HAL-MINES ParisTech
Not a member yet
27348 research outputs found
Sort by
Error analysis of a compositional score-based algorithm for simulation-based inference
No full textInternational audienceSimulation-based inference (SBI) has become a widely used framework in applied sciences for estimating the parameters of stochastic models that best explain experimental observations. A central question in this setting is how to effectively combine multiple observations in order to improve parameter inference and obtain sharper posterior distributions. Recent advances in score-based diffusion methods address this problem by constructing a compositional score, obtained by aggregating individual posterior scores within the diffusion process. While it is natural to suspect that the accumulation of individual errors may significantly degrade sampling quality as the number of observations grows, this important theoretical issue has so far remained unexplored. In this paper, we study the compositional score produced by the GAUSS algorithm of Linhart et al. (2024) and establish an upper bound on its mean squared error in terms of both the individual score errors and the number of observations. We illustrate our theoretical findings on a Gaussian example, where all analytical expressions can be derived in a closed form
TUNNNG HYDROPHILICITY OF PECTIN AEROGELS VIA HETEROGENEOUS AMIDATION
No full textInternational audienceAerogels are solid materials with nanometer-scale pores showing interesting properties (e.g., high porosity, low density, and high specific surface area) for a range of applications. The field of aerogels based on natural resources, such as polysaccharides, is gaining more and more attention due to the need for sustainable alternatives to conventional, non-renewable counterparts. Among bio-based aerogels, pectin aerogels possess a density of around 0.1 g -1 cm -3 and a very fine network with meso-and small macropores.[1] Thanks to these properties, the thermal conductivity of pectin aerogels is below that of air (< 0.025 W m -1 K -1 ) making these materials perfect candidates for thermal insulation applications.[2] However, pectin aerogels are highly hygroscopic: they adsorb moisture from the air leading to network collapse and loss of the initial properties.In this work, heterogeneous pectin modification was applied to tune the wettability of the aerogels. Pectin was chemically modified at the gel state via amidation of the carboxyl groups using different amines and reaction conditions. The effect of the solvent, the reaction time, and the amine concentration on the degree of substitution were investigated. The best conditions regarding the degree of substitution and the gel morphology were chosen to fabricate grafted pectin aerogels, using six amines with different alkyl or aryl chains.The modified pectin aerogels had density 0.1 -0.4 g cm -3 and specific surface area 140 -490 m 2 g -1 . The hydrophilicity was evaluated with water contact angles that were in the range of 48 -134°, strongly dependent on the lipophilicity of the amine used for the modification. The thermal superinsulating properties of pectin aerogels were maintained after the modification for several formulations, with very low thermal conductivity (around 0.016 W m -1 K -1 ).</div
Middle Miocene climate evolution in the northern Mediterranean region (Digne–Valensole basin, SE France)
No full textInternational audienceAbstract. During the Middle Miocene, the Earth shifted from a warm state, the Miocene Climatic Optimum (MCO; 16.9–14.7 Ma), to a colder state associated with the formation of extensive and permanent ice sheets on Antarctica. This climatic shift, the Middle Miocene Climatic Transition (MMCT; 14.7–13.8 Ma) strongly affected the composition and structure of major biomes, ocean circulation, and precipitation patterns. Although Middle Miocene climate dynamics are well documented in marine records, our knowledge of terrestrial climate change is not well constrained. Here we present a long-term (23–13 Ma) stable (δ13C, δ18O) and clumped (Δ47) isotope record of soil carbonates from a northern Mediterranean Alpine foreland basin: the Digne–Valensole basin (DVB), France. Δ47-derived soil carbonate formation temperatures indicate a highly dynamic dry season temperature pattern that is consistent with multiple periods of reorganization of atmospheric circulation during the MCO. We propose that changes in atmospheric circulation patterns modified the seasonality of precipitation and, ultimately, the timing of pedogenic carbonate formation. Consequently, Δ47 soil carbonate temperature data record the combined effects of long-term regional temperature and carbonate formation seasonality change. The data are consistent with the existence of a proto-Mediterranean climate already during certain MCO time intervals. Following the MMCT, the stable and clumped isotope record displays pronounced cooling after 13.8 Ma accompanied by a rather large (−5.0 %) decrease in soil water δ18O values. Our northern Mediterranean foreland basin climate record shares strong similarities with time-equivalent records from the terrestrial European mid-latitudes and the global oceans and enhances our understanding of the circum-Alpine Middle Miocene terrestrial climate dynamics
Effet des conditions de trempe et de revenu d'un acier faiblement allié pour réacteur sous pression (A508 Gr.3) sur l'amorçage de la rupture fragile et sur sa ténacité
No full textThe fracture toughness and the brittle fracture initiation of a quenched/tempered A508 Gr.3 low-alloy steel were studied to reveal the effects of heat treatment conditions. The results show better fracture toughness properties (i.e. lower fracture toughness reference temperature T0 (according to ASTM E1921 standard)) at higher quenching rate and lower tempering parameter, as expected. Three types of brittle fracture initiation mechanisms have been observed at the microscopic scale: at grain boundary carbides; on inclusions; at grain boundaries with no detectable particle. The proportion of occurrences of crack initiations cases observed at inclusions was higher when the reference temperature T0 is lower, either with a smaller tempering parameter or a higher quenching rate. Overall, the mechanism of crack initiation on a carbide as observed in this study was predominant
