HAL Portal IOGS (nstitut d'Optique Graduate School)
Not a member yet
12589 research outputs found
Sort by
Physics-informed Machine Learning for Better Understanding Laser-Matter Interaction
No full textInternational audiencePhysics-informed machine learning typically assumes that the underlying physical laws are known and abundant training data is available. These assumptions do not hold in the context of self-organization of matter, a phenomenon that leads to the emergence of patterns when a surface is irradiated with an ultrafast laser beam. Indeed, due to the constraints of the electronic data acquisition devices, the creation of large datasets is made impossible. Moreover, modeling this dynamic process is challenging as it involves coupling between electromagnetism, thermodynamics and fluid mechanics under far-from-equilibrium conditions that are not yet fully understood. This paper aims at taking a step forward towards a better understanding of this complex phenomenon. We specifically focus on the laser energy absorption of the surface, which is governed by the distinctive characteristics of Maxwell's equations in an inhomogeneous lossy medium. This involves modelling physics at the nano scale and incurs high simulation costs that make any exploration impractical. To address this major issue, we investigate different physics-informed learning models. In this low data regime, our study reveals that learning a simple U-Net-based surrogate model surpasses (i) more sophisticated neural architectures and (ii) the FDTD-based solver in speed by several orders of magnitude. Interestingly, our study highlights a link between the formation of patterns and the magnitude of absorbed energy
Impact of gate metallization for Total Ionizing Dose Testing of MOS capacitors
No full textInternational audienc
Inscription par laser d’images en couleur dans des cartes en polycarbonate
No full textInternational audienceL’identification et la sécurisation des produits sont des problématiques actuelles de plus en plus prégnantes. Pour répondre à ces besoins, le laboratoire Hubert Curien et la société HID Global ont développé une nouvelle technologie d’inscription d’images sans contact avec des propriétés optiques uniques [1]. Elle est basée sur le comportement optique d’un film mince nanocomposite (TiO2 : Ag) qui est exposé à des rayonnements laser dans l’UV et le visible. Lors de cette communication, les auteurs présenteront une voie d’élaboration de film mésoporeux compatible avec des substrats thermosensibles. Cette voie implique une formulation utilisant des précurseurs de titane, et des traitements à basse température. Après imprégnation par une solution de sel métallique, séchage et stabilisation sous UV, la couche obtenue est laminée avec plusieurs substrats en polycarbonate conduisant à une carte plastique. L’irradiation du matériau par un laser nanoseconde induit la création de nanostructures bien définies, à l’origine de couleurs variées, dans le film nanocomposite. Le contrôle de ces couleurs par ordinateur permet l’inscription rapide d’images personnalisées en couleur
Optical diffraction properties of three superimposed self-organized nanostructures induced by laser process
No full textInternational audienceControlling the diffraction properties of materials over a large area holds great promise for a wide range of optical applications. Laser-based techniques have emerged as a viable solution to address this need. Here, we present the diffraction properties of laser-induced selforganized structures, which consist of three interlaced grating-like structures: self-organized nanoparticles, self-organized cracks and laser marking lines. Under normal incidence external illumination, the sample exhibits an asymmetric diffraction pattern. However, when the incidence angle is tilted, circular diffraction patterns are observed in the plane perpendicular to both the sample and the incidence plane. These phenomena are attributed to the combination effect of the diffraction gratings. To elucidate the underlying physics of multiple diffraction, we use rigorous coupled-wave analysis (RCWA) and grating equations written in direction cosine space, extended to account for the presence of three superimposed gratings. Exploiting the laser-induced diffraction properties of these samples may have great potential for various industrial implementations, including security, display or design.</div
Reconstruire l'invisible: GRIOT pour l'Imputation de Graphes Attribués par Transport Optimal
