HAL-CEA
Not a member yet
145450 research outputs found
Sort by
La confiance avec le contrôle : spécification et vérification d’hyperpropriétés sur réseaux de neurones
No full textInternational audienceL’apprentissage automatique consiste à produire des programmes qui respectent certaines propriétés statistiques sur une distribution de probabilité. Ce formalisme probabiliste se traduit notamment par la notion de « confiance » qu’un programme porte sur sa sortie. Ainsi, un réseau de neurones peut modéliser ses sorties comme une distribution de probabilité. La vérification de réseaux de neurones se concentre encore majoritairement sur des propriétés dites « locales » - c’est-à-dire ne portant que sur des points précis et sur leur voisinage. Nous proposons dans cet article d’employer la notion de score de confiance pour écrire des spécifications globales. Plus précisément, nous proposons de spécifier des propriétés de robustesse sur des réseaux de neurones étant donné une certaine probabilité affectée à une sortie - autrement dit, le niveau de confiance que le programme accorde à sa prédiction. Contrairement aux propriétés de robustesse locale communément rencontrées dans la littérature, notre approche permet une spécification globale, c’est-à-dire pour tout point de l’espace d’entrée qui respecte l’hypothèse de confiance. Cette approche ouvre ainsi la voie à la vérification de propriétés portant sur tout l’espace d’entrée d’un réseau. Nous présentons un travail de réécriture nécessaire pour traduire ces spécifications vers des prouveurs sur étagère, et évaluons les performances et limites de notre approche sur plusieurs cas d’étude
Investigation of Al-Si-Cu alloys as phase change materials for high temperature thermal energy storage
No full textInternational audienceThe present work explores the suitability of Al-Cu-Si ternary alloys as hightemperature metallic phase change materials (PCMs) for thermal energy storage systems (TESS) operating between 550ºC and 850ºC. While prior research has primarily focused on thermodynamic modeling or thermal property characterization below 600°C, this work provides a comprehensive experimental assessment of selected invariant compositions within the Al-Cu-Si system. CALPHAD calculations were performed using FactSage 8.2 software (SGTE and FTLite database) to guide alloy selection, followed by synthesis and detailed characterization of melting point, latent heat, specific heat, thermal diffusivity, and thermal conductivity. Critically, the thermal reliability of these materials was evaluated under repeated solid-liquid cycling up to 100 cycles in oxidizing atmospheres, revealing their stability and degradation profiles. Additionally, dilatometry and density analysis were conducted to provide an in-depth understanding of the alloys and practical properties for end users. Among the tested alloys, several demonstrated high volumetric energy densities (over 500 kWh/m³ for a temperature difference of 300°C) and good thermal durability, establishing Al-Cu-Si alloys as promising PCM candidates for industrial-scale high-temperature energy storage applications. This study fills a notable gap in the literature by combining phase selection, comprehensive thermophysical property evaluation, and long-term cycling analysis above 600°C
A new genome-scale metabolic model of oleaginous microalgae with refined lipid metabolism clarifies Microchloropsis gaditana mutant phenotypes
No full textInternational audienceThe oleaginous microalga Microchloropsis gaditana (formerly Nannochloropsis gaditana) has gained large interest due to its potential to produce lipids for a wide range of biotechnological applications. Computational models such as genome-scale metabolic models constitute powerful tools for unravelling microorganism metabolism. In this work we present iMgadit23, a new genome-scale metabolic model for M. gaditana that encompasses 2330 reactions involving 1977 metabolites and associated with 889 genes. A key advancement of iMgadit23 over other Microchloropsis models is its fully detailed lipid metabolism: pathways involved in membrane and storage glycerolipid biosynthesis and degradation that have undergone thorough manual curation and have been comprehensively described based on current knowledge of M. gaditana lipid metabolism.iMgadit23 successfully reproduced in vivo growth phenotypes under diverse environmental and genetic conditions, including heterotrophic growth on multiple carbon sources and photoautotrophic growth under varying nitrogen sources and concentrations. The model was also able to capture the role of the Bubblegum acyl-CoA synthetase in remodeling M. gaditana lipid metabolism, with predicted flux distributions consistent with the altered lipid profile observed in a mutant strain.In addition, we developed a comprehensive 2D-pathway map of model content to provide a systems-level visualization of M. gaditana metabolism.Together, iMgadit23 and its 2D map constitute valuable tools for advancing our understanding of M. gaditana metabolism and for deciphering mutant phenotypes, particularly in the context of lipid metabolism, allowing optimization of cultivation strategies, enhancement of lipid productivity, and guidance of strain engineering for biofuel production and other biotechnological processe
Exploring the science of radon adsorption: Materials, methodologies, and emerging directions
No full textInternational audienceRadon, a radioactive noble gas, poses significant health risks due to its link to lung cancer and presents a major challenge in low-background physics experiments, where its decay products contribute to background noise. Developing efficient adsorbent materials for radon mitigation is therefore crucial for both environmental safety and scientific research. However, designing effective radon adsorbents remains challenging due to the low concentration of the gas, its inert nature, and associated health risks. This review provides a comprehensive analysis of radon adsorption on porous materials, with a focus on experimental methodologies, structure-property relationships, and emerging trends in material design. We highlight key material classes, including activated carbons, zeolites, metal-organic frameworks, and other carbon-based adsorbents, evaluating their adsorption performance and practical considerations. Additionally, we explore future directions in radon adsorbent development, discuss improved experimental approaches, and examine insights from xenon adsorption studies as a predictive tool for radon capture. By consolidating current knowledge and identifying critical gaps, this review aims to guide the design of next-generation materials for efficient radon mitigation in environmental, industrial, and scientific applications
