Search

Financial organizations collect a huge amount of data about clients that typically has a temporal (sequential) structure and is collected from various sources (modalities). Due to privacy issues, there are no large-scale open-source multimodal datasets of event sequences, which significantly limits the research in this area. In this paper, we present the industrial-scale publicly available multimodal banking dataset, MBD, that contains more than 1.5M corporate clients with several modalities: 950M bank transactions, 1B geo position events, 5M embeddings of dialogues with technical support and monthly aggregated purchases of four bank's products. All entries are properly anonymized from real proprietary bank data. Using this dataset, we introduce a novel benchmark with two business tasks: campaigning (purchase prediction in the next month) and matching of clients. We provide numerical results that demonstrate the superiority of our multi-modal baselines over single-modal techniques for each task. As a result, the proposed dataset can open new perspectives and facilitate the future development of practically important large-scale multimodal algorithms for event sequences. HuggingFace Link: https://huggingface.co/datasets/ai-lab/MBD Github Link: https://github.com/Dzhambo/MBD

Machine learning (ML) models trained on datasets owned by different organizations and physically located in remote databases offer benefits in many real-world use cases. State regulations or business requirements often prevent data transfer to a central location, making it difficult to utilize standard machine learning algorithms. Federated Learning (FL) is a technique that enables models to learn from distributed datasets without revealing the original data. Vertical Federated learning (VFL) is a type of FL where data samples are divided by features across several data owners. For instance, in a recommendation task, a user can interact with various sets of items, and the logs of these interactions are stored by different organizations. In this demo paper, we present \emph{Stalactite} - an open-source framework for VFL that provides the necessary functionality for building prototypes of VFL systems. It has several advantages over the existing frameworks. In particular, it allows researchers to focus on the algorithmic side rather than engineering and to easily deploy learning in a distributed environment. It implements several VFL algorithms and has a built-in homomorphic encryption layer. We demonstrate its use on a real-world recommendation datasets.

The quantum Hall effect, which exhibits a number of unusual properties, is studied in a gated 1000-nm-thick HgTe film, nominally a three-dimensional system. A weak zero plateau of Hall resistance, accompanied by a relatively small value of Rxx of the order of h/e^2, is found around the point of charge neutrality. It is shown that the zero plateau is formed by the counter-propagating chiral electron-hole edge channels, the scattering between which is suppressed. So, phenomenologically, the quantum spin Hall effect is reproduced, but with preserved ballisticity on macroscopic scales (larger than 1mm). It is shown that the formation of the QHE occurs in a two-dimensional (2D) accumulation layer near the gate, while the bulk carriers play the role of an electron reservoir. Due to the exchange of carriers between the reservoir and the 2D layer, an anomalous scaling of the QHE is observed not with respect to the CNP, but with respect to the first electron plateau., Comment: 12 pages, 10 figures

Observations of quasars show that the polarization position angle of the emission coming from them varies greatly over time, including periods called rotations during which the angle changes in an orderly manner. The study proposes a method for identifying such events and assessing their statistical significance. The operation of the method is demonstrated using the example of long-term polarimetric observations of the blazars CTA 102, 3C 454.3, and OT 081. During the analysis of light curves, 51 rotations of the polarization position angle were found and it was shown that for CTA 102 and 3C 454.3 the rotations are predominantly oriented in one direction., Comment: 21 pages, 12 figures

Forecasting future events over extended periods, known as long-horizon prediction, is a fundamental task in various domains, including retail, finance, healthcare, and social networks. Traditional methods, such as Marked Temporal Point Processes (MTPP), typically use autoregressive models to predict multiple future events. However, these models frequently encounter issues such as converging to constant or repetitive outputs, which significantly limits their effectiveness and applicability. To overcome these limitations, we propose DeTPP (Detection-based Temporal Point Processes), a novel approach inspired by object detection methods from computer vision. DeTPP utilizes a novel matching-based loss function that selectively focuses on reliably predictable events, enhancing both training robustness and inference diversity. Our method sets a new state-of-the-art in long-horizon event prediction, significantly outperforming existing MTPP and next-K approaches. The implementation of DeTPP is publicly available on GitHub.

The expansion in automation of increasingly fast applications and low-power edge devices poses a particular challenge for optimization based control algorithms, like model predictive control. Our proposed machine-learning supported approach addresses this by utilizing a feed-forward neural network to reduce the computation load of the online-optimization. We propose approximating part of the problem horizon, while maintaining safety guarantees -- constraint satisfaction -- via the remaining optimization part of the controller. The approach is validated in simulation, demonstrating an improvement in computational efficiency, while maintaining guarantees and near-optimal performance. The proposed MPC scheme can be applied to a wide range of applications, including those requiring a rapid control response, such as robotics and embedded applications with limited computational resources., Comment: 6 pages, 4 figures, 4 tables, American Control Conference (ACC) 2024

The digital landscape provides a dynamic platform for political discourse crucial for understanding shifts in public opinion and engagement especially under authoritarian governments This study examines YouTube user behavior during the Russian-Ukrainian war analyzing 2168 videos with over 36000 comments from January 2022 to February 2024 We observe distinct patterns of participation and gender dynamics that correlate with major political and military events Notably females were more active in antigovernment channels especially during peak conflict periods Contrary to assumptions about online engagement in authoritarian contexts our findings suggest a complex interplay where women emerge as pivotal digital communicators This highlights online platforms role in facilitating political expression under authoritarian regimes demonstrating its potential as a barometer for public sentiment.

