Search

In the context of the GRAVITY+ upgrade, the adaptive optics (AO) systems of the GRAVITY interferometer are undergoing a major lifting. The current CILAS deformable mirrors (DM, 90 actuators) will be replaced by ALPAO kilo-DMs (43x43, 1432 actuators). On top of the already existing 9x9 Shack-Hartmann wavefront sensors (SH-WFS) for infrared (IR) natural guide star (NGS), new 40x40 SH-WFSs for visible (VIS) NGS will be deployed. Lasers will also be installed on the four units of the Very Large Telescope to provide a laser guide star (LGS) option with 30x30 SH-WFSs and with the choice to either use the 9x9 IR-WFSs or 2x2 VIS-WFSs for low order sensing. Thus, four modes will be available for the GRAVITY+ AO system (GPAO): IR-NGS, IR-LGS, VIS-NGS and VIS-LGS. To prepare the instrument commissioning and help the observers to plan their observations, a tool is needed to predict the performances of the different modes and for different observing conditions (NGS magnitude, science object magnitude, turbulence conditions...) We developed models based on a Mar{\'e}chal approximation to predict the Strehl ratio of the four GPAO modes in order to feed the already existing tool that simulates the GRAVITY performances. Waiting for commissioning data, our model was validated and calibrated using the TIPTOP toolbox, a Point Spread Function simulator based on the computation of Power Spectrum Densities. In this work, we present our models of the NGS modes of GPAO and their calibration with TIPTOP.

In the last decade, EU became aware of its dependency on mining countries for mineral resources. In that context in 2022, France launched new airborne geophysical surveys in the Limousin region (northwestern French Massif Central) which is historically renown for its gold, uranium, tungsten and tin production. Thanks to these new data, it has become possible to both explore the Saint-Goussaud - Auriat zone, centered on the Puy-les-Vignes tungsten deposit and to investigate the numerous W occurrences. To do so, direct geophysical modelling using gravimetric, airborne magnetic and electromagnetic data has allowed to create a 3D regional model. It mainly illustrates locations where evolved granites deepen, possibly associated with hidden granitic cupolas. Besides, electromagnetic data have been analysed to highlight conductive anomalies. The lineament analysis has contributed to identify a conductive corridor, interpreted as a regional structure. Conductive vertical conduits, which are spatially close to mineralized occurrences and stream-sediment anomalies have also been spotted. This new method has thus allowed generating new elements which could be combined and could have implications for granite-related tungsten exploration in this region.

We derive the full next-to-eikonal (NEik) corrections to the gluon propagator from before to after traversing a highly boosted gluon background field, including corrections both beyond the shockwave limit and beyond the static limit in particular. After summarizing the results of the full NEik corrections to the before-to-after quark propagator computed in our earlier works, we also derive the before-to-inside, inside-to-inside and inside-to-after quark and gluon propagators, which are building blocks to calculate high-energy scattering processes at NEik order. Using these results and also including the NEik corrections that stem from interactions with the target via t-channel quark exchanges, we compute inclusive cross sections for quark and gluon production at forward rapidities in quark-nucleus and gluon-nucleus scatterings at NEik accuracy., Comment: 46 pages, 8 figures

The rheology of coextruded layered films of polystyrene/poly(methyl methacrylate) (PS/PMMA) has been studied with small and large amplitude oscillations at a temperature above their glass transition. While the complex viscosity remains constant over the experimental time window for the micron-sized layered films, a decrease has been observed for the nanolayered films. The rheological behavior has then been correlated to the morphological evolution of the multilayer films: while the nanolayers dewet. Layer breakup followed by retraction and coalescence leading to a lamellar-like blend morphology succeeded by a nodular-like morphology has been evidenced in the nanolayer films, for all compositions and conditions tested. The analysis of the microscopic images of the film cross-sections also provided the droplet size distribution. The nodular morphology is achieved more rapidly when the initial layers are the thinnest at low strains, while at high strains the formation of these droplets is prevented., Comment: 16 pages, 10 figure + supplementary information

We introduce a framework for entanglement-assisted quantum error correcting codes that unifies the three original frameworks for such codes called EAQEC, EAOQEC, and EACQ under a single umbrella. The unification is arrived at by viewing entanglement-assisted codes from the operator algebra quantum error correction perspective, and it is built upon a recently established extension of the stabilizer formalism to that setting. We denote the framework by EAOAQEC, and we prove a general error correction theorem for such codes, derived from the algebraic perspective, that generalizes each of the earlier results. This leads us to a natural notion of distance for such codes, and we derive a number of distance results for subclasses of the codes. We show how EACQ codes form a proper subclass of the entanglement-assisted subspace codes defined by EAOAQEC. We identify and construct new classes of entanglement-assisted subsystem codes and entanglement-assisted hybrid classical-quantum codes that are found outside of the earlier approaches., Comment: 38 pages, 1 figure

In this paper, we prove the existence and the uniqueness of a weak and mild solution of the following nonlinear parabolic problem involving the porous $p$-fractional Laplacian: \begin{equation*} \begin{cases} \partial_t u+(-\Delta)^s_p(|u|^{m-1}u)=h(t,x,|u|^{m-1}u) & \text{in} \; (0,T)\times \Omega,\\ u=0 & \text{in} \; (0,T) \times \mathbb{R}^d\backslash \Omega, \\ u(0,\cdot)=u_0 & \text{in} \; \Omega . \end{cases}\ \end{equation*} We also study further the the homogeneous case $h(u)=|u|^{q-1}u$ with $q>0$. In particular we investigate global time existence, uniqueness, global behaviour of weak solutions and stabilization.

