ATLANTIS

ATLANTIS - 2025

2025Activity report‌‌Project-TeamATLANTIS

RNSR: 202023535Z

Research center Inria Centre‌ at Université Côte d'Azur‌
In partnership with:CNRS,‌‌ Université Côte d'Azur
Team name: modeling and numerical‌ methods for computATionaL wave-mAtter‌ iNteracTIons at the nanoScale‌‌
In collaboration with:Laboratoire Jean-Alexandre Dieudonné (JAD)

Creation‌ of the Project-Team: 2020‌ February 01

Each year,‌‌ Inria research teams publish‌ an Activity Report presenting their work and results‌ over the reporting period. These reports follow a‌ common structure, with some optional sections depending on‌ the specific team. They typically begin by outlining‌ the overall objectives and research programme, including the‌ main research themes, goals, and methodological approaches. They‌ also describe the application domains targeted by the‌ team, highlighting the scientific or societal contexts in‌ which their work is situated.

The reports then‌ present the highlights of the year, covering major‌ scientific achievements, software developments, or teaching contributions. When‌ relevant, they include sections on software, platforms, and‌ open data, detailing the tools developed and how‌ they are shared. A substantial part is dedicated‌ to new results, where scientific contributions are described‌ in detail, often with subsections specifying participants and‌ associated keywords.

Finally, the Activity Report addresses funding,‌ contracts, partnerships, and collaborations at various levels, from‌ industrial agreements to international cooperations. It also covers‌ dissemination and teaching activities, such as participation in‌ scientific events, outreach, and supervision. The document concludes‌ with a presentation of scientific production, including major‌ publications and those produced during the year.

Keywords‌

Computer Science and Digital Science

A1.1.4. High performance‌ computing
A1.1.5. Exascale
A6.1.5. Multiphysics modeling
A6.2.1. Numerical‌ analysis of PDE and ODE
A6.2.6. Optimization
A6.2.7.‌ HPC for machine learning
A6.3.4. Model reduction
A6.5.4.‌ Waves
A9.2.5. Bayesian methods
A9.2.6. Neural networks
A9.2.8.‌ Deep learning

1 Team members, visitors, external collaborators

Research‌ Scientists

Stéphane Lanteri [Team leader, INRIA‌, Senior Researcher, HDR]
Mahmoud Elsawy‌ [INRIA, ISFP, HDR]

Faculty‌ Members

Stéphane Descombes [Université Côte d'Azur,‌ Professor, HDR]
Claire Scheid [Université‌ Côte d'Azur, Associate Professor Delegation, from‌ Sep 2025, HDR]
Claire Scheid [‌Université Côte d'Azur, Associate Professor, until‌ Aug 2025, HDR]

Post-Doctoral Fellows

Ayoub‌ Bellouch [Inria, Post-Doctoral Fellow, until‌ Sep 2025]
Alemayehu Getahun Kumela [Inria‌, Post-Doctoral Fellow]
Francisco Teixeira Orlandini [‌Inria, Post-Doctoral Fellow, from Sep 2025‌]

PhD Students

Arthur Clini De Souza [‌Solnil, CIFRE]
Carlotta Filippin [Inria‌]
Roman Gelly [Université Côte d'Azur]‌
Daria Hrebenshchykova [Inria]
Enzo Isnard [‌Thales Research & Technology, CIFRE, until‌ Oct 2025]
Thibault Laufroy [Université Côte‌ d'Azur]
Cedric Legrand [Université Côte d'Azur‌]
Martin Lepers [STMicroelectronics, CIFRE]
Huynh‌ Thanh Phuong Lê [Inria, from Jul‌ 2025]
Florentin Proust [Université Côte d'Azur‌, from Nov 2025]
Florentin Proust [‌Inria, until Oct 2025]
Alexandre Pugin‌ [Inria]

Technical Staff

Alexis Gobe [‌Inria, Engineer]
Arthur Gouinguenet [CNRS‌, Engineer, until May 2025]
Enzo Isnard [Inria,‌ Engineer, from Dec‌ 2025]
Alan Youssef‌‌ [Inria, Engineer, until Jun 2025‌]

Interns and Apprentices‌

Julien Noel [Inria‌‌, Apprentice, until Aug 2025]

Administrative‌ Assistant

Delphine Robache [‌Inria]

2 Overall‌‌ objectives

Nanostructuring of materials has paved the way‌ for manipulating and enhancing‌ wave-matter interactions, thereby opening‌‌ the door for the full control of these‌ interactions at the nanoscale.‌ In particular, the interaction‌‌ of light waves (or more general optical waves‌) with matter is‌ a subject of rapidly‌‌ increasing scientific importance and technological relevance. Indeed, the‌ corresponding science, referred to‌ as nanophotonics59,‌‌ aims at using nanoscale light-matter interactions to achieve‌ an unprecedented level of‌ control on light. Nanophotonics‌‌ encompasses a wide variety of topics, including metamaterials,‌ plasmonics, high resolution imaging,‌ quantum nanophotonics and functional‌‌ photonic materials. Previously viewed as a largely academic‌ field, nanophotonics is now‌ entering the mainstream, and‌‌ will play a major role in the development‌ of exciting new products,‌ ranging from high efficiency‌‌ solar cells, to personalized health monitoring devices able‌ to detect the chemical‌ composition of molecules at‌‌ ultralow concentrations. Plasmonics 62 is a field closely‌ related to nanophotonics. Metallic‌ nanostructures whose optical scattering‌‌ is dominated by the response of the conduction‌ electrons are considered as‌ plasmonic media. If the‌‌ metallic structure presents an interface with a positive‌ dielectric permittivity, collective oscillations‌ of surface electrons create‌‌ waves (called surface plasmons) that are guided along‌ the interface, with the‌ unique characteristic of subwavelength-scale‌‌ confinement. Nanofabricated systems that exploit these plasmon waves‌ offer fascinating opportunities for‌ crafting and controlling the‌‌ propagation of light in matter. In particular, it‌ can be used to‌ channel light efficiently into‌‌ nanometer-scale volumes. As light is squeezed down into‌ nanoscale volumes, field enhancement‌ effects occur resulting in‌‌ new optical phenomena that can be exploited to‌ challenge existing technological limits‌ and deliver superior photonic‌‌ devices. The resulting enhanced sensitivity of light to‌ external parameters (for example,‌ an applied electric field‌‌ or the dielectric constant of an adsorbed molecular‌ layer) shows also great‌ promises for applications in‌‌ sensing and switching.

In ATLANTIS, our research activities‌ aim at studying and‌ impacting some scientific and‌‌ technological challenges raised by physical problems involving optical‌ waves in interaction with‌ nanostructured matter. A crucial‌‌ component in the implementation of this scientific endeavor‌ lies in a close‌ networking with physicists who‌‌ bring the experimental counterpart of the proposed research.‌ Driven by a number‌ of nanophotonics-related physical drivers,‌‌ our overall objectives are to design and develop‌ innovative numerical methodologies for‌ the simulation of nanoscale‌‌ light-matter interactions and to demonstrate their capabilities by‌ studying challenging applications in‌ close collaboration with our‌‌ physicist partners. On the methodological side, the Discontinous‌ Galerkin (DG) family of‌ methods is a cornerstone‌‌ of our contributions. In particular, we study various‌ variants of DG methods‌ that can deal with‌‌ complex material models and‌ coupled PDE systems that are relevant to the‌ study of nanoscale light-matter interactions. Moreover, mathematical modeling‌ is a central activity of the team, in‌ particular for shaping initial and boundary value problems‌ in view of devising accurate, efficient and robust‌ numerical methods in the presence of multiple space‌ and time scales or/and geometrical singularities. Additional methodological‌ topics that are considered in close collaboration with‌ colleagues from other Inria teams or external applied‌ mathematics research groups are model order reduction, inverse‌ design. Novel methodological contributions on these topics in‌ the context of the physical problems studied in‌ ATLANTIS are eventually implemented in the DIOGENeS software‌ suite, which is a unique software plaform dedicated‌ to computational nanophotonics.

3 Research program

3.1 Driving‌ physical fields

Our research activities eventually materialize as‌ innovative computational techniques for studying concrete questions and‌ applications that are tightly linked to specific physical‌ fields (driving physical fields) related to nanophotonics and‌ plasmonics. In most cases, these scientific topics and‌ applications are addressed in close collaboration with physicists.‌

Quantum plasmonics. The physical phenomena involved in the‌ deep confinement of light when interacting with matter‌ opens a major route for novel nanoscale devices‌ design. Indeed, the recent progress of fabrication at‌ the nanoscale makes it possible to conceive metallic‌ structures with increasingly large size mismatch, in which‌ microscale devices can be characterized by sub-nanometer features‌ 55. These advances have also allowed to‌ achieve spatial separation between metallic elements of only‌ few nanometers 53. At such sizes quantum‌ effects become non-negligible, producing huge variations in the‌ macroscopic optical response. Following this evolution, the quantum‌ plasmonics field has emerged, and with it the‌ possibility of building quantum-controled devices, such as single‌ photon sources, transistors and ultra-compact circuitry at the‌ nanoscale. In ATLANTIS, we study novel numerical modeling‌ methods for solving some semi-classical models of quantum‌ plasmonic effects such as in the context of‌ the PhD work of Nikolkai Schmitt 69-‌24.

Planar optics. Nanostructuring of matter can‌ be tailored to shape, control wavefront and achieve‌ unusual device operations. Recent years have seen tremendous‌ advances in the fabrication and understanding of two-dimensional‌ (2D) materials, giving rise to the field of‌ planar optics. In particular, the concept of quasi-2D‌ metasurfaces has started to develop into an exciting‌ research area, where nanostructured surfaces are designed for‌ novel functionalities 60-54-56.‌ Metasurfaces are planar metamaterials with subwavelength thickness, consisting‌ of single-layer or few-layer stacks of nanostructures. They‌ can be readily fabricated using lithography and nanoprinting‌ methods, and the ultrathin thickness in the wave‌ propagation direction can greatly suppress the undesirable losses.‌ Metasurfaces enable a spatially varying optical response (e.g.‌ scattering amplitude, phase, and polarization). They mold optical‌ wavefronts into shapes that can be designed at‌ will, and facilitate the integration of functional materials‌ to accomplish active control and greatly enhanced nonlinear‌ response. Our first contributions on this topic have been obtained in the‌ context of the ANR‌ OPERA project (completed in‌‌ September 2022) and are concerned with numerical modeling‌ methods for the inverge‌ design of metasurfaces 11‌‌-9 and metalenses 10.

