Publications

2023

• Mouvements de ligne de contact et solidification
  
  • Grivet Rodolphe
  , 2023. Ce manuscrit porte sur l’interaction entre les phénomènes de solidification et le mouvement de lignes de contact dans le contexte plus général de la solidification d’écoulements capillaires. Au delà de leur élégance esthétique, ces phénomènes sont d’une importance capitale dans différents domaines industriels, tels que le givrage des aéronefs, les recouvrements métallurgiques ou encore la fabrication additive. Ils sont également très présents dans des contextes naturels, lors de la formation de cascades de glace ou de stalagmites par exemple.La ligne de contact déterminant des conditions aux limites géométriques pour l’écoulement, son rôle est essentiel dans le couplage entre l’écoulement et la solidification. Elle est cependant également affectée par cette dernière. Son comportement est donc examiné dans différentes situations expérimentales.Dans le cas du dépôt de gouttes sur des surfaces froides, nous montrons que la ligne de contact est arrêtée par la croissance de cristaux au contact de la surface, ce qui modifie la taille du dépôt final. Nous proposons un modèle simplifié permettant de comprendre la dynamique de ces cristaux. Dans le cas d’un impact de gouttes d’eau sur des surface froides, nous montrons que l’inertie couplée à la solidification rend progressivement le système superhydrophobe, un comportement qui a des conséquences drastiques sur la morphologie de l’impact et la transition au splash. Ce changement de comportement est caractérisé non seulement par l’évolution de la vitesse de transition au splash, mais aussi par un changement de dynamique des gouttelettes secondaires. Nous présentons également des résultats expérimentaux sur l'impact d'une goutte d'eau chaude sur de la glace et montrons le rôle de la fusion de la glace dans l'augmentation du rayon d'étalement maximal de a goutte. Ensuite, dans le cas d’une ligne de contact en recul sur de la glace en croissance (ménisque dynamique avec solidification), nous montrons qu’il est possible de maintenir une ligne de contact stationnaire à n’importe quel nombre capillaire, ce qui est confirmé par l’intégration numérique d’un modèle théorique. Ces dernières expériences suggèrent également un angle de contact non nul entre l’eau et la glace, indépendamment du changement de phase. Enfin, nous présentons des résultats expérimentaux montrant la dynamique d'accrochage-décrochage non-triviale d'une ligne de contact avançant sur de la glace.
• Guaranteed approximations of arbitrarily quantified reachability problems
  
  • Goubault Eric
  • Putot Sylvie
  , 2023. We propose an approach to compute inner and outer-approximations of the sets of values satisfying constraints expressed as arbitrarily quantified formulas. Such formulas arise for instance when specifying important problems in control such as robustness, motion planning or controllers comparison. We propose an interval-based method which allows for tractable but tight approximations. We demonstrate its applicability through a series of examples and benchmarks using a prototype implementation. (10.48550/arXiv.2309.07662)
  DOI : 10.48550/arXiv.2309.07662
• Chip-Interleaved DSSS for Energy-Efficient Physical Layer Encryption
  
  • Leroy Clément
  • Arbi Tarak
  • Pasquero Oudomsack Pierre
  • Geller Benoit
  , 2023. Traditional methods for ensuring secure communications typically rely on encryption, which necessitates some form of coordination between the transmitter and the receiver. The Direct Sequence Spread Spectrum modulation can introduce some security at the physical layer but this technique must be combined with other technologies. Physical layer encryption techniques offer an alternative approach by taking advantage of transmit arrays to introduce beamformed noise. These techniques guarantee a certain level of confidentiality without requiring neither a heavy coordination nor losing time for secret key establishment, which make these techniques very attractive in particular for military and secured communications. However, the proposed solutions in the literature suffer from several limitations, and in particular, poor energy efficiency. To address this issue, we propose an original solution for physical layer encryption that does not spoil any energy with a signal mask by taking advantage of the wiretap channel inherent degradation introduced by the spatial chip interleaving over the transmitter antennas. Our simulations confirm that our proposal guarantees a considerable level of confidentiality without any energy loss.
• Numerical simulation of the impact of a gas jet on a free water surface
  