Search for long-lived particles using displaced vertices of oppositely charged leptons in 140 fb of pp collisions at TeV with the ATLAS detector
No full textInternational audienceA search is presented for long-lived particles decaying into an oppositely charged lepton pair, , , or , that form a vertex within the inner tracking system of the ATLAS detector at the Large Hadron Collider, displaced from the primary proton-proton interaction region. The analysis uses the 140 fb of Run-2 data collected at TeV by the ATLAS experiment in 2015-2018. The results of the analysis are interpreted in the context of three benchmark models covering masses from 0.1 to 2.2 TeV and a range of mean proper lifetimes times the speed of light from 1 to 10000 mm. The first model is a generic boson pair-produced by a new heavy scalar, with the decaying into lepton pairs. The remaining two models are -parity violating supersymmetric models in which the lightest neutralino decays into (, ). The models differ by the mode of production of the , which can be produced via the decay of pairs of gluinos or of pairs of charginos and neutralinos (, , or ). Although each benchmark sample includes pair-produced LLPs, only a single vertex is required to be reconstructed. No dilepton displaced vertex candidate is observed and the results are presented as upper limits on the production cross-sections. This analysis sets leading limits on the production cross-sections for multiple models, including parameter space that has never been directly probed
Reconstruction of atmospheric neutrinos in DUNE's horizontal-drift far-detector module
No full textInternational audienceThis paper reports on the capabilities in reconstructing and identifying atmospheric neutrino interactions in one of the Deep Underground Neutrino Experiment's (DUNE) far detector modules, a liquid argon time projection chamber (LArTPC) with horizontal drift (FD-HD) of ionization electrons. The reconstruction is based upon the workflow developed for DUNE's long-baseline oscillation analysis, with some necessary machine-learning models' retraining and the addition of features relevant only to atmospheric neutrinos such as the neutrino direction reconstruction. Where relevant, the impact of the detection of the charged particles of the hadronic system is emphasized, and comparisons are carried out between the case when lepton-only information is considered in the reconstruction (as is the case for many neutrino oscillation experiments), versus when all particles identified in the LArTPC were included. Three neutrino direction reconstruction methods have been developed and studied for the atmospheric analyses: using lepton-only information, using all reconstructed particles, and using only correlations from reconstructed hits. The results indicate that incorporating more than just lepton information significantly improves the resolution of both neutrino direction and energy reconstruction. The angle reconstruction algorithms developed in this work result in no strong dependence on particle direction for reconstruction efficiencies or neutrino flavor identification. This comprehensive review of the reconstruction of atmospheric neutrinos in DUNE's FD-HD LArTPC is the first step towards developing a first neutrino oscillation sensitivity analysis, which will ready DUNE for its first measurements
Cluster properties as a function of dynamical state in the DESI Legacy x UNIONS surveys
No full textInternational audienceWe investigate how the dynamical state of galaxy clusters influences their galaxy populations and mass distributions. Using photometrically selected clusters from the DESI Legacy Imaging Survey cross-matched with the UNIONS galaxy shear catalogue, we classify clusters as evolved or evolving based on their rest-frame r-band magnitude gaps and stellar mass ratios between the brightest cluster galaxies (BCGs) and bright satellites. We measure the stellar mass functions, weak-lensing profiles, and radial number density and red-fraction profiles of stacked clusters in both subsamples. Evolved clusters exhibit more concentrated lensing profiles, bimodal stellar mass functions dominated by massive BCGs, and a deficit of intermediate-mass satellites, while evolving clusters show flatter central lensing signals and an excess of massive satellites. Applying the same selection to IllustrisTNG clusters reproduces these trends and links the observed differences to distinct mass accretion histories. These results demonstrate the close link between cluster galaxy populations and the overall dynamical state of their underlying dark matter halo
Amplified Arctic–boreal fire regimes from permafrost thaw feedbacks
No full textInternational audienc
Lipo-oligo/polysaccharide biorecognition at the nano-bio interface: insights into LA27 aptamer binding to biomimetic bacterial membranes
No full textInternational audienceGram-negative bacteria feature an asymmetric outer membrane, composed of phospholipids in the inner leaflet and lipopolysaccharides (LPS) or lipooligosaccharides (LOS) in the outer leaflet, which underpins virulence and antibiotic resistance. Conventional LPS assays like the Limulus amebocyte lysate test face ethical and practical challenges, motivating the development of aptamer-based detection. Here, we probe the interaction of the LA27 DNA aptamer (Kd ≈ 46 nM) with biomimetic membranes incorporating LOS/LPS from Akkermansia muciniphila, Flavobacterium sp. Root935, and Paenalcaligenes hominis strains. Asymmetric supported lipid bilayers (SLBs) and symmetric large unilamellar vesicles (LUVs) were characterized by neutron reflectometry (NR), small-angle neutron scattering (SANS), and dynamic light scattering (DLS). NR confirmed preserved bilayer asymmetry and revealed that LA27 deeply penetrates Akkermansia LOS bilayers, evidenced by increased core hydration and structural disruption, whereas interactions with Flavobacterium LOS are moderate, and those with Paenalcaligene
