6)\times 10^9~M_{\odot}$). The FIR dust continuum emission revealed in 9 galaxies yields IR luminosities of $(5.9-28.3)\times 10^{11}~L_{\odot}$ and large dust masses barely produced by SNe within the 10~Myr timescale. The CO emission detected in 8 galaxies evidence large molecular gas masses with a mean molecular gas fraction of 82%. The corresponding star formation efficiencies reach $\gtrsim 40$%, with amazingly short molecular gas depletion timescales between <13 Myr and 71 Myr. These unique properties never reported in previously studied galaxies highlight that these galaxies are likely caught at the very beginning of their stellar mass build-up and undergo a very efficient and fast conversion of gas into stars that can only result from the gas collapse within very short free-fall times. We find that the feedback-free starburst model seems to be able to explain the formation of these galaxies. To reconcile the co-spatial FIR dust emission with the UV-bright unattenuated emission, we speculate about radiation-driven outflows that can temporarily remove dust at the location of the starburst and expel dust at large distances in line with the measured large FIR effective radii ($\rm 1.7~kpc - 5~kpc$) in comparison to very compact stellar radii., Comment: 21 pages, 8 figures, 4 tables, 1 Appendix. Astronomy & Astrophysics in press

The safety alignment ability of Vision-Language Models (VLMs) is prone to be degraded by the integration of the vision module compared to its LLM backbone. We investigate this phenomenon, dubbed as ''safety alignment degradation'' in this paper, and show that the challenge arises from the representation gap that emerges when introducing vision modality to VLMs. In particular, we show that the representations of multi-modal inputs shift away from that of text-only inputs which represent the distribution that the LLM backbone is optimized for. At the same time, the safety alignment capabilities, initially developed within the textual embedding space, do not successfully transfer to this new multi-modal representation space. To reduce safety alignment degradation, we introduce Cross-Modality Representation Manipulation (CMRM), an inference time representation intervention method for recovering the safety alignment ability that is inherent in the LLM backbone of VLMs, while simultaneously preserving the functional capabilities of VLMs. The empirical results show that our framework significantly recovers the alignment ability that is inherited from the LLM backbone with minimal impact on the fluency and linguistic capabilities of pre-trained VLMs even without additional training. Specifically, the unsafe rate of LLaVA-7B on multi-modal input can be reduced from 61.53% to as low as 3.15% with only inference-time intervention. WARNING: This paper contains examples of toxic or harmful language., Comment: Preprint

We study large classes of real-valued analytic functions that naturally emerge in the understanding of Dulac's problem, which addresses the finiteness of limit cycles in planar differential equations. Building on a Maximum Modulus-type result we got, our main statement essentially follows. Namely, for any function belonging to these classes, the following dichotomy holds: either it has isolated zeros or it coincides with the identity. As an application, we prove that the non-accumulation of limit cycles holds around a specific class of the so-called superreal polycycles., Comment: 19 pages, 4 figures

6G, the next generation of mobile networks, is set to offer even higher data rates, ultra-reliability, and lower latency than 5G. New 6G services will increase the load and dynamism of the network. Network Function Virtualization (NFV) aids with this increased load and dynamism by eliminating hardware dependency. It aims to boost the flexibility and scalability of network deployment services by separating network functions from their specific proprietary forms so that they can run as virtual network functions (VNFs) on commodity hardware. It is essential to design an NFV orchestration and management framework to support these services. However, deploying bulky monolithic VNFs on the network is difficult, especially when underlying resources are scarce, resulting in ineffective resource management. To address this, microservices-based NFV approaches are proposed. In this approach, monolithic VNFs are decomposed into micro VNFs, increasing the likelihood of their successful placement and resulting in more efficient resource management. This article discusses the proposed framework for resource allocation for microservices-based services to provide end-to-end Quality of Service (QoS) using the Double Deep Q Learning (DDQL) approach. Furthermore, to enhance this resource allocation approach, we discussed and addressed two crucial sub-problems: the need for a dynamic priority technique and the presence of the low-priority starvation problem. Using the Deep Deterministic Policy Gradient (DDPG) model, an Adaptive Scheduling model is developed that effectively mitigates the starvation problem. Additionally, the impact of incorporating traffic load considerations into deployment and scheduling is thoroughly investigated.