  • Haegeman Ward
  • Le Touze Clément
  • Dupays Joël
  • Massot Marc
  Multiphase Science and Technology, Begell House, 2023, 36 (1), pp.27-48. In this work, we are interested in the numerical simulation of a high-speed hot jet impinging on a free liquid surface at rest by means of diffuse interface models. We first consider the case of a low-temperature subsonic jet; a 4-equation model is used on a 2D axi-symmetric setup. Turbulence is accounted for by solving the Reynolds averaged equations and using a k-ω turbulence model. Numerical results are evaluated by comparing the depth of the cavity formed in the liquid surface to the predicted values using theoretical models from the literature. We then consider the case of a high-temperature jet. We start by showing equilibrium assumption between the liquid and gas phases which is no longer valid. A 5-equation model that does not rely on this assumption is presented. Both models are compared numerically on a simplified set-up. (10.1615/MultScienTechn.2023047916)
  DOI : 10.1615/MultScienTechn.2023047916
• Reinforcement learning with formal performance metrics for quadcopter attitude control under non-nominal contexts
  
  • Bernini Nicola
  • Bessa Mikhail
  • Delmas Rémi
  • Gold Arthur
  • Goubault Eric
  • Pennec Romain
  • Putot Sylvie
  • Sillion François
  Engineering Applications of Artificial Intelligence, Elsevier, 2023, 127. We explore the reinforcement learning approach to designing controllers by extensively discussing the case of a quadcopter attitude controller. We provide all details allowing to reproduce our approach, starting with a model of the dynamics of a crazyflie 2.0 under various nominal and non-nominal conditions, including partial motor failures and wind gusts. We develop a robust form of a signal temporal logic to quantitatively evaluate the vehicle's behavior and measure the performance of controllers. The paper thoroughly describes the choices in training algorithms, neural net architecture, hyperparameters, observation space in view of the different performance metrics we have introduced. We discuss the robustness of the obtained controllers, both to partial loss of power for one rotor and to wind gusts and finish by drawing conclusions on practical controller design by reinforcement learning. (10.1016/j.engappai.2023.107090)
  DOI : 10.1016/j.engappai.2023.107090
• Two Classes of Equatorial Magnetotail Dipolarization Fronts Observed by Magnetospheric Multiscale Mission: A Statistical Overview
  
  • Alqeeq Soboh
  • Le Contel O.
  • Canu P.
  • Retinò A.
  • Chust T.
  • Mirioni L.
  • Chuvatin A.
  • Nakamura R.
  • Ahmadi N.
  • Wilder F.
  • Gershman D.
  • Khotyaintsev Yu.
  • Lindqvist P.‐a.
  • Ergun R.
  • Burch J.
  • Torbert R.
  • Fuselier S.
  • Russell C.
  • Wei H.
  • Strangeway R.
  • Bromund K.
  • Fischer D.
  • Giles B.
  • Saito Y.
  Journal of Geophysical Research Space Physics, American Geophysical Union/Wiley, 2023, 128 (10). Magnetotail earthward fast plasma flows (Baumjohann et al., 1990) or bursty bulk flows (BBF, Angelopoulos et al., 1992) play a major role in the energy, plasma and magnetic flux transport from the magnetotail to the inner magnetosphere (e.g., Angelopoulos et al., 1994). They are often, although not always (Richard et al., 2022), accompanied by a sharp and transient increase of the northward component of the magnetic field called dipolarization fronts (DFs). DFs are considered as tangential discontinuities (velocity and magnetic field variations are tangential to the front so with no normal component of the magnetic field and no plasma flux flowing through it) separating a relatively cold dense plasma at rest from a hot tenuous plasma in rapid motion (e.g., H. S. Fu et al., 2012a; Sergeev et al., 2009). The origin of the fast flows and their related DFs is still a matter of debate. The main formation mechanisms currently studied are magnetic reconnection (e.g., (10.1029/2023JA031738)
  DOI : 10.1029/2023JA031738
• Scattered wavefield in the stochastic homogenization regime
  
