Portail HAL Ensta
Not a member yet
11080 research outputs found
Sort by
Multidomain FEM-BEM coupling for acoustic scattering
No full textInternational audienceWe model time-harmonic acoustic scattering by an object composed of piece-wise homogeneous parts and an arbitrarily heterogeneous part. We propose and analyze new formulations that couple, adopting a Costabel-type approach, boundary integral equations for the homogeneous subdomains with volume variational formulations for the heterogeneous subdomain. This is an extension of the Costabel FEM-BEM coupling to a multi-domain configuration, with cross-points allowed, i.e. points where three or more subdomains are adjacent. While generally just the exterior unbounded subdomain is treated with the BEM, here we wish to exploit the advantages of BEM whenever it is applicable, that is, for all the homogeneous parts of the scattering object. Our formulation is based on the multi-trace formalism, which initially was introduced for acoustic scattering by piece-wise homogeneous objects. Instead, here we allow the wavenumber to vary arbitrarily in a part of the domain. We prove that the bilinear form associated with the proposed formulation satisfies a Gårding coercivity inequality, which ensures stability of the variational problem if it is uniquely solvable. We identify conditions for injectivity and construct modified versions immune to spurious resonances
Path differences between quasistatic and fatigue cracks in anisotropic media
No full textInternational audienceThe propagation path of quasistatic cracks under monotonic loading is known to be stronglyinfluenced by the anisotropy of the fracture energy in crystalline solids or engineered materials witha regular microstructure. Such cracks generally follow directions close to minima of the fractureenergy. Here we demonstrate both experimentally and computationally that fatigue cracks undercyclic loading follow dramatically different paths that are predominantly dictated by the symmetryof the loading with the microstructure playing a negligible or subdominant role
Shaping the infrared luminescence of colloidal nanocrystals using a dielectric microcavity
No full textInternational audienceAs they have gained maturity, colloidal nanocrystals (NCs) have also expand the spectral range over of which they could be used for photonic and optoelectronic applications. In particular, the infrared use of NCs has become of utmost interest to develop cost‐effective alternatives to current technologies. It is then critical not to let the material dictate the light–matter interaction, which is why the coupling of NCs to photonic cavities has been proposed. For infrared NCs, this approach has first been devoted to the control of absorption with in mind the increase of the signal magnitude for detectors. A Lot of efforts have been focused on the use of metallic metasurfaces. However, these generate significant optical losses and yield low quality factor. Here, this study rather focus on the coupling of infrared NCs to a dielectric mirror cavity. HgTe/CdS core‐shell NCs are used and integrated into a cavity made of aperiodic dielectric mirrors. The effect of the substrate is systematically study on spectral linewidth, carrier dynamic, and emission directivity. The cavity is shown to narrow the PL by a factor 10, while focusing the emission over a 12° angle. Monitoring the power dependence of the emission, this study shows that the cavity leads to 250 K increase in the effective electronic temperature
45 W average-power, high-energy deep-UV generation at 257 nm in LBO
No full textInternational audienceWe report on high average-power, high-energy picosecond fourth-harmonic generation in LBO. The first stages of a Yb:YAG laser chain operating at 1 kHz repetition rate generate few-picosecond 220 mJ chirped pulses at 1030 nm fundamental wavelength. They are frequency-converted in a cascade of three LBO crystals to generate the second-, third-, and fourth harmonics at 515 nm, 343 nm and 257 nm respectively. Crystals thicknesses and angular phase matching detuning were calculated as a function of pulse duration through broadband nonlinear optical numerical simulations. Last crystal is both conduction-cooled on edge and surface-cooled at center through forced-air flow to mitigate heating due to nonlinear absorption in the deep-UV and reduce temperature gradients. Chirped-pulse duration was experimentally adjusted to achieve stable 20% overall conversion efficiency. Near-field beam profiles were continuously recorded at 10 Hz, for all four wavelengths involved, together with corresponding energies, showing no significant beam degradation over 50 hours. Temperatures of the two last crystals were monitored and will help optimize surface cooling for future power ramping-up