Thermoplasmonics. Plasmonic‌ resonances can be exploited‌ for many applications 62‌‌. In particular, the strong local field enhancement‌ associated with the plasmonic‌ resonances of a metallic‌‌ nanostructure, together with the absorption properties of the‌ metal, induce a photo-thermal‌ energy conversion. Thus, in‌‌ the vicinity of the nanostructure, the temperature increases.‌ These effects, viewed as‌ ohmic losses, have been‌‌ for a long time considered as a severe‌ drawback for the realization‌ of efficient devices. However,‌‌ the possibility to control this temperature rise with‌ the illumination wavelength or‌ polarization has gathered strong‌‌ interest in the nano-optics community, establishing the basis‌ of thermoplasmonics 51.‌ By increasing temperature in‌‌ their surroundings, metal nanostructures can be used as‌ integrated heat nanosources. Decisive‌ advances are foreseen in‌‌ nanomedicine with applications in photothermal cancer therapy, nano-surgery,‌ drug delivery, photothermal imaging,‌ protein tracking, photoacoustic imaging,‌‌ but also in nano-chemistry, optofluidics, solar and thermal‌ energy harvesting (thermophotovoltaics).

Optoelectronics‌ and nanoelectronics. Semiconductors also‌‌ play a major role in leveraging nanoscale light-matter‌ interactions. Emission or absorption‌ of light by a‌‌ semiconductor is at the heart of optoelectronics, which‌ is concerned with devices‌ that source, detect or‌‌ control light. Photodiodes, solar cells, light emitting diodes‌ (LEDs), optical fibers and‌ semiconductor lasers are some‌‌ typical examples of optoelectronic devices. The attractive properties‌ of these devices is‌ based on their efficiency‌‌ in converting light into electrical signals (or vice‌ versa). Using a structuration‌ with low dimensional materials‌‌ and carrier-photons interaction, optoelectronics aims at improving the‌ quality of these systems.‌ A closeby field is‌‌ nanoelectronics 63, i.e., the physical field that,‌ while incorporating manufacturing constraints,‌ tries to describe and‌‌ understand the influence of the nanostructuration of electronic‌ devices on their electronic‌ properties. This area has‌‌ quickly evolved with the increasing fabrication capabilities. One‌ striking motivating example is‌ the drastic increase of‌‌ the number of transistors (of a few nanometer‌ size) per chip on‌ integrated circuits. At the‌‌ achieved nanostructuration scales, inter-atomic forces, tunneling or quantum‌ mechanical properties have a‌ non-negligible impact. A full‌‌ understanding of these effects is mandatory for exploiting‌ them in the design‌ of electronic components, thereby‌‌ improving their characteristics.

3.2 Research agenda

The processes‌ that underly the above-described‌ physical fields raise a‌‌ number of modeling challenges that motivate our research‌ agenda:

They exhibit multiple‌ space and time scales;‌‌
They are highly sensitive to exquisite geometrical features‌ of nanostructures and matter‌ nanostructuring;
They impose dealing‌‌ with unconventional material models;
They may require to‌ leave the comfortable setting‌ of linear differential models;‌‌
Some of them are inherently multiphysics processes.

3.2.1‌ Core research topics

Our‌ research activities are organized‌‌ around core theoretical and methodological topics to address‌ the above-listed modeling challenges.‌

High order DG methods.‌‌ Designing numerical schemes that‌ are high order accurate on general meshes, i.e.,‌ unstructured or hybrid structured/unstructured meshes, is a major‌ objective of our core research activities in ATLANTIS.‌ We focus on the family of Discontinuous Galerkin‌ (DG) methods that has been extensively developed for‌ wave propagation problems during the last 15 years.‌ We investigate several variants, namely nodal DGTD for‌ time-domain problems, and HDG (Hybridized DG) for frequency-domain‌ problems, with the general goal of devising, analyzing‌ and developing extensions of these methods in order‌ to deal with the above-mentioned physical drivers: nonlinear‌ features, in particular in relation with generation of‌ higher order harmonics in electromagnetic wave interaction with‌ nonlinear materials, and nonlinear models of electronic response‌ in metallic and semiconductor materials; multiphysic couplings such‌ as for instance when considering PDE models relevant‌ to thermoplasmonics, optoelectronics and nanoelectronics. There are to‌ date very few works promoting DG-type methods for‌ these situations. Our methodological contributions of these methods‌ eventually materialize in the DIOGENeS software suite.

Time‌ integration for multiscale problems. Multiscale physical problems with‌ complex geometries or heterogeneous media are extremely challenging‌ for conventional numerical simulations. Adaptive mesh refinement is‌ an attractive technique for treating such problems and‌ will be developed in our research activities in‌ ATLANTIS. Local mesh refinement imposes a severe stability‌ condition on explicit time integration since the allowed‌ maximal time step size is constrained by the‌ smallest element in the mesh. We consider different‌ ways to overcome this stability condition, especially by‌ using implicit-explicit (IMEX) methods where a time implicit‌ scheme is used only for the refined part‌ of the mesh, and a time explicit scheme‌ is used for the other part.

Reduced-order and‌ surrogate modeling. Reduced-order modeling aims at reducing the‌ computational requirements of costly high-fidelity solution methods while‌ maintaining an acceptable level of accuracy. One of‌ the most studied methods for establishing the reduced-order‌ model is the Proper Orthogonal Decomposition (POD), also‌ known as Karhunen-Loéve decomposition, principal component analysis, or‌ singular value decomposition, which uses the solutions of‌ high fidelity numerical simulations or experiments at certain‌ time instants, typically called snapshots, to compute‌ a set of POD basis vectors spanning a‌ low-dimensional space. POD is very popular in the‌ computational fluid dynamics field. However, the development of‌ POD for electromagnetics has been more scarce. We‌ study POD-based reduced-order modeling strategies in the context‌ of a long term collaboration that has started‌ in 2018 with researchers at the School of‌ Mathematical Sciences of the University of Electronic Science‌ and Technology of China and Southwest University of‌ Finance & Economics, which are both located in‌ Chengdu. In the context of this collabortaion, we‌ have proposed and developed several reduced-order modeling techniques,‌ from intrusive to fully data-driven and non-intrusive methods,‌ for time-domain electromagnetics. Alternatively, several works in the‌ recent years have promoted highly efficient surrogate modeling‌ approaches based on Deep Neural Networks (DNNs) to‌ achieve non-linear reduced-order modeling. This is also a novel direction of investigation‌ that we have started‌ to consider in 2023‌‌ in the team.

Scientific Machine Learning. Scientific Machine‌ Learning (SciML) is a‌ relatively new research field‌‌ based on both machine learning (ML) and scientific‌ computing tools. Its aim‌ is the development of‌‌ new methods to solve several kinds of problems,‌ which can be forward‌ solution of multidimensional partial‌‌ differential equations, identification of parameters, or inverse problems.‌ The methods that are‌ investigated in this context‌‌ must be robust, reliable and interpretable. These new‌ SciML tools should also‌ allow the natural inclusion‌‌ of data in the numerical simulation in order‌ to generate new results.‌ We initiated in 2022‌‌ a new research direction on a particular family‌ of DNNs referred as‌ Physics-Informed Neural Networks (PINNs)‌‌ 67 that we investigate for PDE models that‌ are relevant to nanophotonics‌ with the goal of‌‌ designing non-intrusive surrogate modeling approaches that require a‌ minimal amount of training‌ data.

Dealing with complex‌‌ materials. Physically relevant simulations deal with increasing levels‌ of complexity in the‌ geometrical and/or physical characteristics‌‌ of nanostructures, as well as their interaction with‌ light. Standard simulation methods‌ may fail to reproduce‌‌ the underlying physical phenomena, therefore motivating the search‌ for more sophisticated light-matter‌ interaction numerical modeling strategies.‌‌ A first direction consists in refining classical linear‌ dispersion models and we‌ put a special focus‌‌ on deriving a complete hierarchy of models, that‌ will encompass standard linear‌ models to more complex‌‌ and nonlinear ones (such as Kerr-type materials, nonlinear‌ quantum hydrodynamic theory models,‌ etc.). One possible approach‌‌ relies on an accurate description of the Hamiltonian‌ dynamics with intricate kinetic‌ and exchange correlation energies,‌‌ for different modeling purposes. A second direction is‌ motivated by the study‌ of 2D materials. A‌‌ major concern is centered around the choice of‌ the modeling approach between‌ a full costly 3D‌‌ modeling and the use of equivalent boundary conditions,‌ that could in all‌ generality be nonlinear. Assessing‌‌ these two directions requires efficient dedicated numerical algorithms‌ that are able to‌ tackle several types of‌‌ nonlinearities and scales.

Dealing with coupled models. Several‌ of our target physical‌ fields are multiphysics in‌‌ essence and require going beyond the sole description‌ of the electromagnetic response.‌ In thermoplasmonics, the various‌‌ phenomena (heat transfer through light concentration, bubbles formation‌ and dynamics) call for‌ different kinds of governing‌‌ PDEs (Maxwell, conduction, fluid dynamics). Since, in addition,‌ these phenomena can occur‌ in significatively different space‌‌ and time scales, drawing a quite complete picture‌ of the underlying physics‌ is a challenging task,‌‌ both in terms of modeling and numerical treatment.‌ In the nanoelectronics field,‌ an accurate description of‌‌ the electronic properties involves including quantum effects. A‌ coupling between Maxwell’s and‌ Schrödinger's equations (again at‌‌ significantly different time and space scales) is a‌ possible relevant scenario. In‌ the optoelectronics field, the‌‌ accurate prediction of semiconductors optical properties is a‌ major concern. A possible‌ strategy may require to‌‌ solve both the electromagnetic‌ and the drift-diffusion equations. In all these aforementioned‌ examples, difficulties mainly arise both from the differences‌ in physical nature as well as in the‌ time/space scales at which each physical phenomenon occurs.‌ Accurately modeling/solving their coupled interactions remains a formidable‌ challenge.