  • Garnier Josselin
  • Giovangigli Laure
  • Goepfert Quentin
  • Millien Pierre
  , 2023. In the context of providing a mathematical framework for the propagation of ultrasound waves in a random multiscale medium, we consider the scattering of classical waves (modeled by a divergence form scalar Helmholtz equation) by a bounded object with a random composite micro-structure embedded in an unbounded homogeneous background medium. Using quantitative stochastic homogenization techniques, we provide asymptotic expansions of the scattered field in the background medium with respect to a scaling parameter describing the spatial random oscillations of the micro-structure. Introducing a boundary layer corrector to compensate the breakdown of stationarity assumptions at the boundary of the scattering medium, we prove quantitative $L^2$- and $H^1$- error estimates for the asymptotic first-order expansion. The theoretical results are supported by numerical experiments.
• Multititration: The New Method for Implementing Ultrasensitive and Quantitative Multiplexed In-Field Immunoassays Despite Cross-Reactivity?
  
  • Mousseau Fanny
  • Féraudet Tarisse Cécile
  • Simon Stéphanie
  • Gacoin Thierry
  • Alexandrou Antigoni
  • Bouzigues Cédric Ismael
  Analytical Chemistry, American Chemical Society, 2023, 95 (36), pp.13509-13518. The accurate in-field titration of multiple pathogens is essential to efficiently describe and monitor environmental or biological contamination, isolate, act, and treat adequately. This underscores the requirement of portable, fast, quantitative, and multiplexed detection technologies, which, however, have not been properly developed so far, notably because it has been hindered by the phenomenon of cross-reactivity. In this work, we proposed a new analytical method based on the imaging through a portable device of lanthanide-based nanoparticles (YVO 4 :Eu) for spatially multiplexed detection, relying on a multiparameter analysis, i.e., a simultaneous analysis of all of the luminescence signals through the comparison to a calibration surface built in the presence of multiple analytes of interest. We then demonstrated the possibility to simultaneously quantify by multiplexed lateral flow assay (xLFA) the three enterotoxins SEG, SEH, and SEI in unknown mixtures, over two concentration decades (from a dozen of pg•mL −1 to few ng•mL −1). Assays were performed in less than an hour (25 min of strip migration followed by 30 min of drying at room temperature), the time during which the presence of the operator was not required for more than 5 min, in order to dip the strip and have it imaged by the reader. The concepts of nominal concentration recovery, coefficient of variation (CV), limit of blank (LOB), and limit of detection (LOD) were discussed in detail in the context of multiplexed assays. With our new definitions, quantitative results demonstrated a high recovery of the nominal concentrations (115%), reliability (CV = 20%), and sensitivity (LOBs of 3, 27, and 6 pg•mL −1 for SEG, SEH, and SEI respectively, and LODs of 6, 48, and 11 pg•mL −1 for SEG, SEH, and SEI, respectively). Based on this method, we observed an increase in sensitivity of 100 compared to the other multiplexed LFA labeled with gold particles and we approached the sensitivity of the simplex enzyme-linked immunosorbent assay (ELISA) performed with the same capture and detection antibodies. To conclude, our results, which are applicable to virtually any kind of multiplexed test, pave the way to the next generation of in-field analytical immunoassays by providing fast, quantitative, and highly sensitive multiplexed detection of biomarkers or pathogens. (10.1021/acs.analchem.3c01846)
  DOI : 10.1021/acs.analchem.3c01846
• Estimating the Coverage Measure and the Area Explored by a Line-Sweep Sensor on the Plane
  
  • Vianna Maria Costa
  • Goubault Eric
  • Jaulin Luc
  • Putot Sylvie
  , 2023. This paper presents a method for determining the area explored by a line-sweep sensor during an area-covering mission in a two-dimensional plane. Accurate knowledge of the explored area is crucial for various applications in robotics, such as mapping, surveillance, and coverage optimization. The proposed method leverages the concept of coverage measure of the environment and its relation to the topological degree in the plane, to estimate the extent of the explored region. In addition, we extend the approach to uncertain coverage measure values using interval analysis. This last contribution allows for a guaranteed characterization of the explored area, essential considering the often critical character of area-covering missions. Finally, this paper also proposes a novel algorithm for computing the topological degree in the 2-dimensional plane, for all the points inside an area of interest, which differs from existing solutions that compute the topological degree for single points. The applicability of the method is evaluated through a real-world experiment. (10.1016/j.ijar.2024.109162)
  DOI : 10.1016/j.ijar.2024.109162
• Magnetospheric Time History in Storm‐Time Magnetic Flux Dynamics
  
