Search

We give a topological classification of Lawson's bipolar minimal surfaces corresponding to his $\xi$- and $\eta$-family. Therefrom we deduce upper as well as lower bounds on the area of these surfaces, and find that they are not embedded., Comment: typos corrected, minor changes in the introduction

Sequence-to-sequence models based on LSTM and GRU are a most popular choice for forecasting time series data reaching state-of-the-art performance. Training such models can be delicate though. The two most common training strategies within this context are teacher forcing (TF) and free running (FR). TF can be used to help the model to converge faster but may provoke an exposure bias issue due to a discrepancy between training and inference phase. FR helps to avoid this but does not necessarily lead to better results, since it tends to make the training slow and unstable instead. Scheduled sampling was the first approach tackling these issues by picking the best from both worlds and combining it into a curriculum learning (CL) strategy. Although scheduled sampling seems to be a convincing alternative to FR and TF, we found that, even if parametrized carefully, scheduled sampling may lead to premature termination of the training when applied for time series forecasting. To mitigate the problems of the above approaches we formalize CL strategies along the training as well as the training iteration scale. We propose several new curricula, and systematically evaluate their performance in two experimental sets. For our experiments, we utilize six datasets generated from prominent chaotic systems. We found that the newly proposed increasing training scale curricula with a probabilistic iteration scale curriculum consistently outperforms previous training strategies yielding an NRMSE improvement of up to 81% over FR or TF training. For some datasets we additionally observe a reduced number of training iterations. We observed that all models trained with the new curricula yield higher prediction stability allowing for longer prediction horizons., Comment: Published in Transactions on Machine Learning Research (10/2022)

Traceability approves trace links among software artifacts based on whether two artifacts are related by system functionalities. The traces are valuable for software development, but are difficult to obtain manually. To cope with the costly and fallible manual recovery, automated approaches are proposed to recover traces through textual similarities among software artifacts, such as those based on Information Retrieval (IR). However, the low quality & quantity of artifact texts negatively impact the calculated IR values, thus greatly hindering the performance of IR-based approaches. In this study, we propose to extract co-occurred word pairs from the text structures of both requirements and code (i.e., consensual biterms) to improve IR-based traceability recovery. We first collect a set of biterms based on the part-of-speech of requirement texts, and then filter them through the code texts. We then use these consensual biterms to both enrich the input corpus for IR techniques and enhance the calculations of IR values. A nine-system-based evaluation shows that in general, when solely used to enhance IR techniques, our approach can outperform pure IR-based approaches and another baseline by 21.9% & 21.8% in AP, and 9.3% & 7.2% in MAP, respectively. Moreover, when used to collaborate with another enhancing strategy from different perspectives, it can outperform this baseline by 5.9% in AP and 4.8% in MAP., Comment: Accepted by the 37th IEEE/ACM International Conference on Automated Software Engineering (ASE 2022)

Every moment, countless meteoroids enter our atmosphere unseen. The detection and measurement of meteors offer the unique opportunity to gain insights into the composition of our solar systems' celestial bodies. Researchers, therefore, carry out a wide-area-sky-monitoring to secure 360-degree video material, saving every single entry of a meteor. Existing machine intelligence cannot accurately recognize events of meteors intersecting the earth's atmosphere due to a lack of high-quality training data publicly available. This work presents four reusable open source solutions for researchers trained on data we collected due to the lack of available labeled high-quality training data. We refer to the proposed dataset as the NightSkyUCP dataset, consisting of a balanced set of 10,000 meteor- and 10,000 non-meteor-events. Our solutions apply various machine learning techniques, namely classification, feature learning, anomaly detection, and extrapolation. For the classification task, a mean accuracy of 99.1\% is achieved. The code and data are made public at figshare with DOI: 10.6084/m9.figshare.16451625, Comment: 11 Pages, 10 Figures, Accepted for publication in MNRAS Journal

Transformer networks such as CodeBERT already achieve outstanding results for code clone detection in benchmark datasets, so one could assume that this task has already been solved. However, code clone detection is not a trivial task. Semantic code clones, in particular, are challenging to detect. We show that the generalizability of CodeBERT decreases by evaluating two different subsets of Java code clones from BigCloneBench. We observe a significant drop in F1 score when we evaluate different code snippets and functionality IDs than those used for model building.