Méthodes de solutions exactes pour les problèmes de localisation discrète de p-installations à grande échelle
No full textThis thesis focuses on the exact solution of the NP-hard problems p-median and p-center, combinatorial optimization problems that quickly become difficult to solve as the instance size increases. These discrete location problems involve opening a defined number p of facilities and then allocating to them a set of clients according to an objective function to be minimized.First, we study the p-median problem, which seeks to minimize the sum of distances between clients and the open facilities to which they are allocated. We develop an algorithm based on Benders decomposition that outperforms state-of-the-art exact methods. The algorithm considers a two-stage approach and an efficient algorithm for separating Benders cuts. The method has been evaluated on over 230 benchmark instances with up to 238025 clients and sites. Many instances are solved to optimality for the first time or have their best known solution improved.Secondly, we explore the p-center problem, which seeks to minimize the largest distance between a client and its nearest open facility. We first compare the five main MILP formulations in the literature. We study the Benders decomposition and also propose an exact algorithm based on a client clustering procedure based on the structure of the problem. All the proposed methods are compared with the state-of-the-art on benchmark instances. The results obtained are analyzed, highlighting the advantages and disadvantages of each method.Finally, we study a robust two-stage p-center problem with uncertainty on node demands and distances. We introduce the robust reformulation of the problem based on the five main deterministic MILP formulations in the literature. We prove that only a finite subset of scenarios from the infinite uncertainty set can be considered without losing optimality. We also propose a column and constraint generation algorithm and a branch-and-cut algorithm to efficiently solve this problem. We show how these algorithms can also be adapted to solve the robust single-stage problem. The different proposed formulations are tested on randomly generated instances and on a case study drawn from the literature.Cette thèse porte sur la solution exacte des problèmes NP-difficiles du p-median et du p-centre, des problèmes d'optimisation combinatoire qui deviennent rapidement difficiles à résoudre lorsque la taille de l'instance augmente. Ces problèmes de localisation discrète consistent à ouvrir un nombre défini p d'installations, puis à leur affecter un ensemble de clients selon une fonction objectif à minimiser.Tout d'abord, nous étudions le problème du p-median qui cherche à minimiser la somme des distances entre les clients et les installations ouvertes auxquelles ils sont affectés. Nous développons un algorithme basé sur la décomposition de Benders qui surpasse les méthodes exactes de l'état de l'art. L'algorithme considère une approche en deux étapes et ainsi qu'un algorithme efficace pour la séparation des coupes de Benders. Cette méthode est évaluée sur plus de 230 instances de benchmark avec jusqu'à 238025 clients et sites. De nombreuses instances sont résolues à l'optimalité pour la première fois ou ont leur meilleure solution connue améliorée.Deuxièmement, nous explorons le problème du p-centre qui cherche à minimiser la plus grande distance entre un client et l'installation ouverte qui en est la plus proche. Nous comparons d'abord les cinq principales formulations MILP de la littérature. Nous étudions la décomposition de Benders et nous proposons également un algorithme exact basé sur une procédure de partionnement des clients reposant sur la structure du problème. Toutes les méthodes proposées sont comparées à l'état de l'art dans des instances de benchmark. Les résultats obtenus sont analysés, mettant en évidence les avantages et les inconvénients de chaque méthode.Enfin, nous étudions un problème robuste du p-centre en deux étapes avec une incertitude sur les demandes et les distances des nœuds. Nous introduisons la reformulation robuste du problème basée sur les cinq principales formulations déterministes MILP de la littérature. Nous prouvons que seul un sous-ensemble fini de scénarios de l'ensemble d'incertitude infini peut être pris en compte sans perdre l'optimalité. Nous proposons également un algorithme de génération de colonnes et de contraintes et ainsi qu'un algorithme de branch-and-cut pour résoudre efficacement ce problème. Nous montrons comment ces algorithmes peuvent également être adaptés pour résoudre le problème robuste d'une seule étape. Les différentes formulations proposées sont testées sur des instances générées aléatoirement et sur un cas d'étude de la littérature