  • Akhavan-Tafti M.
  • Atilaw T.
  • Fontaine D.
  • Le Contel O.
  • Slavin J.
  • Pulkkinen T.
  Journal of Geophysical Research Space Physics, American Geophysical Union/Wiley, 2023, 128 (9). Abstract Magnetospheric magnetic flux dynamics is quantified in 29 geomagnetic storms between 2015 and 2019, using near‐equatorial Van Allen Probe, GOES, and Magnetospheric Multiscale satellites. For the first time, concurrent, multi‐probe observations are utilized to preserve magnetospheric time history, defined as the state of the magnetosphere leading up to an observation. It is revealed that, relative to pre‐storm conditions, (a) during the storm sudden commencement (SSC), magnetic flux uniformly increases Δ Ψ = +15% throughout the magnetosphere, except in the nightside inner magnetosphere where Δ Ψ = −30%, and (b) during storm main and recovery phases, Δ Ψ = −30% and −15%, respectively, in the nightside magnetosphere, at radial distances 5 ≤ r [ R E ] < 8. It is found that a symmetric ring current is likely formed in the nightside, early in the storm process (localized during SSC), which then broadens during the main phase, before weakening during the recovery phase. The current system on the dayside shows a distinct dawn‐dusk asymmetry. (10.1029/2023JA031832)
  DOI : 10.1029/2023JA031832
• An acoustic-transport splitting method for the barotropic Baer-Nunziato two-phase flow model
  
  • Ait-Ameur Katia
  • Kokh Samuel
  • Massot Marc
  • Pelanti Marica
  • Pichard Teddy
  ESAIM: Proceedings and Surveys, EDP Sciences, 2023, 72, pp.93-116. This work focuses on the numerical approximation of the barotropic Baer-Nunziato twophase flow model. The scheme relies on an operator splitting method corresponding to a separate treatment of fast propagation phenomena due to the acoustic waves on the one hand and slow propagation phenomena due to the fluid motion on the other. We propose to extend the implicit-explicit schemes developed in [7]. These methods enable the use of time steps that are no longer constrained by the sound velocity thanks to an implicit treatment of the acoustic waves, and maintain accuracy in the subsonic regime thanks to an explicit treatment of the material waves. In the present setting, a particular attention will be also given to the discretization of the non conservative terms in the Baer-Nunziato model. We prove that the proposed numerical strategy preserve positive values of the volume fractions and densities and we illustrate its behaviour with several relevant test cases. (10.1051/proc/202372093)
  DOI : 10.1051/proc/202372093
• On-the-Fly Control of Unknown Systems: From Side Information to Performance Guarantees Through Reachability
  
  • Djeumou Franck
  • Vinod Abraham P
  • Goubault Eric
  • Putot Sylvie
  • Topcu Ufuk
  IEEE Transactions on Automatic Control, Institute of Electrical and Electronics Engineers, 2023, 68, pp.4857 - 4872. (10.1109/tac.2022.3217260)
  DOI : 10.1109/tac.2022.3217260
• Automated approach for source location in shallow waters
  
  • Niclas Angèle
  • Garnier Josselin
  , 2023. This paper proposes a fully automated method for recovering the location of a source and medium parameters in shallow waters. The scenario involves an unknown source emitting low-frequency sound waves in a shallow water environment, and a single hydrophone recording the signal. Firstly, theoretical tools are introduced to understand the robustness of the warping method and to propose and analyze an automated way to separate the modal components of the recorded signal. Secondly, using the spectrogram of each modal component, the paper investigates the best way to recover the modal travel times and provides stability estimates. Finally, a penalized minimization algorithm is presented to recover estimates of the source location and medium parameters. The proposed method is tested on experimental data of right whale gunshot and combustive sound sources, demonstrating its effectiveness in real-world scenarios.
• Nanosecond Surface Dielectric Barrier Discharge: Experimental investigation of streamer to filament transition in high-pressure gases
  