Generating synthetic text addresses the challenge of data availability in privacy-sensitive domains such as healthcare. This study explores the applicability of synthetic data in real-world medical settings. We introduce MedSyn, a novel medical text generation framework that integrates large language models with a Medical Knowledge Graph (MKG). We use MKG to sample prior medical information for the prompt and generate synthetic clinical notes with GPT-4 and fine-tuned LLaMA models. We assess the benefit of synthetic data through application in the ICD code prediction task. Our research indicates that synthetic data can increase the classification accuracy of vital and challenging codes by up to 17.8% compared to settings without synthetic data. Furthermore, to provide new data for further research in the healthcare domain, we present the largest open-source synthetic dataset of clinical notes for the Russian language, comprising over 41k samples covering 219 ICD-10 codes., Comment: 16 pages, accepted to ECML PKDD 2024

In this paper, we describe the results of the HSEmotion team in two tasks of the seventh Affective Behavior Analysis in-the-wild (ABAW) competition, namely, multi-task learning for simultaneous prediction of facial expression, valence, arousal, and detection of action units, and compound expression recognition. We propose an efficient pipeline based on frame-level facial feature extractors pre-trained in multi-task settings to estimate valence-arousal and basic facial expressions given a facial photo. We ensure the privacy-awareness of our techniques by using the lightweight architectures of neural networks, such as MT-EmotiDDAMFN, MT-EmotiEffNet, and MT-EmotiMobileFaceNet, that can run even on a mobile device without the need to send facial video to a remote server. It was demonstrated that a significant step in improving the overall accuracy is the smoothing of neural network output scores using Gaussian or box filters. It was experimentally demonstrated that such a simple post-processing of predictions from simple blending of two top visual models improves the F1-score of facial expression recognition up to 7%. At the same time, the mean Concordance Correlation Coefficient (CCC) of valence and arousal is increased by up to 1.25 times compared to each model's frame-level predictions. As a result, our final performance score on the validation set from the multi-task learning challenge is 4.5 times higher than the baseline (1.494 vs 0.32)., Comment: 16 pages, 7 figures, 9 tables

Using the free fermions technique and bosonization rules we introduce the multicomponent DKP hierarchy as a generating bilinear integral equation for the tau-function. A number of bilinear equations of the Hirota-Miwa type are obtained as its corollaries. We also consider the dispersionless version of the hierarchy as a set of nonlinear differential equations for the dispersionless limit of logarithm of the tau-function (the $F$-function). We show that there is an elliptic curve built in the structure of the hierarchy, with the elliptic modulus being a dynamical variable. This curve can be uniformized by elliptic functions, and in the elliptic parametrization many dispersionless equations of the Hirota-Miwa type become equivalent to a single equation having a nice form., Comment: 22 pages, no figures

We present multiwavelength polarization measurements of the luminous blazar Mrk~501 over a 14-month period. The 2--8 keV X-ray polarization was measured with the Imaging X-ray Polarimetry Explorer (IXPE) with six 100-ks observations spanning from 2022 March to 2023 April. Each IXPE observation was accompanied by simultaneous X-ray data from NuSTAR, Swift/XRT, and/or XMM-Newton. Complementary optical-infrared polarization measurements were also available in the B, V, R, I, and J bands, as were radio polarization measurements from 4.85 GHz to 225.5 GHz. Among the first five IXPE observations, we did not find significant variability in the X-ray polarization degree and angle with IXPE. However, the most recent sixth observation found an elevated polarization degree at $>3\sigma$ above the average of the other five observations. The optical and radio measurements show no apparent correlations with the X-ray polarization properties. Throughout the six IXPE observations, the X-ray polarization degree remained higher than, or similar to, the R-band optical polarization degree, which remained higher than the radio value. This is consistent with the energy-stratified shock scenario proposed to explain the first two IXPE observations, in which the polarized X-ray, optical, and radio emission arises from different regions., Comment: Accepted for publication in ApJ

Event sequences (ESs) arise in many practical domains including finance, retail, social networks, and healthcare. In the context of machine learning, event sequences can be seen as a special type of tabular data with annotated timestamps. Despite the importance of ESs modeling and analysis, little effort was made in adapting large language models (LLMs) to the ESs domain. In this paper, we highlight the common difficulties of ESs processing and propose a novel solution capable of solving multiple downstream tasks with little or no finetuning. In particular, we solve the problem of working with long sequences and improve time and numeric features processing. The resulting method, called ESQA, effectively utilizes the power of LLMs and, according to extensive experiments, achieves state-of-the-art results in the ESs domain., Comment: 25 pages, 3 figures

Context: There are currently three pulsars firmly detected by imaging atmospheric Cherenkov telescopes (IACTs), two of them reaching TeV energies, challenging models of very-high-energy (VHE) emission in pulsars. More precise observations are needed to better characterize pulsar emission at these energies. The LST-1 is the prototype of the Large-Sized Telescope, that will be part of the Cherenkov Telescope Array Observatory (CTAO). Its improved performance over previous IACTs makes it well suited for studying pulsars. Aims: To study the Crab pulsar emission with the LST-1, improving and complementing the results from other telescopes. These observations can also be used to characterize the potential of the LST-1 to study other pulsars and detect new ones. Methods: We analyzed a total of $\sim$103 hours of gamma-ray observations of the Crab pulsar conducted with the LST-1 in the period from September 2020 to January 2023. The observations were carried out at zenith angles less than 50 degrees. A new analysis of the Fermi-LAT data was also performed, including $\sim$14 years of observations. Results: The Crab pulsar phaseogram, long-term light-curve, and phase-resolved spectra are reconstructed with the LST-1 from 20 GeV to 450 GeV for P1 and up to 700 GeV for P2. The pulsed emission is detected with a significance of 15.2$\sigma$. The two characteristic emission peaks of the Crab pulsar are clearly detected (>10$\sigma$), as well as the so-called bridge emission (5.7$\sigma$). We find that both peaks are well described by power laws, with spectral indices of $\sim$3.44 and $\sim$3.03 respectively. The joint analysis of Fermi-LAT and LST-1 data shows a good agreement between both instruments in the overlapping energy range. The detailed results obtained in the first observations of the Crab pulsar with LST-1 show the potential that CTAO will have to study this type of sources., Comment: Accepted by A&A