Finding antibodies in cryo-EM maps with CrAI
No full textInternational audienceMotivation: Therapeutic antibodies have emerged as a prominent class of new drugs due to their high specificity and their ability to bind to several protein targets. Once an initial antibody has been identified, its design and characteristics are refined using structural information, when it is available. Cryo-EM is currently the most effective method to obtain 3D structures. It relies on well-established methods to process raw data into a 3D map, which may, however, be noisy and contain artifacts. To fully interpret these maps the number, position, and structure of antibodies and other proteins present must be determined. Unfortunately, existing automated methods addressing this step have limited accuracy, require additional inputs and high-resolution maps, and exhibit long running times. Results: We propose the first fully automatic and efficient method dedicated to finding antibodies in cryo-EM maps: CrAI. This machine learning approach leverages the conserved structure of antibodies and a dedicated novel database that we built to solve this problem. Running a prediction takes only a few seconds, instead of hours, and requires nothing but the cryo-EM map, seamlessly integrating within automated analysis pipelines. Our method can find the location and pose of both Fabs and VHHs at resolutions up to 10 Å and is significantly more reliable than existing approaches.Availability and implementation: We make our method available both in open source github.com/Sanofi-Public/crai and as a ChimeraX bundle (crai).Results: We propose the first fully automatic and efficient method dedicated to finding antibodies in cryo-EM maps: CrAI. This machine learning approach leverages the conserved structure of antibodies and a dedicated novel database that we built to solve this problem. Running a prediction takes only a few seconds, instead of hours, and requires nothing but the cryo-EM map, seamlessly integrating within automated analysis pipelines. Our method can find the location and pose of both Fabs and VHHs at resolutions up to 10 Å and is significantly more reliable than existing approaches. Availability and implementation: We make our method available both in open source github.com/Sanofi-Public/crai and as a ChimeraX bundle (crai)
Trajectoires hydrologiques du bassin de la Seine
No full textNational audienceFor several decades, the effects of climate change on water resources have become increasingly tangible. The resulting pressure is so significant that water management has emerged as one of the greatest challenges of the twenty-first century. Although uncertainties persist regarding the quantification of climate change impacts on water resources, the number of people exposed to hydrological stress is expected to rise in the near future.Extreme events may well become the new normal. What collective strategies can be adopted to guard against these risks, to mitigate the impacts of climate change, and ultimately to adapt? Such questions weigh heavily on the minds of citizens, policymakers, researchers, and managers of water and aquatic environments. To provide quantitative insights for regional and local initiatives in water and ecosystem management, researchers from the PIREN-Seine program have undertaken studies reconstructing past and future trajectories to assess the impacts of climate change on water resources in the Seine basin. Unlike southern French territories, substantial uncertainties remain for the Seine basin, with recurrent disagreements among models used in international intercomparison exercises.The studies presented in this volume employ advanced numerical methods to reconstruct and evaluate representative hydroclimatic forcings and to quantify their impacts using the CaWaQS hydrological model. They highlight the growing tension on water resources throughout the twentieth century, arising from the combined effects of climate variability and anthropogenic pressures. Prospective analyses, though still subject to considerable uncertainty, already reaffirm the critical importance of mitigating greenhouse gas emissions in line with the Paris Climate Agreement.Nevertheless, uncertainties remain profound, potentially leaving decision-makers confronted with adaptation choices across multiple dimensions—from resource scarcity during low-flow periods to the prospect of more frequent flood events in winter and spring. Catastrophic scenarios could ensue, such as the collapse of groundwater reserves (with the disappearance of aquifers and upstream rivers dependent on them), leading to a generalized breakdown of the socio-ecosystem of the basin. Multi-annual hydroclimatic situations that could precipitate such collapse will be analyzed during Phase 9 of the PIREN-Seine program, with the aim of further supporting water management policies in the basin.Depuis plusieurs décénnies, les effets du changement climatique sur la ressource en eau sont de plus en plus tangibles. Cette pression est telle, que la gestion de l’eau est devenue l’un des plus grands défis du XXIe siècle. Bien que des incertitudes persistent quant à la quantification des impacts du changement climatique sur les ressources en eau le nombre de personnes exposées à un stress hydrologique augmentera dans un avenir proche.Ces phénomènes intenses, que la normale qualifiait auparavant de rares, sont-ils amenés à devenir la norme dans le futur ? Quelles stratégies collectives pouvons-nous adopter pour nous en prémunir, pour atténuer les effets du changement climatique et ultimement, comment nous y adapter ? Telles sont des questions prégnantes à l’esprit des citoyens, politiques, chercheurs et gestionnaires de l’eau et des milieux aquatiques.Au vu de l’urgence de la situation et pour éclairer quantitativement les initiatives régionales ou locales de gestion de l’eau et des milieux aquatiques, les chercheurs du PIREN-Seine ont réalisé des travaux de construction de trajectoires passées et futures pour évaluer les impacts du changement climatique sur la ressource en eau du bassin de