  • Lafaurie V
  • Starikovskaia S
  , 2023. In high-pressure nanosecond surface dielectric barrier discharges (nSDBD) it is common to observe two different type of streamer development: a diffuse mode and a filamentary mode [1-2]. The transition between these modes was defined as the apparition of 2 to 5 filaments with varying applied voltage. In this study, the effects of photoionisation and electronegativity of the gas mixture on the transition voltage were compared for a wide range of pressures and concentrations. In the positive polarity, it was found that photoionisation played a stabilising role on the diffusive discharge for all studied pressures, while electronegativity appeared to reduce the threshold for transition at lower pressures. In the negative polarity, the photoionisation effect was no longer observable while the filament threshold was lowered for more electronegative mixtures. Fundemental understanding of these processes is of great importance in many applications, most of all in plasma-assisted combustion [3-4].
• G-quadruplexes in the evolution of hepatitis B virus
  
  • Brázda Václav
  • Dobrovolná Michaela
  • Bohálová Natália
  • Mergny Jean-Louis
  Nucleic Acids Research, Oxford University Press, 2023, 51. Abstract Hepatitis B virus (HBV) is one of the most dangerous human pathogenic viruses found in all corners of the world. Recent sequencing of ancient HBV viruses revealed that these viruses have accompanied humanity for several millenia. As G-quadruplexes are considered to be potential therapeutic targets in virology, we examined G-quadruplex-forming sequences (PQS) in modern and ancient HBV genomes. Our analyses showed the presence of PQS in all 232 tested HBV genomes, with a total number of 1258 motifs and an average frequency of 1.69 PQS per kbp. Notably, the PQS with the highest G4Hunter score in the reference genome is the most highly conserved. Interestingly, the density of PQS motifs is lower in ancient HBV genomes than in their modern counterparts (1.5 and 1.9/kb, respectively). This modern frequency of 1.90 is very close to the PQS frequency of the human genome (1.93) using identical parameters. This indicates that the PQS content in HBV increased over time to become closer to the PQS frequency in the human genome. No statistically significant differences were found between PQS densities in HBV lineages found in different continents. These results, which constitute the first paleogenomics analysis of G4 propensity, are in agreement with our hypothesis that, for viruses causing chronic infections, their PQS frequencies tend to converge evolutionarily with those of their hosts, as a kind of ‘genetic camouflage’ to both hijack host cell transcriptional regulatory systems and to avoid recognition as foreign material. (10.1093/nar/gkad556)
  DOI : 10.1093/nar/gkad556
• Étude de traitements omniphobes pour les textiles
  
  • Cheibas Cristina
  • El Kaim Laurent
  • Guillard Marianne
  • Six Yvan
  • Zard Samir
  , 2023.
• Fireball - Fourth Industrial REvolution: Blockchain & Artificial inteLLigence
  
  • Lebert Didier
  , 2023.
• Semi-Simplicial Set Models for Distributed Knowledge
  
  • Goubault Éric
  • Kniazev Roman
  • Ledent Jérémy
  • Rajsbaum Sergio
  , 2023. In recent years, a new class of models for multiagent epistemic logic has emerged, based on simplicial complexes. Since then, many variants of these simplicial models have been investigated, giving rise to different logics and axiomatizations. In this paper, we present a further generalization, which encompasses all previously studied variants of simplicial models. Geometrically, this is achieved by generalizing beyond simplicial complexes, and considering instead semi-simplicial sets. By doing so, we define a new semantics for epistemic logic with distributed knowledge, where a group of agents may distinguish two worlds, even though each individual agent in the group is unable to distinguish them. As it turns out, these models are the geometric counterpart of a generalization of Kripke models, called "pseudomodels". We show how to recover the previously defined variants of simplicial models as sub-classes of our models; and give a sound and complete axiomatization for each of them. (10.1109/lics56636.2023.10175737)
  DOI : 10.1109/lics56636.2023.10175737
• On the 3D global dynamics of terrestrial bow-shock rippling in a quasi-perpendicular interaction with steady solar wind
  