Variability can be the pathway to understanding the physical processes in astrophysical jets, however, the high-cadence observations required to test particle acceleration models are still missing. Here we report on the first attempt to produce continuous, >24 hour polarization light curves of blazars using telescopes distributed across the globe and the rotation of the Earth to avoid the rising Sun. Our campaign involved 16 telescopes in Asia, Europe, and North America. We observed BL Lacertae and CGRaBS J0211+1051 for a combined 685 telescope hours. We find large variations in the polarization degree and angle for both sources in sub-hour timescales as well as a ~180 degree rotation of the polarization angle in CGRaBS J0211+1051 in less than two days. We compared our high-cadence observations to Particle-In-Cell magnetic reconnection and turbulent plasma simulations. We find that although the state of the art simulation frameworks can produce a large fraction of the polarization properties, they do not account for the entirety of the observed polarization behavior in blazar jets., Comment: 20 pages, 15 figures, 2 tables, accepted for publication in A&A. The data used in the paper are available here: https://doi.org/10.7910/DVN/IETSXS

In sequential event prediction, which finds applications in finance, retail, social networks, and healthcare, a crucial task is forecasting multiple future events within a specified time horizon. Traditionally, this has been addressed through autoregressive generation using next-event prediction models, such as Marked Temporal Point Processes. However, autoregressive methods use their own output for future predictions, potentially reducing quality as the prediction horizon extends. In this paper, we challenge traditional approaches by introducing a novel benchmark, HoTPP, specifically designed to evaluate a model's ability to predict event sequences over a horizon. This benchmark features a new metric inspired by object detection in computer vision, addressing the limitations of existing metrics in assessing models with imprecise time-step predictions. Our evaluations on established datasets employing various models demonstrate that high accuracy in next-event prediction does not necessarily translate to superior horizon prediction, and vice versa. HoTPP aims to serve as a valuable tool for developing more robust event sequence prediction methods, ultimately paving the way for further advancements in the field.

Behavior study experiments are an important part of society modeling and understanding human interactions. In practice, many behavioral experiments encounter challenges related to internal and external validity, reproducibility, and social bias due to the complexity of social interactions and cooperation in human user studies. Recent advances in Large Language Models (LLMs) have provided researchers with a new promising tool for the simulation of human behavior. However, existing LLM-based simulations operate under the unproven hypothesis that LLM agents behave similarly to humans as well as ignore a crucial factor in human decision-making: emotions. In this paper, we introduce a novel methodology and the framework to study both, the decision-making of LLMs and their alignment with human behavior under emotional states. Experiments with GPT-3.5 and GPT-4 on four games from two different classes of behavioral game theory showed that emotions profoundly impact the performance of LLMs, leading to the development of more optimal strategies. While there is a strong alignment between the behavioral responses of GPT-3.5 and human participants, particularly evident in bargaining games, GPT-4 exhibits consistent behavior, ignoring induced emotions for rationality decisions. Surprisingly, emotional prompting, particularly with `anger' emotion, can disrupt the "superhuman" alignment of GPT-4, resembling human emotional responses.

The nearby elliptical galaxy M87 contains one of the only two supermassive black holes whose emission surrounding the event horizon has been imaged by the Event Horizon Telescope (EHT). In 2018, more than two dozen multi-wavelength (MWL) facilities (from radio to gamma-ray energies) took part in the second M87 EHT campaign. The goal of this extensive MWL campaign was to better understand the physics of the accreting black hole M87*, the relationship between the inflow and inner jets, and the high-energy particle acceleration. Understanding the complex astrophysics is also a necessary first step towards performing further tests of general relativity. The MWL campaign took place in April 2018, overlapping with the EHT M87* observations. We present a new, contemporaneous spectral energy distribution (SED) ranging from radio to very high energy (VHE) gamma-rays, as well as details of the individual observations and light curves. We also conduct phenomenological modelling to investigate the basic source properties. We present the first VHE gamma-ray flare from M87 detected since 2010. The flux above 350 GeV has more than doubled within a period of about 36 hours. We find that the X-ray flux is enhanced by about a factor of two compared to 2017, while the radio and millimetre core fluxes are consistent between 2017 and 2018. We detect evidence for a monotonically increasing jet position angle that corresponds to variations in the bright spot of the EHT image. Our results show the value of continued MWL monitoring together with precision imaging for addressing the origins of high-energy particle acceleration. While we cannot currently pinpoint the precise location where such acceleration takes place, the new VHE gamma-ray flare already presents a challenge to simple one-zone leptonic emission model approaches, and emphasises the need for combined image and spectral modelling., Comment: 46 pages, 23 figures, accepted by Astronomy & Astrophysics on August. 29, 2024