Type IIb supernovae are believed to originate from core-collapse progenitors having kept only a very thin hydrogen envelope. We aim to explore how some physical factors, such as rotation, metallicity, overshooting, and the initial orbital period in binaries, significantly affect the Roche lobe overflow and the formation of type IIb supernovae. It is found that binaries are the main channel that capable of producing type typeIIb supernovae progenitors in the mass range for initial masses below 20 $M_{\odot}$. The formation of type IIb supernova progenitors is extremely sensitive to the initial orbital period. A less massive hydrogen indicates smaller radius and a higher effective temperatures, and vice versa. Binary systems with initial periods between 300 and 720 days produce type IIb progenitors that are a red supergiant. Those with an initial period between 50 and 300 days produce yellow supergiant progenitors and those with initial periods shorter than 50 days, blue supergiant progenitors. Both rapid rotation and larger overshooting can enlarge the carbon-oxygen core mass and lead to higher core temperature and lower central density at the pre-collapse phase. They are also beneficial to surface nitrogen enrichment but restrict the efficiency of the first dredge-up. SN IIb progenitors with low metallicity have smaller hydrogen envelope masses and radii than the high metallicity counterparts. Ultra-stripped binary models have systematically higher core mass fraction $\rm ^{12}C$ left, which has important influence on the compactness of type IIb progenitors., Comment: Accepted in ApJS

Gradient inversion attacks on federated learning systems reconstruct client training data from exchanged gradient information. To defend against such attacks, a variety of defense mechanisms were proposed. However, they usually lead to an unacceptable trade-off between privacy and model utility. Recent observations suggest that dropout could mitigate gradient leakage and improve model utility if added to neural networks. Unfortunately, this phenomenon has not been systematically researched yet. In this work, we thoroughly analyze the effect of dropout on iterative gradient inversion attacks. We find that state of the art attacks are not able to reconstruct the client data due to the stochasticity induced by dropout during model training. Nonetheless, we argue that dropout does not offer reliable protection if the dropout induced stochasticity is adequately modeled during attack optimization. Consequently, we propose a novel Dropout Inversion Attack (DIA) that jointly optimizes for client data and dropout masks to approximate the stochastic client model. We conduct an extensive systematic evaluation of our attack on four seminal model architectures and three image classification datasets of increasing complexity. We find that our proposed attack bypasses the protection seemingly induced by dropout and reconstructs client data with high fidelity. Our work demonstrates that privacy inducing changes to model architectures alone cannot be assumed to reliably protect from gradient leakage and therefore should be combined with complementary defense mechanisms., Comment: 25 pages, 17 figures, 9 tables (supplementary material included)

Gradient Inversion (GI) attacks are a ubiquitous threat in Federated Learning (FL) as they exploit gradient leakage to reconstruct supposedly private training data. Common defense mechanisms such as Differential Privacy (DP) or stochastic Privacy Modules (PMs) introduce randomness during gradient computation to prevent such attacks. However, we pose that if an attacker effectively mimics a client's stochastic gradient computation, the attacker can circumvent the defense and reconstruct clients' private training data. This paper introduces several targeted GI attacks that leverage this principle to bypass common defense mechanisms. As a result, we demonstrate that no individual defense provides sufficient privacy protection. To address this issue, we propose to combine multiple defenses. We conduct an extensive ablation study to evaluate the influence of various combinations of defenses on privacy protection and model utility. We observe that only the combination of DP and a stochastic PM was sufficient to decrease the Attack Success Rate (ASR) from 100% to 0%, thus preserving privacy. Moreover, we found that this combination of defenses consistently achieves the best trade-off between privacy and model utility., Comment: This version represents a comprehensive rework of the initial study, including substantial updates to the methodology, analysis, and conclusions. 26 pages, 2 figures, 5 tables

[Context and motivation] Ecosystems developed as Open Source Software (OSS) are considered to be highly innovative and reactive to new market trends due to their openness and wide-ranging contributor base. Participation in OSS often implies opening up of the software development process and exposure towards new stakeholders. [Question/Problem] Firms considering to engage in such an environment should carefully consider potential opportunities and challenges upfront. The openness may lead to higher innovation potential but also to frictional losses for engaged firms. Further, as an ecosystem progresses, power structures and influence on feature selection may fluctuate accordingly. [Principal ideas/results] We analyze the Apache Hadoop ecosystem in a quantitative longitudinal case study to investigate changing stakeholder influence and collaboration patterns. Further, we investigate how its innovation and time-to-market evolve at the same time. [Contribution] Findings show collaborations between and influence shifting among rivaling and non-competing firms. Network analysis proves valuable on how an awareness of past, present and emerging stakeholders, in regards to power structure and collaborations may be created. Furthermore, the ecosystem's innovation and time-to-market show strong variations among the release history. Indications were also found that these characteristics are influenced by the way how stakeholders collaborate with each other.