  • Cazzola Emanuele
  • Fontaine D.
  • Savoini P.
  Journal of Atmospheric and Solar-Terrestrial Physics, Elsevier, 2023, 246, pp.106053. The phenomenon of bow-shock surface rippling has been studied throughout multidimensional simulations of ad hoc planar shock fronts. However, the investigation of global bow-shock behaviour with a 3D curved scenario has been poorly addressed thus far. In this work, we present an analysis of this scenario occurring during a low-β quasiperpendicular interaction with the interplanetary magnetic field by means of kinetic 3D computer simulations. The analysis was carried out with 3D hybrid simulations properly set to reproduce the interaction between solar wind and a realistic near-Earth environment. We have found that the ripples behave as IMFperpendicular elongated structures extending along the bow-shock meridian plane and propagating parallel to the IMF orientation from the nose towards the flanks with a constant velocity (in the case studied here ∼ 8 times the upstream Alfvén speed). We have also confirmed that these ripples feature a broad range of wavelengths along the entire travel path, as locally observed with past simulations and observations (in this case ≥ 8 d i). Moreover, from a kinetic analysis of the velocity distribution across the bow-shock nose, we have observed global signatures of the occurrence of shock-front reformation. We suggest that, among other kinetic mechanisms, shock-front reformation in the nose region can play an important role in the perturbation of the bow-shock surface, leading to the generation of modulations ultimately propagating along the bow-shock surface as MHD waves. (10.1016/j.jastp.2023.106053)
  DOI : 10.1016/j.jastp.2023.106053
• Microstructure, Mechanical Properties, and Thermal Stability of Al-Al2O3 Nanocomposites Consolidated by ECAP or SPS from Milled Powders
  
  • Lacour-Gogny-Goubert Antoine
  • Doquet Véronique
  • Novelli Marc
  • Tanguy Alexandre
  • Hallais Simon
  • Bourgon Julie
  • Villeroy Benjamin
  • Massion Roxane
  Metals, MDPI, 2023, 13 (5), pp.825. Ultrafine-grained Al matrix nanocomposites, reinforced with Al2O3 nanoparticles, were produced from milled powders, either by equal channel angular pressing (ECAP), at room or high temperature, with or without back pressure, or by spark plasma sintering (SPS). Their microstructures, mechanical properties (compression, hardness, and sliding wear), and thermal stabilities (thermally induced softening and cracking) were compared, and the advantages and limitations of each process discussed on a scientific but also practical point of view. For the most successful set of process parameters, the yield stress in compression reached 380 MPa, the hardness, HV = 139, remained stable up to 500 °C, and the resistance to sliding wear was comparable to that of Al 5083, and better than that of Al 7075-T6. While the samples consolidated at high temperatures (by ECAP or SPS) showed a good thermal stability, those consolidated by ECAP at room temperature were prone to thermally induced softening and cracking, which was related to trapped and pressurized gases. (10.3390/met13050825)
  DOI : 10.3390/met13050825
• Packed-Ensembles for Efficient Uncertainty Estimation
  
  • Laurent Olivier
  • Lafage Adrien
  • Tartaglione Enzo
  • Daniel Geoffrey
  • Martinez Jean-Marc
  • Bursuc Andrei
  • Franchi Gianni
  , 2022. Deep Ensembles (DE) are a prominent approach for achieving excellent performance on key metrics such as accuracy, calibration, uncertainty estimation, and out-of-distribution detection. However, hardware limitations of real-world systems constrain to smaller ensembles and lower-capacity networks, significantlydeteriorating their performance and properties. We introduce Packed-Ensembles(PE), a strategy to design and train lightweight structured ensembles by carefully modulating the dimension of their encoding space. We leverage grouped convolu-tions to parallelize the ensemble into a single shared backbone and forward pass to improve training and inference speeds. PE is designed to operate within the memory limits of a standard neural network. Our extensive research indicates that PE accurately preserves the properties of DE, such as diversity, and performs equally well in terms of accuracy, calibration, out-of-distribution detection, and robustness to distribution shift. We make our code available at github.com/ENSTA-U2IS/torch-uncertainty.
• Waveform Inversion with a Data Driven Estimate of the Internal Wave
  
