HAL-CEA
Not a member yet
145450 research outputs found
Sort by
Enhanced generative model evaluation with Clipped Density and Coverage
No full textAlthough generative models have made remarkable progress in recent years, their use in critical applications has been hindered by their incapacity to reliably evaluate sample quality. Quality refers to at least two complementary concepts: fidelity and coverage. Current quality metrics often lack reliable, interpretable values due to an absence of calibration or insufficient robustness to outliers. To address these shortcomings, we introduce two novel metrics, Clipped Density and Clipped Coverage. By clipping individual sample contributions and, for fidelity, the radii of nearest neighbor balls, our metrics prevent out-of-distribution samples from biasing the aggregated values. Through analytical and empirical calibration, these metrics exhibit linear score degradation as the proportion of poor samples increases. Thus, they can be straightforwardly interpreted as equivalent proportions of good samples. Extensive experiments on synthetic and real-world datasets demonstrate that Clipped Density and Clipped Coverage outperform existing methods in terms of robustness, sensitivity, and interpretability for evaluating generative models
"Dormant” opto-thermo-mechanical properties in the isotropic phase of fluids
No full textInternational audienceThere is little literature on the flow properties of the isotropic phase of liquid crystalline fluids. Yet, this phase is an ideal tool to bridge the gap between liquid crystal physics and that of (ordinary) fluids. Optical and mechanical studies are presented, demonstrating that far from any phase transition, the isotropic phase of lyotropic liquid crystalline (LC) thermotropic molecules (colloidal solutions) can function as an optical oscillator at low-frequency mechanical excitation, establishing the elastic origin of the flow birefringence and “visualizing” the very existence of the elastic nature of the liquid state [1]. Moreover, mimicking the excellent anchoring capacity of liquid crystals, a new rheological protocol optimizing fluid/substrate interfaces is presented to access it and reveal that the oldest physicalproperty identified in condensed matter, shear elasticity, is also a fundamental property of mesoscopic liquids. Long considered a solid characteristic, it has been shown that mesoscopic fluids and colloids also have (shear) elasticity [2]. In other words, mesoscopic liquids propagate collective waves of compression and shear, they have the capacity to change temperature under the effect of mechanical stress
Integrative Multiparametric Analysis of Circulating Cell‐Free Nucleic Acids of Plasma in Healthy Individuals During Aging
No full textInternational audienceABSTRACT Plasma circulating cell‐free nucleic acids (ccfNAs) provide an exceptional source of information about an individual's health, yet their biology in healthy individuals during aging remains poorly understood. Here, we present the first integrative multiparametric analysis of the major types of plasma ccfNAs, including nuclear (ccfnDNA) and mitochondrial (ccfmtDNA) DNA, as well as ribosomal (ccfrRNA), messenger (ccfmRNA) and micro‐RNA (ccfmiRNA) in 139 healthy donors aged 19–66 years. We focused on quantity, integrity, and DNA methylation using an optimized experimental workflow that combines highly sensitive analytical methods with the detection of highly repetitive DNA and highly abundant RNA sequences, thereby reducing the required amount of ccfNAs per analysis. We showed a highly significant increase in ccfnDNA levels during aging ( p < 0.001), associated with a decrease in its integrity ( p < 0.05), while no significant changes were detected in ccfmtDNA levels and ccfDNA methylation. Moreover, a significant increase in ccfmRNA and ccfrRNA ( p < 0.05), as well as miR‐483‐5p ( p < 0.001) levels was detected during aging, but without any changes in ccfRNA integrity. Finally, we also showed that ccfDNA and ccfRNA levels were correlated ( p < 0.001), and a similar pattern was observed for ccfmtDNA and ccfRNA levels, suggesting a possible common release, maintenance, and/or clearance mechanism. Therefore, our study provides an optimized workflow for the global analysis of ccfNAs, enhances the understanding of their biology during aging, and identifies several potential ccfNA‐based biomarkers of aging
Characterization of challenging forensic DNA traces using advanced molecular technologies