The problem of imputing multivariate time series spans a wide range of fields, from clinical healthcare to multi-sensor systems. Initially, Recurrent Neural Networks (RNNs) were employed for this task; however, their error accumulation issues led to the adoption of Transformers, leveraging attention mechanisms to mitigate these problems. Concurrently, the promising results of diffusion models in capturing original distributions have positioned them at the forefront of current research, often in conjunction with Transformers. In this paper, we propose replacing time-oriented Transformers with State-Space Models (SSM), which are better suited for temporal data modeling. Specifically, we utilize the latest SSM variant, S6, which incorporates attention-like mechanisms. By embedding S6 within Mamba blocks, we develop a model that integrates SSM, Graph Neural Networks, and node-oriented Transformers to achieve enhanced spatiotemporal representations. Implementing these architectural modifications, previously unexplored in this field, we present Time series Imputation with Bi-directional mamba blocks and diffusion models (TIMBA). TIMBA achieves superior performance in almost all benchmark scenarios and performs comparably in others across a diverse range of missing value situations and three real-world datasets. We also evaluate how the performance of our model varies with different amounts of missing values and analyse its performance on downstream tasks. In addition, we provide the original code to replicate the results., Comment: 14 pages, 7 tables and 2 figures

This research presents a framework for analyzing the dynamics of online communities in social media platforms, utilizing a temporal fusion of text and network data. By combining text classification and dynamic social network analysis, we uncover mechanisms driving community formation and evolution, revealing the influence of real-world events. We introduced fourteen key elements based on social science theories to evaluate social media dynamics, validating our framework through a case study of Twitter data during major U.S. events in 2020. Our analysis centers on discrimination discourse, identifying sexism, racism, xenophobia, ableism, homophobia, and religious intolerance as main fragments. Results demonstrate rapid community emergence and dissolution cycles representative of discourse fragments. We reveal how real-world circumstances impact discourse dominance and how social media contributes to echo chamber formation and societal polarization. Our comprehensive approach provides insights into discourse fragmentation, opinion dynamics, and structural aspects of online communities, offering a methodology for understanding the complex interplay between online interactions and societal trends.

Some giant radio galaxies selected at X-rays with an AGN show signs of a restarted nuclear activity. One object in this peculiar class is Mrk1498, a giant low-frequency double radio source that shows extended emission in [OIII]. This emission is likely related to the history of the nuclear activity of the galaxy. We investigate whether this bubble-like emission might trace an outflow from either present or past AGN activity. Using MEGARA/GTC, medium-resolution spectroscopy (R 10000) data, we obtained the kinematics and fluxes of the ionised gas from modelling the [OIII] and Hbeta features.with three kinematic components. All the components show an overall blue to red velocity pattern, with similar peak-to-peak velocities but a different velocity dispersion. At a galactocentric distance of 2.3 kpc, we found a blob with a velocity up to 100km/s, and a high velocity dispersion (170km/s) that is spatially coincident with the direction of the radio jet. The observed [OIII]/Hbeta line ratio indicates possible ionisation from AGN or shocks nearly everywhere. The clumpy structure visibile in HST images at kpc scales show the lowest values of log[OIII]/Hbeta , which is likely not related to the photoionisation by the AGN. Taking optical and radio activity into account, we propose a scenario of two different ionised gas features over the radio AGN lifecycle of Mrk 1498. The radio emission suggests at least two main radio activity episodes: an old episode at Mpc scales (formed during a time span of 100Myr), and a new episode from the core (>2000yr ago). At optical wavelengths, we observe clumps and a blob that are likely associated with fossil outflow. The latter is likely powered by past episodes of the flickering AGN activity that may have occurred between the two main radio phases., Comment: 10 pages, 6 Figures

Recent observations on the large-scale structure of the universe indicate that the cosmological constant cannot be the definitive answer to the nature of dark energy. Therefore, it is a good time to propose alternatives to understand the late-accelerated expansion of the universe. In this work we study the possibility that the acceleration of space-time is due to the topology of the universe. We assume that the topology of the universe is a principal fiber bundle whose base space is our 4-dimensional spacetime and the fiber is an $N$-dimensional Lie group that evolves with time. For the base space we consider a homogeneous and isotropic spacetime, we find that the base space is currently accelerating for $1 < N$ compact semi-simple Lie groups whose scale-factors are equal and for $1 < N$ non-Abelian Lie groups whose scale-factors are different and as long as its structure constants satisfy some conditions. However when we study the evolution of the density parameters these differ from the evolution within the $\Lambda$CDM model, this led us to think in the possibility of use a different group as fiber in order to obtain the right evolution of the density parameters. We conclude that it is possible that the accelerated expansion of the universe is due to consider a different topology of the universe as a principal fiber bundle., Comment: 16 pages, 5 figures