High performance computing (HPC). HPC is transversal‌ to almost all the other research topics considered‌ in the team, and is concerned with both‌ numerical algorithm design and software development. We work‌ toward taking advantage of fine grain massively parallel‌ processing offered by GPUs in modern exascale architectures,‌ by revisiting the algorithmic structure of the computationally‌ intensive numerical kernels of the high order DG-based‌ solvers that we develop in the framework of‌ the DIOGENeS software suite.

3.2.2 Complementary topics

Beside‌ the above-discussed core research topics, we have also‌ identified additional topics that are important or compulsory‌ in view of maximizing the impact in nanophotonics‌ or nanophononics of our core activities and methodological‌ contributions.

Numerical optimization. Inverse design has emerged rather‌ recently in nanophotonics, and is currently the subject‌ of intense research as witnessed by several reviews‌ 64. Artificial Intelligence (AI) techniques are also‌ increasingly investigated within this context 70. In‌ ATLANTIS, we will extend the modeling capabilities of‌ the DIOGENeS software suite by using statistical learning‌ techniques for the inverse design of nanophotonic devices.‌ When it is linked to the simulation of‌ a realistic 3D problem making use of one‌ of the high order DG and HDG solvers‌ we develop, the evaluation of a figure of‌ merit is a costly process. Since a sufficiently‌ large input data set of candidate designs, as‌ required by using Deep Learning (DL), is generally‌ not available, global optimization strategies relying on Gaussian‌ Process (GP) models are considered in the first‌ place. This activity will be conducted in close‌ collaboration with researchers of the ACUMES project-team. In‌ particular, we investigate GP-based inverse design strategies that‌ were initially developed for optimization studies in relation‌ with fluid flow problems 57-58 and‌ fluid-structure interaction problems 68.

Uncertainty analysis and‌ quantification. The automatic inverse design of nanophotonic devices‌ enables scientists and engineers to explore a wide‌ design space and to maximize a device performance.‌ However, due to the large uncertainty in the‌ nanofabrication process, one may not be able to‌ obtain a deterministic value of the objective, and‌ the objective may vary dramatically with respect to‌ a small variation in uncertain parameters. Therefore, one‌ has to take into account the uncertainty in‌ simulations and adopt a robust design model 61‌. We study this topic in close collaboration‌ with researchers of the ACUMES project-team one on‌ hand, and researchers at TU Braunschweig in Germany.‌

Numerical linear algebra. Sparse linear systems routinely appear‌ when discretizing frequency-domain wave-matter interaction PDE problems. In‌ the past, we have considered direct methods, as‌ well as domain decomposition preconditioning coupled with iterative‌ algorithms to solve such linear systems 21. In the future, we‌ would like to further‌ enhance the efficiency of‌‌ our solvers by considering state-of-the-art linear algebra techniques‌ such as block Krylov‌ subspace methods 50,‌‌ or low-rank compression techniques 66. We will‌ also focus on multi-incidence‌ problems in periodic structures,‌‌ that are relevant to metagrating or metasurface design.‌ Indeed, such problems lead‌ to the resolution of‌‌ several sparse linear systems that slightly differ from‌ one another and could‌ benefit from dedicated solution‌‌ algorithms. We will collaborate with researchers of the‌ CONCACE (Inria center at‌ Université de Bordeaux) industrial‌‌ project-team to develop efficient and scalable solution strategies‌ for such questions.

4‌ Application domains

Nanoscale wave-matter‌‌ interactions find many applications of industrial and societal‌ relevance. The applications discussed‌ in this section are‌‌ those that we address in the first place‌ in the short- to‌ medium-term. Our general goal‌‌ is to impact scientific discovery and technological development‌ in these application topics‌ by leveraging our methodological‌‌ contributions for the numerical modeling of nanoscale wave-matter‌ interactions, and working in‌ close collaboration with external‌‌ partners either from the academic or the industrial‌ world. Each of these‌ applications is linked to‌‌ one or more of the driving physical fields‌ described in section 3.1‌ except nanoelectronics that we‌‌ consider as a more prospective, hence long-term application.‌

4.1 Nanostructures for sunlight‌ harvesting

Photovoltaics (PV) converts‌‌ photon energy from the sun into electric energy.‌ One of the major‌ challenges of the PV‌‌ sector is to achieve high conversion efficiencies at‌ low cost. Indeed, the‌ ultimate success of PV‌‌ cell technology requires substantial progress in both cost‌ reduction and efficiency improvement.‌ An actively studied approach‌‌ to simultaneously achieve both objectives is to exploit‌ light trapping schemes. Light‌ trapping enables solar cells‌‌ absorption using an active material layer much thinner‌ than the material intrinsic‌ absorption length. This then‌‌ reduces the amount of materials used in PV‌ cells, cuts cell cost,‌ facilitates mass production of‌‌ these cells that are based on less abundant‌ material and moreover can‌ improve cell efficiency (due‌‌ to better collection of photogenerated charge carriers). Enhancing‌ the light absorption in‌ ultrathin film silicon solar‌‌ cells is thus of paramount importance for improving‌ efficiency and reducing costs.‌ Our activities in relation‌‌ with this application field aim at precisely studying‌ light absorption in nanostructured‌ solar cell structures (see‌‌ Fig. 1). We consider both the accurate‌ simulation of light trapping‌ for a given texturing‌‌ of material layers, and the goal-oriented inverse design‌ of the geometrical characteristics‌ of nanostructuring. This application‌‌ domain is studied in collaboration with physicists from‌ C2N (Centre for Nanosciences‌ and Nanotechnology) in Campus‌‌ Paris-Saclay), from LAAS (Laboratoire d'Analyse et d'Architecture des‌ Systèmes) in Toulouse and‌ from the Fraunhofer-Institut für‌‌ Solare Energiesysteme ISE in Freiburg, Germany.

Figure 1: Example of a PV cell‌ with nanocone gratings. Field map of the module‌ of the electric field.

4.2 Metasurfaces for light‌ front shaping

In the last decade metasurfaces have‌ revolutionized the field of optics with the promise‌ to replace bulky and difficult-to-align optical components with‌ ultrathin and flat devices like metagratings, metalenses and‌ metaholograms, which can also implement new functionalities in‌ terms of aberrations correction and arbitrary wavefront shaping.‌ Metasurfaces produce abrupt changes over the scale of‌ the free-space wavelength in the phase, amplitude and/or‌ polarization of a light beam. Metasurfaces are generally‌ created by assembling arrays of miniature, anisotropic light‌ scatterers, e.g., resonators such as optical antennas. The‌ spacing between antennas and their dimensions are much‌ smaller than the wavelength. As a result the‌ metasurfaces, on account of Huygens principle, are able‌ to mould optical wavefronts into arbitrary shapes with‌ subwavelength resolution by introducing spatial variations in the‌ optical response of the light scatterers (see Fig.‌ 2). Designing metasurfaces for realistic applications such‌ as metalenses is a challenging inverse problem. In‌ this context, an important line of research of‌ the team during the last years has been‌ dedicated to improve the capabilities of these numerical‌ tools to produce novel inverse design methodologies for‌ optical metasurfaces. This application domain is studied in‌ collaboration with several groups of physicists in France‌ and abroad, in particular from CRHEA (Centre de‌ Recherche sur l'Hétéro-Epitaxie et ses Applications) in Sophia‌ Antipolis, MPQ (Matériaux et Phénomènes Quantiques) at Université‌ Paris Cité and EPFL in Lausanne, Switzerland.

Figure 2: Different light‌ front shaping schemes with a metasurface.

4.3 Nanostructuring‌ for THz wave generation

Recent research on the‌ interaction of short optical pulses with semiconductors has‌ stimulated the development of low power terahertz (THz)‌ radiation transmitters. The THz spectral range of electromagnetic‌ waves (0.1 to 10 THz) is of great‌ interest. In particular, it includes the excitation frequencies‌ of semiconductors and dielectrics, as well as rotational‌ and vibrational resonances of complex molecules. As a‌ result, THz waves have many applications in areas‌ ranging from the detection of dangerous or illicit substances and biological sensing‌ to diagnosis and diseases‌ treatment in medicine. The‌‌ most common mecanism of THz generation is based‌ on the use of‌ THz photoconductive antennas (PCA),‌‌ consisting of two electrodes spaced by a given‌ gap and placed onto‌ a semiconductor surface. The‌‌ excitation of the gap by a femtosecond optical‌ pulse induces a sharp‌ increase of the concentration‌‌ of charge carriers for a short period of‌ time, and a THz‌ pulse is generated. Computer‌‌ simulation plays a central role in understanding and‌ mastering these phenomena in‌ order to improve the‌‌ design of PCA devices. The numerical modeling of‌ a general 3D PCA‌ configuration is a challenging‌‌ task. Indeed, it requires the simultaneous solution of‌ charge transport in the‌ semiconductor substrate and the‌‌ electromagnetic wave radiation from the antenna in fullwave‌ context. The recently-introduced concept‌ of hybrid photoconductive antennas‌‌ leveraging plasmonic effects is even more challenging since‌ it requires to include‌ plasmonic nanostructures in the‌‌ modeling setting. So far, existing simulation approaches are‌ based on the Finite‌ Difference Time-Domain (FDTD) method,‌‌ and are only able to deal with classical‌ PCAs. In relation with‌ the design of photonic‌‌ devices for THz waves generation and manipulation, we‌ intend to develop a‌ multiscale numerical modeling strategy‌‌ for solving the system of Maxwell equations coupled‌ to various models of‌ charge carrier dynamics in‌‌ semiconductors. Our first achievements on this topic have‌ been obtained in the‌ context of the PhD‌‌ thesis of Massimiliano Montone defended in June 2023‌ (see Fig. 3).‌

Figure 3.a — Figure 3: Left: unstructured mesh of‌ a model of a‌ plasmon-enhanced antenna. Right: concurrent‌‌ snapshots of photo-generation and electron concentration at relevant‌ times in the illuminated‌ region of a plasmon-enhanced‌‌ antenna. Times are marked in ps, in ascending‌ order, from the top‌ to the bottom.

