HAL-Polytechnique
Not a member yet
51406 research outputs found
Sort by
Spherical Skin Model: Stratified Co‐Culture of Fibroblasts and Keratinocytes on Spherical Beads Toward Compound Screening
No full textInternational audienceABSTRACT Advanced skin models are critical for pursuing non‐animal approaches in drug and cosmetic testing. However, existing 3D models remain complex and time‐consuming, which limits their adoption. Spherical skin model (SSM) is presented, a platform that balances biological fidelity with experimental robustness. The SSM is based on a core–shell structure where the dermal core is modeled by embedding human fibroblasts into collagen microcarriers (150 ), while the epidermal shell is formed by outer layers of immortalized keratinocytes. The collagen beads are generated using droplet microfluidics to enable rapid and reproducible production. The biological relevance of SSM is revealed through elevated expression of epidermal differentiation markers (loricrin, involucrin, keratin 1, keratin 10) and the dermal–epidermal junction marker collagen VII. The barrier function is validated by permeability assays that show strong exclusion of fluorescent dextran above 4 kDa. Moreover, their usefulness for screening is shown by identifying a dose‐dependent effect of vitamins in reducing oxidative stress and apoptosis against tert‐butyl hydroperoxide. As such, this 3D microphysiological model recapitulates key structural, molecular, and functional features of human skin while offering rapid generation, scalability, and compatibility with high‐throughput applications in dermatological and cosmetic research
Time Series for QFFE: Special Issue of the Journal of Time Series Analysis
No full textInternational audienc
Advances in metal-catalyzed strategies for carbohydrate functionalization
No full textInternational audienceTransition metal catalysis provides a versatile and powerful platform for the construction of structurally complex molecules. In recent years, its application has been successfully extended to the functionalization of carbohydrates, significantly expanding the toolbox of catalytic strategies available for sustainable synthesis. This review offers a comprehensive and up-to-date overview of recent advances in metal-catalyzed carbohydrate transformations, emphasizing key methodological innovations and their relevance to synthetic applications
AtomSurf : Surface Representation for Learning on Protein Structures
No full textInternational audienceWhile there has been significant progress in evaluating and comparing different representations for learning on protein data, the role of surface-based learning approaches remains not well-understood. In particular, there is a lack of direct and fair benchmark comparison between the best available surface-based learning methods against alternative representations such as graphs. Moreover, the few existing surface-based approaches either use surface information in isolation or, at best, perform global pooling between surface and graph-based architectures. In this work, we fill this gap by first adapting a state-of-the-art surface encoder for protein learning tasks. We then perform a direct and fair comparison of the resulting method against alternative approaches within the Atom3D benchmark, highlighting the limitations of pure surface-based learning. Finally, we propose an integrated approach, which allows learned feature sharing between graphs and surface representations on the level of nodes and vertices across all layers. We demonstrate that the resulting architecture achieves state-of-the-art results on all tasks in the Atom3D benchmark, while adhering to the strict benchmark protocol, as well as more broadly on binding site identification and binding pocket classification. Furthermore, we use coarsened surfaces and optimize our approach for efficiency, making our tool competitive in training and inference time with existing techniques. Code can be found online: https://github.com/Vincentx15/atomsur
Pulsed plasma approach for mild or strong ignition of a detonation wave using gradient of atomic species
No full textThis thesis presents a study conducted on the ignition of a detonation wave with non-equilibrium plasma. This form of plasma-assisted combustion has been of recent increased interest with the developments of new forms of detonation-based propulsion and their implementation into real flight structures. A key component of these propulsive devices is the requirement for reliable, efficient, “at will” ignition of a detonation wave. Typically, a detonation wave is formed either by depositing a large amount of energy (of the order of Joules) in a combustible mixture to directly form a detonation, or by igniting a flame with a weak energy source which accelerates before transitioning to a detonation (known as deflagration to detonation transition, or DDT). This work suggests the use of nanosecond plasma to produce a gradient of active species as a means of promoting the DDT process. For this purpose, the thesis is split into two parts.The initial focus is on the development of a new set of electrodes to achieve a gradient of atomic species. Nanosecond pulsed plasma (30 ns FWHM) in the range of -6 to -22.5 kV are used in this study. After an extensive set of possible geometries is tested, a plane-to-plane configuration with a varying gap size (28 to 50 mm over an 80 mm span) is developed and characterised using a variety of diagnostics. The formation of a gradient of atomic oxygen along the span in air at multiple pressures in the range of 100 to 150 mbar is shown by two-photon absorption laser-induced fluorescence (TALIF). The longitudinal reduced electric field is measured through multiple techniques such as optical emission spectroscopy, capacitive probe detector and electric-field induced second-harmonic generation (E-FISH). These give information on both the range of absolute values of field in the plasma (100 to 200 Td) and the early development of the discharge.The discharge is then tested in multiple combustible mixtures (2 H¬2 + O2, C2H2 + 2.5 O2, C2H4 + 4 O2) for a wide range of pressure (100 to 200 mbar) and deposited energy (50 to 600 mJ). The DDT length is recorded. Schlieren imaging is performed and shows the formation of both mild and strong ignition of a detonation wave. This study therefore participates in a deeper understanding of the mechanisms through which plasma can facilitate ignition of a detonation.Cette thèse présente une étude menée sur l'allumage d'une onde de détonation avec un plasma hors-équilibre. Cette forme de combustion assistée par plasma a récemment suscité un intérêt accru avec le développement de nouvelles