  • Borcea Liliana
  • Garnier Josselin
  • Mamonov Alexander
  • Zimmerling Jörn
  SIAM Journal on Imaging Sciences, Society for Industrial and Applied Mathematics, 2023, 16 (1), pp.280-312. We study an inverse problem for the wave equation, concerned with estimating the wave speed from data gathered by an array of sources and receivers that emit probing signals and measure the resulting waves. The typical approach to solving this problem is a nonlinear least squares minimization of the data misfit, over a search space. There are two main impediments to this approach, which manifest as multiple local minima of the objective function: The nonlinearity of the mapping from the wave speed to the data, which accounts for multiple scattering effects, and poor knowledge of the kinematics (smooth part of the wave speed), which causes cycle skipping. We show that the nonlinearity can be mitigated using a data driven estimate of the wave field at points inside the medium, also known as the “internal wave field.” This leads to improved performance of the inversion for a reasonable initial guess of the kinematics. (10.1137/22M1517342)
  DOI : 10.1137/22M1517342
• Waveform inversion via reduced order modeling
  
  • Borcea Liliana
  • Garnier Josselin
  • Mamonov Alexander
  • Zimmerling Jörn
  Geophysics, Society of Exploration Geophysicists, 2023, 88 (2), pp.R175-R191. We introduce a novel approach to waveform inversion based on a data-driven reduced order model (ROM) of the wave operator. The presentation is for the acoustic wave equation, but the approach can be extended to elastic or electromagnetic waves. The data are time resolved measurements of the pressure wave gathered by an acquisition system that probes the unknown medium with pulses and measures the generated waves. We propose to solve the inverse problem of velocity estimation by minimizing the square misfit between the ROM computed from the recorded data and the ROM computed from the modeled data, at the current guess of the velocity. We give a step by step computation of the ROM, which depends nonlinearly on the data and yet can be obtained from them in a noniterative fashion, using efficient methods from linear algebra. We also explain how to make the ROM robust to data inaccuracy. The ROM computation requires the full array response matrix gathered with colocated sources and receivers. However, we find that the computation can deal with an approximation of this matrix, obtained from towed-streamer data using interpolation and reciprocity on-the-fly. Although the full-waveform inversion approach of nonlinear least-squares data fitting is challenging without low-frequency information, due to multiple minima of the data fit objective function, we find that the ROM misfit objective function has better behavior, even for a poor initial guess. We also find by explicit computation of the objective functions in a simple setting that the ROM misfit objective function has convexity properties, whereas the least-squares data fit objective function displays multiple local minima. (10.1190/geo2022-0070.1)
  DOI : 10.1190/geo2022-0070.1
• Making superhydrophobic splashes by surface cooling
  
  • Grivet Rodolphe
  • Huerre Axel
  • Séon Thomas
  • Josserand Christophe
  Physical Review Fluids, American Physical Society, 2023, 8 (6), pp.063603. We study experimentally the enhancement of splashing due to solidification. Investigating the impact of water drops on dry smooth surfaces, we show that the transition velocity to splash can be drastically reduced by cooling the surface below the liquid melting temperature. We find that at very low temperatures (below −60 • C), the splashing behaviour becomes independent of surface undercooling and presents the same characteristics as on ambient temperature superhydrophobic surfaces. This resemblance arises from an increase of the dynamic advancing contact angle of the lamella with surface undercooling, going from the isothermal hydrophilic to the superhydrophobic behaviour. We propose that crystal formation can affect the dynamic contact angle of the lamella, which would explain this surprising transition. Finally, we show that the transition from hydrophilic to superydrophobic behaviour can also be characterized quantitatively on the dynamics of the ejecta. (10.1103/PhysRevFluids.8.063603)
  DOI : 10.1103/PhysRevFluids.8.063603
• Rebalancing gradient to improve self-supervised co-training of depth, odometry and optical flow predictions
  
  • Hariat Marwane
  • Manzanera Antoine
  • Filliat David
  , 2023, pp.1267-1276. We present CoopNet, an approach that improves the cooperation of co-trained networks by dynamically adapting the apportionment of gradient, to ensure equitable learning progress. It is applied to motion-aware self-supervised prediction of depth maps, by introducing a new hybrid loss, based on a distribution model of photo-metric reconstruction errors made by, on the one hand the depth + odometry paired networks, and on the other hand the optical flow network. This model essentially assumes that the pixels from moving objects (that must be discarded for training depth and odometry), correspond to those where the two reconstructions strongly disagree. We justify this model by theoretical considerations and experimental evidences. A comparative evaluation on KITTI and CityScapes datasets shows that CoopNet improves or is comparable to the state-of-the-art in depth, odometry and optical flow predictions. Our code is available here: https://github.com/mhariat/CoopNet.