In the dorsal horn, projection neurons play a significant role in pain processing by transmitting sensory stimuli to supraspinal centers during nociception. Following exposure to intense noxious stimuli, a sensitization process occurs to adapt the dorsal horn functional state. Notably, projection neurons can display a switch in firing pattern from tonic firing to plateau potentials with sustained afterdischarges. For afterdischarges to manifest after this switch, the neuron must develop bistability, which characterizes the ability to show resting or spiking at the same input current depending on the context. In numerous instances, neuronal bistability arises through voltage-gated calcium channels. However, computational studies have demonstrated a trade-off between bistability and the plausibility of its resting states if calcium channels are counterbalanced with voltage-gated potassium channels. Current knowledge leaves a gap in understanding the underlying mechanisms by which robust bistability, plateau potentials, and sustained afterdischarges emerge in neurons via calcium channels. In this study, we used a conductance-based model to explore the mechanisms by which L-type calcium (CaL) channels can achieve bistability when combined with either M-type potassium (KM) channels or inward-rectifying potassium (Kir) channels. Unlike KM channels, Kir channels enhance bistability. Combined with CaL channels, KM and Kir channels promote different types of bistability, with distinct robustness, and function. An analysis of their inward/outward properties revealed that their distinct steady-state currents explain this contrast. Altogether, the complementary of CaL and Kir channels creates a reliable possible pathway for central sensitization in the dorsal horn.

In light of recent observations, we study evidence for thawing quintessence over a cosmological constant as dark energy, with emphasis on the effect of the choice of priors. Working with a parametrization for the equation of state parameter motivated by the theory, we find a preference for thawing quintessence compared to a bare cosmological constant {\it only} if we use priors which are heavily informed by the data itself. If we extend the priors to physically better motivated ranges, the evidence for thawing quintessence disappears., Comment: 12 pages, 8 figures

Reconciliation has become an essential tool in multivariate point forecasting for hierarchical time series. However, there is still a lack of understanding of the theoretical properties of probabilistic Forecast Reconciliation techniques. Meanwhile, Conformal Prediction is a general framework with growing appeal that provides prediction sets with probabilistic guarantees in finite sample. In this paper, we propose a first step towards combining Conformal Prediction and Forecast Reconciliation by analyzing how including a reconciliation step in the Split Conformal Prediction (SCP) procedure enhances the resulting prediction sets. In particular, we show that the validity granted by SCP remains while improving the efficiency of the prediction sets. We also advocate a variation of the theoretical procedure for practical use. Finally, we illustrate these results with simulations., Comment: 14 pages, 2 figures

The universal dependence of hadron suppression, $R_{\rm{AA}}(p_\perp)$, observed at large-$p_\perp$ in heavy ion collisions at RHIC and LHC allows for a systematic determination of the average parton energy loss $\langle \epsilon \rangle$ in quark-gluon plasma (QGP). A simple relation between $\langle \epsilon \rangle$ and the soft particle multiplicity allows for probing the dependence of parton energy loss on the medium path-length. We find that all the available measurements are consistent with $\langle \epsilon \rangle \propto L^\beta$ with $\beta=1.02\pm^{0.09}_{0.06}$, consistent with the pQCD expectation of parton energy loss in a longitudinally expanding QGP. We then show, based on the model predictions, that the data on the azimuthal anisotropy coefficient divided by the collision eccentricity, $v_2/\rm{e}$, follows the same scaling property as $R_{\rm{AA}}$. Finally, a linear relationship between $v_2/\rm{e}$ and the logarithmic derivative of $R_{\rm{AA}}$ at large $p_\perp$ offers a purely data-driven access to the $L$ dependence of parton energy loss. Quite remarkably, both hadron and jet measurements obey this latter relationship, moreover with consistent values of $\beta$. This points to the same parametric path-length dependence of parton and jet energy loss in QGP., Comment: 6 pages, 4 figures. Talk given at the 42nd International Conference on High Energy Physics (ICHEP2024)

The Erd\"{o}s-Moser equation $ \sum_{i=1}^{m - 1} i^k = m^k $ is a longstanding problem in number theory, with the only known solution in positive integers being $ (k, m) = (1, 3) $. This paper investigates the possibility of other solutions by employing approximation methods based on the Euler-MacLaurin formula to extend the discrete sum $ S(m - 1, k) $ to a continuous function $ S_{\mathbb{R}}(m - 1, k) $. Analyzing the approximate polynomial $ P_{\mathbb{R}}(m) = S_{\mathbb{R}}(m - 1, k) - m^k $, we apply the rational root theorem to search for potential integer solutions. Our investigation confirms that for $ k = 1 $, the only solution is $ m = 3 $. For $ k \geq 2 $, the approximation suggests that no additional positive integer solutions exist. However, we acknowledge the limitations of using approximation methods in the context of Diophantine equations, where exactness is crucial. The omission of correction terms in the approximation may overlook valid solutions. Despite these limitations, our work provides insights into the behavior of the Erd\"{o}s-Moser equation and highlights the challenges in finding solutions using analytical methods. We discuss the implications of our findings and suggest directions for future research, emphasizing the need for exact analytical techniques to conclusively address the conjecture.

We establish a central limit theorem for the sum of $\epsilon$-independent random variables, extending both the classical and free probability setting. Central to our approach is the use of graphon limits to characterize the limiting distribution, which depends on the asymptotic structure of the underlying graphs governing $\epsilon$-independence. This framework yields a range of exotic limit laws, interpolating between free and classical cases and allowing for mixtures such as free and classical convolutions of the semi-circle and Gaussian distributions. We provide a complete characterization of the limiting law, captured as the distribution of an operator on the full Fock space. We extend our main result to the multivariate setting as well as the non-identically distributed case. The proof provides insights into the combinatorial structure of $\epsilon$-independence, shedding light on a natural connection between the graph of crossings of the partitions involved and the graphon limit of the graphs of $\epsilon$-independence.