We fit various colour-magnitude diagrams (CMDs) of the high-latitude Galactic globular clusters NGC\,5024 (M53), NGC\,5053, NGC\,5272 (M3), NGC\,5466, and NGC\,7099 (M30) by isochrones from the Dartmouth Stellar Evolution Database and Bag of Stellar Tracks and Isochrones for $\alpha$-enrichment [$\alpha$/Fe]$=+0.4$. For the CMDs, we use data sets from {\it Hubble Space Telescope}, {\it Gaia}, and other sources utilizing, at least, 25 photometric filters for each cluster. We obtain the following characteristics with their statistic uncertainties for NGC\,5024, NGC\,5053, NGC\,5272, NGC\,5466, and NGC\,7099, respectively: metallicities [Fe/H]$=-1.93\pm0.02$, $-2.08\pm0.03$, $-1.60\pm0.02$, $-1.95\pm0.02$, and $-2.07\pm0.04$ dex with their systematic uncertainty 0.1 dex; ages $13.00\pm0.11$, $12.70\pm0.11$, $11.63\pm0.07$, $12.15\pm0.11$, and $12.80\pm0.17$ Gyr with their systematic uncertainty 0.8 Gyr; distances (systematic uncertainty added) $18.22\pm0.06\pm0.60$, $16.99\pm0.06\pm0.56$, $10.08\pm0.04\pm0.33$, $15.59\pm0.03\pm0.51$, and $8.29\pm0.03\pm0.27$ kpc; reddenings $E(B-V)=0.023\pm0.004$, $0.017\pm0.004$, $0.023\pm0.004$, $0.023\pm0.003$, and $0.045\pm0.002$ mag with their systematic uncertainty 0.01 mag; extinctions $A_\mathrm{V}=0.08\pm0.01$, $0.06\pm0.01$, $0.08\pm0.01$, $0.08\pm0.01$, and $0.16\pm0.01$ mag with their systematic uncertainty 0.03 mag, which suggest the total Galactic extinction $A_\mathrm{V}=0.08$ across the whole Galactic dust to extragalactic objects at the North Galactic pole. The horizontal branch morphology difference of these clusters is explained by their different metallicity, age, mass-loss efficiency, and loss of low-mass members in the evolution of the core-collapse cluster NGC\,7099 and loose clusters NGC\,5053 and NGC\,5466., Comment: 9 figures, published in Research in Astronomy and Astrophysics

Effective processing of financial transactions is essential for banking data analysis. However, in this domain, most methods focus on specialized solutions to stand-alone problems instead of constructing universal representations suitable for many problems. We present a representation learning framework that addresses diverse business challenges. We also suggest novel generative models that account for data specifics, and a way to integrate external information into a client's representation, leveraging insights from other customers' actions. Finally, we offer a benchmark, describing representation quality globally, concerning the entire transaction history; locally, reflecting the client's current state; and dynamically, capturing representation evolution over time. Our generative approach demonstrates superior performance in local tasks, with an increase in ROC-AUC of up to 14\% for the next MCC prediction task and up to 46\% for downstream tasks from existing contrastive baselines. Incorporating external information improves the scores by an additional 20\%.

Let $T$ be a tournament of odd order $n\ge 5,$ $c_{m}(T)$ be the number of its $m$-cycles, and $s_{m}(T)$ be the number of its strongly connected $m$-subtournaments. Due to work of L.W. Beineke and F. Harary, it is well known that $s_{m}(T)\le s_{m}(RLT_{n}),$ where $RLT_{n}$ is the regular locally transitive tournament of order $n.$ For $m=3$ and $m=4,$ $c_{m}(T)$ equals $s_{m}(T),$ but it is not so for $m\ge 5.$ As J.W. Moon pointed out in his note in 1966, the problem of determining the maximum of $c_{m}(T)$ seems very difficult in general (i.e. for $m\ge 5$). In the present paper, based on the Komarov-Mackey formula for $c_{5}(T)$ obtained recently, we prove that $c_{5}(T)\le (n+1)n(n-1)(n-2)(n-3)/160$ with equality holding iff $T$ is doubly regular. A formula for $s_{5}(T)$ is also deduced. With the use of it, we show that $s_{5}(T)\le (n+1)n(n-1)(n-3)(11n-47)/1920$ with equality holding iff $T=RLT_{n}$ or $n=7$ and $T$ is regular or $n=5$ and $T$ is strong. It is also proved that for a regular tournament $T$ of (odd) order $n\ge 9,$ a lower bound $(n+1)n(n-1)(n-3)(17n-59)/3840\le s_{5}(T)$ holds with equality iff $T$ is doubly regular. These results are compared with the ones recently obtained by the author for $c_{5}(T).$

The BL Lac object 3C 371 has been observed by the Transiting Exoplanet Survey Satellite (\textit{TESS}) for approximately a year, between July 2019 and July 2020, with an unmatched 2-minute observing cadence. In parallel, the Whole Earth Blazar Telescope (WEBT) Collaboration organized an extensive observing campaign, providing three years of continuous optical monitoring between 2018 and 2020. These datasets allow for a thorough investigation of the variability of the source. The goal of this study is evaluating the optical variability of 3C 371. Taking advantage of the remarkable cadence of \textit{TESS} data, we aim to characterize the intra-day variability (IDV) displayed by the source and identify its shortest variability timescale. With this estimate, constraints on the size of the emitting region and black hole mass can be calculated. Moreover, WEBT data is used to investigate long-term variability (LTV), including understanding spectral behaviour of the source and the polarization variability. Based on the derived characteristics, information on the origin of the variability on different timescales is extracted. We evaluated the variability applying the variability amplitude tool that quantifies how variable the emission is. Moreover, we employed common tools like ANOVA (ANalysis Of VAariance) tests, wavelet and power spectral density (PSD) analyses to characterize the shortest variability timescales present in the emission and the underlying noise affecting the data. Short- and long-term colour behaviours have been evaluated to understand the spectral behaviour of the source. The polarized emission was analyzed, studying its variability and possible rotation patterns of the electric vector position angle (EVPA). Flux distributions of IDV and LTV were also studied with the aim of linking the flux variations to turbulent and/or accretion disc related processes., Comment: Accepted in A&A, 20 pages, 19 figures

