HAL-INSA Toulouse
Not a member yet
34325 research outputs found
Sort by
Thermal Hall conductivity in the strongest cuprate superconductor: Estimate of the mean free path in the trilayer cuprate HgBa 2 Ca 2 Cu 3 O 8 + δ
No full textInternational audienceThe thermal Hall conductivity of the trilayer cuprate HgBaCaCuO (Hg1223) - the superconductor with the highest critical temperature at ambient pressure - was measured at temperatures down to 2 K for three dopings in the underdoped regime ( = 0.09, 0.10, 0.11). By combining a previously introduced simple model and prior theoretical results, we derive a formula for the inverse mean free path, , which allows us to estimate the mean free path of -wave quasiparticles in Hg1223 below . We find that grows as , in agreement with the theoretical expectation for a clean -wave superconductor. Measurements were also conducted on the single layer mercury-based cuprate HgBaCuO (Hg1201), revealing that the mean free path in this compound is roughly half that of its three-layered counterpart at the same doping ( = 0.10). This observation is be attributed to the protective role of the outer planes in Hg1223, which results in a more pristine inner plane. We also report data in an ultraclean crystal of YBaCuO (YBCO) with full oxygen content = 0.18, believed to be the cleanest of any cuprate, and find that is not longer than in Hg1223
Exposing the Illusion of Fairness: Auditing Vulnerabilities to Distributional Manipulation Attacks
No full textProving the compliance of AI algorithms has become an important challenge with the growing deployment of such algorithms for real-life applications. Inspecting possible biased behaviors is mandatory to satisfy the constraints of the regulations of the EU Artificial Intelligence's Act. Regulation-driven audits increasingly rely on global fairness metrics, with Disparate Impact being the most widely used. Yet such global measures depend highly on the distribution of the sample on which the measures are computed. We investigate first how to manipulate data samples to artificially satisfy fairness criteria, creating minimally perturbed datasets that remain statistically indistinguishable from the original distribution while satisfying prescribed fairness constraints. Then we study how to detect such manipulation. Our analysis (i) introduces mathematically sound methods for modifying empirical distributions under fairness constraints using entropic or optimal transport projections, (ii) examines how an auditee could potentially circumvent fairness inspections, and (iii) offers recommendations to help auditors detect such data manipulations. These results are validated through experiments on classical tabular datasets in bias detection
Enhanced Production of (+)-Limonene through Targeted Engineering of Citrus sinensis Limonene Synthase
No full textInternational audienceLimonene is a high-value monoterpene with numerous industrial applications including flavors, fragrances, and pharmaceuticals. Its biosynthesis in microbes is often limited by limonene synthase and can be enhanced through targeted enzyme engineering. This study aims to optimize the Citrus sinensis limonene synthase (CsLS) in order to boost (+)-limonene yield. Site-saturation mutagenesis was carried out to create a library of truncated CsLS variants, which were evaluated for their enzymatic activity, enantioselectivity, and kinetic properties. Through medium-throughput screening, a mutant, tCsLS-Q8K, was found to improve limonene production by 2.1-fold. Bioreactor experiments with this optimized mutant, in combination with the ispA S80F mutation to enhance monoterpene precursor availability, achieved a (+)-limonene titer of 4.9 g/L, representing the highest production level of limonene reported to date. This work not only deepens our understanding of CsLS function but also offers a scalable approach to enhanced (+)-limonene production. The findings underscore the potential of enzyme engineering and pathway optimization to transform biotechnological processes, significantly impacting the commercial viability of the flavor and fragrance industries
Investigation sur les performances d'un liner vibro-acoustique : application au transport aérien
No full textBruit des transports : bruit des aéronefs; GABE - Acoustique du Bâtiment et de l'Environnement: GVB - Vibro acoustique et Contrôle du BruitNational audienceLes traitements acoustiques passifs, utilisés dans les nacelles aéronautiques pour réduire le bruit, présentent un comportement en quart-d’onde. Cependant, leur capacité d’absorption acoustique est naturellement limitée en raison des contraintes en épaisseur. De plus, le comportement de ces traitements est fortement « non linéaire » avec le niveau de bruit et l’écoulement représentatifs de conditions de vol, ce qui les rend plus difficile à dimensionner. Le concept du LEONAR (Long Elastic Open Neck Acoustic Resonator) offre la possibilité d’atteindre une absorption acoustique, en conservant un ratio « épaisseur sur longueur d’onde » faible. Les performances atteintes par le LEONAR sont également peu sensibles au niveau de bruit et à la vitesse de l'écoulement. Toutefois, pour du bruit multi-tonal, il peut être nécessaire de cibler plusieurs bandes de fréquence tout en conservant les contraintes d’épaisseur. L’idée proposée consiste à ajouter une caractéristique de contrôle vibratoire au système de contrôle acoustique classique du bruit applicable dans le domaine aéronautique. Cette approche vibratoire peut néanmoins amplifier la dépendance aux conditions d’essai, telles qu’évoquées ci-dessus, à l’interface avec l’écoulement rasant. Pour s’en affranchir, est étudié analytiquement et expérimentalement un DDOF constitué de deux cavités classiques, complété par un effet mécanique à l’interface entre les deux cavités. Cette modification permet d’ajouter un troisième degré de liberté d’origine vibratoire sans pour autant dégrader les performances acoustiques du DDOF. Le dimensionnement de ce troisième degré de liberté va être étroitement lié aux capacités des moyens de fabrications utilisés. La fabrication additive polymère est couramment utilisée pour réaliser des démonstrateurs de géométries spécifiques. Ce moyen de fabrication présente ses contraintes propres qui vont impacter les propriétés mécaniques du démonstrateur. En ajoutant la possibilité d’utiliser des polymères spécifiques il devient possible d’exploiter cette technologie pour concevoir un démonstrateur vibro-acoustique à trois dégrées de libertés
Développement d'un moyen d'essai expérimental pour l'étude du comportement des stratifiés composites en environnement cryogénique
No full textInternational audienceDéveloppement d'un moyen d'essai expérimental pour l'étude du comportement des stratifiés composites en environnement cryogéniqu
Impact of biochemical properties on the gelation of EPS extracted from aerobic granules.