We obtain estimates on nonlocal quantities appearing in the Volume Preserving Mean Curvature Flow (VPMCF) in the closed, Euclidean setting. As a result we demonstrate that blowups of finite time singularities of VPMCF are ancient solutions to Mean Curvature Flow (MCF), prove that monotonicity methods may always be applied at finite times and obtain information on the asymptotics of the flow., Comment: comments welcome

We demonstrate that the property of being Alexandrov immersed is preserved along mean curvature flow. Furthermore, we demonstrate that mean curvature flow techniques for mean convex embedded flows such as noncollapsing and gradient estimates also hold in this setting. We also indicate the necessary modifications to the work of Brendle--Huisken to allow for mean curvature flow with surgery for the Alexandrov immersed, $2$-dimensional setting., Comment: 19 pages

The natural occurrence of singular spaces in applications has led to recent investigations on performing topological data analysis (TDA) in a stratified framework. In many applications, there is no a priori information on what points should be regarded as singular or regular. For this purpose we describe a fully implementable process that provably approximates the stratification for a large class of two-strata Whitney stratified spaces from sufficiently close non-stratified samples. Additionally, in this work, we establish a notion of persistent stratified homotopy type obtained from a sample with two strata. In analogy to the non-stratified applications in TDA which rely on a series of convenient properties of (persistent) homotopy types of sufficiently regular spaces, we show that our persistent stratified homotopy type behaves much like its non-stratified counterpart and exhibits many properties (such as stability, and inference results) necessary for an application in TDA. In total, our results combine to a sampling theorem guaranteeing the (approximate) inference of (persistent) stratified homotopy types of sufficiently regular two-strata Whitney stratified spaces., Comment: Fixed several typos; Expanded on the introduction with several illustrative examples

Scale invariance is expected in empty Universe models, while the presence of matter tends to suppress it. As shown recently, scale invariance is certainly absent in cosmological models with densities equal to or above the critical value $\varrho_{\mathrm{c}} =3H^2_0/(8 \pi G)$. For models with densities below $\varrho_{\mathrm{c}}$, the possibility of limited effects remains open. If present, scale invariance would be a global cosmological property. Some traces could be observable locally. For the Earth-Moon two-body system, the predicted additional lunar recession would be increased by 0.92 cm/yr, while the tidal interaction would also be slightly increased. The Earth-Moon distance is the most systematically measured distance in the Solar System, thanks to the Lunar Laser Ranging (LLR) experiment active since 1970. The observed lunar recession from LLR amounts to 3.83 ($\pm 0.009$) cm/yr; implying a tidal change of the length-of-the-day (LOD) by 2.395 ms/cy. However, the observed change of the LOD since the Babylonian Antiquity is only 1.78 ms/cy, a result supported by paleontological data, and implying a lunar recession of 2.85 cm/yr. The significant difference of (3.83-2.85) cm/yr = 0.98 cm/yr, already pointed out by several authors over the last two decades, corresponds well to the predictions of the scale-invariant theory, which is also supported by several other astrophysical tests., Comment: 21 pages, 1 figure, submitted as conference contribution to Cosmology on Small Scales (CSS2022) https://css2022.math.cas.cz/

A combined convolutional autoencoder-recurrent neural network machine learning model is presented to analyse and forecast the dynamics and low-order statistics of the local convective heat flux field in a two-dimensional turbulent Rayleigh-B\'{e}nard convection flow at Prandtl number ${\rm Pr}=7$ and Rayleigh number ${\rm Ra}=10^7$. Two recurrent neural networks are applied for the temporal advancement of flow data in the reduced latent data space, a reservoir computing model in the form of an echo state network and a recurrent gated unit. Thereby, the present work exploits the modular combination of three different machine learning algorithms to build a fully data-driven and reduced model for the dynamics of the turbulent heat transfer in a complex thermally driven flow. The convolutional autoencoder with 12 hidden layers is able to reduce the dimensionality of the turbulence data to about 0.2 \% of their original size. Our results indicate a fairly good accuracy in the first- and second-order statistics of the convective heat flux. The algorithm is also able to reproduce the intermittent plume-mixing dynamics at the upper edges of the thermal boundary layers with some deviations. The same holds for the probability density function of the local convective heat flux with differences in the far tails. Furthermore, we demonstrate the noise resilience of the framework which suggests the present model might be applicable as a reduced dynamical model that delivers transport fluxes and their variations to the coarse grid cells of larger-scale computational models, such as global circulation models for the atmosphere and ocean., Comment: 14 pages, 13 figures