The goal of this work is to study occurrences of non-unique solutions in dual-energy CT (DECT) for objects containing water and a contrast agent. Previous studies of the Jacobian of nonlinear systems identified that a vanishing Jacobian determinant indicates the existence of multiple solutions to the system. Vanishing Jacobian determinants are identified for DECT setups by simulating intensity data for practical thickness ranges of water and contrast agent. Once existence is identified, non-unique solutions are found by simulating scan data and finding intensity contours with that intersect multiple times. With this process non-unique solutions are found for DECT setups scanning iodine and gadolinium, including setups using tube potentials in practical ranges. Non-unique solutions demonstrate a large range of differences and can result in significant discrepancies between recovered and true material mapping., Comment: Proceedings paper for accepted contribution to the 8th International Conference on Image Formation in X-Ray Computed Tomography

In the astrophysics community it is common practice to model collisionless dust, entrained in a gas flow, as a pressureless fluid. However a pressureless fluid is fundamentally different from a collisionless fluid - the latter of which generically possess a non-zero anisotropic pressure or stress tensor. In this paper we derive a fluid model for collisionless dust, entrained in a turbulent gas, starting from the equations describing the motion of individual dust grains. We adopt a covariant formulation of our model to allow for the geometry and coordinate systems prevalent in astrophysics, and provide a closure valid for the accretion disc context. We show that the continuum mechanics properties of a dust fluid corresponds to a higher-dimensional anisotropic Maxwell fluid, after the extra dimensions are averaged out, with a dynamically important rheological stress tensor. This higher-dimensional treatment has the advantage of keeping the dust velocity and velocity of the fluid seen, and their respective moments, on the same footing. This results in a simplification of the constitutive relation describing the evolution of the dust Rheological stress., Comment: 51 pages, 7 figures, accepted for publication in JfM

Context. HD140283, or the Methuselah star, is a well-known reference object in stellar evolution. Its peculiar chemical composition, proximity and absence of reddening makes it an interesting case-study of Pop II stars. Thanks to recent observational efforts, we now have precise interferometric and spectroscopic constraints, as well as revised astrometric parallaxes from the Gaia mission. Aims. We aim at determining the age of HD140283 with these lastest constraints, as well as quantifying the impact of systematics from physical inaccuracies in the stellar evolution models. Methods. Using recent spectroscopic abundances from the literature, including 3D non-LTE values for C, O, and Fe, we compute opacity tables specific to HD140283. We then use them in grids of stellar evolution models coupled to a Markov Chain Monte Carlo tool to determine the age of HD140283. Results. With our tailored models we find an age of 12.3Gy. Using a solar-scaled mixture instead results in an age value of 14Gy, in tension with the age of the universe ($13.77\pm0.06$Gy). We also find that reducing the mixing length parameter from its solar calibrated value will lead to an even lower age, in agreement with other recent studies. However, we find no direct evidence to favour a lower mixing length parameter value from our modelling. Conclusions. Taking into account the specific elemental abundances is crucial for the modelling of HD140283, as it leads to significant differences in the inferred age. However, this effect is degenerate with a lowering of the mixing length parameter. In this respect, asteroseismic constraints might play a key role in accurately deriving the mass of HD140283, therefore strongly constraining its age., Comment: Accepted for publication in A&A

AI workloads, particularly those driven by deep learning, are introducing novel usage patterns to high-performance computing (HPC) systems that are not comprehensively captured by standard HPC benchmarks. As one of the largest academic research centers dedicated to deep learning, Mila identified the need to develop a custom benchmarking suite to address the diverse requirements of its community, which consists of over 1,000 researchers. This report introduces Milabench, the resulting benchmarking suite. Its design was informed by an extensive literature review encompassing 867 papers, as well as surveys conducted with Mila researchers. This rigorous process led to the selection of 26 primary benchmarks tailored for procurement evaluations, alongside 16 optional benchmarks for in-depth analysis. We detail the design methodology, the structure of the benchmarking suite, and provide performance evaluations using GPUs from NVIDIA, AMD, and Intel. The Milabench suite is open source and can be accessed at github.com/mila-iqia/milabench.

Combining classical density functional theory (cDFT) with quantum mechanics (QM) methods offers a computationally efficient alternative to traditional QM/molecular mechanics (MM) approaches for modeling mixed quantum-classical systems at finite temperatures. However, both QM/MM and QM/cDFT rely on approximations that introduce ambiguities. This series of papers aims to establish a rigorous theoretical framework to guide the development of a QM/cDFT formulation that clarifies the approximations involved. In this first paper, we establish a comprehensive density functional theory (DFT) framework for mixed quantum (QM)-classical (MM) systems within the canonical ensemble. We derive the equilibrium density matrix for a system comprising N qm QM and N mm MM particles and, based on this derivation, propose a variational formulation of the Helmholtz free energy in terms of the full QM-MM density matrix. By leveraging permutational symmetry and thanks to constrained-search methods, we reformulate the variational problem using functionals of objects having reduced dimensionalities. Three different formulations are presented, each using quantities with progressively lower dimensionalities. This ultimately enables computation of the Helmholtz free energy using only the QM and MM one-body densities, generalizing both cDFT and electronic DFT (eDFT) for QM-MM systems. In a subsequent paper, this formalism will be extended to the grand canonical ensemble, allowing classical particle number fluctuations. This extension, in which cDFT is most naturally formulated, provides a framework for studying solvation problems.

This document summarizes the prospective physics plans of the French QCD and Heavy-Ion community, including the experimental programs at the LHC Run 5 and beyond and future colliders at CERN, within the context of the French contribution to the update of the European Strategy in Particle Physics (ESPPU 2025), as discussed in the workshop on European Strategy for Particle Physics Update 2025 organised by the QCD GdR in Oct. 2024., Comment: 6.5 pages without title page and without references, no figures. Keywords: QCD, Heavy Ions, Quark Gluon Plasma, LHC, FCC

Symmetry-breaking is pivotal for controlling ferroelectric, ferroelastic and/or ferromagnetic functions of materials, which enables applications in sensors, memories, transducers or actuators. Commonly, ferroic phases emerge from descending symmetry-breaking, as the laws of thermodynamics dictate that the ordered low entropy phases form at low temperature, which limits practical applications of many materials at room temperature. Rare examples of ascending symmetry-breakings have been observed, but the driving force remains often unclear. Here, we report on a ferroelastic symmetry-breaking occurring at high temperature in a spin-crossover material, studied by magnetic, DSC and X-ray diffraction measurements. Our DFT calculations and our model, based on the Landau theory of phase transitions, explain how the cooperative thermal switching of molecular spin state drives a ferroelastic symmetry breaking at high temperature, through a coupled Jahn-Teller distortion. Ferroelastic materials have rich properties, with important applications in memory, multifunctional and novel controllable devices. The electronic bistability in soft functional materials represents an important source of entropy gain, capable of overcoming the cost of symmetry-breaking entropy, which opens up new perspectives for stabilizing high-temperature and low-symmetry ferroic functions of advanced materials

