HAL-Polytechnique
Not a member yet
51406 research outputs found
Sort by
Stratégies numériques pour la simulation à l’échelle microscopique des batteries lithium-ion
No full textIn this thesis, we develop and analyze numerical strategies for the simulation of lithium-ion batteries (LIBs) based on their continuum description at the microscale. Our focus is on addressing the computational challenges posed by this inherently multiphysics and multiscale problem, particularly the nonlinearity at the reaction interface and the stiffness of the governing equations.In the first phase of this doctoral work, we tackle the temporal multiscale nature of LIBs by decoupling the domains into subproblems that can be solved independently. Building upon an adaptive high-order coupling strategy, we implement this approach in a Python code. The effectiveness and performance of the method are demonstrated through 1D LIB half-cell simulations. Additionally, we discuss how this promising numerical strategy can be extended to 3D LIB simulations.The multiphysics nature of the LIB model further motivates us to explore adaptive methods in both space and time to reduce computational costs. To extend our study to higher dimensions, we employ a C++ framework with a monolithic solution strategy. Specifically, we implement a multiresolution-based adaptive mesh refinement (AMR) technique using SAMURAI and an adaptive high-order implicit time integrator using PETSc, examining their performance when used together. Using this fully adaptive implementation, we conduct parametric studies to evaluate the impact of interdigitated electrodes on the performance and behavior of 2D LIB half cells.The thesis concludes by synthesizing the two numerical strategies developed to address the computational challenges of LIB microscale simulations. As a natural extension of this work, we propose a unified framework that integrates these approaches, providing a robust foundation for tackling additional complexities that may arise in future microscale LIB model developments.Dans cette thèse, nous développons et analysons des stratégies numériques pour la simulation des batteries lithium-ion (LIBs) basées sur leur description continue à l’échelle microscopique. Notre objectif est de relever les défis computationnels posés par ce problème intrinsèquement multiphysique et multi-échelle, en particulier la non-linéarité à l’interface de réaction et la raideur des équations aux dérivées partielles associées.Dans la première phase de ce travail doctoral, nous abordons la nature multi-échelle temporelle des LIBs en découplant les domaines en sous-problèmes pouvant être résolus indépendamment. En nous appuyant sur une stratégie de couplage adaptatif d’ordre élevé, nous implémentons cette approche dans un code Python. L’efficacité et les performances de cette méthode sont démontrées à travers des simulations de demi-cellules LIB en 1D. De plus, nous discutons des perspectives d’extension de cette stratégie numérique prometteuse vers des simulations de LIB en 3D.La nature multiphysique du modèle LIB nous pousse également à explorer des méthodes adaptatives en espace et en temps afin de réduire les coûts computationnels. Pour étendre notre étude aux dimensions supérieures, nous utilisons un code C++ avec une stratégie de solution monolithique. Plus précisément, nous implémentons une technique de raffinement de maillage adaptatif (AMR) basée sur la multi-résolution avec texttt{SAMURAI}, ainsi qu’un intégrateur temporel implicite adaptatif d’ordre élevé avec texttt{PETSc}, et nous examinons leurs performances lorsqu’ils sont utilisés ensemble. Grâce à cette implémentation entièrement adaptative, nous menons des études paramétriques pour évaluer l’impact des électrodes interdigitées sur les performances et le comportement des demi-cellules LIB en 2D.La thèse se conclut en synthétisant les deux stratégies numériques développées pour relever les défis computationnels des simulations à l’échelle microscopique des LIB. Dans le prolongement naturel de ce travail, nous proposons un cadre unifié intégrant ces approches, offrant ainsi une base solide pour aborder les complexités supplémentaires susceptibles de survenir dans les développements futurs des modèles de LIB à l’échelle microscopique
Understanding the Importance of HDO and D/H on Venus
No full textInternational audienceThe study of HDO and the deuterium/hydrogen (D/H) ratio plays a critical role in reconstructing the past and present climate of Venus. These isotopic tracers offer valuable insights into the planet’s hydrological history, atmospheric escape processes, and potential ancient reservoirs of water. In particular, variations in D/H can shed light on the mechanisms that led Venus to become the hot, arid world we observe today.In this work, we present the first fully three-dimensional simulation of the HDO cycle on Venus, by implementing HDO in both gas and liquid phases within the Venus Planetary Climate Model (VPCM). Our model allows us to explore the spatial and temporal behavior of HDO, and how it interacts with the atmospheric dynamics and cloud chemistry of Venus.One of our key findings is that the vertical distribution of D/H is strongly influenced by processes previously neglected in earlier models. Notably, we show that the presence of deuterated sulfuric acid (HDSO₄) in Venusian clouds can significantly alter the D/H profile and must be taken into account in future studies.We also assess the individual effects of isotope fractionation during three key processes: condensation, molecular diffusion, and photolysis. Our analysis reveals that photolysis-induced fractionation dominates the D/H ratio in the upper atmosphere, driving its increase to approximately 480 × VSMOW (Vienna Standard Mean Ocean Water) at altitudes near 130 km.In addition, we observe notable diurnal variations in the abundances of H₂O, HDO, and D/H in the upper atmosphere, consistent with expectations based on solar-driven atmospheric dynamics.Overall, our results highlight the complex interplay of physical and chemical processes controlling the distribution of water isotopologues on Venus. These findings not only refine our understanding of the present Venusian atmosphere but also provide essential constraints for reconstructing its climatic evolution and water loss history
A tasty introduction to packing problems: Understanding how particles of all kinds fill space has applications in physics, engineering, materials design, and even machine learning.