A new parasitic, mixed-field, neutron-dominated irradiation station has been recently commissioned at the European Laboratory for Particle Physics (CERN). The station is installed in the Neutron Time-Of-Flight (n\_TOF) facility, taking advantage of the secondary radiation produced by the neutron spallation target. The new station allows radiation damage studies to be performed in irradiation conditions that are closer to the ones encountered during the operation of particle accelerators; the irradiation tests carried out in the station will be complementary to the standard tests on materials, usually performed with gamma sources. Samples will be exposed to neutron doses in the MGy range per year, with minimal impact on the n TOF facility operation. The station has twenty-four irradiation positions, each hosting up to 100 cm3 of sample material. In view of its proximity to the n\_TOF target, inside protective shielding, the irradiation station and its operating procedures have been carefully developed taking into account the safety of personnel and to avoid any unwanted impact on the operation of the n\_TOF facility and experiments. Due to the residual radioactivity of the whole area around the n\_TOF target and of the irradiated samples, access to the irradiation station is forbidden to human operators even when the n\_TOF facility is not in operation. Robots are used for the remote installation and retrieval of the samples, and other optimizations of the handling procedures were developed in compliance with radiation protection regulations and the aim of minimizing doses to personnel.

In this paper, a deep learning approach is presented for direction of arrival estimation using automotive-grade ultrasonic sensors which are used for driving assistance systems such as automatic parking. A study and implementation of the state of the art deterministic direction of arrival estimation algorithms is used as a benchmark for the performance of the proposed approach. Analysis of the performance of the proposed algorithms against the existing algorithms is carried out over simulation data as well as data from a measurement campaign done using automotive-grade ultrasonic sensors. Both sets of results clearly show the superiority of the proposed approach under realistic conditions such as noise from the environment as well as eventual errors in measurements. It is demonstrated as well how the proposed approach can overcome some of the known limitations of the existing algorithms such as precision dilution of triangulation and aliasing., Comment: Published in the international conference of intelligent vehicles (ICOIV 2021)

We present a summary of the main results within the Scale Invariant Vacuum (SIV) paradigm as related to the Weyl Integrable Geometry. After a brief review of the mathematical framework, we will highlight the main results related to inflation within the SIV [9], the growth of the density fluctuations [8], and the application of the SIV to scale-invariant dynamics of Galaxies, MOND, Dark Matter, and the Dwarf Spheroidals [7]. The connection of the weak-field SIV results to the un-proper time parametrization within the re-parametrization paradigm is also discussed [14]., Comment: 10 pages, 3 figures, summary of the talk presented at the conference Alternative Gravities and Fundamental Cosmology, Organized by the University of Szczecin, Poland, 6-10 September 2021

Context: Grids of stellar models, computed with the same physical ingredients, allow one to study the impact of a given physics on a broad range of initial conditions and are a key ingredient for modeling the evolution of galaxies. Aims: We present a grid of single star models for masses between 0.8 and 120 $M_\odot$, with and without rotation for a mass fraction of heavy element Z=0.006, representative of the Large Magellanic Cloud (LMC). Methods: We used the Geneva stellar evolution code. The evolution was computed until the end of the central carbon-burning phase, the early asymptotic giant branch phase, or the core helium-flash for massive, intermediate, and low mass stars, respectively. Results: The outputs of the present stellar models are well framed by the outputs of the two grids obtained by our group for metallicities above and below the one considered here. The models of the present work provide a good fit to the nitrogen surface enrichments observed during the main sequence for stars in the LMC with initial masses around 15 $M_\odot$. They also reproduce the slope of the luminosity function of red supergiants of the LMC well, which is a feature that is sensitive to the time-averaged mass loss rate over the red supergiant phase. The most massive black hole that can be formed from the present models at Z=0.006 is around 55 $M_\odot$. No model in the range of mass considered will enter into the pair-instability supernova regime, while the minimal mass to enter the region of pair pulsation instability is around 60 $M_\odot$ for the rotating models and 85 $M_\odot$ for the nonrotating ones. Conclusions: The present models are of particular interest for comparisons with observations in the LMC and also in the outer regions of the Milky Way. We provide public access to numerical tables that can be used for computing interpolated tracks and for population synthesis studies., Comment: 11 pages, 11 figures, published in A&A