We consider the problem of text-to-video generation tasks with precise control for various applications such as camera movement control and video-to-video editing. Most methods tacking this problem rely on providing user-defined controls, such as binary masks or camera movement embeddings. In our approach we propose OnlyFlow, an approach leveraging the optical flow firstly extracted from an input video to condition the motion of generated videos. Using a text prompt and an input video, OnlyFlow allows the user to generate videos that respect the motion of the input video as well as the text prompt. This is implemented through an optical flow estimation model applied on the input video, which is then fed to a trainable optical flow encoder. The output feature maps are then injected into the text-to-video backbone model. We perform quantitative, qualitative and user preference studies to show that OnlyFlow positively compares to state-of-the-art methods on a wide range of tasks, even though OnlyFlow was not specifically trained for such tasks. OnlyFlow thus constitutes a versatile, lightweight yet efficient method for controlling motion in text-to-video generation. Models and code will be made available on GitHub and HuggingFace.

We report on the dynamics of a soliton propagating on the surface of a fluid in a 4-m-long canal with a random or periodic bottom topography. Using a full space-and-time resolved wavefield measurement, we evidence, for the first time experimentally, how the soliton is affected by the disorder, in the context of Anderson localization, and how localization depends on nonlinearity. For weak soliton amplitudes, the localization length is found in quantitative agreement with a linear shallow-water theory. For higher amplitudes, this spatial attenuation of the soliton amplitude is found to be enhanced. Behind the leading soliton slowed down by the topography, different experimentally unreported dynamics occur: Fission into backward and forward nondispersive pulses for the periodic case, and scattering into dispersive waves for the random case. Our findings open doors to potential applications regarding ocean coastal protection against large-amplitude waves., Comment: in press in Physical Review Letters (2024)

Context: Timing of pulsar PSR J0337+1715 provides a unique opportunity to test the strong equivalence principle (SEP) with a strongly self-gravitating object. This is due to its unique situation in a triple stellar system with two white dwarfs. Aims: Our previous study suggested the presence of a strong low-frequency signal in the timing residuals. We set out to model it on a longer dataset in order to determine its nature and improve accuracy. Methods: Three models are considered: chromatic or achromatic red-noise, and a small planet in a hierarchical orbit with the triple stellar system. These models are implemented in our numerical timing model. We perform Bayesian inference of posterior distributions. Best fits are compared using information-theoretic criteria. Results: Chromatic red noise from dispersion-measure variations is ruled out. Achromatic red noise or a planet in keplerian orbit provide the best fits. If it is red noise then it appears exceptionally strong. Assuming the presence of a planet, we obtain a marginal detection of mutual interactions which allows us to constrain its mass to $\sim 0.5 M_{\rm Moon}$ as well as its inclination. The latter is intriguingly coincident with a Kozai resonance. We show that a longer observation span will ultimately lead to a clear signature of the planet model due to its mutual interactions with the triple system. We produce new limits on SEP violation: $|\Delta| < 1.5\cdot 10^{-6}$ or $|\Delta| < 2.3\cdot 10^{-6}$ at 95\% confidence level under the planet or red-noise hypothesis, respectively. This model dependence emphasises the need for additional data and model selection. As a by-product, we estimate a rather low supernova kick velocity of $\sim 110-125 \rm km/s$, strengthening the idea that it is a necessary condition for the formation of pulsar triple systems., Comment: Astronomy and Astrophysics - A\&A, In press

Inertial waves in convective regions of stars exhibit topological properties linked to a Chern number of 1. The first of these is a unique, unidirectional, prograde oscillation mode within the cavity, which propagates at arbitrarily low frequencies for moderate azimuthal wavenumbers. The second one are phase singularities around which the phase winds in Fourier space, with winding numbers of $\pm 1$ depending on the hemisphere. Phase winding is a collective effect over waves propagating in all directions that is strongly robust to noise. This suggests a topology-based method for wave detection in noisy observational data., Comment: 8 pages, 9 figures. Accepted for publication in Physical Review Research

Spectroscopy of Earth-like exoplanets and ultra-faint galaxies are priority science cases for the coming decades. Here, broadband source flux rates are measured in photons per square meter per hour, imposing extreme demands on detector performance, including dark currents lower than \mbox{1 e-/pixel/kilosecond}, read noise less than \mbox{1 e-/pixel/frame}, and large formats. There are currently no infrared detectors that meet these requirements. The University of Hawai'i and industrial partners are developing one promising technology, linear mode avalanche photodiodes (LmAPDs), which is on track to meet the above-mentioned requirements. We present progress towards developing a science-grade, megapixel format linear-mode avalanche photodiode array for low background shortwave (1 - 2.4 um) infrared astronomy. Our latest results show outstanding performance, with dark current \textless 1e-4 electrons/pixel/second and read noise reducing by 30\% per volt of bias, reaching less than 1e-/pixel/frame in correlated double-sampling, and able to average down to $\sim$0.3 e-/pixel/frame when using multiple non-destructive reads. We present some on-sky data as well as comment on prospects for photon number resolving capability., Comment: SPIE Astronomical Telescopes + Instrumentation 2024, PROC#13103-25. arXiv admin note: text overlap with arXiv:2208.11834

The SMT (Satisfiability Modulo Theories) theory of arrays is well-established and widely used, with variousdecision procedures and extensions developed for it. However, recent works suggest that developing tailoredreasoning for some theories, such as sequences and strings, is more efficient than reasoning over them throughaxiomatization over the theory of arrays. In this paper, we are interested in reasoning over n-indexed sequences asthey are found in some programming languages, such as Ada. We propose an SMT theory of n-indexed sequencesand explore different ways to represent and reason over n-indexed sequences using existing theories, as well astailored calculi for the theory., Comment: 22nd International Workshop on Satisfiability Modulo Theories, Jul 2024, Montreal, Canada, France