Figure 3.b — Figure 3: Left: unstructured mesh of‌ a model of a‌ plasmon-enhanced antenna. Right: concurrent‌‌ snapshots of photo-generation and electron concentration at relevant‌ times in the illuminated‌ region of a plasmon-enhanced‌‌ antenna. Times are marked in ps, in ascending‌ order, from the top‌ to the bottom.

4.4‌‌ Plasmonic nanostructures for nanoscale sensing

The propagation of‌ light in a slit‌ between metals is known‌‌ to give rise to guided modes. When the‌ slit is of nanometric‌ size, plasmonic effects must‌‌ be taken into account, since most of the‌ mode propagates inside the‌ metal. Indeed, light experiences‌‌ an important slowing-down in the slit, the resulting‌ mode being called gap‌ plasmon. Hence, a metallic‌‌ structure presenting a nanometric slit can act as‌ a light trap, i.e.‌ light will accumulate in‌‌ a reduced space and lead to very intense,‌ localized fields. Nanocubes are‌ extensively studied in this‌‌ context and have been shown to support such‌ gap plasmon modes (see‌ Fig. 4). At‌‌ visible frequencies, the lossy‌ behavior of metals will cause the progressive absorption‌ of the trapped electromagnetic field, turning the metallic‌ nanocubes into efficient absorbers. The frequencies at which‌ this absorption occurs can be tuned by adjusting‌ the dimensions of the nanocube and the spacer.‌ Such metallic nanocubes can be used for a‌ broad range of applications including plasmonic sensing, surface‌ enhanced Raman scattering (SERS), metamaterials, catalysis, and bionanotechnology.‌ We aim at devising a numerical methodology for‌ characterizing the impact of geometrical parameters such as‌ the dimensions of the cube, the rounding of‌ nanocube corners or the size of the slit‌ separating the cube and the substrate, on the‌ overall performance of these absorbers. In practice, this‌ leads us to address two main modeling issues.‌ First, as the size of the slit is‌ decreased, spatial dispersion effects have to be taken‌ into account when dealing with plasmonic structures. For‌ this purpose, we consider a fluid model in‌ the form of a nonlocal hydrodynamic Drude model,‌ which materializes as a system of PDEs coupled‌ to Maxwell's equations. The second issue is concerned‌ with the assessment of geometrical uncertainties and their‌ role in the development of spatial dispersion effects.‌ This application domain is currently studied in collaboration‌ with physicists from the Pascal Institute at Université‌ Clermont Auvergne in Clermont-Ferrand.

Figure 4: Left: scanning electronic microscopy image‌ of silver nanocubes. Right: nanocube setup consisting of‌ an infinite gold ground layer, a dielectric spacer‌ and the cube surrounded by vacuum.

4.5 Plasmonic‌ nanostructures for photothermal effects

Plasmonic resonances can be‌ exploited for many applications. In particular, the strong‌ local field enhancement associated with the plasmonic resonances‌ of a metallic nanostructure or a dimer of‌ metallic nanostructures (see Fig. 5), together with‌ the absorption properties of the metal, induce a‌ photothermal energy conversion. Thus, in the vicinity of‌ the nanostructure, the temperature increases. These effects, viewed‌ as ohmic losses, have been for a long‌ time considered as a severe drawback for the‌ realization of efficient devices. However, the possibility to‌ control this temperature rise with the illumination wavelength‌ or polarization has gathered strong interest in the‌ nano-optics community, establishing the basis of thermoplasmonics. By‌ increasing temperature in their surroundings, metal nanostructures can‌ be used as integrated heat nanosources. Decisive advances‌ are foreseen in nanomedicine with applications in photothermal‌ cancer therapy, nano-surgery, drug delivery, photothermal imaging, protein‌ tracking, photoacoustic imaging, but also in nano-chemistry, optofluidics,‌ solar and thermal energy harvesting (thermophotovoltaics). Modeling realistic thermoplasmonics effects is a‌ highly multiscale and challenging‌ task. Indeed, the irradiation‌‌ of a metallic nanoparticle embedded in water with‌ ultrashort laser pulses rapidly‌ excite plasmons that, in‌‌ return, excite the particle's phonons, which then act‌ as hot carriers and‌ heat the surrounding of‌‌ the particle through conduction at the metal interface.‌ This process takes place‌ in $≃$ 10-100 fs‌‌1. Furthermore, the concentration of the electromagnetic‌ field in a small‌ volume near the particle‌‌ can lead to the excitation of a nanoscale‌ plasma in $≃$ 100‌ fs – 5 ps.‌‌ The plasma energy is then transferred to the‌ water molecules in a‌ few picoseconds, leading to‌‌ high stress and thermal confinement. The resulting extreme‌ temperature and pressure induce‌ cavitation in $≃$ 1‌‌ ns, leading to the formation, growth and collapse‌ of 0.1–10 $μ$ m‌ diameter bubbles, with lifetimes‌‌ ranging from 100 ps up to 100 ns.‌ As of today, the‌ computational methods that have‌‌ been developed do not enable, as such, efficient‌ prototyping for applications. Our‌ ambition on this topic‌‌ is to propose innovative numerical methods that are‌ able to deal accurately‌ and efficiently with all‌‌ the peculiarities of thermoplasmonics phenomena.

Figure 5‌: Plasmonic sphere dimer‌‌ demonstrating field enhancement in the gap between two‌ glod nanospheres.

4.6 Light‌ management by disordered nanostructures‌‌

With recent advances in nanophotonics and nanofabrication, disordered‌ nanostructures are studied for‌ the design of different‌‌ optical systems with unique features that otherwise cannot‌ be realized by their‌ periodic counterpart, including generation‌‌ of colors, broadband transmission enhancement, perfect focusing, broadband‌ light trapping and broadband‌ energy harvesting. The disparity‌‌ of involved length scales, with features on a‌ nanoscale and device characteristics‌ possibly on a millimeter‌‌ or centimeter scale, renders the quantitative description of‌ the emerging phenomena extremely‌ challenging. Our works on‌‌ this topic aim at producing fast and accurate‌ optical modeling methods enabling‌ the rigorous calculation of‌‌ the scattering by a large ensemble of cylindrical‌ scattering centers embedded in‌ a thin film or‌‌ a bulk medium depending on the target application.‌

5 Social and environmental‌ responsibility

5.1 Impact of‌‌ research results

The parts of our research activities‌ that are addressing the‌ design of nanostructures for‌‌ sunlight harvesting on one‌ hand, and of nanostructures for photothermal effects on‌ the other hand, target applications concerned with production‌ of renewable energy and biomedical engineering (ranging from‌ light controlled drug-release to the battle against Covid-19‌ in the context of the ANR SWAG-P project).‌

6 Highlights of the year

This year, the‌ team has initiated the development of a novel‌ software project named POSEidSON (PhOtonic SolvErs at the‌ nanOscale with Neural networks). In the future, POSEidSON‌ will be the flagship software of the team‌ for hosting all our methodological contributions on modeling‌ methods for photonics leveraging Deep Neural Networks.

7‌ Latest software developments, platforms, open data

7.1 Latest‌ software developments

7.1.1 DIOGENeS

Name:
DIscOntinuous GalErkin Nanoscale‌ Solvers
Keywords:
High-Performance Computing, Computational electromagnetics, Discontinuous Galerkin,‌ Computational nanophotonics
Functional Description:
The DIOGENeS software suite‌ provides several tools and solvers for the numerical‌ resolution of light-matter interactions at nanometer scales. A‌ choice can be made between time-domain (DGTD solver)‌ and frequency-domain (HDGFD solver) depending on the problem.‌ The available sources, material laws and observables are‌ very well suited to nano-optics and nano-plasmonics (interaction‌ with metals). A parallel implementation allows to consider‌ large problems on dedicated cluster-like architectures.
URL:
https://diogenes.inria.fr/‌
Contact:
Stéphane Lanteri
Participants:
Stéphane Lanteri, Alexis Gobe,‌ Guillaume Leroy

7.1.2 POSEidON

Name:
PhOtonic SolvErs at‌ the nanOscale with Neural networks
Keywords:
Deep learning,‌ Neural networks, Computational nanophotonics
Functional Description:
POSEidSON is‌ a software platform that implements several tools for‌ the development of surrogate models and inverse design‌ models based on Deep Neural Networks (DNNs) for‌ the design of photonics devices. Different Deep Learning‌ (DL) models are considered, ranging from purely data-driven‌ models to physics-based models. It provides users with‌ an intuitive interface application programming interface for constructing‌ and experimenting with DL models efficiently. This enables‌ rapid prototyping and testing of approaches for leveraging‌ DNNs.
Contact:
Stéphane Lanteri
Participants:
Alexis Gobe, Enzo‌ Isnard

7.1.3 Geom4PhotoPigments

Name:
Geometrical models for the‌ study of photonic pigments
Keyword:
Computational nanophotonics
Scientific‌ Description:
Geom4PhotoPigments is a suite of scripts (plugins)‌ that are compatible with the GFactory component of‌ the DIOGENeS software suite and the GMSH tetrahedral‌ mesh generation tool, for building geometric models that‌ can be used to simulate the optical properties‌ of photonic pigments.
Functional Description:
Geom4PhotoPigments is a‌ suite of scripts (plugins) that are compatible with‌ the GFactory component of the DIOGENeS software suite‌ and the GMSH tetrahedral mesh generation tool, for‌ building geometric models that can be used to‌ simulate the optical properties of photonic pigments.
URL:‌
http://www-sop.inria.fr/atlantis/
Contact:
Alexis Gobe
Participants:
Alexis Gobe, Stéphane‌ Lanteri

8 New results

8.1 High order methods‌ for complex problems

In this section, we present‌ ongoing studies aiming at designing, analyzing and developing‌ high order numerical methods for solving PDE systems‌ modeling nanoscale light-matter interactions in relation with the‌ physical settings and applications presented in section 3‌. We focus on the family of Discontinuous‌ Galerkin (DG) methods. In the time-domain setting, the starting point of these‌ works is the DGTD‌ (Discontinuous Galerkin Time-Domain) method‌‌ introduced in 15. In the frequency-domain setting,‌ the HDGFD (Hybridized Discontinuous‌ Galerkin Frequency-Domain) method 1‌‌ is considered as the basis of our works.‌

8.1.1 Time-domain numerical modeling‌ of gain media

Participants:‌‌ Stéphane Descombes, Stéphane Lanteri, Cédric Legrand‌, Gian Luca Lippi‌ [INPHYNI laboratory, Sophia Antipolis]‌‌.