We present a grid of stellar models at super-solar metallicity (Z = 0.020) extending the previous grids of Geneva models at solar and sub-solar metallicities. A metallicity of Z = 0.020 was chosen to match that of the inner Galactic disk. A modest increase of 43% (=0.02/0.014) in metallicity compared to solar models means that the models evolve similarly to solar models but with slightly larger mass loss. Mass loss limits the final total masses of the super-solar models to 35 M$_\odot$ even for stars with initial masses much larger than 100 M$_\odot$. Mass loss is strong enough in stars above 20 M$_\odot$ for rotating stars (25 M$_\odot$ for non-rotating stars) to remove the entire hydrogen-rich envelope. Our models thus predict SNII below 20 M$_\odot$ for rotating stars (25 M$_\odot$ for non-rotating stars) and SNIb (possibly SNIc) above that. We computed both isochrones and synthetic clusters to compare our super-solar models to the Westerlund 1 (Wd1) massive young cluster. A synthetic cluster combining rotating and non-rotating models with an age spread between log10 (age/yr) = 6.7 and 7.0 is able to reproduce qualitatively the observed populations of WR, RSG and YSG stars in Wd1, in particular their simultaneous presence at log10(L/L$_\odot$) = 5-5.5. The quantitative agreement is imperfect and we discuss the likely causes: synthetic cluster parameters, binary interactions, mass loss and their related uncertainties. In particular, mass loss in the cool part of the HRD plays a key role., Comment: 15 pages, 12 figures. Accepted for publication in MNRAS

Recently it was found from Cassini data that the mean recession speed of Titan from Saturn is $v=11.3\pm 2.0$ cm/yr which corresponds to a tidal quality factor of Saturn $Q\cong 100$ while the standard estimate yields $Q\ge 6\cdot 10^4$. It was assumed that such a large speed $v$ is due to a resonance locking mechanism of five inner mid-sized moons of Saturn. In this paper, we show that an essential part of $v$ may come from a local Hubble expansion, where the Hubble-Lema\^{\i}tre constant $H_0$ recalculated to the Saturn-Titan distance $D$ is 8.15 cm/(yr $D$). Our hypothesis is based on many other observations showing a slight expansion of the Solar system and also of our Galaxy at a rate comparable with $H_0$. We demonstrate that the large disproportion in estimating the $Q$ factor can be just caused by the local expansion effect., Comment: 20 pages, 1 figure, 2 tables, to appear in the journal Gravitation and Cosmology Vol. 28, Issue 2 (2022)

Microscale dynamic testing is vital to the understanding of material behavior at application relevant strain rates. However, despite two decades of intense micromechanics research, the testing of microscale metals has been largely limited to quasi-static strain rates. Here we report the dynamic compression testing of pristine 3D printed copper micropillars at strain rates from $\sim0.001$ s$^{-1}$ to $\sim500$ s$^{-1}$. It was identified that microcrystalline copper micropillars deform in a single-shear like manner exhibiting a weak strain rate dependence at all strain rates. Ultrafine grained (UFG) copper micropillars, however, deform homogenously via barreling and show strong rate-dependence and small activation volumes at strain rates up to $\sim0.1$ s$^{-1}$, suggesting dislocation nucleation as the deformation mechanism. At higher strain rates, yield stress saturates remarkably, resulting in a decrease of strain rate sensitivity by two orders of magnitude and a four-fold increase in activation volume, implying a transition in deformation mechanism to collective dislocation nucleation., Comment: Corresponding authors with equal contribution are: Szilvia Kal\'acska and Rajaprakash Ramachandramoorthy. Accepted manuscript: Applied Materials Today