Magnetic skyrmions are topologically nontrivial spin configurations that possess particle-like properties. Earlier research was mainly focused on a specific type of skyrmion with topological charge Q = -1. However, theoretical analyses of two-dimensional chiral magnets have predicted the existence of skyrmion bags -- solitons with arbitrary positive or negative topological charge. Although such spin textures are metastable states, recent experimental observations have confirmed the stability of isolated skyrmion bags in a limited range of applied magnetic fields. Here, by utilizing Lorentz transmission electron microscopy, we show the extraordinary stability of skyrmion bags in thin plates of B20-type FeGe. In particular, we show that skyrmion bags embedded within a skyrmion lattice remain stable even in zero or inverted external magnetic fields. A robust protocol for nucleating such embedded skyrmion bags is provided. Our results agree perfectly with micromagnetic simulations and establish thin plates of cubic chiral magnets as a powerful platform for exploring a broad spectrum of topological magnetic solitons., Comment: 17 pages, 15 figures

Forthcoming large-scale structure (LSS) Stage IV surveys will provide us with unprecedented data to probe the nature of dark matter and dark energy. However, analysing these data with conventional Markov Chain Monte Carlo (MCMC) methods will be challenging, due to the increase in the number of nuisance parameters and the presence of intractable likelihoods. In light of this, we present the first application of Marginal Neural Ratio Estimation (MNRE) (a recent approach in simulation-based inference) to LSS photometric probes: weak lensing, galaxy clustering and the cross-correlation power spectra. In order to analyse the hundreds of spectra simultaneously, we find that a pre-compression of data using principal component analysis, as well as parameter-specific data summaries lead to highly accurate results. Using expected Stage IV experimental noise, we are able to recover the posterior distribution for the cosmological parameters with a speedup factor of $\sim 40-60$ compared to classical MCMC methods. To illustrate that the performance of MNRE is not impeded when posteriors are highly non-Gaussian, we test a scenario of two-body decaying dark matter, finding that Stage IV surveys can improve current bounds on the model by up to one order of magnitude. This result supports that MNRE is a powerful framework to constrain the standard cosmological model and its extensions with next-generation LSS surveys., Comment: 25 pages, 9 figures, 2 tables. Comments welcome!

This article presents our results for the sixth Affective Behavior Analysis in-the-wild (ABAW) competition. To improve the trustworthiness of facial analysis, we study the possibility of using pre-trained deep models that extract reliable emotional features without the need to fine-tune the neural networks for a downstream task. In particular, we introduce several lightweight models based on MobileViT, MobileFaceNet, EfficientNet, and DDAMFN architectures trained in multi-task scenarios to recognize facial expressions, valence, and arousal on static photos. These neural networks extract frame-level features fed into a simple classifier, e.g., linear feed-forward neural network, to predict emotion intensity, compound expressions, action units, facial expressions, and valence/arousal. Experimental results for five tasks from the sixth ABAW challenge demonstrate that our approach lets us significantly improve quality metrics on validation sets compared to existing non-ensemble techniques., Comment: 10 pages, 1 figure, 8 tables

For a regular tournament $T$ of order $n,$ denote by $c_{8}(T)$ the number of cycles of length $8$ in $T.$ Let $DR_{n}$ be a doubly-regular tournament of order $n\equiv 3\mod4$ (so, the out-sets and in-sets of its vertices are also regular and hence, contain the maximum possible number of cyclic triples) and $RLT_{n}$ be the unique regular locally transitive tournament of (odd) order $n$ (so, the out-sets and in-sets of its vertices are transitive and hence, contain no cyclic triples, at all). Some arguments based on the spectral properties of tournaments allow us to suggest that $c_{8}(T) \le c_{8}(RLT_{n}),$ where $n$ is sufficiently large. This restriction on $n$ is essential because our computer processing of B. McKay's file of tournaments implies that for $n=9,11,13,$ the maximum of $c_{8}(T)$ is attained at tournaments with regular structure of the out and in-sets of their vertices. In the present paper, we show that $c_{8}(DR_{n})$ does not depend on a particular choice of $DR_{n}$ and determine expressions for $c_{8}(DR_{n})$ and $c_{8}(RLT_{n}).$ They are both polynomials of degree $8$ in $n.$ Comparing $c_{8}(DR_{n})$ with $c_{8}(RLT_{n})$ yields the inequality $c_{8}(DR_{n})>c_{8}(RLT_{n})$ for $11\le n\le 35,$ while $c_{8}(RLT_{n}) > c_{8}(DR_{n})$ for $n\ge 39.$ This allows us to treat the value $n=39$ as the point of phase transition in the local properties of maximizers and minimizers of $c_{8}(T)$ in the class of regular tournaments of order $n.$

PG 1553+113 is one of the few blazars with a convincing quasi-periodic emission in the gamma-ray band. The source is also a very high-energy (VHE; >100 GeV) gamma-ray emitter. To better understand its properties and identify the underlying physical processes driving its variability, the MAGIC Collaboration initiated a multiyear, multiwavelength monitoring campaign in 2015 involving the OVRO 40-m and Medicina radio telescopes, REM, KVA, and the MAGIC telescopes, Swift and Fermi satellites, and the WEBT network. The analysis presented in this paper uses data until 2017 and focuses on the characterization of the variability. The gamma-ray data show a (hint of a) periodic signal compatible with literature, but the X-ray and VHE gamma-ray data do not show statistical evidence for a periodic signal. In other bands, the data are compatible with the gamma-ray period, but with a relatively high p-value. The complex connection between the low and high-energy emission and the non-monochromatic modulation and changes in flux suggests that a simple one-zone model is unable to explain all the variability. Instead, a model including a periodic component along with multiple emission zones is required., Comment: Accepted for publication in Monthly Notices of the Royal Astronomical Society. 19 pages, 9 figures. Corresponding authors: Elisa Prandini, Antonio Stamerra, Talvikki Hovatta