In laser physics, gain or amplification is‌ a process where the‌ medium transfers part of‌‌ its energy to an incident electromagnetic radiation, resulting‌ in an increase in‌ optical power. This is‌‌ the basic principle of all lasers. Quantitatively, gain‌ is a measure of‌ the ability of a‌‌ laser medium to increase optical power. Modeling optical‌ gain requires to study‌ the interaction of the‌‌ atomic structure of the medium with the incident‌ electromagnetic wave. Indeed, electrons‌ and their interactions with‌‌ electromagnetic fields are important in our understanding of‌ chemistry and physics. In‌ the classical view, the‌‌ energy of an electron orbiting an atomic nucleus‌ is larger for orbits‌ further from the nucleus‌‌ of an atom. However, quantum mechanical effects force‌ electrons to take on‌ discrete positions in orbitals.‌‌ Thus, electrons are found in specific energy levels‌ of an atom. In‌ a semiclassical setting, such‌‌ transitions between atomic energy levels are generally described‌ by the so-called rate‌ equations. These rate‌‌ equations model the behavior of a gain material,‌ and they need to‌ be solved self-consistently with‌‌ the system of Maxwell equations. So far, the‌ resulting coupled system of‌ Maxwell-rate equations has mostly‌‌ been considered in a time-domain setting using the‌ FDTD method for which‌ several extensions have been‌‌ proposed. In the context of the PhD of‌ Cédric Legrand, we study‌ an alternative numerical modeling‌‌ approach based on a high order DGTD method‌ 43. This year,‌ we have formulated and‌‌ developed explicit fourth–order Runge–Kutta (RK4) temporal scheme and‌ combine it with a‌ DG spatial discretization (RK-DGTD‌‌ method) based on centered or upwind fluxes. Moreover,‌ we have developed a‌ new temporal integration scheme‌‌ to address the computational efficiency issue linked to‌ the multiscale in time‌ nature of the coupled‌‌ Maxwell-rate equations. We propose to improve the RK–DGTD‌ method by introducing a‌ multirate approach that treats‌‌ the fast dynamics of the fields separately from‌ the slower evolution of‌ the electronic densities, using‌‌ the MultiRate Generalized Additive Runge–Kutta (MR–GARK) framework. The‌ presented method preserves high-order‌ accuracy while notably reducing‌‌ the simulation time.

8.1.2 Time-domain numerical modeling of‌ plasmonic-based nanoscale heating

Participants:‌ Yves D'Angelo [LJAD, Université‌‌ Côte d'Azur], Thibault Laufroy, Claire Scheid‌.

Due to the‌ various scales and phenomena‌‌ that come into play in realistic thermoplasmonics problems,‌ accurate numerical modeling is‌ challenging. Laser illumination first‌‌ excites a plasmon oscillation (reaction of the electrons‌ of the metal) that‌ relaxes to a thermal‌‌ equilibrium and in turn excites the metal lattice‌ (phonons). The latter is‌ then responsible for heating‌‌ the surroundings. A relevant‌ modeling approach thus consists in describing the electron-phonon‌ coupling through the evolution of their respective temperature.‌ Maxwell's equations are then coupled to a set‌ of coupled nonlinear hyperbolic (or parabolic) equations for‌ the temperatures of respectively electrons, phonons and environment.‌ The nonlinearities and the different time scales at‌ which each thermalization occurs make the numerical approximation‌ of these equations quite challenging. In the context‌ of the PhD of Thibault Laufroy, which has‌ started in October 2020, we propose to develop‌ a suitable numerical framework for studying thermoplasmonics. As‌ a first step, we have reviewed the models‌ used in thermoplasmonics that are most often based‌ on strong or weak (nonlinear) couplings of Maxwell's‌ equations with nonlinear equations modeling heat transfer (hyperbolic‌ or parabolic). We first especially targeted the hyperbolic‌ version of the model and proposed an implementation‌ in 2D, based on a Discontinuous Galerkin approximation‌ in space. We used specific strategies for time‌ integration to account for the multiple time scales‌ of the problem. This has been validated on‌ academical test cases, but also on more concrete‌ cases. A theoretical stability study has also been‌ achieved. A preprint with these results has been‌ submitted for publication. Moreover, the PhD of Thibault‌ Laufroy has been defended in December 2025 45‌.

8.1.3 Time-domain numerical modeling of liquid crystal‌ materials

Participants: Mahmoud Elsawy, Stéphane Descombes,‌ Stéphane Lanteri, Claire Scheid.

Liquid crystal‌ (LC)-based reconfigurable metasurfaces offer significant potential for advancing‌ active metasurface technologies. This potential stems from their‌ cost-efficiency, ease of fabrication, and ability to operate‌ across a wide frequency spectrum. Among the various‌ types of LCs, nematic liquid crystals are particularly‌ prominent due to their favorable properties for such‌ applications. Traditional LC modeling typically focuses on simulating‌ the static response of the anisotropic permittivity tensor,‌ often assuming that the permittivity simply transitions between‌ two distinct states. However, this static approach neglects‌ the dynamic time-dependent behavior of LCs, a critical‌ factor that is overlooked in the majority of‌ existing studies. Such neglect can lead to inaccuracies‌ in predicting modulation times. Moreover, when LCs are‌ integrated into highly resonant structures, the strong optical‌ confinement within these structures can further influence the‌ LC response time, necessitating more sophisticated modeling. In‌ this work, our first goal is to develop‌ a comprehensive dynamic model of LC behavior. By‌ coupling the differential equations governing LC orientation under‌ time-varying applied voltages with Maxwell's equations, the foreseen‌ model will provide an accurate representation of the‌ time-dependent transitions in LC orientation. This approach should‌ surpass conventional odelin approaches by capturing the intricate‌ dynamics of LC responses under varying excitation conditions.‌ Consequently, this model will enable precise temporal control‌ of unit cell responses, making it particularly well-suited‌ for the design of reflective and transmissive multiresonant‌ metasurfaces in advanced technological applications. As a next‌ step, in the context of the PhD thesis‌ of Roman Gelly, we will devise an appropriate discretization of this dynamic‌ model of LC behavior.‌

8.1.4 Numerical modeling of‌‌ time-modulated metasurfaces

Participants: Mahmoud Elsawy, Roman Gelly‌, Stéphane Lanteri.‌

Active metasurfaces represent a‌‌ transformative platform for the dynamic modulation of electromagnetic‌ wavefronts, achieved through precise‌ spatial control of the‌‌ phase and amplitude of scattered light by leveraging‌ arrays of subwavelength scatterers‌ under external stimuli. This‌‌ spatial modulation imparts tailored momentum to the outgoing‌ light, enabling advanced functionalities‌ such as beam steering,‌‌ focusing, and holography. Moreover, the periodic temporal modulation‌ of these metasurfaces offers‌ the ability to manipulate‌‌ the frequency content of the scattered light, adding‌ a new dimension of‌ control and unlocking novel‌‌ possibilities in optical signal processing and frequency-domain applications.‌ However, the accurate numerical‌ modeling of time-modulated metasurfaces‌‌ poses significant challenges. Temporal modulation introduces dynamic variations‌ in the electromagnetic properties‌ of the metasurface, specifically‌‌ in the permittivity, which must be seamlessly integrated‌ into Maxwell's equations to‌ account for the interplay‌‌ between spatial and temporal effects. Such modeling is‌ crucial to predict and‌ optimize the metasurface performance,‌‌ particularly for applications involving complex temporal waveforms or‌ high-speed modulation. In our‌ work, we rely on‌‌ the Discontinuous Galerkin Time-Domain (DGTD) to address these‌ challenges. In this context,‌ the DGTD approach allows‌‌ for the discretization of Maxwell's equations in both‌ space and time, enabling‌ accurate modeling of the‌‌ temporal modulation effects while maintaining the spatial fidelity‌ required to capture subwavelength‌ features. By exploiting DGTD,‌‌ it is possible to capture the full dynamics‌ of time-modulated metasurfaces, including‌ the generation of frequency‌‌ sidebands and their spatial diffraction. This capability is‌ essential for designing metasurfaces‌ that integrate both spatial‌‌ and temporal modulation, enabling functionalities such as frequency‌ mixing, harmonic beam steering,‌ and the controlled breaking‌‌ of Lorentz reciprocity. Furthermore, the DGTD approach is‌ well-equipped to handle the‌ computational demands of large-scale‌‌ metasurfaces operating at high frequencies, ensuring accurate predictions‌ and facilitating experimental validation.‌ This year, we have‌‌ finalized the 2D implementation of a LF-DGTD method‌ (DG method with centered‌ fluxes combined to a‌‌ Leap-Frog time-stepping). This LF-DGTD method has been validated‌ on realistic physical uses‌ cases extracted from related‌‌ publications 42-31. Moreover, we have‌ also defined an inverse‌ design approach that combines‌‌ this LF-DGTD fullwave solver with a statistical learning-based‌ global optimization method. By‌ using this methodology, we‌‌ have unveiled a realistic space-time modulated silicon-based metasurface‌ that achieves nearly 80%‌ frequency-conversion efficiency and precise‌‌ harmonic beam steering in the near-infrared. Meanwhile, by‌ pushing the optimization into‌ higher modulation frequency regimes,‌‌ we explored exotic functionalities such as asymmetric frequency‌ mixing, highlighting the method’s‌ versatility for the design‌‌ of spacetime-modulated metasurfaces and paving the way for‌ a new generation of‌ space-time reconfigurable metasurfaces 35‌‌. This work is carried out in the‌ context of the PhD‌ thesis of Roman Gelly.‌‌

8.1.5 Efficient approximation of high-frequency Helmholtz problems

Participants:‌ Théophile Chaumont-Frelet [RAPSODI project-team,‌ Centre Inria de l'Université‌‌ de Lille], Victorita‌ Dolean [TU Eindhoven, The Netherlands], Maxime Ingremeau‌ [Institut Fourier, Université Grenoble-Alpes], Florentin Proust.‌

Helmholtz problems describe the time-harmonic solutions of the‌ wave equation (possibly in a heterogeneous medium, in‌ a bounded medium, with boundary conditions, etc.). In‌ general, there is no explicit solution to such‌ an equation, and an approximate solution of the‌ equation must be computed numerically. All the existing‌ methods (finite elements, finite differences, etc.) have in‌ common that they become more and more expensive‌ when the frequency of the waves increases. In‌ 52, we study new finite-dimensional spaces specifically‌ designed to approximate the solutions to high-frequency Helmholtz‌ problems with smooth variable coefficients. These discretization spaces‌ are spanned by Gaussian coherent states, that have‌ the key property to be localized in phase‌ space. We carefully select the Gaussian coherent states‌ spanning the approximation space by exploiting the (known)‌ micro-localization properties of the solution. This work is‌ conducted in the context of the Inria POPEG‌ Exploratory Research Action and the topic is also‌ at the heart of the PhD thesis of‌ Florentin Proust.