Second-order optimization has been shown to accelerate the training of deep neural networks in many applications, often yielding faster progress per iteration on the training loss compared to first-order optimizers. However, the generalization properties of second-order methods are still being debated. Theoretical investigations have proved difficult to carry out outside the tractable settings of heavily simplified model classes -- thus, the relevance of existing theories to practical deep learning applications remains unclear. Similarly, empirical studies in large-scale models and real datasets are significantly confounded by the necessity to approximate second-order updates in practice. It is often unclear whether the observed generalization behaviour arises specifically from the second-order nature of the parameter updates, or instead reflects the specific structured (e.g.\ Kronecker) approximations used or any damping-based interpolation towards first-order updates. Here, we show for the first time that exact Gauss-Newton (GN) updates take on a tractable form in a class of deep reversible architectures that are sufficiently expressive to be meaningfully applied to common benchmark datasets. We exploit this novel setting to study the training and generalization properties of the GN optimizer. We find that exact GN generalizes poorly. In the mini-batch training setting, this manifests as rapidly saturating progress even on the \emph{training} loss, with parameter updates found to overfit each mini-batchatch without producing the features that would support generalization to other mini-batches. We show that our experiments run in the ``lazy'' regime, in which the neural tangent kernel (NTK) changes very little during the course of training. This behaviour is associated with having no significant changes in neural representations, explaining the lack of generalization., Comment: Accepted at NeurIPS 2024

Aims. Recently, both the presence of multiple stellar chemo-kinematic components and rotation in the Sculptor dwarf spheroidal galaxy have been put into question. Therefore, we re-examine the chemo-kinematic properties of this galaxy making use of the best spectroscopic data-set available containing both line-of-sight velocities and metallicities of individual stars. Methods. We carry out a detailed, quantitative analysis on the recent spectroscopic data-set from Tolstoy et al. (2023) that contains high precision velocities and metallicities for 1339 members of Sculptor. In particular, we assess whether Sculptor is best represented by a single stellar population with a negative metallicity gradient or by the super-position of two or more components with different mean metallicity, spatial distribution and kinematic properties. For this analysis, we also include the incompleteness of the spectroscopic data-set. Results. We find that Sculptor is better described by a two-populations model than by a single-population model with a metallicity gradient. Moreover, given the assumptions of the current modeling, we find evidence of a third population, composed of few stars, that is more extended and metal-poor than the two other populations. This very metal-poor group of stars shows a shift of around 15 km/s in its average l.o.s. velocity (vlos) with respect to the rest of the galaxy. We discuss several possible origins for this new population, finding a minor merger as the most likely one. We also find a vlos gradient of 4.0 +1.5 -1.5 km s-1 deg-1 but its statistical evidence is inconclusive and, moreover, its detection is partially driven by the group of stars with off-set velocities., Comment: 17 pages, 9 figures, Accepted for publication in Astronomy & Astrophysics

Detecting and tracking emerging trends and weak signals in large, evolving text corpora is vital for applications such as monitoring scientific literature, managing brand reputation, surveilling critical infrastructure and more generally to any kind of text-based event detection. Existing solutions often fail to capture the nuanced context or dynamically track evolving patterns over time. BERTrend, a novel method, addresses these limitations using neural topic modeling in an online setting. It introduces a new metric to quantify topic popularity over time by considering both the number of documents and update frequency. This metric classifies topics as noise, weak, or strong signals, flagging emerging, rapidly growing topics for further investigation. Experimentation on two large real-world datasets demonstrates BERTrend's ability to accurately detect and track meaningful weak signals while filtering out noise, offering a comprehensive solution for monitoring emerging trends in large-scale, evolving text corpora. The method can also be used for retrospective analysis of past events. In addition, the use of Large Language Models together with BERTrend offers efficient means for the interpretability of trends of events., Comment: 17 pages, 12 figures, FuturED 2024: Workshop on Future of Event Detection (CoLocated with EMNLP 2024)

We compare methods for signal classification applied to voltage traces from transition edge sensors (TES) which are photon-number resolving detectors fundamental for accessing quantum advantages in information processing, communication and metrology. We quantify the impact of numerical analysis on the distinction of such signals. Furthermore, we explore dimensionality reduction techniques to create interpretable and precise photon number embeddings. We demonstrate that the preservation of local data structures of some nonlinear methods is an accurate way to achieve unsupervised classification of TES traces. We do so by considering a confidence metric that quantifies the overlap of the photon number clusters inside a latent space. Furthermore, we demonstrate that for our dataset previous methods such as the signal's area and principal component analysis can resolve up to 16 photons with confidence above $90\%$ while nonlinear techniques can resolve up to 21 with the same confidence threshold. Also, we showcase implementations of neural networks to leverage information within local structures, aiming to increase confidence in assigning photon numbers. Finally, we demonstrate the advantage of some nonlinear methods to detect and remove outlier signals., Comment: Code is publicly available at https://github.com/polyquantique/Photon-Number-Classification. Data is publicly available at https://doi.org/10.5281/zenodo.14042152 and https://doi.org/10.5281/zenodo.14101974

In real-world applications where confidence is key, like autonomous driving, the accurate detection and appropriate handling of classes differing from those used during training are crucial. Despite the proposal of various unknown object detection approaches, we have observed widespread inconsistencies among them regarding the datasets, metrics, and scenarios used, alongside a notable absence of a clear definition for unknown objects, which hampers meaningful evaluation. To counter these issues, we introduce two benchmarks: a unified VOC-COCO evaluation, and the new OpenImagesRoad benchmark which provides clear hierarchical object definition besides new evaluation metrics. Complementing the benchmark, we exploit recent self-supervised Vision Transformers performance, to improve pseudo-labeling-based OpenSet Object Detection (OSOD), through OW-DETR++. State-of-the-art methods are extensively evaluated on the proposed benchmarks. This study provides a clear problem definition, ensures consistent evaluations, and draws new conclusions about effectiveness of OSOD strategies., Comment: Accepted at ECCV 2024 Workshop: "The 3rd Workshop for Out-of-Distribution Generalization in Computer Vision Foundation Models"