Building an intelligent and efficient medical assistant is still a challenging AI problem. The major limitation comes from the data modality scarceness, which reduces comprehensive patient perception. This demo paper presents the GigaPevt, the first multimodal medical assistant that combines the dialog capabilities of large language models with specialized medical models. Such an approach shows immediate advantages in dialog quality and metric performance, with a 1.18% accuracy improvement in the question-answering task., Comment: IJCAI 2024, 4 pages, 2 figures, 2 tables

High-energy neutrinos could be produced in the interaction of charged cosmic rays with matter or radiation surrounding astrophysical sources. To look for transient sources associated with neutrino emission, a follow-up program of neutrino alerts has been operating within the ANTARES Collaboration since 2009. This program, named TAToO, has triggered robotic optical telescopes (MASTER, TAROT, ROTSE and the SVOM ground based telescopes) immediately after the detection of any relevant neutrino candidate and scheduled several observations in the weeks following the detection. A subset of ANTARES events with highest probabilities of being of cosmic origin has also been followed by the Swift and the INTEGRAL satellites, the Murchison Widefield Array radio telescope and the H.E.S.S. high-energy gamma-ray telescope. The results of twelve years of observations are reported. No optical counterpart has been significantly associated with an ANTARES candidate neutrino signal during image analysis. Constraints on transient neutrino emission have been set. In September 2015, ANTARES issued a neutrino alert and during the follow-up, a potential transient counterpart was identified by Swift and MASTER. A multi-wavelength follow-up campaign has allowed to identify the nature of this source and has proven its fortuitous association with the neutrino. The return of experience is particularly important for the design of the alert system of KM3NeT, the next generation neutrino telescope in the Mediterranean Sea., Comment: 27 pages, 14 figures, submitted to JCAP

In the rapidly evolving domain of Recommender Systems (RecSys), new algorithms frequently claim state-of-the-art performance based on evaluations over a limited set of arbitrarily selected datasets. However, this approach may fail to holistically reflect their effectiveness due to the significant impact of dataset characteristics on algorithm performance. Addressing this deficiency, this paper introduces a novel benchmarking methodology to facilitate a fair and robust comparison of RecSys algorithms, thereby advancing evaluation practices. By utilizing a diverse set of $30$ open datasets, including two introduced in this work, and evaluating $11$ collaborative filtering algorithms across $9$ metrics, we critically examine the influence of dataset characteristics on algorithm performance. We further investigate the feasibility of aggregating outcomes from multiple datasets into a unified ranking. Through rigorous experimental analysis, we validate the reliability of our methodology under the variability of datasets, offering a benchmarking strategy that balances quality and computational demands. This methodology enables a fair yet effective means of evaluating RecSys algorithms, providing valuable guidance for future research endeavors., Comment: 8 pages with 11 figures

Spiral structure can contribute significantly to a galaxy's luminosity. However, only rarely are proper photometric models of spiral arms used in decompositions. As we show in the previous work, including the spirals as a separate component in a photometric model of a galaxy would both allow to obtain their structural parameters, and reduce the systematic errors in estimating the parameters of other components. Doing so in different wavebands, one can explore how their properties vary with the wavelength. In this paper, second in this series, we perform decomposition of M 51 in 17 bands, from the far UV to far IR, using imaging from the DustPedia project. We use the same 2D photometric model of spiral structure where each arm is modelled independently. The complex and asymmetric spiral structure in M 51 is reproduced relatively well with our model. We analyze the differences between models with and without spiral arms, and investigate how the fit parameters change with wavelength. In particular, we find that the spiral arms demonstrate the largest width in the optical, whereas their contribution to the galaxy luminosity is most significant in the UV. The disk central intensity drops by a factor of 1.25--3 and its exponential scale changes by 5--10\% when spiral arms are included, depending on wavelength. Taking into account the full light distribution across the arms, we do not observe the signs of a long-lived density wave in the spiral pattern of M 51 as a whole., Comment: Accepted for publication in MNRAS

We fit various colour-magnitude diagrams (CMDs) of the Galactic globular clusters NGC\,6397 and NGC\,6809 (M55) by isochrones from the Dartmouth Stellar Evolution Database (DSED) and Bag of Stellar Tracks and Isochrones (BaSTI) for $\alpha$-enhanced [$\alpha$/Fe]$=+0.4$. For the CMDs, we use data sets from {\it HST}, {\it Gaia}, VISTA, and other sources utilizing 32 and 23 photometric filters for NGC\,6397 and NGC\,6809, respectively, from the ultraviolet to mid-infrared. We obtain the following characteristics for NGC\,6397 and NGC\,6809, respectively: metallicities [Fe/H]$=-1.84\pm0.02\pm0.1$ and $-1.78\pm0.02\pm0.1$ (statistic and systematic uncertainties); distances $2.45\pm0.02\pm0.06$ and $5.24\pm0.02\pm0.18$ kpc; ages $12.9\pm0.1\pm0.8$ and $13.0\pm0.1\pm0.8$ Gyr; reddenings $E(B-V)=0.178\pm0.006\pm0.01$ and $0.118\pm0.004\pm0.01$ mag; extinctions $A_\mathrm{V}=0.59\pm0.01\pm0.02$ and $0.37\pm0.01\pm0.04$ mag; extinction-to-reddening ratio $R_\mathrm{V}=3.32^{+0.32}_{-0.28}$ and $3.16^{+0.66}_{-0.56}$. Our estimates agree with most estimates from the literature. BaSTI gives systematically higher [Fe/H] and lower reddenings than DSED. Despite nearly the same metallicity, age, and helium enrichment, these clusters show a considerable horizontal branch (HB) morphology difference, which must therefore be described by another parameter. This parameter must predominantly explain why the least massive HB stars (0.58-0.63 solar masses) are only found within NGC 6809. Probably they have been lost by the core-collapse cluster NGC\,6397 during its dynamical evolution and mass segregation. In contrast, NGC\,6809 has a very low central concentration and, hence, did not undergo this process., Comment: 19 figures. arXiv admin note: text overlap with arXiv:2211.12684