Digitalization in seaports dovetails the IT infrastructure of various actors (e.g., shipping companies, terminals, customs, port authorities) to process complex workflows for shipping containers. The security of these workflows relies not only on the security of each individual actor but actors must also provide additional guarantees to other actors like, for instance, respecting obligations related to received data or checking the integrity of workflows observed so far. This paper analyses global security requirements (e.g., accountability, confidentiality) of the workflows and decomposes them - according to the way workflow data is stored and distributed - into requirements and obligations for the individual actors. Security mechanisms are presented to satisfy the resulting requirements, which together with the guarantees of all individual actors will guarantee the security of the overall workflow., Comment: Presented at the 29th Conference of the International Association of Maritime Economists, Rotterdam, November 2021

The observed late-type WC Wolf-Rayet stars (WC7-9) with low luminosity below $\rm \log L/L_{\odot} < 5.4$ in the HR diagram cannot be reproduced satisfactorily by the evolutionary track of single stars. The mass transfer due to Roche lobe overflow drastically modifies the internal structure and surface compositions of two components. Therefore, binaries provide a very promising evolutionary channel to produce these WC stars., Comment: Accept for publication in A&A

Moving object detection and segmentation is an essential task in the Autonomous Driving pipeline. Detecting and isolating static and moving components of a vehicle's surroundings are particularly crucial in path planning and localization tasks. This paper proposes a novel real-time architecture for motion segmentation of Light Detection and Ranging (LiDAR) data. We use three successive scans of LiDAR data in 2D Bird's Eye View (BEV) representation to perform pixel-wise classification as static or moving. Furthermore, we propose a novel data augmentation technique to reduce the significant class imbalance between static and moving objects. We achieve this by artificially synthesizing moving objects by cutting and pasting static vehicles. We demonstrate a low latency of 8 ms on a commonly used automotive embedded platform, namely Nvidia Jetson Xavier. To the best of our knowledge, this is the first work directly performing motion segmentation in LiDAR BEV space. We provide quantitative results on the challenging SemanticKITTI dataset, and qualitative results are provided in https://youtu.be/2aJ-cL8b0LI., Comment: Accepted for Presentation at International Conference on Computer Vision Theory and Applications (VISAPP 2022)

We study complete minimal surfaces in $\mathbb{R}^n$ with finite total curvature and embedded planar ends. After conformal compactification via inversion, these yield examples of surfaces stationary for the Willmore bending energy $\mathcal{W}: =\frac{1}{4} \int|\vec H|^2$. In codimension one, we prove that the $\mathcal{W}$-Morse index for any inverted minimal sphere or real projective plane with $m$ such ends is exactly $m-3=\frac{\mathcal{W}}{4\pi}-3$. We also consider several geometric properties -- for example, the property that all $m$ asymptotic planes meet at a single point -- of these minimal surfaces and explore their relation to the $\mathcal{W}$-Morse index of their inverted surfaces., Comment: Comments welcome

Speech-to-speech translation combines machine translation with speech synthesis, introducing evaluation challenges not present in either task alone. How to automatically evaluate speech-to-speech translation is an open question which has not previously been explored. Translating to speech rather than to text is often motivated by unwritten languages or languages without standardized orthographies. However, we show that the previously used automatic metric for this task is best equipped for standardized high-resource languages only. In this work, we first evaluate current metrics for speech-to-speech translation, and second assess how translation to dialectal variants rather than to standardized languages impacts various evaluation methods., Comment: ASRU 2021

We review the problem of the consistency between the observed values of the lunar recession from Lunar Laser Ranging (LLR) and of the increase of the length-of-the-day (LOD). From observations of lunar occultations completed by recent IERS data, we derive a variation rate of the LOD equal to 1.09 ms/cy from 1680 to 2020, which compares well with McCarthy and Babcock (1986) and Sidorenkov (2005). This rate is lower than the mean rate of 1.78 ms/cy derived by Stephenson et al. (2016) on the basis of eclipses in the Antiquity and Middle Age. The difference in the two observed rates starts at the epoch of a major change in the data accuracy with telescopic observations. The observed lunar recession appears too large when compared to the tidal slowing down of the Earth determined from eclipses in the Antiquity and Middle Age and even much more when determined from lunar occultations and IERS data from 1680 to 2020. With a proper account of the tidal effects and of the detailed studies on the atmospheric effects, the melting from icefields, the changes of the sea level, the glacial isostatic adjustment, and the core-mantle coupling, we conclude that the long-standing problem of the presence or absence of a local cosmological expansion is still an open question., Comment: 25 pages, 5 figures, 3 tables, accepted for publication in Astrophysics and Space Science with manuscript number ASTR-D-21-00252