In the beginning of this thesis,‌ such a method had been implemented for a‌ simple one-dimensional problem. Even in this very simple‌ case, it became clear that Gaussian coherent states‌ could not be used in practice because they‌ were strongly ill-conditioned. However, if the discretization spaces‌ are now spanned by some particular linear combinations‌ of Gaussian coherent states forming a so-called Wilson‌ basis, then this problem disappears. In 52,‌ it had been mathematically proved that using Gaussian‌ coherent states to solve high-frequency Helmholtz problems had‌ some advantages. In 2023, we had started to‌ prove similar - and even broader - results‌ about Wilson basis. In 2024, we continued to‌ work on these theoretical aspects. We also developed‌ a Python code to solve one-dimensional Helmholtz problems‌ with Gaussian coherent states or with a Wilson‌ basis. More precisely, we first focused on a‌ one-dimensional equation with constant coefficients, and then we‌ started to generalize to one-dimensional equations with variable‌ coefficients.

In 2025, Florentin Proust continued to work‌ on the Python code. However, it was decided‌ to implement the method in C++ (still for‌ one-dimensional problems) to try to get a more‌ efficient code (this implementation in C++ was mainly‌ done by Théophile Chaumont-Frelet). We also finished to‌ write the proofs of the theoretical results. Furthermore,‌ in February 2025, Florentin Proust gave a talk‌ at the Conference on Mathematics of Wave Phenomena‌ 2025 in Karlsruhe, Germany 29.

In early‌ 2026, we plan to submit an article containing‌ all the theoretical aspects and the one-dimensional numerical‌ results.

8.2 Data-driven reduced-order modeling

In short, reduced-order‌ modeling (ROM) allows to construct simplifications of high‌ fidelity, complex models. The resulting lower fidelity (also‌ referred as surrogate) models capture the salient features‌ of the source models so that one can‌ quickly study a system's dominant effects using minimal computational resources.

8.2.1 POD-based‌ ROM methods for parameterized‌ electromagnetic problems

Participants: Stéphane‌‌ Lanteri, Kun Li [SUFEC, Chengdu, China],‌ Liang Li [UESTC, Chengdu,‌ China].

In collaboration‌‌ with researchers at the University of Electronic Science‌ and Technology of China‌ (UESTC) and the Southwestern‌‌ University of Finance (SUFEC) and Economics, which are‌ both located in Chengdu,‌ we study ROM for‌‌ time-domain electromagnetics and nanophotonics. Most of our works‌ so far are based‌ on the proper orthogonal‌‌ decomposition (POD) technique. Our main contributions are described‌ in 16-18‌, where we have‌‌ proposed POD approach for building a reduced subspace‌ with a significantly smaller‌ dimension given a set‌‌ of space-time snapshots that are extracted from simulations‌ with a high order‌ DGTD method. Subsequently, we‌‌ have designed several fully data-driven non-intrusive POD-based ROM‌ approaches. In 17,‌ we have proposed the‌‌ POD-CSI (POD combined to Cubic Spline Interpolation) method‌ in the context of‌ parameterized time-domain electromagnetic scattering‌‌ problems. The considered parameters are the dielectric electric‌ permittivity and the temporal‌ variable. Then in 13‌‌-22 we have designed improved versions of‌ the POD-CSI method respectively‌ refererred as POD-CSI-CAE (using‌‌ a Convolutional Auto-Encoder for a further reduction of‌ the POD-based model) and‌ POD-DMD-RBF (based on Dynamic‌‌ Mode Decomposition and Radial Basis Function).

This year‌ we have proposed a‌ parametric geometry-based non-intrusive model‌‌ order reduction (NIMOR) model for electromagnetic simulation in‌ domains of different shapes.‌ The free-from-deformation (FFD) method‌‌ is introduced to adjust the mesh when the‌ shape changes. This NIMOR‌ model generates a set‌‌ of reduced-order basis (RB) functions by applying a‌ two-step randomized singular value‌ decomposition (SVD) method with‌‌ an auto-rank generator (ARRSVD) to the matrix composed‌ of the full-order solutions.‌ A map between the‌‌ time/geometry parameters and the projection coefficients is approximated‌ by the cubic-spline interpolation‌ (CSI) approach. This reduced-order‌‌ model (ROM) is trained in the offline stage,‌ while the RB solutions‌ for new parameters can‌‌ be quickly recovered in the online stage. Numerical‌ experiments show that the‌ proposed NIMOR method can‌‌ achieve near real-time electromagnetic scattering simulation on domains‌ of different shapes while‌ guaranteeing that the error‌‌ between high-fidelity and reduced-order solutions is below an‌ acceptable threshold 25.‌

8.2.2 Nonlinear ROM with‌‌ Graph Convolutional Autoencoder

Participants: Carlotta Filippin, Stéphane‌ Lanteri, Federico Pichi‌ [EPFL, Switzerland], Claire‌‌ Scheid, Maria Strazzullo [Politecnico di Torino, Italy]‌.

Although the POD-CSI‌ method introduced in 17‌‌ provides encouraging results, it is not as efficient‌ and robust as one‌ would expect from a‌‌ ROM perspective. Indeed, the hyperbolic nature of the‌ underlying PDE system, i.e.,‌ the system of time-domain‌‌ Maxwell equations, is known to represent a challenging‌ issue for linear reduction‌ methods such as POD.‌‌ In practice, a large number of modes is‌ required therefore hampering the‌ obtention of an efficient‌‌ ROM strategy. One possible path to address this‌ problem, which is currently‌ investigated by several groups‌‌ worldwide, relies on nonlinear‌ reduction techniques. We initiated this year a study‌ on nonlinear ROM for the time-domain Maxwell equations.‌ More precisely, we study the approach recently proposed‌ in 65, which proposes a nonlinear model‌ order reduction based on a Graph Convolutional Autoencoder‌ (GCA-ROM). This year, in the context of the‌ PhD thesis of Carlotta Filippin, we have developed‌ a GCA-ROM approach for physically parameterized time-domain electromagnetics‌ in 2D and 3D, with training data provided‌ by high-fidelity DGTD simulations. Our first achievements have‌ been presented at the MORTech 2025 conference 36‌, which is one of the major international‌ scientific events dedicated to model order reduction. Moreover,‌ this research is conducted in collaboration with Federico‌ Pichi (EPFL, Switzerland) and Maria Strazzullo (Politecnico di‌ Torino, Italy).

8.3 Numerical optimization and inverse design‌ approaches

Participants: Ayoub Bellouch, Mickaël Binois [ACUMES‌ project-team, Centre Inria d'Université Côte d'Azur], Régis‌ Duvigneau [ACUMES project-team, Centre Inria d'Université Côte d'Azur]‌, Mahmoud Elsawy, Stéphane Lanteri.

Developing‌ inverse design approaches for discovering non-intuitive nanostructures or‌ material nanostructuring for harvesting and tailoring the interaction‌ of light with matter on the nanoscale is‌ an important methdological objective of the team.

For‌ this, one has typically to deal with two‌ numerical ingredients: on the one hand, a numerical‌ method for characterizing the optical performance of a‌ given design of the photonic device at hand,‌ which is generally referred to as the forward‌ problem; on the other hand, the so-called‌ inverse problem calls for a numerical optimization algorithm‌ 12, which has to be compatible with‌ the required number of design parameters and the‌ computational cost of the evaluation of a single‌ design with the numerical characterization method. For what‌ concern the former task, mathematical modeling is generally‌ based on the system of three-dimensional (3D) Maxwell‌ equations formulated in the time-domain or frequency-domain, which‌ is coupled to an appropriate model of frequency-dependent‌ material response. In the recent years, we have‌ developed a computational framework that combines high order‌ Discontinuous Galerkin (DG) methods for solving the system‌ of time-domain 15 or frequency-domain 1 Maxwell equations‌ in 3D with an efficient global optimization technique‌ that belongs to the class of Bayesian optimization.‌

In our works, we use one of the‌ most advanced optimization techniques based on a statistical‌ learning-based approach, which is known as Efficient Global‌ Optimization (EGO). The EGO algorithm is a global‌ optimization algorithm that substitutes the complex and costly‌ iterative electromagnetic evaluation process with a simpler and‌ cheaper metamodel. EGO is related to the class‌ of Bayesian optimization. Contrary to the traditional common‌ global optimization strategies like genetic algorithms, EGO is‌ not based on adaptive sampling but on a‌ surrogate model, which is constructed on the basis‌ of available objective function evaluations. This surrogate model‌ utilizes a statistical learning criterion related to the‌ optimization target (usually called merit function) in‌ order to identify which design (set of parameters) should be tested in‌ the next iteration that‌ would provide better results‌‌ close to the predefined goal.