Proximal Point Type Algorithms with Relaxed and Inertial Effects Beyond Convexity
No full textInternational audienceWe show that the recent relaxed-inertial proximal point algorithm due to Attouch and Cabot remains convergent when the function to be minimizedis not convex, being only endowed with certain generalized convexity properties. Numerical experiments showcase the improvements brought bythe relaxation and inertia features to the standard proximal point method in this setting, too
Solvability results for the transient acoustic scattering by an elastic obstacle
No full textInternational audienceThe well-posedness of the linear evolution problem governing the transient scattering of acoustic waves by an elastic obstacle is investigated. After using linear superposition in the acoustic domain, the analysis focuses on an equivalent causal transmission problem. The proposed analysis provides existence and uniqueness results, as well as continuous data-to-solution maps. Solvability results are established for three cases, which differ by the assumed regularity in space on the transmission data on the acoustic-elastic interface Γ. The first two results consider data with "standard" H −1/2 (Γ) and improved H 1/2 (Γ) regularity in space, respectively, and are established using the Hille-Yosida theorem and energy identities. The third result assumes data with L 2 (Γ) regularity in space and follows by Sobolev interpolation. Obtaining the latter result was motivated by the key role it plays (in a separate study) in the justification of an iterative numerical solution method based on domain decomposition. A numerical example is presented to emphasize the latter point
Mode of action of the natural herbicide radulanin A as an inhibitor of photosystem II
No full textInternational audienceBackground: Radulanin A is a natural 2,5-dihydrobenzoxepin synthesized by several liverworts of the Radula genus. Breakthroughs in the total synthesis of radulanin A paved the way to the discovery of its phytotoxic activity. Nevertheless, its mode of action remained so far unknown and was investigated in Arabidopsis thaliana.Results: Radulanin A phytotoxicity was associated with cell death and partially depended on light exposure. Photosynthesis measurements based on chlorophyll a fluorescence evidenced that radulanin A and a Radula chromene inhibited photosynthetic electron transport with IC50 of 95 μM and 100 μM, respectively. We established a strong correlation between inhibition of photosynthesis and phytotoxicity for a range of radulanin A analogs. Based on these data, we also determined that radulanin A phytotoxicity was abolished when the hydroxyl group was modified and was modulated by the presence of the heterocycle and its aliphatic chain. Thermoluminescence studies highlighted that radulanin A targeted the QB site of the Photosystem II with a similar mode of action as 3-(3,4-dichloropheny)-1,1-dimethylurea (DCMU).Conclusion: We establish that radulanin A targets Photosystem II, expanding QB sites inhibitors to bibenzyl compounds. The identification of an easy-to-synthesize analog of radulanin A with similar MoA and efficiency might be useful for future herbicide development. This article is protected by copyright. All rights reserved
Degenerate McKean-Vlasov equations with drift in anisotropic negative Besov spaces
No full textThe paper is concerned with a McKean-Vlasov type SDE with drift in anisotropic Besov spaces with negative regularity and with degenerate diffusion matrix under the weak Hörmander condition. The main result is of existence and uniqueness of a solution in law for the McKean-Vlasov equation, which is formulated as a suitable martingale problem. All analytical tools needed are derived in the paper, such as the well-posedness of the Fokker-Planck and Kolmogorov PDEs with distributional drift, as well as continuity dependence on the coefficients. The solutions to these PDEs naturally live in anisotropic Besov spaces, for which we developed suitable analytical inequalities, such as Schauder estimates
A PDE WITH DRIFT OF NEGATIVE BESOV INDEX AND LINEAR GROWTH SOLUTIONS
No full textInternational audienceThis paper investigates a class of PDEs with coefficients in negative Besov spaces and whose solutions have linear growth. We show existence and uniqueness of mild and weak solutions, which are equivalent in this setting, and several continuity results. To this aim, we introduce ad-hoc Besov-Hölder type spaces that allow for linear growth, and investigate the action of the heat semigroup on them. We conclude the paper by introducing a special subclass of these spaces which has the useful property to be separable