formes de propulsion basées sur la détonation et leur mise en œuvre dans des structures de vol réelles. Un élément clé de ces dispositifs de propulsion est l'exigence d'un allumage fiable et efficace d'une onde de détonation. Généralement, une onde de détonation est formée soit en déposant une grande quantité d'énergie (de l'ordre de Joules) dans un mélange combustible pour former directement une détonation, soit en allumant une flamme avec une faible source d'énergie qui accélère avant de transiter vers une détonation (connue sous le nom de transition déflagration à détonation, ou TDD). Ces travaux suggèrent l'utilisation d'un décharge nanoseconde afin de promouvoir TDD. À cette fin, la thèse est divisée en deux parties.La première porte sur le développement d'un nouvel ensemble d'électrodes pour obtenir un gradient d'espèces atomiques. Un plasma pulsé nanoseconde (30 ns FWHM) dans la gamme de -6 à -22,5 kV est utilisé dans cette étude. Après avoir testé un vaste ensemble de géométries possibles, une configuration plan-plan avec une distance inter-électrode variable (28 à 50 mm sur une longueur de 80 mm) a été mise au point et caractérisée à l'aide d'une variété de diagnostics. La formation d'un gradient d'oxygène atomique le long de la cellule de décharge dans l'air à des pressions multiples dans la gamme de 100 à 150 mbar est démontrée par la fluorescence induite par two-photon absorbed laser-induced fluorescence (TALIF). Le champ électrique réduit longitudinal est mesuré par de multiples techniques telles que la spectroscopie d'émission optique, un détecteur à sonde capacitive et electric-field induced second-harmonic generation (E-FISH). Ces mesures donnent des informations sur la gamme des valeurs absolues du champ dans le plasma (100 à 200 Td) et sur les premiers stades de développement de la décharge.La décharge est ensuite testée dans plusieurs mélanges combustibles (2 H2 + O2, C2H2 + 2,5 O2, C2H4 + 4 O2) pour une large gamme de pression (100 à 200 mbar) et d'énergie déposée (50 à 600 mJ). La longueur du DDT est enregistrée. De l’imagerie Schlieren est réalisée et montre la formation d'une onde de détonation faible ou forte. Cette étude permet donc de mieux comprendre les mécanismes par lesquels le plasma peut faciliter l'allumage d'une détonation
3D Monte Carlo Radiative Transfer for Parameter Retrieval in Planetary Atmospheres
No full textInternational audienceRetrieving planetary atmospheric parameters from observational data is particularly challenging under large observation angles, in thick and highly scattering media (such as Titan and Venus), and in the presence of horizontal heterogeneities, like clouds and hazes. Traditional radiative transfer models, often based on plane-parallel or pseudo-spherical approximations, typically assume horizontally homogeneous layers, which limits their applicability in such scenarios.To overcome these limitations, we have developed a novel 3D radiative transfer solver, htrdr-planets, based on the Monte Carlo method that solves models considering spherical and heterogeneous atmospheres[1]. This solver leverages recent advances from the computer graphics and statistical physics communities to ensure computational efficiency.htrdr-planets supports arbitrary ground geometry, represented as triangular meshes with user-defined surface materials, and atmospheric properties defined on unstructured tetrahedral meshes. Gas absorption is modeled using the k-distribution method, and multiple aerosol and cloud populations with their own radiative properties can be described on separate spatial grids.Critically, we address the need for gradients (i.e., sensitivities) in parameter retrieval. Since conventional finite-difference methods are inefficient or infeasible with Monte Carlo, we differentiate the Monte Carlo estimator itself [2]. By reusing the same radiative paths, we construct a Monte Carlo estimator that computes both the radiance and its gradient with respect to atmospheric and surface parameters at negligible additional time cost.We apply this method to Titan and Venus, producing spatially resolved maps of sensitivity with respect to scattering, absorption, and surface-reflection properties. This framework enables retrievals in geometrically complex cases that defy traditional models, including Titan’s polar cloud structure and haze distribution, using Cassini and JWST datasets. This work is supported by the Agence National de la Recherche (ANR) through the RaD3-net project (ANR-21-CE49-0020-01).[1] htrdr-planets, https://www.meso-star.com/projects/htrdr/htrdr.html[2] He, Zili, et al. "Simultaneous Estimation of Radiance and its Sensitivities to Radiative Properties in a Spherical-Heterogeneous Atmospheric Radiative Transfer Model by Monte Carlo: Application to Titan." (Submitted to Journal of Quantitative Spectroscopy and Radiative Transfer.
PhysioBlocks: an Opensource Python Library for Simulating Block Diagrams of Dynamical Physiological Systems
No full textThe PhysioBlocks Python library is designed to simulate the dynamics of physiological systems (in particular cardiovascular systems) represented by block diagrams, in order to provide built-in modularity. Accordingly, a system is represented by a network of modules (blocks) connected by nodes in which they share quantities (degrees of freedom) and exchange fluxes. The user can easily create a new network by combining existing blocks. At a more advanced level, new blocks can be defined. The library is distributed under the LGPL-3.0-only license, and the initial distribution focuses on providing building blocks associated with lumped-parameter models of the cardiovascular system
T-REGS: Minimum Spanning Tree Regularization for Self-Supervised Learning
No full textInternational audienceSelf-supervised learning (SSL) has emerged as a powerful paradigm for learning representations without labeled data, often by enforcing invariance to input transformations such as rotations or blurring. Recent studies have highlighted two pivotal properties for effective representations: (i) avoiding dimensional collapse-where the learned features occupy only a low-dimensional subspace, and (ii) enhancing uniformity of the induced distribution. In this work, we introduce T-REGS, a simple regularization framework for SSL based on the length of the Minimum Spanning Tree (MST) over the learned representation. We provide theoretical analysis demonstrating that T-REGS simultaneously mitigates dimensional collapse and promotes distribution uniformity on arbitrary compact Riemannian manifolds. Several experiments on synthetic data and on classical SSL benchmarks validate the effectiveness of our approach at enhancing representation quality