We present new three-dimensional (3D) interstellar extinction maps in the $V$ and Gaia $G$ filters within 2 kpc of the Sun, a 3D differential extinction (dust spatial distribution density) map along the lines of sight in the same space, a 3D map of variations in the ratio of the extinctions in the $V$ and Gaia $G$ filters within 800 pc of the Sun, and a 2D map of total Galactic extinction through the entire dust half-layer from the Sun to extragalactic space for Galactic latitudes $|b|>13^{\circ}$. The 3D maps have a transverse resolution from 3.6 to 11.6 pc and a radial resolution of 50 pc. The 2D map has an angular resolution of 6.1 arcmin. We have produced these maps based on the Gaia DR3 parallaxes and Gaia, Pan-STARRS1, SkyMapper, 2MASS, and WISE photometry for nearly 100 million stars. We have paid special attention to the space within 200 pc of the Sun and high Galactic latitudes as regions where the extinction estimates have had a large relative uncertainty so far. Our maps estimate the extinction within the Galactic dust layer from the Sun to an extended object or through the entire dust half-layer from the Sun to extragalactic space with a precision $\sigma(A_\mathrm{V})=0.06$ mag. This gives a high relative precision of extinction estimates even at high Galactic latitudes, where, according to our estimates, the median total Galactic extinction through the entire dust half-layer from the Sun to extragalactic objects is $A_\mathrm{V}=0.12\pm0.06$ mag. We have shown that the presented maps are among the best ones in data amount, space size, resolution, precision, and other properties., Comment: 15 pages, 4 tables

We report on the optical realization of the magneto-intersubband oscillations that have been measured in the sub-terahertz transmittance of a GaAs quantum well with two subbands occupied. Following their dc analogue, the oscillations are periodic in the inverse magnetic field with the period governed by the subband gap. Their magnitude and polarization dependence accurately follow the presented simplified version of the dynamic magneto-intersubband oscillations equation that naturally combines dc magneto-intersabband oscillations with microwave-induced resistance oscillations (MIRO). Simultaneously measured photoresistance also reveals its strong sensitivity to the sign of the circular polarization, proving the used theoretical modeling., Comment: 5 pages, 2 figures

We report on dynamic Shubnikov - de Haas (SdH) oscillations that are measured in the optical response, sub - terahertz transmittance of two-dimensional systems, and reveal two distinct types of oscillation nodes: "universal" nodes at integer ratios of radiation and cyclotron frequencies and "tunable" nodes at positions sensitive to all parameters of the structure. The nodes in both real and imaginary parts of the measured complex transmittance are analyzed using a dynamic version of the static Lifshitz-Kosevich formula. These results demonstrate that the node structure of the dynamic SdH oscillations provides an all-optical access to quantization- and interaction-induced renormalization effects, in addition to parameters one can obtain from the static SdH oscillations., Comment: Main text plus supplemental information

We present a sample of 950 edge-on spiral galaxies found with the use of an artificial neural network in the Hubble Space Telescope COSMOS field. This is currently the largest sample of distant edge-on galaxies. For all galaxies we analyzed the 2D brightness distributions in the F814W filter and measured the radial and vertical exponential scales ($h$ and $h_z$ correspondingly) of the brightness distribution. By comparing the characteristics of distant galaxies with those of nearby objects, we conclude that thin stellar discs with $h/h_z \geq 10$ at $z \approx 0.5$ should be rarer than today. Both exponential scales of the stellar disc show evidence of luminosity-dependent evolution: in faint galaxies the $h$ and $h_z$ values do not change with $z$, in bright (and massive) spiral galaxies both scales, on average, grow towards our epoch., Comment: 5 pages, accepted for publication in MNRAS Letters

Observations of linear polarization in the 2-8 keV energy range with the Imaging X-ray Polarimetry Explorer (IXPE) explore the magnetic field geometry and dynamics of the regions generating non-thermal radiation in relativistic jets of blazars. These jets, particularly in blazars whose spectral energy distribution peaks at X-ray energies, emit X-rays via synchrotron radiation from high-energy particles within the jet. IXPE observations of the X-ray selected BL Lac-type blazar 1ES 1959+650 in 2022 May 3-4 showed a significant linear polarization degree of $\Pi_\mathrm{x} = 8.0\% \pm 2.3\%$ at an electric-vector position angle $\psi_\mathrm{x} = 123^\circ \pm 8^\circ$. However, in 2022 June 9-12, only an upper limit of $\Pi_\mathrm{x} \leq 5.1\%$ could be derived (at the 99% confidence level). The degree of optical polarization at that time $\Pi_\mathrm{O} \sim 5\%$ is comparable to the X-ray measurement. We investigate possible scenarios for these findings, including temporal and geometrical depolarization effects. Unlike some other X-ray selected BL Lac objects, there is no significant chromatic dependence of the measured polarization in 1ES 1959+650, and its low X-ray polarization may be attributed to turbulence in the jet flow with dynamical timescales shorter than 1 day., Comment: 16 pages, 4 figures, accepted for publication in The Astrophysical Journal