This paper discusses the use and popularity of a rhythm, which henceforth is referred to as "Tresillo rhythm". We first define and formalizes the Tresillo rhythm. Given a mathematical representation of the rhythm, it is then traced in the US Billboard Top 20 Charts of the last 20 years. To detect and determine the use of this rhythm in a song, we compute the similarity of a song with this rhythm. The calculated similarity, then indicates how similar the rhythm of a pop song is compared to the prior defined Tresillo rhythm. To assert and cross-validate the computer rhythm similarity, two different formalizations of the Tresillo rhythm have been compiled and several different approaches to calculated rhythm similarities have been tested and compared. This similarity measure is then used to do an empirical study on the usage of the Tresillo rhythm in the US Billboard Top 20 Charts of the past 20 years (1999-2019). Finally, we argue about some of the possible reasons for the observed trend.

Nonclassical light sources are highly sought after as they are an integral part of quantum communication and quantum computation devices. Typical sources rely on bulk crystals that are not compact and have limited bandwidth due to phase-matching conditions. In this work, we demonstrate the generation of photon pairs from a free-standing lithium niobate microcube at the telecommunication wavelength through the spontaneous parametric down-conversion process. The maximum photon pair generation rate obtained from a single microcube with the size of ~4 microns is ~80 Hz, resulting in an efficiency of ~1.2 GHz/Wm per unit volume, which is an order of magnitude higher than the efficiency of photon-pair generation in bulky nonlinear crystals. The microcubes are synthesized through a solvothermal method, offering the possibility for scalable devices via bottom-up assembly. Our work constitutes an important step forward in the realization of compact nonclassical light sources with broadband tunability for various applications in quantum communication, quantum computing, and quantum metrology., Comment: 19 pages, 4 figures

Sigma Ori E, a massive helium B-type star, shows a high surface rotation and a strong surface magnetic field potentially challenging the process of wind magnetic braking. The Gaia satellite provides an accurate distance to that star and confirms its membership to the sigma Ori cluster. We account for these two key pieces of information to investigate whether single star models can reproduce the observed properties of sigma Ori E and provide new estimates for its metallicity, mass, and age. We compute rotating stellar models accounting for wind magnetic braking and magnetic quenching of the mass loss. We obtain that sigma Ori E is a very young star (age less than 1 Myr) with an initial mass around 9 Msol, a surface equatorial magnetic field around 7 kG and having a metallicity Z (mass fraction of heavy elements) around 0.020. No solution is obtained with the present models for a metallicity Z=0.014. The initial rotation of the models fitting sigma Ori E is not much constrained and can be anywhere in the range studied in the present work. Because of its very young age, models predict no observable changes of the surface abundances due to rotational mixing. The simultaneous high surface rotation and high surface magnetic field of sigma Ori E may simply be a consequence of its young age. This young age implies that the processes responsible for producing the chemical inhomogeneities that are observed at its surface should be rapid. Thus for explaining the properties of sigma Ori E, there is no necessity to invoke a merging event although such a scenario cannot be discarded. Other stars (HR 5907, HR 7355, HR 345439, HD 2347, CPD -50^{o}3509$) showing similar properties as sigma Ori E (fast rotation and strong surface magnetic field) may also be very young stars, although determination of the braking timescales is needed to confirm such a conclusion., Comment: 13 pages, 8 figures, paper accepted for publication in Astronomy and Astrophysics

Collaborative training of neural networks leverages distributed data by exchanging gradient information between different clients. Although training data entirely resides with the clients, recent work shows that training data can be reconstructed from such exchanged gradient information. To enhance privacy, gradient perturbation techniques have been proposed. However, they come at the cost of reduced model performance, increased convergence time, or increased data demand. In this paper, we introduce PRECODE, a PRivacy EnhanCing mODulE that can be used as generic extension for arbitrary model architectures. We propose a simple yet effective realization of PRECODE using variational modeling. The stochastic sampling induced by variational modeling effectively prevents privacy leakage from gradients and in turn preserves privacy of data owners. We evaluate PRECODE using state of the art gradient inversion attacks on two different model architectures trained on three datasets. In contrast to commonly used defense mechanisms, we find that our proposed modification consistently reduces the attack success rate to 0% while having almost no negative impact on model training and final performance. As a result, PRECODE reveals a promising path towards privacy enhancing model extensions., Comment: 16 pages, 16 figures, 5 tables (supplementary material included). Accepted at WACV 2022. Code available at https://github.com/SECSY-Group/PRECODE