In general, the‌ EGO is based on‌ two phases. The first‌‌ one is the Design Of Experiment (DOE), in‌ which an initial database‌ is generated. In essence,‌‌ a uniform sampling strategy (e.g. Latin Hypercube Sampling)‌ is deployed in order‌ to generate different designs‌‌ in which the cost function is evaluated using‌ an electromagnetic solver. In‌ the second phase, using‌‌ the data obtained from the DOE, a Gaussian‌ Process (GP) model, is‌ constructed to fit these‌‌ data. This GP model allows us to predict‌ the values of the‌ cost function in the‌‌ parameter space without the need to perform additional‌ electromagnetic simulations. Once this‌ GP model is determined,‌‌ one can estimate at any point of the‌ design space, the objective‌ function (mean of the‌‌ GP model) and an uncertainty value (variance of‌ the GP model). The‌ mean and the variance‌‌ are used together to determine a statistical merit‌ function. In our case,‌ we rely on the‌‌ expected improvement, which is a function whose‌ maximum defines the next‌ design parameters set to‌‌ be evaluated. That is to say, in the‌ search parameter space where‌ this function is maximized,‌‌ we extract the corresponding parameter values, and the‌ corresponding design will be‌ simulated using our electromagnetic‌‌ solver. Then the database is updated accounting for‌ this new observation (construction‌ of a new GP‌‌ model based on the updated database). We repeat‌ this process until a‌ predefined convergence criterion is‌‌ reached, or when the expected improvement is sufficiently‌ small.

In the recent‌ past, in close collaboration‌‌ with researchers from the ACUMES project-team, we have‌ developed inverse design approaches‌ for single-objective optimization of‌‌ phase gradient metasurfaces operating at visible wavelengths 11‌. Then, in 10‌, we presented for‌‌ the first time to the metasurface community a‌ multiobjective inverse design approach‌ leveraging the EGO method.‌‌ Finally, in 9, we addressed the optimization‌ of metasurface designs taking‌ into account uncertainties due‌‌ to fabrication errors still in the framework of‌ the EGO method. All‌ these achievements clearly demonstrate‌‌ the versatility of the EGO method for inverse‌ design of nanoscale photonic‌ devices 39.

In‌‌ the context of the AEROCOM project, we have‌ developed an advanced optimization‌ workflow for the design‌‌ of a metadeflector-based microwave antenna for satellite communication.‌ Beam steering antennas, based‌ on subwavelength deflectors, aim‌‌ to facilitate information exchange with satellite communication (SatCom)‌ on-the-move systems. Recent beam‌ deflection solutions are often‌‌ based on the phase-gradient metasurfaces concept to generate‌ the desired wavefront. However,‌ the performance remains inadequate‌‌ for advanced communication technologies due to bandwidth limitations‌ and insufficient efficiency. In‌ the context of the‌‌ AEROCOM project, we have designed a highly efficient‌ broadband beam deflection system‌ for SatCom applications, which‌‌ has been fabricated and characterized experimentally by the‌ other partners of this‌ project namely, the Nanoe‌‌ company and Thales Research‌ & Technology. This non-trivial broadband design leverages a‌ sophisticated coupling between adjacent subwavelength elements, together with‌ varying thicknesses along the propagation direction to manipulate‌ the incoming beam. The beam deflector is illuminated‌ with a circularly polarized wave and is designed‌ to operate in the Ka-Tx band from 27.5‌ GHz to 31 GHz. The optimized structure demonstrates‌ exceptionally high performance compared to the conventional phase-gradient‌ synthesizing counterparts. Experimental validation demonstrates an excellent agreement‌ with numerical simulation, highlighting the potential of this‌ optimization-assisted design to advance metasurface technology for various‌ beam-steering applications 27- 33.

This year,‌ we have also made some progress on multifidelity‌ optimization, still in the framework of the EGO‌ algorithm. In this research, the overarching goal is‌ to substantially reduce the computational cost of a‌ complex numerical optimization scenario by wisely exploiting different‌ levels of fidelity, i.e., accuracy in our context,‌ in the numerical characterization (forward step) of a‌ device design. For this, we can rely on‌ the flexibility of our DG-based fullwave solvers by‌ playing with the mesh resolution level or/and the‌ degree of the polynomial interpolation method used to‌ approximate the electromagnetic field components cell-wise. Aletrnatively, we‌ can also leverage reduced-order models (see Section 8.2‌). Our preliminary results on the development of‌ a MF-EGO (MultiFidelity-EGO) method have been presented at‌ the International Symposium on Electromagnetic Theory (EMTS 2025)‌ 28.

8.4 Deep Learning methods

We initiated‌ in 2022 a novel research direction on alternative‌ numerical modeling methods based on Neural Networks (NN).‌ We investigate both data-driven and model-driven approaches for‌ dealing with the system of Maxwell equations possibly‌ coupled with various material models of interest to‌ nanophotonics. One first question that we want to‌ address is whether DL methods can yield highly‌ efficient surrogate models of 3D time-domain electromagnetic wave‌ propagation problems. Besides, we are also interested in‌ devising DL-based methods for dealing with problems which‌ are more difficult to handle with traditional numerical‌ methods such as electromagnetic wave interaction with space-time‌ adaptive materials or nonlinear media. For this purpose,‌ we study the formulation and application of Physics-Informed‌ Neural Networks (PINNs) that can accurately and efficiently‌ deal with the modeling characteristics of these problems.‌

8.4.1 PINNs for the parametric Maxwell equations

Participants:‌ Stéphane Descombes, Stéphane Lanteri, Alexandre Pugin‌, Mathieu Riou [Thales Research & Technology, Palaiseau,‌ France].

Numerical simulations of electromagnetic wave propagation‌ problems primarily rely on discretization of the system‌ of time-domain Maxwell equations using finite difference or‌ finite element type methods. For complex and realistic‌ three-dimensional situations, such a process can be computationally‌ prohibitive, especially in view of many-query analyses (e.g.,‌ optimization design and uncertainty quantification). Therefore, developing cost-effective‌ surrogate models is of great practical significance. Among‌ the different possible approaches for building a surrogate‌ model of a given PDE system in a‌ non-intrusive way (i.e., with minimal modifications to an‌ existing discretization-based simulation methodology), approaches based on neural networks and Deep Learning‌ (DL) has recently shown‌ new promises due to‌‌ their capability of handling nonlinear or/and high dimensional‌ problems. In the present‌ study, we propose to‌‌ focus on the particular case of Physics-Informed Neural‌ Networks (PINNs) introduced in‌ 67. PINNs are‌‌ neural networks trained to solve supervised learning tasks‌ while respecting any given‌ laws of physics described‌‌ by a general (possibly nonlinear) PDE system. They‌ seamlessly integrate the information‌ from both the measurements‌‌ and partial differential equations (PDEs) by embedding the‌ PDEs into the loss‌ function of a neural‌‌ network using automatic differentiation. In 2022, we have‌ initiated a study dedicated‌ to the applicability of‌‌ PINNs for building efficient surrogate models of the‌ parametric Maxwell equations. This‌ year, we have progressed‌‌ on this objective in the context of the‌ PhD thesis of Alexandre‌ Pugin by conducting a‌‌ detailed analysis of the PINN concept for dealing‌ with the 2D time-domaine‌ Maxwell equations 44.‌‌ For dealing with heterogenous problems with a piecewise‌ constant dielectric permittivity, we‌ have designed a domain‌‌ decomposition approach with a loss function that includes‌ a residual term for‌ the jump relations at‌‌ an interface between two media, in addition to‌ the usual residual terms‌ associated to Maxwell equations,‌‌ and initial and boundary conditions. This approach notably‌ improves the accuracy of‌ PINN predictions for such‌‌ heterogeneous problems. We are now working on a‌ similar study for the‌ 2D frequency-domain Maxwell equations.‌‌

8.4.2 Multilevel and distributed PINNs for the Helmholtz‌ equation

Participants: Victorita Dolean‌ [TU Eindhoven, The Netherlands]‌‌, Daria Hrebenshchykova, Stéphane Lanteri, Victor‌ Michel-Dansac [MACARON project-team, Centre‌ Inria de l'Université de‌‌ Lorraine].

In the context of the PhD‌ thesis of Daria Hrebenshchykova,‌ which has started in‌‌ November 2024, we investigate recently proposed extensions of‌ PINNs, namely Finite Basis‌ PINNs (FBPINNs) and Multilevel‌‌ Finite Basis PINNs (MFBPINNs) for building fast surrogates‌ of high frequency and‌ multiscale frequency-domain wave propagation‌‌ problems. As a first step, we consider problems‌ that can be modeled‌ by a 2D Helmholtz‌‌ equation. Through numerical simulations, we compare the efficiency,‌ accuracy, and scalability of‌ these methods. While FBPINN‌‌ exhibit good performance for low frequency problems, the‌ multilevel extension outperforms the‌ FBPINN method for high‌‌ frequency problems. This work also introduces novel distributed‌ NN architectures and associated‌ training schemes for FBPINNs‌‌ and MFBPINNs, including architectures based on single and‌ multiple optimizers, integrated into‌ the ScimBa library developed‌‌ by the MACARON project-team of Centre Inria de‌ l'Université de Lorraine. Moreover,‌ this year, we have‌‌ extended the PINN, FBPINN and MLFBPINN formulations to‌ deal with the 2D‌ Hemholtz equations with PMLs‌‌ (Perfectly Matched Layers) for artificial truncation of the‌ computational domain. These achievements‌ have been presented at‌‌ the International Domain Decomposition Conference (DD28) 30.‌

8.4.3 Efficient deep learning‌ methodology for large-scale metalens‌‌

Participants: Marco Abbarchi [Solnil, Marseille, France], Arthur‌ Clini de Souza,‌ Mahmoud Elsawy, Hugo‌‌ Enrique Hernandez-Figueroa [University of‌ Campinas (Unicamp), Campinas, Sao Paulo, Brazil], Badre‌ Kerzabi [Solnil, Marseille, France], Stéphane Lanteri,‌ Plaoma Pellegrini [University of Campinas (Unicamp), Campinas, Sao‌ Paulo, Brazil].