No full textInternational audienceIn recent years, there has been a significant surge in machine learning techniques, particularly in the domain of deep learning, tailored for handling attributed graphs. Nevertheless, to work, these methods assume that the attribute values are fully known, which is not realistic in numerous real-world applications. This paper explores the potential of Optimal Transport (OT) to impute missing attribute values on graphs. To proceed, we design a novel multi-view OT loss function that can encompass both node feature data and the underlying topological structure of the graph by utilizing multiple graph representations. We then utilize this novel loss to train efficiently a Graph Convolutional Neural Network (GCN) architecture capable of imputing all missing values over the graph at once. We evaluate the interest of our approach with experiments both on synthetic data and real-world graphs, including different missingness mechanisms and a wide range of missing data. These experiments demonstrate that our method is competitive with the state-of-the-art in all cases and of particular interest on weakly homophilic graphs.Ces dernières années, les techniques d’apprentissage automatique ont connu un essor considérable, en particulier dans le domaine de l’apprentissage profond, notamment adaptées à la gestion des graphes attribués. Néanmoins, pour fonctionner, ces méthodes supposent que les valeurs des attributs sont entièrement connues, ce qui n’est pas réaliste dans de nombreuses applications du monde réel. Cet article explore le potentiel du transport optimal (TO) pour imputer les valeurs d'attributs manquantes sur les graphes. Pour ce faire, nous concevons une nouvelle fonction de perte basée sur du TO multi-vues qui peut englober à la fois les attributs des nœuds et la structure topologique sous-jacente du graphe à travers plusieurs représentations (vues) de ce dernier. Nous utilisons ensuite cette nouvelle fonction de perte pour former efficacement une architecture de réseau de neurones convolutif en graphes (GCN) capable d'imputer simultanément toutes les valeurs manquantes du graphe. Nous évaluons l'intérêt de notre approche avec des expérimentations à la fois sur des données synthétiques et sur des graphiques du monde réel, incluant différents mécanismes et une grande plage de données manquantes. Ces expériences démontrent que notre méthode est compétitive avec l’état de l’art dans tous les cas et particulièrement intéressante sur les graphes faiblement homophiles
Advancing Laser-Induced Nanoscale Surface Self-Organization with Machine Learning Guidance
No full textInternational audienc
Selective Modal Excitation of FBGs in FMF Through Inscription Techniques and the use of a Spatial Multiplexer
No full textInternational audienc
Very-High-Energy Heavy Ion Beam Dosimetry Using Solid State Detectors for Electronics Testing
No full textInternational audienc
Compressive stress triggers fibroblasts spreading over cancer cells to generate carcinoma in situ organization
No full textInternational audienceAt the early stage of tumor progression, fibroblasts are located at the outer edges of the tumor, forming an encasing layer around it. In this work, we have developed a 3D in vitro model where fibroblasts’ layout resembles the structure seen in carcinoma in situ. We use a microfluidic encapsulation technology to co-culture fibroblasts and cancer cells within hollow, permeable, and elastic alginate shells. We find that in the absence of spatial constraint, fibroblasts and cancer cells do not mix but segregate into distinct aggregates composed of individual cell types. However, upon confinement, fibroblasts enwrap cancer cell spheroid. Using a combination of biophysical methods and live imaging, we find that buildup of compressive stress is required to induce fibroblasts spreading over the aggregates of tumor cells. We propose that compressive stress generated by the tumor growth might be a mechanism that prompts fibroblasts to form a capsule around the tumor
Beyond Total Locking : Demonstrating and Measuring Mutual Influence on a RO-Based True Random Number Generator on an FPGA
No full textInternational audienceRing oscillator-based true random number generators are of interest because of their well-known and wellcharacterised conversion of analog noise into random numbers.The main drawback of using ring oscillators as a source of randomness is their tendency to be influenced by their environment.In particular, oscillators may lock to another signal at a frequency close to the nominal frequency of the ring, or two or more rings may even lock to each other. This is particularly dangerous for generators with multiple rings, which require the rings to be independent of each other. Furthermore, to reduce the risk of manipulable global noise sources, the rings should have the same structure and topology, making them even more vulnerable to locking. The metrics commonly used to quantify the degree of locking have limitations that can lead to erroneous conclusions as to whether a ring is locked or not. This is why we prefer to use the term mutual influence. In this paper, we propose a clear definition of the mutual influence between ring oscillators used as sources of randomness. One of the advantages of this definition is that it can easily be extended to include the case of the total locking of rings. Based on this definition, we introduce a new metric to quantify the mutual influence, which evaluates a statistical distance between the current distribution of phase differences and uniform distribution. The experimental results of several FPGA implementations of ring oscillators highlighted the suitability of the Kolmogorov-Smirnov test as a metric for detecting mutual influence