No full textInternational audienc
Diffusive Nature of Housing Prices
No full textInternational audienceWe analyze the French housing market prices in the period 1970-2022, with high-resolution data from 2018 to 2022. The spatial correlation of the observed price field exhibits logarithmic decay characteristic of the twodimensional random diffusion equationlocal interactions may create long-range correlations. We introduce a stylized model, used in the past to model spatial regularities in voting patterns, that accounts for both spatial and temporal correlations with reasonable values of parameters, some fitted on impulse response data. Our analysis reveals that price shocks are persistent in time and their amplitude is strongly heterogeneous in space. Our study quantifies the diffusive nature of housing prices that was anticipated long ago [1, 2], albeit on much restricted, local data sets
Robust ML Auditing using Prior Knowledge
No full textInternational audienceAmong the many technical challenges to enforcing AI regulations, one crucial yet underexplored problem is the risk of audit manipulation. This manipulation occurs when a platform deliberately alters its answers to a regulator to pass an audit without modifying its answers to other users. In this paper, we introduce a novel approach to manipulation-proof auditing by taking into account the auditor's prior knowledge of the task solved by the platform. We first demonstrate that regulators must not rely on public priors (e.g. a public dataset), as platforms could easily fool the auditor in such cases. We then formally establish the conditions under which an auditor can prevent audit manipulations using prior knowledge about the ground truth. Finally, our experiments with two standard datasets illustrate the maximum level of unfairness a platform can hide before being detected as malicious. Our formalization and generalization of manipulation-proof auditing with a prior opens up new research directions for more robust fairness audits
Etude du rôle de la myosine II dans la migration cellulaire collective au cours de la gastrulation du poisson-zèbre Danio rerio
No full textCollective cell migration, where each cell depends on its neighbors for direction, is a fundamental process during embryonic development, wound healing, and certain metastatic processes. The central question of this thesis is: how do cells migrating as a group orient themselves? To answer this, this thesis used a model organism of choice: zebrafish gastrulation (Danio rerio). During this key stage of embryonic development, a dorsal structure called the axial mesoderm extends toward the animal pole, driven by a group of cells known as the polster. These cells lead the migration and contribute to the elongation of the embryo. Previous work has shown that polster cells are guided by their immediate following cells, and that this guidance relies on mechanotransduction via α-catenin, which adopts an open conformation, suggesting that tension is exerted between the cells of the polster. But what is the origin of these tensions? This thesis focused on the role of non-muscle myosin II, a well-known motor protein that generates intracellular tension. Using various technical approaches, such as functional inhibition of myosin II through dominant-negative kinase forms like MLCK and ROCK, pharmacological treatments, and cell transplantation, this thesis demonstrated that inhibition of myosin II is necessary for the non-autonomous orientation of polster cells, meaning that it operates through its effects on neighboring cells. At the same time, the use of a transgenic line marking myosin II, combined with an in vitro imaging system and laser ablation by two-photon microscopy, revealed an autonomous role of myosin II in the contractility of protrusions, particularly in the retrograde flow of actin. Furthermore, the use of a tension-sensitive antibody on α-catenin showed that myosin II also acts in a non-autonomous manner to put tension on adherens junctions. Together, these results support the idea that follower cells orient leading cells by exerting traction via contractile protrusions, dependent on myosin II. This guiding mechanism, based on forces exerted by the trailing cells, enables efficient coordination of collective movements without relying on pre-established chemical gradients. This work thus positions myosin II as a key player in the mechanical coordination of collective migrations, with implications not only for embryonic development but also for pathological contexts such as cancer.La migration collective des cellules, où chaque cellule dépend de ses voisines pour se diriger, est un processus fondamental présent lors du développement embryonnaire, de la cicatrisation et dans certains processus métastatiques. La question de fond de cette thèse est : comment des cellules qui migrent en groupe s’orientent-elles ? Pour y répondre, cette thèse a utilisé un modèle de choix : la gastrulation de poisson-zèbre (Danio rerio). Lors de cette étape clé du développement embryonnaire, une structure dorsale appelée mésoderme axial s’étend vers le pôle animal, conduite par un groupe de cellules appelées polster. Ces cellules migrent en tête, et contribuent à l’allongement de l’embryon. Les travaux précédents ont montré que les cellules de polster sont guidées par leurs cellules suiveuses immédiates et que ce guidage repose sur de la mécanotransduction via l’α-caténine, qui adopte une conformation ouverte, suggérant que des tensions sont exercées entre les cellules du polster. Mais quelle est l’origine de ces tensions ? Cette thèse s’est concentrée sur le rôle de la myosine II non musculaire (myosine II), une protéine motrice bien connue pour générer des tensions intracellulaires. À l’aide de diverses approches techniques, telles que l’inhibition fonctionnelle de la myosine II par des formes dominantes négatives de kinases comme MLCK et ROCK, des traitements pharmacologiques et des transplantations cellulaires, cette