No full textInternational audienceAbstract The majority of crime scenes contain DNA that is either present in small amounts or degraded, making it difficult to obtain usable DNA profiles using conventional technologies. The current standard for analyzing casework samples is the specific amplification of short tandem repeats (STR), which is limited by DNA quality and quantity. Since the goal of forensic science is to identify a suspect or victim regardless of trace quality, we evaluated three technological approaches to better characterize and exploit these traces: (i) ultra-sensitive pulse-field electrophoresis on a Femto Pulse System (FPS) to visualize DNA content, (ii) real-time quantitative PCR based on Alu repeats to quantify human DNA and analyze its integrity, and (iii) 16S ribosomal RNA gene (16S rRNA) amplicon sequencing to identify microbiota. We optimized FPS analysis using DNA from model traces (blood, saliva, semen, touch DNA, and vaginal swabs) and applied the protocol to 100 casework samples. We found differences between the FPS profiles of model and casework samples, showing a variation in fragment size and distribution, suggesting the presence of non-human DNA. Using Alu -qPCR and 16S rRNA amplicon sequencing, we determined the amount and proportion of human and non-human DNA. Human DNA was detected in 84% of traces with an average of 70 pg per trace, while 16S rRNA revealed microbial DNA as the most abundant DNA in traces. These analyses provide new insights into forensic trace composition, allowing better sorting and profiling of traces
Evidence of well-dispersed ionomer in the cathode catalyst layer of proton exchange membrane fuel cell by small angle scattering, and effect of its content on performance
No full textInternational audienceThe cathode catalyst layer plays a major role for the performance of Proton Exchange Membrane Fuel Cells (PEMFCs). It consists of catalyst (platinum nanoparticles deposited on carbon nanoparticles) and ionomer. The I/C ratio (ionomer mass / carbon mass) is a key parameter for the optimization of catalyst layers. In this work, catalyst layers with different I/C ratios were structurally and electrochemically characterized. The ratio I/C=1.1 showed an optimum of performance and protonic resistance. Transmission electron microscopy (TEM), small-angle scattering using X-ray (SAXS) and neutrons (SANS) were used to characterize the structure of the ionomer in the catalyst layer. In this work, an innovative data processing method allowed to obtain the same information on the ionomer structure by SAXS than by SANS. All these characterizations showed that for all the I/Cs tested, the ionomer was mostly well dispersed. Thus, increasing the I/C results in a more wide-spread ionomer network, which explains the increase in performance and the decrease in Rp. Scanning Electron Microscopy (SEM) highlighted that an I/C ratio of 1.3 or higher, the porosity decreased sharply, which may explain the decrease in performance with increasing I/C
Geochemical fingerprinting of permanent magnet along the value chain using high precision Nd isotopic analysis
No full textInternational audienceThe transition towards a carbon-neutral economy may drastically increase our dependence on metals [1]. Permanent magnets play an important role in this transition, as they are intensively used in wind turbine generators and electric vehicles [2]. The rare earth supply chain involves more than a hundred actors and keep track of the material during the different steps of production is a real challenge. Many initiatives have been taken in order to bring more transparency in the complex network of the permanent magnet supply chain, but the actual certification methods suffer from a lack of reliable control over the different data collected. The MaDiTraCe project funded by the European Union aims at developing high performance analytical methods for critical raw material traceability. The present work focuses on Nd isotopic analysis as a geochemical tool for the traceability of permanent magnets. The objective of this study is to investigate whether the initial Nd isotopic composition of the rare earth ore is preserved throughout the different steps of magnet manufacturing, by measuring samples from the entire value chain. Rare earth ores, targeted ores and concentrates from all over the world were collected, as well as permanent magnets. Metallic samples from the value chain were provided by the Permanent Magnet Platform (CEA, Grenoble, France). Isotopic analyses were performed using both MC-ICP-MS (Ghent University, A&MS group, Belgium) and TIMS (BRGM, France) for inter-lab cross-validation. Several isolation protocols were tested and optimized in order to be able to address the complex variety of the different samples. The isotope ratio obtained using TIMS and MC-ICP-MS are very close to each other, validating the two approaches for accurate and precise isotopic analysis. The eight magnet samples display an average ɛ 143Nd of – 18.50 (±0.79, 2sd). The results for the value chain samples don’t show significant isotopic fractionation between the samples, suggesting that no isotopic fractionation occurs during magnet manufacturing, despite the complex metallurgical processes involved. These results provide new insights into Nd isotopic analysis as a possible tool for permanent magnet source tracking