We present a framework for learning a single policy capable of producing all quadruped gaits and transitions. The framework consists of a policy trained with deep reinforcement learning (DRL) to modulate the parameters of a system of abstract oscillators (i.e. Central Pattern Generator), whose output is mapped to joint commands through a pattern formation layer that sets the gait style, i.e. body height, swing foot ground clearance height, and foot offset. Different gaits are formed by changing the coupling between different oscillators, which can be instantaneously selected at any velocity by a user. With this framework, we systematically investigate which gait should be used at which velocity, and when gait transitions should occur from a Cost of Transport (COT), i.e. energy-efficiency, point of view. Additionally, we note how gait style changes as a function of locomotion speed for each gait to keep the most energy-efficient locomotion. While the currently most popular gait (trot) does not result in the lowest COT, we find that considering different co-dependent metrics such as mean base velocity and joint acceleration result in different `optimal' gaits than those that minimize COT. We deploy our controller in various hardware experiments, showing all 9 typical quadruped animal gaits, and demonstrate generalizability to unseen gaits during training, and robustness to leg failures. Video results can be found at https://youtu.be/OLoWSX_R868.

Training robot policies in simulation is becoming increasingly popular; nevertheless, a precise, reliable, and easy-to-use tactile simulator for contact-rich manipulation tasks is still missing. To close this gap, we develop TacEx -- a modular tactile simulation framework. We embed a state-of-the-art soft-body simulator for contacts named GIPC and vision-based tactile simulators Taxim and FOTS into Isaac Sim to achieve robust and plausible simulation of the visuotactile sensor GelSight Mini. We implement several Isaac Lab environments for Reinforcement Learning (RL) leveraging our TacEx simulation, including object pushing, lifting, and pole balancing. We validate that the simulation is stable and that the high-dimensional observations, such as the gel deformation and the RGB images from the GelSight camera, can be used for training. The code, videos, and additional results will be released online https://sites.google.com/view/tacex., Comment: 11 pages, accepted at "CoRL Workshop on Learning Robot Fine and Dexterous Manipulation: Perception and Control"

New physics and systematic errors can lead to deviations between the models used to analyze gravitational wave data and the actual signal. Such deviations will generally be correlated between detectors and manifest differently across the gravitational wave source parameter space. The previously introduced \ttt{SCoRe} framework uses these features to distinguish these deviations from noise and extract physical information from their source-dependent variation. In this work, we further analyze the hierarchical component of the method -- we include the expected dependence of the deviations on the source parameters into the inference process, obtaining more physically informative results. As a specific example, we study a deviation that scales as a power law of the mass scale of black hole binaries -- as, for example, in Effective Field Theory of gravity. We show how the signal-to-noise ratio of the cross-correlated residual power can be used to recover the power-law index. We demonstrate how both the distribution in source and deviation strength determine which region of source parameter space influences the inference most. Finally, we forecast the constraint on the power law index for a network of two Cosmic Explorer-like detectors with a year of observation period., Comment: 18 pages, 10 figures

This document is comprised of a collection of updated preliminary parameters for the key parts of the muon collider. The updated preliminary parameters follow on from the October 2023 Tentative Parameters Report. Particular attention has been given to regions of the facility that are believed to hold greater technical uncertainty in their design and that have a strong impact on the cost and power consumption of the facility. The data is collected from a collaborative spreadsheet and transferred to overleaf.

The design space for the fabrication of photonic lanterns is vast, which allows for a great diversity of designs but also complicates finding the right parameters for a given component. Coupled-mode theory is a great tool to sort through the parameter space and find the right designs for a multiplicity of photonic lantern types, including conventional, mode-selective, and anything in-between. It correctly models the behavior of the fabricated component, but it also simplifies the task of finding the right fabrication sequence by computing the all-important adiabatic criteria, which dictate which transverse modes will couple, and at which step of the recipe. Here, using coupled-mode theory, we predict and test experimentally the behavior of four types of 3X1 photonic lanterns ranging from conventional to mode-specific and hybrid configurations, and explain mode-coupling, operation bandwidth, and excess loss mechanisms., Comment: 8 pages, 6 figures, This work has been submitted to the IEEE for possible publication

Choices in the semantics and the signature of a theory are integral in determining how the theory is used and how challenging it is to reason over it. Our interest in this paper lies in the SMT theory of sequences. Various versions of it exist in the literature and in state-of-the-art SMT solvers, but it has not yet been standardized in the SMT-LIB. We reflect on its existing variants, and we define a set of theory design criteria to help determine what makes one variant of a theory better than another. The criteria we define can be used to appraise theory proposals for other theories as well. Based on these criteria, we propose a series of changes to the SMT theory of sequences as a contribution to the discussion regarding its standardization.

The concept of 6G distributed integrated sensing and communications (DISAC) builds upon the functionality of integrated sensing and communications (ISAC) by integrating distributed architectures, significantly enhancing both sensing and communication coverage and performance. In 6G DISAC systems, tracking target trajectories requires base stations (BSs) to hand over their tracked targets to neighboring BSs. Determining what information to share, where, how, and when is critical to effective handover. This paper addresses the target handover challenge in DISAC systems and introduces a method enabling BSs to share essential target trajectory information at appropriate time steps, facilitating seamless handovers to other BSs. The target tracking problem is tackled using the standard trajectory Poisson multi-Bernoulli mixture (TPMBM) filter, enhanced with the proposed handover algorithm. Simulation results confirm the effectiveness of the implemented tracking solution., Comment: Submitted to ICC 2025

