HAL-Polytechnique
Not a member yet
51406 research outputs found
Sort by
Sparse recovery of Diffusion Dynamics: Handling High-Dimensionality in Repeated Short Trajectories
No full textViewed as systems of interacting particles, high-dimensional stochastic differential equations encode complex interaction structures within their drift component. We propose a novel approach to estimate this drift from independent high-frequency trajectory data observed over a short time horizon. Each trajectory is modelled as the solution of a Brownian-driven stochastic differential equation, while the number of time points within each path tends to infinity. We further assume that the drift function governing the dynamics can be expressed as a linear combination of a growing number of Lipschitz basis functions. To promote accurate recovery of the underlying dynamics under sparsity constraints, we propose a Lasso-regularised likelihood criterion. Under suitable regularity conditions, we establish convergence rates for the resulting estimator and emphasise how they depend on the dimensional parameters of the problem, in particular on the number of observed trajectories. We assess the performance of the estimator on synthetic datasets, both from an estimation and a generative perspective. Finally, we illustrate the practical relevance of the approach on a real-world climate dataset, highlighting its ability to perform variable selection
Leveraging Contrastive Learning for a Similarity-Guided Tampered Document Data Generation Pipeline
No full textDetecting tampered text in document images is a challenging task due to data scarcity. To address this, previous work has attempted to generate tampered documents using rule-based methods. However, the resulting documents often suffer from limited variety and poor visual quality, typically leaving highly visible artifacts that are rarely observed in real-world manipulations. This undermines the model's ability to learn robust, generalizable features and results in poor performance on real-world data. Motivated by this discrepancy, we propose a novel method for generating high-quality tampered document images. We first train an auxiliary network to compare text crops, leveraging contrastive learning with a novel strategy for defining positive pairs and their corresponding negatives. We also train a second auxiliary network to evaluate whether a crop tightly encloses the intended characters, without cutting off parts of characters or including parts of adjacent ones. Using a carefully designed generation pipeline that leverages both networks, we introduce a framework capable of producing diverse, high-quality tampered document images. We assess the effectiveness of our data generation pipeline by training multiple models on datasets derived from the same source images, generated using our method and existing approaches, under identical training protocols. Evaluating these models on various open-source datasets shows that our pipeline yields consistent performance improvements across architectures and datasets
The injective norm of CSS quantum error-correcting codes
No full textIn this paper, we compute the injective norm - a.k.a. geometric entanglement - of standard basis states of CSS quantum error-correcting codes. The injective norm of a quantum state is a measure of genuine multipartite entanglement. Computing this measure is generically NP-hard. However, it has been computed exactly in condensed-matter theory - notably in the context of topological phases - for the Kitaev code and its extensions, in works by Orús and collaborators. We extend these results to all CSS codes and thereby obtain the injective norm for a nontrivial, infinite family of quantum states. In doing so, we uncover an interesting connection to matroid theory and Edmonds' intersection theorem
Observation of the rare baryonic decay and measurement of its weak decay parameter
No full textInternational audienceThe first observation of the decay is presented, using proton-proton collision data collected by the LHCb experiment between 2016 and 2018 at a center-of-mass energy of 13 TeV, corresponding to an integrated luminosity of . The signal significance exceeds seven standard deviations. Using the decay as a normalization channel, the branching fraction is measured and combined with previous LHCb results based on data collected at 7 and 8 TeV in 2011 and 2012, yielding , where the first uncertainty is statistical, the second is systematic, and the third comes from the uncertainty on the branching fraction of the normalization channel. The weak decay parameter is measured to be , indicating the presence of comparable S-wave and P-wave decay amplitudes
Efficient and Scalable Search for Statistics
No full textInternational audienceInformed public debate needs high-quality data. In this context, high-quality statistical data sources are a valuable category of reference information based on which a claim can be checked. To facilitate the work of journalists or other fact-checkers, users’ questions about a specific claim should be automatically answered based on statistical tables. This task is complicated by the large number, size, and variety of statistical datasets.We introduce the statistical table discovery problem (STD, in short), which aims, given a natural language question and a set of statistic datasets (multidimensional tables), to find the tables most relevant for the question. We then describe STAR, an algorithm for solving the STD problem. Unlike existing table discovery (TD) solutions aimed at relational tables, STAR is devised specifically for multidimensional ones. Further, STAR treats the space and time dimensions of statistical datasets separately. We experimentally show that these features, together, make STAR outperform state-of-the-art TD systems adapted to the STD problem, in terms of scalability, search quality, preprocessing and question answering time
Micro-Poro-Mechanical Modeling of The Lung Parenchyma: Theoretical Modeling and Parameters Identification
No full textInternational audienceMicro-poro-mechanical approaches can be employed to simulate the behavior of porous media, such as lung parenchyma, with respect to their microscopic morphological and mechanical features. In this work, we propose a general micromechanical framework to describe the behavior of a porous hyperelastic material in large strains, including surface tension, and adapt its parameters to reproduce lung parenchyma behavior. We illustrate the method on a 2D periodic microstructure. The modeling framework is adaptable to any microstructure and any combination of stress, strain and pressure loadings.The identification of the model parameters in the context of lung parenchyma, based on existing experimental morphological and pressure-volume data, is performed sequentially. 12 parameters related to morphology, alveolar wall constitutive behavior and surface tension are calibrated to reproduce pressure-volume curves in various conditions, for a porosity in the unloaded state set to Φf0 =63%. The calibrated alveolar diameter is Dalv = 54 μm. The identifiability of the Neohookean and Ogden-Ciarlet-Geymonat hyperelastic potential parameters is studied; their values are β1 = 94.3 Pa, β2 = 16.9 Pa, β3 = 619 Pa and α = 3.154. The hysteretic response of lung to pressure is reproduced thanks to the formulation of a surface-dependent surface tension. This work paves the way for a better understanding of the relationship between microscopic features and the macroscopic response of lung, in healthy and pathological conditions. Further experimental investigations could help confirming the ranges of parameters obtained in this study