Measurements and interpretations of production cross-sections in collisions at 13 TeV with the ATLAS detector
No full textInternational audienceMeasurements of integrated and differential cross-sections for production in proton-proton collisions are presented. The data collected by the ATLAS detector at the Large Hadron Collider from 2015 to 2018 at a centre-of-mass energy of 13 TeV are used, corresponding to an integrated luminosity of 140 fb. The candidate events are reconstructed using leptonic decay modes of the gauge bosons into electrons or muons. The integrated cross-section per lepton flavour for the production of is measured in the detector fiducial region with a relative precision of 4%. The measured value is compared with the Standard Model prediction at a precision of up to next-to-next-to-leading-order in QCD and next-to-leading-order in electroweak. Cross-sections for and production and their ratio are presented. The production is also measured differentially as functions of various kinematic variables, including new observables sensitive to CP-violation effects. All measurements are compared with state-of-the-art Standard Model predictions from fixed-order calculations or Monte Carlo generators based on next-to-leading-order matrix elements interfaced with parton showers. An effective field theory interpretation of the measurements is performed, considering both CP-conserving and CP-violating dimension-6 operators modifying the production. In the absence of observed deviations from the Standard Model, limits on CP-conserving Wilson coefficients are extracted using the transverse mass of the system. For CP-violating coefficients a machine learning approach is used to construct an observable with enhanced sensitivity to CP-violation effects
Alpha-synuclein seed amplification assays: Data sharing, standardization needed for clinical use
No full textInternational audienceAlpha-synuclein seed amplification assays can improve neurodegenerative disease diagnosis and care, but widespread use depends on a framework that standardizes protocols and encourages data sharing
Development of a PID algorithm for the CALIFA detector to study multinucleon knockout reactions in exotic nuclei
No full textInternational audiencePrevious studies have shown a reduction in cross sections relative to theoretical predictions for single-nucleon knockout reactions, with varying conclusions about the dependence of this reduction on the N/Z ratio of the projectile. The (p,pX) knockout reactions studied with the R3B setup offer a unique opportunity for kinematically complete measurements using inverse kinematics. This work focuses on the development of an algorithm for performing particle identification using the CALIFA detector, of vital importance for the study of knocked-out clusters such as deuterium
Climate mitigation potential for targeted forestation after considering climate change, fires, and albedo
No full textInternational audienceAfforestation and reforestation, both of which refer to forestation strategies, are widely promoted as key tools to mitigate anthropogenic warming. However, the carbon sequestration potential of these efforts remains uncertain in satellite-based assessments, particularly when accounting for dynamic climate conditions, vegetation-climate feedback, fire-dominated disturbance, and the trade-offs associated with surface albedo changes. Leveraging a coupled Earth system model, we estimated that global forestation mitigates 31.3 to 69.2 Pg C eq (carbon equivalent) during 2021–2100 under a sustainable shared socioeconomic pathway. Regionally, the highest carbon mitigation potential of forestation concentrates in tropical areas, while mid-high-latitude regions demonstrate higher heterogeneity, highlighting the need for region-specific strategies and further refinement of nature-based mitigation plans. Our findings underscore the importance of considering disturbances and minimizing adverse albedo changes when estimating the carbon mitigation potential of forestation initiatives. We also advocate for the development of consistent, high-resolution maps of suitable areas for targeted forestation, avoiding environmentally sensitive lands and potential conflicts with other human activities