la Seine. Contrairement aux territoires du sud de la France, des incertitudes importantes demeurent au sujet du bassin de la Seine, avec des désaccords récurrents entre modèles mobilisés dans le cadre des inter-comparaisons internationales.Les travaux, relatés dans ce fascicule, mobilisent des méthodes numériques avancées pour reconstruire, évaluer des forçages hydroclimatiques représentatifs et quantifier leurs impacts via le modèle hydrologique CaWaQS. Ils mettent en évidence une tension grandissante sur la ressource en eau au cours du XXe siècle, en croisant effets climatiques et pression anthropique. La prospective, quant à elle, est encore soumise à des incertitudes, mais permet d’ores et déjà de réaffirmer l’importance de l’atténuation des émissions de gaz à effet de serre par le respect des accords de Paris sur le climat. Pour autant, les incertitudes restent fortes, laissant possiblement les décideurs face à des choix d’adaptation sur tous les plans depuis la raréfaction de la ressource en période d’étiages jusqu’à de possibles événements de crue plus fréquents en hiver et au printemps. Des situations catastrophiques pourraient alors se produire, comme un effondrement des ressources souterraines (disparition d’aquifères et des cours d’eau amont qui en dépendent) et, en conséquence, un effondrement généralisé du socio-écosystème du bassin. Les situations hydroclimatiques pluri-annuelles menant à cet effondrement seront analysées dans le cadre de la prochaine phase 9 du PIREN-Seine afin de poursuivre l’accompagnement des politiques de l’eau sur le bassin
Improving Consistency in Vehicle Trajectory Prediction Through Preference Optimization
No full textInternational audienceTrajectory prediction is an essential step in the pipeline of an autonomous vehicle. Inaccurate or inconsistent predictions regarding the movement of agents in its surroundings lead to poorly planned maneuvers and potentially dangerous situations for the end-user. Current state-of-the-art deep-learningbased trajectory prediction models can achieve excellent accuracy on public datasets. However, when used in more complex, interactive scenarios, they often fail to capture important interdependencies between agents, leading to inconsistent predictions among agents in the traffic scene. Inspired by the efficacy of incorporating human preference into large language models, this work fine-tunes trajectory prediction models in multiagent settings using preference optimization. By taking as input automatically calculated preference rankings among predicted futures in the fine-tuning process, our experiments-using stateof-the-art models on three separate datasets-show that we are able to significantly improve scene consistency while minimally sacrificing trajectory prediction accuracy and without adding any excess computational requirements at inference time.</div
Differential flatness by pure prolongation: Necessary and sufficient conditions
No full textInternational audienceIn this article, we introduce the notion of differential flatness by pure prolongation: loosely speaking, a system admits this property if, and only if, there exists a pure prolongation of finite order such that the prolonged system is feedback linearizable. We obtain Lie-algebraic necessary and sufficient conditions for a general nonlinear multi-input system to satisfy this property. These conditions are comprised of the involutivity and relative invariance of a pair of filtrations of distributions of vector fields. An algorithm computing the minimal prolongation lengths of the input channels that achieve the system linearization, yielding the associated flat outputs, is deduced. Examples that show the efficiency and computational tractability of the approach are then presented.</div
Modeling in tribology: Recent advances, applications, and open questions
No full textInternational audienceRecent advances in modeling have enhanced our ability to make quantitative predictions for tribological phenomena, thereby unraveling relevant mechanisms. Algorithmic innovations, including those based on multiscale methods and machine learning, have been especially impactful, for example in overcoming long-standing bottlenecks that hinder simulations of systems with strong coupling across disparate scales. However, traditional modeling approaches, such as boundary-element techniques, have also progressed and continue to yield new insights. This article reviews developments from the past decade, examining how both new and established methods have deepened our understanding of experimental results and have furthered theoretical approaches in key tribological areas, including contact mechanics, lubrication, metal friction, and tribo-chemistry. Selected applications, such as tunable interfaces and energy harvesting, illustrate the broad influence of recent developments on the field
Anthraquinone grafting on carbon aerogels for supercapacitors applications
No full textInternational audienceNew materials are constantly emerging to respond to the desire of having more performant energy storage devices. Among all energy storage devices, supercapacitors also called electrochemical capacitors, have the ability of quickly accumulating charges but lacks the important energy density of batteries 1 . Carbon aerogels (CA) are three-dimensional porous materials made of interconnected carbon particles through sol-gel processes or pyrolysis of a carbon-rich structure. They are interesting materials as supercapacitors electrodes for the controllability of pore size, their high surface area and low electrical resistance 2 . Surface modification through diazonium salts is a well-known technique for the grafting of molecules on carbons 3 . The covalent attachment of electroactive molecules onto carbon can allow an increase of the capacitance by adding faradaic process to the double layer capacitance of the electrodes. Here, we reported the grafting of 1,4-anthraquinone on aerogels made from resorcinol-formaldehyde condensation. Different types of aerogels showing different porosity were grafted and characterized through electrochemical and elementary analysis to ensure the grafting and analyze the impact of the grafting on the surface