We report the results of a spectroscopic survey of candidate T subdwarfs identified by the Backyard Worlds: Planet 9 program. Near-infrared spectra of 31 sources with red $J-W2$ colors and large $J$-band reduced proper motions show varying signatures of subsolar metallicity, including strong collision-induced H$_2$ absorption, obscured methane and water features, and weak K I absorption. These metallicity signatures are supported by spectral model fits and 3D velocities, indicating thick disk and halo population membership for several sources. We identify three new metal-poor T subdwarfs ([M/H] $\lesssim$ $-$0.5), CWISE J062316.19+071505.6, WISEA J152443.14$-$262001.8, and CWISE J211250.11-052925.2; and 19 new "mild" subdwarfs with modest metal deficiency ([M/H] $\lesssim$ $-$0.25). We also identify three metal-rich brown dwarfs with thick disk kinematics. We provide kinematic evidence that the extreme L subdwarf 2MASS J053253.46+824646.5 and the mild T subdwarf CWISE J113010.07+313944.7 may be part of the Thamnos population, while the T subdwarf CWISE J155349.96+693355.2 may be part of the Helmi stream. We define a metallicity classification system for T dwarfs that adds mild subdwarfs (d/sdT), subdwarfs (sdT), and extreme subdwarfs (esdT) to the existing dwarf sequence. We also define a metallicity spectral index that correlates with metallicities inferred from spectral model fits and iron abundances from stellar primaries of benchmark T dwarf companions. This expansion of the T dwarf classification system supports investigations of ancient, metal-poor brown dwarfs now being uncovered in deep imaging and spectroscopic surveys., Comment: 82 pages, 19 figures, accepted to ApJS

We construct a novel orientifold of type IIB string theory that breaks all supersymmetries. It is a closed string theory without open sector and it can be understood as a Scherk-Schwarz deformation in which supersymmetry is restored at infinite radius. We conjecture that it is realised in F-theory as a compactification on a freely acting orbifold that acts as the reflection on the elliptic fibre. The SL(2,Z) selfduality is manifest in the F-theory formulation. We construct explicitly the D-branes in this model and find that stable D-branes match the geometric prediction in M-theory. This theory has the salient feature that the O-planes couple only to the massive twisted states of the theory. We call them twisted O-planes. We describe supersymmetric examples of such twisted O-planes and argue that they are similar in nature to combinations of O+ and O- planes with vanishing total charge., Comment: 51 pages

Given an operator $\phi:X\rightarrow Y$ between Banach spaces, we consider its tensor powers $\phi^{\otimes k}$ as operators from the $k$-fold injective tensor product of $X$ to the $k$-fold projective tensor product of $Y$. We show that after taking the $k$th root, the operator norm of $\phi^{\otimes k}$ converges to the $2$-dominated norm $\gamma^*_2(\phi)$, one of the standard operator ideal norms., Comment: 11 pages, no figures

An appropriate description of the lithiation/delithiation dynamics in bi-phasic primary cathode particles of Li-ion batteries requires an accurate treatment of the conditions holding at the interface between the particle and the surrounding liquid electrolyte. We propose a phase field model based on the Allen-Cahn approach within which the particle-electrolyte interface is smooth (Smoothed Boundary Method - SBM), in order to simulate arbitrarily shaped particles. Surface terms are added to the evolution equations, and SBM calculations are compared with benchmark simulations for which the boundary conditions are explicitly imposed at the borders of the calculation domain. We find that strengthening factors for the surface terms are needed in order to achieve the desired conditions for the phase and elastic fields, and to correctly reproduce the level of stress within the particle. This reveals crucial in the context of Li-ion batteries for an accurate prediction of the Li insertion/extraction rate and the Li diffusion behavior. We perform also a simulation under potentio-static conditions with a full coupling of the different physical processes at play. It illustrates the applicability of our approach and demonstrates the capabilities of the SBM for a simulation of lithiation/delithiation dynamics with coupled electrochemistry and mechanics., Comment: 17 pages, 15 figures

We present a new analytical model for the attenuation to Epoch of Reionization (EoR) galaxies by proximate neutral hydrogen gas. Many galaxy spectra in the EoR taken by JWST have shown a flux deficit at wavelengths just redward of the Lyman break, and this has been regarded as resulting from Ly$\alpha$ damping wing absorption by the increasing amount of neutral hydrogen in the line-of-sight. However, previous attenuation models for the intergalactic medium (IGM) commonly used in photometric redshift template-fitting codes assume that the Lyman break is rather sharp, which leads to systematic overestimation of photometric redshifts at $z>7$. In this letter, we build and empirically calibrate a new attenuation model that takes the increased Ly$\alpha$ damping wing absorption into account. Our model consists of the canonical IGM absorption and an additional absorption component due to dense neutral hydrogen gas clouds proximate to the galaxy, and we derive the redshift evolution of HI column density of the proximate clouds by calibrating the model using CANUCS JWST observations. The resulting total transmission curve resolves the photometric redshift bias at $z>7$, an improvement that is robust to choice of template-fitting code, template set, and photometric catalog used. The new attenuation model can be easily implemented in existing template-fitting codes, and significantly improves the photometric redshift performance in the EoR., Comment: Submitted to ApJL. Comments are welcome. The new attenuation model has been implemented in the latest version EAzY (v0.8 or later) and Phosphoros (v2.0.2 or later)

Lifelong Multi-Agent Path Finding (LMAPF) is a variant of MAPF where agents are continually assigned new goals, necessitating frequent re-planning to accommodate these dynamic changes. Recently, this field has embraced learning-based methods, which reactively generate single-step actions based on individual local observations. However, it is still challenging for them to match the performance of the best search-based algorithms, especially in large-scale settings. This work proposes an imitation-learning-based LMAPF solver that introduces a novel communication module and systematic single-step collision resolution and global guidance techniques. Our proposed solver, Scalable Imitation Learning for LMAPF (SILLM), inherits the fast reasoning speed of learning-based methods and the high solution quality of search-based methods with the help of modern GPUs. Across six large-scale maps with up to 10,000 agents and varying obstacle structures, SILLM surpasses the best learning- and search-based baselines, achieving average throughput improvements of 137.7% and 16.0%, respectively. Furthermore, SILLM also beats the winning solution of the 2023 League of Robot Runners, an international LMAPF competition sponsored by Amazon Robotics. Finally, we validated SILLM with 10 real robots and 100 virtual robots in a mockup warehouse environment., Comment: Submitted to ICRA 2025