Giant flares, short explosive events releasing up to 10$^{47}$ erg of energy in the gamma-ray band in less than one second, are the most spectacular manifestation of magnetars, young neutron stars powered by a very strong magnetic field, 10$^{14-15}$ G in the magnetosphere and possibly higher in the star interior. The rate of occurrence of these rare flares is poorly constrained, as only three have been seen from three different magnetars in the Milky Way and in the Large Magellanic Cloud in about 50 years since the beginning of gamma-ray astronomy. This sample can be enlarged by the discovery of extragalactic events, since for a fraction of a second giant flares reach peak luminosities above 10$^{46}$ erg/s, which makes them visible by current instruments up to a few tens of Mpc. However, at these distances they appear similar to, and difficult to distinguish from, regular short gamma-ray bursts (GRBs). The latter are much more energetic events, 10$^{50-53}$ erg, produced by compact binary mergers and originating at much larger distances. Indeed, only a few short GRBs have been proposed, with different levels of confidence, as magnetar giant flare candidates in nearby galaxies. Here we report the discovery of a short GRB positionally coincident with the central region of the starburst galaxy M82. Its spectral and timing properties, together with the limits on its X-ray and optical counterparts obtained a few hours after the event and the lack of an associated gravitational wave signal, qualify with high confidence this event as a giant flare from a magnetar in M82., Comment: Submitted version. New figures. Accepted for publication in Nature with minor modifications

In this paper, we look to analyse the spiral features of grand-design, multiarmed, and flocculent spiral galaxies using deep optical imaging from DESI Legacy Imaging Surveys. We explore the resulting distributions of various characteristics of spiral structure beyond the optical radius, such as the distributions of azimuthal angle, the extent of spiral arms, and of the spiral arm widths for the aforementioned galaxy classes. We also compare the measured properties for isolated galaxies and galaxies in groups and clusters. We find that, on average, compared to multiarmed and flocculent spiral galaxies, the spiral arms of grand-design galaxies exhibit slightly larger azimuthal angles, greater extent, and larger widths in the periphery of the galaxy. Furthermore, on average, isolated galaxies tend to have slightly smaller widths of outer spiral arms compared to galaxies in tight environments, which is likely related to the tidally-induced mechanism for generating wider outer spiral arms. We also report that breaks of the disc surface brightness profiles are often related to the truncation of spiral arms in galaxies., Comment: 16 pages, 12 figures, accepted for publication in MNRAS

We present an application of an artificial neural network methodology to a modern wide-field sky survey Pan-STARRS1 in order to build a high-quality sample of disk galaxies visible in edge-on orientation. Such galaxies play an important role in the study of the vertical distribution of stars, gas and dust, which is usually not available to study in other galaxies outside the Milky Way. We give a detailed description of the network architecture and the learning process. The method demonstrates good effectiveness with detection rate about 97\% and it works equally well for galaxies over a wide range of brightnesses and sizes, which resulted in a creation of a catalogue of edge-on galaxies with $10^5$ of objects. The catalogue is published on-line with an open access., Comment: 15 pages, 11 figures

Interactions of cosmic ray protons and nuclei in their sources and in the interstellar medium produce "hadronic" gamma-ray emission. Gamma-rays can also be of "leptonic" origin, i.e. originating from high-energy electrons accelerated together with protons. It is difficult to distinguish between hadronic and leptonic emission mechanisms based on gamma-ray data alone. This can be done via detection of neutrinos, because only hadronic processes lead to neutrino production. We use publicly available ten-year IceCube neutrino telescope dataset to demonstrate the hadronic nature of high-energy emission from the direction of Cygnus region of the Milky Way. We find a 3-sigma excess of neutrino events from an extended Cygnus Cocoon, with the flux comparable to the flux of gamma-rays in the multi-TeV energy range seen by HAWC and LHAASO telescopes., Comment: 4 pages 3 figures, version accepted to Phys. Rev. D

Let $T$ be a strongly connected tournament of order $n\ge 4$ whose diameter does not exceed $d\ge 3.$ Denote by $c_{\ell}(T)$ the number of circuits of length $\ell$ in $T.$ In our recent paper, we construct a strongly connected tournament $T_{d,n}$ of order $n$ with diameter $d$ and conjecture that $c_{\ell}(T)\ge c_{\ell}(T_{d,n})$ for any $\ell=3,...,n.$ In particular, for $d=n-1,$ this inequality is true and yields the known Moon (lower) bound $c_{\ell}(T)\ge n-\ell+1.$ Moreover, we suggest that if $n+3\le 2d,$ then for any given $\ell$ taken in the range $n-d+3,...,d,$ the equality $c_{\ell}(T)=c_{\ell}(T_{d,n})$ implies that $T$ is isomorphic to $T_{d,n}$ or its converse $T_{d,n}^{-}.$ For $d=n-1,$ the corresponding particular statement is nothing else than Las Vergnas' theorem. Recently, we have confirmed the posed conjecture for the case $d=n-2.$ In the present paper, we show that it is also true for $d=n-3.$

In the present paper, for a given (possibly, infinite) strongly connected digraph $\cal{D},$ we consider the class $\cal{S}_{<}({\cal D})$ of all truthly substochastic weightings of ${\cal D}$ (here, the word "truthly" means that there exists a vertex whose out-weight is strictly less than $1$). For a finite subdigraph $\cal{F}$ of $\cal{D}$ weighted by $S\in {\cal S}_{<}({\cal D}),$ let $\ell_{max}(\cal{F})$ be the length of its longest directed cycle and $\lambda_{S}(\cal{F})$ be the Perron root (spectral radius) of its weighted adjacency matrix. We prove that the infimum of $\ell_{max}(\cal{F})\bigl(1-\lambda_{S}(\cal{F})\bigr)$ taken over all $\cal{F}$ is positive for every $S\in \cal{S}_{<}({\cal D})$ if and only if $\cal{D}$ admits a finite cycle transversal. The result obtained provides general theorems on the set ${\cal T}({\cal D})$ of transient weightings of ${\cal D}.$ In particular, we present a theorem of alternatives for finite approximations to elements of ${\cal T}({\cal D})$ and simply reprove V. Cyr's criterion for ${\cal T}({\cal D})$ to be empty.