No full textInternational audienc
Bathymetry estimations in river networks from altimetry measurements
No full textInferring unmeasured bathymetry from sparse observations of geophysical fluid surface elevations is a challenging and important problem, particularly in river hydraulics. In this work, we propose a solution to this inverse problem that is well-posed and robust under clearly identified contextual assumptions. Our method relies on a steady-state 1D diffusive-wave equation-defining a nonlinear second-order elliptic boundary value problem-that links bathymetry to water surface elevation, independently of the friction parameterization. While the associated inverse problem is ill-posed, we demonstrate that its solution can be uniquely determined given a single scalar bathymetry value. We provide theoretical analysis and numerical validation using realistic synthetic test cases, showing that bathymetry can be reconstructed up to a finite spatial frequency. Furthermore, the inversion proves robust to observational noise when embedded in a variational data assimilation (VDA) framework. The approach is extended to networks of interconnected river segments, where we show that a single measurement within the network can propagate the correct bathymetry throughout the entire system. This enables scalable inversion across large river networks. Our results offer new insights into the structure and solvability of bathymetry inversion problems in river hydraulics, particularly in sparse spatial observation contexts. Finally, this study opens perspectives for hybrid methods combining Bayesian inferences, neural networks, and variational frameworks to enhance the estimation of river characteristics from sparse measurements
Preliminary study of the anti-corrosion performance of different protection systems applied to two World War II aluminium alloy archaeological objects with different surface finishes
No full textInternational audienceThis article looks at the conservation of historic aircraft from the Second World War, testing different coatings to protect their aluminium alloy structures from corrosion. Five different protective protection systems, including ParaloidTM B72, wax, DinitrolTM , Multi Matt Clear LesonalTM and carboxylates, were evaluated by being applied to artefacts from WWII aircraft wrecks, in particular a wing fragment from a Supermarine Spitfire and a propeller blade from a P38 Lightning, which presented different surface condi- tions. The protected objects were exposed to real-life conditions (uncontrolled indoor environment). Ad- vanced analytical techniques such as electrochemical impedance spectroscopy (EIS), Fourier transform in- frared spectroscopy (FTIR), optical coherence tomography (OCT) and scanning electron microscopy (SEM) were used to characterize the evolution and effectiveness of the protection systems. The results showed different degrees of effectiveness for different coatings. DinitrolTM and LesonalTM demonstrated the best protective properties, forming thin but effective layers that enhance corrosion resistance. Carboxylates, on the other hand, proved ineffective, while ParaloidTM B72 and wax were not very effective and unsuitable for objects with original paint residues
WHITENING SPHERICAL GAUSSIAN MIXTURES IN THE LARGE-DIMENSIONAL REGIME
No full textWhitening is a classical technique in unsupervised learning that can facilitate estimation tasks by standardizing data. An important application is the estimation of latent variable models via the decomposition of tensors built from high-order moments. In particular, whitening orthogonalizes the means of a spherical Gaussian mixture model (GMM), thereby making the corresponding moment tensor orthogonally decomposable, hence easier to decompose. However, in the large-dimensional regime (LDR) where data are high-dimensional and scarce, the standard whitening matrix built from the sample covariance becomes ineffective because the latter is spectrally distorted. Consequently, whitened means of a spherical GMM are no longer orthogonal. Using random matrix theory, we derive exact limits for their dot products, which are generally nonzero in the LDR. As our main contribution, we then construct a corrected whitening matrix that restores asymptotic orthogonality, allowing for performance gains in spherical GMM estimation