Interhemispheric and Latitudinal Variability of Ionospheric Disturbances During the 19-20 December 2015 Geomagnetic Storm: Insights from the South American Sector
No full textInternational audienceThis study investigates the response of the South American ionosphere to the December 19-20, 2015, geomagnetic storm using multi-instrument satellite and ground-based data. An interplanetary shock triggered prompt penetration electric fields (PPEFs) during the storm's initial phase, initiating rapid DP2 currents and significant equatorial electrojet (EEJ) fluctuations, particularly in Jicamarca, Peru. Our analysis shows that dayside field-aligned currents (FACs) are the primary drivers of the observed DP2 fluctuations. The South Atlantic Magnetic Anomaly (SAMA) also induced a westward geomagnetic gradient-induced current at the equatorward boundary, which suppressed the EEJ current at São Luiz during the magnetically quiet period. During the storm's main phase, eastward PPEFs enhanced ionospheric irregularities at several Global Positioning Satellite (GPS) stations: Sao Luiz (14.8%), Cuiaba (11.4%), and Jatai (15.5%) in Brazil; Tucuma (33.6%) and Rio Grande (33.2%) in Argentina; and all Peruvian GPS stations (5.8 -57.6%), with Arequipa showing the highest percentage. Conversely, irregularities were inhibited at Eusebio (-31.5%), São José dos Campos (-11.8%), and Campo Grande (-4.8%) in Brazil, and Rosario (-9.4%) and Villegas (-0.8%) in Argentina, relative to a magnetically quiet day, which was attributed to the interplay between westward disturbance dynamo electric fields (DDEF) and PPEF. Joule heating, peaking at 14:35 UT on 20 December drove equatorward disturbance winds that generated the DDEF and modulated irregularities during the recovery phase. Westward DDEF suppressed post-sunset irregularities across most Brazilian and Argentine stations (-50.2% to -10.1%), but slightly enhanced post-midnight ionospheric irregularities at Cuiabá (+4.3%) and Tucuma (+21.1%). All Peruvian stations recorded post-midnight enhancements (+14.8% to +136.7%), with Arequipa showing the highest increase. We quantified the delayed magnetic response of the disturbance dynamo (Ddyn) relative to Joule heating, revealing propagation delays of approximately 6, 8.5, and 8.2 h in Brazil, Argentina, and Peru, respectively. Ddyn exhibited strong spatiotemporal variability, including spatial anomalies associated with the SAMA. The Peruvian sector dominated the Ddyn power (59.0%), followed by Brazil (35.1%) and Argentina (5.9%). Evidence of interhemispheric FAC asymmetry, attributed to seasonal variability, was also observed. These findings significantly advance our understanding of storm-time lowlatitude electrodynamics and their regional variability
A MeerKAT search for persistent radio sources towards twenty-five localised Fast Radio Bursts
No full textInternational audienceThe discovery of persistent radio sources (PRSs) associated with repeating fast radio bursts (FRBs) has shed light on the immediate environments and possible progenitors of these FRBs. The confirmed PRSs may support the theory that FRB progenitors are compact central engines, whilst the non-detections suggest diversity of FRB's local environment. We perform a subarcsecond-resolution MeerKAT search at 1.28 GHz on 25 well-localised FRB positions provided by ASKAP and MeerTRAP. We detect 14 radio sources and provide flux upper limits for 12 non-detections (both these numbers include a source that was detected during two epochs of observation, and not detected during one epoch, adding up to 26). One radio source shows variability as seen in flux variations over three epochs of observation. Archival optical data reveal excesses in the direction of 13 detected radio sources. Similarly for four sources in the X-ray band, with one possibly being a high-energy signature of a radio galaxy core. Since we cannot definitively classify our detected radio sources as PRSs, future high-resolution observations with e-MERLIN will be required to resolve the radio emission and pronounce on the presence of compact PRSs associated with the 14 detected sources presented here
Polynomialization of ordinary differential equations given by straight-line programs
No full textGiven a system of ordinary differential equations represented by a straight-line program, we show how to compute an equivalent ordinary differential system represented by a straight-line program which only uses ring operations. Under mild assumptions, our method essentially runs in linear time
Critical and multicritical Lee-Yang fixed points in the local potential approximation
No full textInternational audienceThe multicritical generalizations of the Lee-Yang universality class arise as renormalization-group fixed points of scalar field theories with complex interaction, , just below their upper critical dimension. It has been recently conjectured that their continuation to two dimensions corresponds to the non-unitary conformal minimal models . Motivated by that, we revisit the functional renormalization group approach to complex -symmetric scalar field theories in the Local Potential Approximation, without or with wavefunction renormalization (LPA and LPA' respectively), aiming to explore the fate of the theories from their upper critical dimension to two dimensions. The fixed points are identified using a perturbative expansion of the functional fixed-point equation near their upper critical dimensions, and they are followed to lower dimensions by numerical integration of the full equation. A peculiar feature of the complex -symmetric potentials is that the fixed points are characterized by real but negative anomalous dimensions , and in low dimension , this can lead to a change of sign of the scaling dimensions , thus requiring a novel analysis of the analytical properties of the functional fixed-point equations. We are able to follow the Lee-Yang universality class () down to two dimensions, and numerically determine the scaling dimension of the fundamental field as a function of . On the other hand, within the LPA', multicritical Lee-Yang fixed points with cannot be continued to due to the existence of unexpected non-perturbative fixed points that annihilate with the fixed points