Imaging in real time the complete dynamics of a process as fundamental as photoemission has long been out of reach due to the difficulty of combining attosecond temporal resolution with fine spectral and angular resolutions. Here, we achieve full decoding of the intricate angle-dependent dynamics of a photoemission process in helium, spectrally and anisotropically structured by twophoton transitions through intermediate bound states. Using spectrally- and angularly-resolved attosecond electron interferometry, we characterize the complex-valued transition probability amplitude towards the photoelectron quantum state. This allows reconstructing in space, time and energy the complete formation of the photoionized wavepacket., Comment: 7 pages, 4 figures

Maxwell equations and the equations of General Relativity are scale invariant in empty space. The presence of charge or currents in electromagnetism or the presence of matter in cosmology are preventing scale invariance. The question arises on how much matter within the horizon is necessary to kill scale invariance. The scale invariant field equation, first written by Dirac in 1973 and then revisited by Canuto et al. in 1977, provides the starting point to address this question. The resulting cosmological models show that, as soon as matter is present, the effects of scale invariance rapidly decline from \rho=0 to \rho_c and are forbidden for densities above \rho_c. The absence of scale invariance in this case is consistent with considerations about causal connection. Below \rho_c, scale invariance appears as an open possibility, which also depends on the occurrence of inflation in the scale invariant context. In the present approach, we identify the scalar field of the empty space in the Scale Invariant Vacuum (SIV) context to the scalar field \phi in the energy density expression of the vacuum at inflation. This leads to some constraints on the potential. This identification also solves the so-called ``cosmological constant problem''. In the framework of scale invariance,an inflation with a large number of e-foldings is also predicted. We conclude that scale invariance for models with densities below \rho_c is an open possibility; the final answer may come from high redshift observations, where differences from the LCDM models appear., Comment: 12 pages, 1 figure

A 360{\deg} perception of scene geometry is essential for automated driving, notably for parking and urban driving scenarios. Typically, it is achieved using surround-view fisheye cameras, focusing on the near-field area around the vehicle. The majority of current depth estimation approaches focus on employing just a single camera, which cannot be straightforwardly generalized to multiple cameras. The depth estimation model must be tested on a variety of cameras equipped to millions of cars with varying camera geometries. Even within a single car, intrinsics vary due to manufacturing tolerances. Deep learning models are sensitive to these changes, and it is practically infeasible to train and test on each camera variant. As a result, we present novel camera-geometry adaptive multi-scale convolutions which utilize the camera parameters as a conditional input, enabling the model to generalize to previously unseen fisheye cameras. Additionally, we improve the distance estimation by pairwise and patchwise vector-based self-attention encoder networks. We evaluate our approach on the Fisheye WoodScape surround-view dataset, significantly improving over previous approaches. We also show a generalization of our approach across different camera viewing angles and perform extensive experiments to support our contributions. To enable comparison with other approaches, we evaluate the front camera data on the KITTI dataset (pinhole camera images) and achieve state-of-the-art performance among self-supervised monocular methods. An overview video with qualitative results is provided at https://youtu.be/bmX0UcU9wtA. Baseline code and dataset will be made public., Comment: To be published at IEEE Transactions on Intelligent Transportation Systems

The present article revisits the Diffusion Operator Integral (DOI) variance reduction technique originally proposed in Heath and Platen (2002) and extends its theoretical concept to the pricing of American-style options under (time-homogeneous) L\'evy stochastic differential equations. The resulting Jump Diffusion Operator Integral (JDOI) method can be combined with numerous Monte Carlo based stopping-time algorithms, including the ubiquitous least-squares Monte Carlo (LSMC) algorithm of Longstaff and Schwartz (cf. Carriere (1996), Longstaff and Schwartz (2001)). We exemplify the usefulness of our theoretical derivations under a concrete, though very general jump-diffusion stochastic volatility dynamics and test the resulting LSMC based version of the JDOI method. The results provide evidence of a strong variance reduction when compared with a simple application of the LSMC algorithm and proves that applying our technique on top of Monte Carlo based pricing schemes provides a powerful way to speed-up these methods.