A current trend in HPC systems is the utilization of architectures with SIMD or vector extensions to exploit data parallelism. There are several ways to take advantage of such modern vector architectures, each with a different impact on the code and its portability. For example, the use of intrinsics, guided vectorization via pragmas, or compiler autovectorization. Our objectives are to maximize vectorization efficiency and minimize code specialization. To achieve these objectives, we rely on compiler autovectorization. We leverage a set of hardware and software tools that allow us to analyze in detail where autovectorization is suboptimal. Thus, we apply an iterative methodology that allows us to incrementally improve the efficient use of the underlying hardware. In this paper, we apply this methodology to a CFD production code. We evaluate the performance on an innovative configurable platform powered by a RISC-V core coupled with a wide vector unit capable of operating with up to 256 double precision elements. Following the vectorization process, we demonstrate a single-core speedup of 7.6$\times$ compared to its scalar implementation. Furthermore, we show that code portability is not compromised, as our solution continues to exhibit performance benefits, or at the very least, no drawbacks, on other HPC architectures such as Intel x86 and NEC SX-Aurora., Comment: Main track paper, presented at IPDPS 2024

Self-interested individuals often fail to cooperate, posing a fundamental challenge for multi-agent learning. How can we achieve cooperation among self-interested, independent learning agents? Promising recent work has shown that in certain tasks cooperation can be established between learning-aware agents who model the learning dynamics of each other. Here, we present the first unbiased, higher-derivative-free policy gradient algorithm for learning-aware reinforcement learning, which takes into account that other agents are themselves learning through trial and error based on multiple noisy trials. We then leverage efficient sequence models to condition behavior on long observation histories that contain traces of the learning dynamics of other agents. Training long-context policies with our algorithm leads to cooperative behavior and high returns on standard social dilemmas, including a challenging environment where temporally-extended action coordination is required. Finally, we derive from the iterated prisoner's dilemma a novel explanation for how and when cooperation arises among self-interested learning-aware agents.

Assuming X is a random vector and A a non-invertible matrix, one sometimes need to perform inference while only having access to samples of Y = AX. The corresponding likelihood is typically intractable. One may still be able to perform exact Bayesian inference using a pseudo-marginal sampler, but this requires an unbiased estimator of the intractable likelihood. We propose saddlepoint Monte Carlo, a method for obtaining an unbiased estimate of the density of Y with very low variance, for any model belonging to an exponential family. Our method relies on importance sampling of the characteristic function, with insights brought by the standard saddlepoint approximation scheme with exponential tilting. We show that saddlepoint Monte Carlo makes it possible to perform exact inference on particularly challenging problems and datasets. We focus on the ecological inference problem, where one observes only aggregates at a fine level. We present in particular a study of the carryover of votes between the two rounds of various French elections, using the finest available data (number of votes for each candidate in about 60,000 polling stations over most of the French territory). We show that existing, popular approximate methods for ecological inference can lead to substantial bias, which saddlepoint Monte Carlo is immune from. We also present original results for the 2024 legislative elections on political centre-to-left and left-to-centre conversion rates when the far-right is present in the second round. Finally, we discuss other exciting applications for saddlepoint Monte Carlo, such as dealing with aggregate data in privacy or inverse problems., Comment: 27 pages, 9 figures, 3 tables

With the explosion in popularity of the Rust programming language, a wealth of tools have recently been developed to analyze, verify, and test Rust programs. Alas, the Rust ecosystem remains relatively young, meaning that every one of these tools has had to re-implement difficult, time-consuming machinery to interface with the Rust compiler and its cargo build system, to hook into the Rust compiler's internal representation, and to expose an abstract syntax tree (AST) that is suitable for analysis rather than optimized for efficiency. We address this missing building block of the Rust ecosystem, and propose Charon, an analysis framework for Rust. Charon acts as a swiss-army knife for analyzing Rust programs, and deals with all of the tedium above, providing clients with a clean, stable AST that can serve as the foundation of many analyses. We demonstrate the usefulness of Charon through a series of case studies, ranging from a Rust verification framework (Aeneas), a compiler from Rust to C (Eurydice), and a novel taint-checker for cryptographic code. To drive the point home, we also re-implement a popular existing analysis (Rudra), and show that it can be replicated by leveraging the Charon framework.

In this work, we present conception and study of gallium nitride (GaN) nanostructures on a gallium arsenide (GaAs) substrate with (111)A orientation. The nanostructures were designed by GaN droplet epitaxy and studied in-situ by X-ray photoelectron spectroscopy and ex-situ by atomic force microscopy, scanning electron microscopy and transmission electron microscopy. These studies were coupled with kinetic Monte Carlo simulations to precisely understand the phenomena occurring during the nitridation and to find the optimum conditions for complete nitridation of gallium droplets. The HRTEM observation showed a cubic (zinc blende) crystal structure of the GaN nanodots for a nitridation at 300{\deg}C. Ramping the temperature from 100{\deg}C to 350{\deg}C during droplet nitridation enabled to obtain a very high density (>1011cm-2) of GaN nanodots with the zinc blende crystallinity.

Autonomous robot exploration requires a robot to efficiently explore and map unknown environments. Compared to conventional methods that can only optimize paths based on the current robot belief, learning-based methods show the potential to achieve improved performance by drawing on past experiences to reason about unknown areas. In this paper, we propose DARE, a novel generative approach that leverages diffusion models trained on expert demonstrations, which can explicitly generate an exploration path through one-time inference. We build DARE upon an attention-based encoder and a diffusion policy model, and introduce ground truth optimal demonstrations for training to learn better patterns for exploration. The trained planner can reason about the partial belief to recognize the potential structure in unknown areas and consider these areas during path planning. Our experiments demonstrate that DARE achieves on-par performance with both conventional and learning-based state-of-the-art exploration planners, as well as good generalizability in both simulations and real-life scenarios.