thèse a démontré que l’inhibition de la myosine II est nécessaire à l’orientation des cellules de polster de manière non-autonome, c’est-à-dire par l’intermédiaire de ses effets sur les cellules voisines. Parallèlement, l’utilisation d’une lignée transgénique marquant la myosine II, combinée à un système d’imagerie in vitro, et à des ablations laser par microscopie à 2-photons, a révélé un rôle autonome de la myosine II dans la contractilité des protrusions, et en particulier dans le flux rétrograde d’actine. De plus, l’utilisation d’un anticorps sensible à la tension sur l’α-caténine a montré que la myosine II intervient également de manière non-autonome pour mettre sous tension les jonctions adhérentes. Ensemble, ces résultats soutiennent l’idée que les cellules suiveuses orientent les cellules meneuses en exerçant une traction via des protrusions contractiles, dépendantes de la myosine II. Ce mécanisme de guidage, fondé sur les forces exercées par les cellules arrière, permet une coordination efficace des mouvements collectifs sans avoir recours à des gradients chimiques préétablis. Ce travail positionne ainsi la myosine II comme un acteur clé dans la coordination mécanique des migrations collectives, avec des implications non seulement pour le développement embryonnaire, mais aussi pour des contextes pathologiques comme le cancer
RDF Query Answering in the Presence of Access Restrictions
No full textNote: also informally presented at BDA 2025International audienceIn this work, we explore algorithms for answering conjunctive RDF queries in the presence of RDFS ontologies and access control. We consider an access control setting where by default all users have access to the complete graph, and a restriction can forbid user a user's access to specific IRIs. Here, restricting for user u the access to an IRI i entails that: no answer to a query by u may contain the IRI i; no triple containing i can be used to compute an answer for a query by i, nor to entail such a triple via reasoning with the ontology. We present a set of query answering algorithms for this novel context, and prove that five among them are correct, i.e., sound and complete, with respect to both the ontology and the access restrictions in place. We have implemented all our algorithms and present experiments comparing their performance.Note: also informally presented at BDA 202
Strict hierarchy between -wise measurement simulability, compatibility structures, and multi-copy compatibility
No full textThe incompatibility of quantum measurements, i.e. the fact that certain observable quantities cannot be measured jointly is widely regarded as a distinctive quantum feature with important implications for the foundations and the applications of quantum information theory. While the standard incompatibility of multiple measurements has been the focus of attention since the inception of quantum theory, its generalizations, such as measurement simulability, -wise incompatibility, and mulit-copy incompatibility have only been proposed recently. Here, we point out that all these generalizations are differing notions of the question of how many measurements are genuinely contained in a measurement device. We then show, that all notions do differ not only in their operational meaning but also mathematically in the set of measurement assemblages they describe. We then fully resolve the relations between these different generalizations, by showing a strict hierarchy between these notions. Hence, we provide a general framework for generalized measurement incompatibility. Finally, we consider the implications our results have for recent works using these different notions
Political Brinkmanship and Compromise
No full textInternational audienceWe study how do-or-die threats ending negotiations affect gridlock and welfare when two opposing parties bargain. Failure to agree on a deal in any period implies a continuation of the negotiation. However, under brinkmanship, agreement failure in any period may precipitate a crisis with a small chance, i.e. an outcome worse than the status-quo and any possible deal. In equilibrium, such brinkmanship threats improve the probability of an agreement but also increase the risk of crisis. Brinkmanship reduces welfare when one might think it is most needed: severe gridlock. In this case, despite this global welfare loss, a party has incentives to use brinkmanship strategically to obtain a favorable bargaining position
CaAdam: Improving Adam optimizer using connection aware methods
No full textInternational audienceWe introduce a new method inspired by Adam that enhances convergence speed and achieves better loss function minima. Traditional optimizers, includingAdam, apply uniform or globally adjusted learning rates across neural networks without considering their architectural specifics. This architecture-agnosticapproach is deeply embedded in most deep learning frameworks, where optimizers are implemented as standalone modules without direct access to the network’s structural information. For instance, in popular frameworks like Keras or PyTorch, optimizersoperate solely on gradients and parameters, without knowledge of layer connectivity or network topology. Our algorithm, CaAdam, explores this overlooked area by introducing connection-aware optimization through carefully designed proxies of architectural information. We propose multiple scaling methodologies that dynamically adjust learning rates based on easily accessible structural properties such as layer depth, connection counts, and gradient distributions. This approach enables more granular optimization while working within the constraints of current deep learning frameworks. Empirical evaluations on standard datasets (e.g., CIFAR-10, Fashion MNIST) show that our method consistently achieves faster convergence and higher accuracy compared to standard Adam optimizer, demonstrating the potential benefits of incorporating architectural awareness in optimization strategies