Metasurfaces are the 2D equivalent‌ of metamaterials, having wavelength-sized elements that leverage various‌ physical phenomena to control the wavefront of the‌ transmitted and reflected beams. Building on our recent‌ advancements in deep learning (DL) 6, we‌ develop an efficient deep learning strategy for designing‌ large-scale metalenses in reflection. Our strategy is based‌ on optimizing several beam deflectors for a wide‌ range of diffracted angles. This eliminates the need‌ for individual optimization for each angle, similar to‌ our earlier work on color filter designs in‌ 6. The proposed structure consists of supercells‌ with two distinct ridges made of TiO $_{2‌}$ (see Fig. 6). The widths, heights and‌ spacings between these ridges are optimized parameters, ensuring‌ enhanced performance of the diffracted beams. We simulated‌ 10 000 different unit cells configurations to build‌ a dataset. Afterwards, we trained a deep neural‌ network surrogate model, which defines two functions. Firstly,‌ it can be exploited as a fast solver‌ to replace costly fullwave simulations. Secondly, it represents‌ a differentiable solver, enabling efficient gradient optimization. The‌ next step is training a Multi-Valued Artificial Neural‌ Network (MVANN) to predict the possible designs responsible‌ for generating a target response. The last step‌ is using a network, hereby called the condenser,‌ to filter out the spurious solutions of the‌ MVANN and post-process the geometrical parameters, such as‌ uniforming the height across all solutions and applying‌ hard constraints. This approach allowed precise selection of‌ the period and corresponding deflection angles. Several metalenses‌ with different diameters have been optimized ranging from‌ 50 µm to 1 mm demonstrating scalability of‌ our approach. Our academic partner in Brazil is‌ in the process of fabricating the optimized metalens‌ structures. The results of this study have been‌ submitted for publication this year.

Figure 6.a — Figure 6: Panel (a) shows‌ the representation of the considered structure with the‌ optimization parameters. Panel (b) presents the gathering technique‌ to form highly performance metalens in reflection. Panel‌ (c) displays numerical simulation of a metalens with‌ numerical aperture in the range of 0.5 with‌ a diameter of 50 $μ$ m.

Figure 6.b — Figure 6: Panel (a) shows‌ the representation of the considered structure with the‌ optimization parameters. Panel (b) presents the gathering technique‌ to form highly performance metalens in reflection. Panel‌ (c) displays numerical simulation of a metalens with‌ numerical aperture in the range of 0.5 with‌ a diameter of 50 $μ$ m.

8.5 Discovering‌ novel nanoscale structures

At the creation of the‌ team in February 2020, our collaborations with physicists‌ from the nanophotonics domain aimed at leveraging our‌ developed numerical modeling methodologies in order to study‌ specific topics in relation with concrete applications (see‌ Section 8.7 for more details). An evolution that‌ started in 2022 was our will to adapt‌ and exploit these numerical methodologies to discover nanostructure organizations exhibiting behaviors and‌ performances opening the road‌ to new application perspectives.‌‌

8.5.1 Trimer metasurfaces for highly sensitive biomedical sensors‌

Participants: Haogang Cai [Department‌ of Biomedical Engineering, New‌‌ York University, USA], Mahmoud Elsawy, Stéphane‌ Lanteri, Hao Wang‌ [Department of Biomedical Engineering,‌‌ New York University, USA].

In this work,‌ which is conducted in‌ collaboration with Haogang Cai‌‌ at the Department of Biomedical Engineering, New York‌ University, USA, we have‌ designed a highly sensitive‌‌ metasurface for biomedical sensing relying on toroidal dipoles.‌ Through numerical simulations (see‌ Fig. 7) and‌‌ experimental validation, we unveil that these metasurfaces offer‌ remarkable sensitivity, especially across‌ the visible spectrum. This‌‌ high sensitivity enables the detection of refractive index‌ variations with high precision,‌ which would have great‌‌ value in many biomedical applications. The fabrication and‌ the experimental characterization have‌ been done and showed‌‌ a very good agreement with our numerical simulations‌ 32. These results‌ are currently assessed for‌‌ a potential joint patenting between Inria and NYU.‌ Moreover, we are preparing‌ a paper to be‌‌ submitted in early 2026.

Figure 7: Panel (a) shows the transmission‌ response of the trimer‌ metasurface. Panel (b) presents‌‌ the sensitivity analysis of the three main modes‌ together the field profile‌ of the toroidal mode.‌‌

8.5.2 Metasurface for quantum information processing

Participants: Mahmoud‌ Elsawy, Alemayehu Getahun‌ Kumela, Stéphane Lanteri‌‌.

The design of nonlinear metasurfaces for quantum‌ applications has usually focused‌ on classical merits like‌‌ tuning resonance at the pump and target frequencies.‌ These approaches can improve‌ nonlinear efficiency but they‌‌ do not ensure the phase control needed for‌ high quality quantum state.‌ In the context of‌‌ the META4QIP AEx, we have developed a hybrid‌ quantum-classical multiobjective inverse design‌ framework that directly includes‌‌ quantum measures like fidelity in the optimization process.‌ By doing so, we‌ increase both entanglement fidelity‌‌ and spontaneous parametric down conversion (SPDC) efficiency, ensuring‌ that high brightness and‌ quantum purity are achieved‌‌ together. Traditional appropaches to improve SPDC often rely‌ on bound state in‌ the continuum (BIC) modes‌‌ or sharp resonances. These methods are generally sensitive‌ to fabrication errors and‌ hard to control across‌‌ polarizations leading to low quantum purity, uncontrolled correlations,‌ and lower entanglement fidelity.‌ Our approach shows that‌‌ reliable, easy-to-fabricate metasurfaces can achieve better quantum performance‌ without these constraints. When‌ applied to an AlGaAs‌‌ nanohole metasurface, our design‌ reaches a fidelity of F=0.9969 and SPDC collection‌ efficiency of 55 Hz for a wide angular‌ collection, exceeding the state-of-the-art devices hence establishing a‌ practical path toward scalable, on-demand quantum light sources‌ 48.

8.6 Software developments in DIOGENeS

Participants:‌ Alexis Gobé, Arthur Gouinguenet, Guillaume Leroy‌, Stéphane Lanteri, Alan Youssef.

In‌ order to maximize the impact of our research‌ activities described in section 3, a modern‌ software platform is necessary. For that purpose, the‌ team develops the DIOGENeS (DIscOntinuous GalErkin Nanoscale Solvers)‌ software suite, which is dedicated to the numerical‌ modeling of nanoscale light-matter interactions in the 3D‌ case. This suite is organized around the several‌ components depicted in Figure 8. The core‌ library and the fullwave solvers are based on‌ an object-oriented architecture implemented in Fortran 2008. The‌ DGTD and HDGFD solvers are adapted to high‌ performance computing platforms by relying on a partitioning‌ of the computational mesh and a parallel programming‌ based on the message exchange model using the‌ MPI standard.

This year, several novel features have‌ been developed that are concerned with the GFactory‌ and Observer components (see Fig. 8). In‌ addition, we have initiated the development of the‌ Surrogates component, which integrates our contributions on fully‌ data-driven ROM methods. Moreover, a GPU accelerated version‌ of the DGTD fullwave solver has been developed,‌ thus drastically enhancing the high performance computing capabilities‌ of this simulation tool.

Figure 8.a — Figure 8: Architecture of the DIOGENeS software‌ suite. DGTD and HDGFD are the high order‌ DG-based fullwave solvers for time-domain and frequency-domain modeling‌ settings. GFactory is the geometrical modeling component that‌ exploits the Python API of the GMSH mesh‌ generation tool. Observer is the base component for‌ developing post-processing scripts of simulation results. Optim is‌ the base component for developing inverse design workflows‌ in Python by using statistical learning global optimization‌ algorithms from external frameworks such as Trieste.

8.7‌ Applications

8.7.1 Modeling of centimeter-scale metasurfaces in imaging‌ systems

Participants: Mahmoud Elsawy, Sébastien Héron [Thales‌ Research & Technology, Palaiseau, France], Enzo Isnard‌, Stéphane Lanteri.

Metasurfaces are 2D optical‌ components structured at sub-wavelength scale that locally control‌ the properties of incident light. They can perform‌ various functions such as beam steering, polarization control‌ and focusing. However, metasurfaces are still difficult to‌ incorporate into imaging systems due to the difficulty‌ of modeling their behavior together with other components.‌ For a traditional optical system composed of mirrors or refractive lenses, ray-tracing‌ tools are used to‌ predict the imaging performances.‌‌ For metasurfaces, one cannot use these tools, as‌ geometrical optics is no‌ longer valid for describing‌‌ the interactions of light with sub-wavelength elements. To‌ accurately simulate them, Maxwell's‌ equations have to be‌‌ solved using finite difference or finite element methods.‌ These methods are computionally‌ expensive and are only‌‌ adapted for components of size around ten wavelengths.‌ Thus, they are of‌ no practical use to‌‌ simulate metasurfaces inside imaging optical systems (see Fig.‌ 9), which are‌ in the centimeter scale‌‌ for industrial applications in imaging. Beyond this size‌ limitation, one may need‌ to integrate a metasurface‌‌ neither at the entrance nor at the output‌ of the system. This‌ positionning often leads to‌‌ a smaller incident angle and/or to a decrease‌ of the component area‌ because of its relative‌‌ position with respect to the system stop and‌ pupils. Besides, this also‌ leads to modeling issues:‌‌ a) the incidents fields are not plane waves‌ anymore, and b) computed‌ electromagnetic fields after the‌‌ metasurface need to be recast as rays. To‌ address these issues, we‌ develop a novel numerical‌‌ methodology to couple a fullwave solver with a‌ ray tracing tool in‌ order to simulate a‌‌ whole system containing mesurfaces and refractive components 37‌. The main objective‌ is to fully simulate‌‌ and optimize the whole system including the metasurface‌ to achieve various functionalities.‌ This work is implemented‌‌ in the frame of the PhD of Enzo‌ Isnard. Moreover, it has‌ been accepted for publication‌‌ Optics Express journal 26.

Figure 9‌‌: Example of an imaging system containing a‌ metasurface. To simulate the‌ whole system, we need‌‌ to convert rays into electromagnetic fields and vice‌ versa. As the metasurface‌ lies in the middle‌‌ of the system, the incident wavefront is not‌ necessarily planar.

8.7.2 Optimization‌ of light trapping in‌‌ nanostructured solar cells

Participants: Stéphane Collin [Sunlit team,‌ C2N-CNRS, Paris-Saclay, France],‌ Alexis Gobé, Henning‌‌ Helmers [Fraunhofer-Institut für Solare Energiesysteme ISE, Freiburg, Germany]‌, Oliver Höhn [Fraunhofer-Institut‌ für Solare Energiesysteme ISE,‌‌ Freiburg, Germany], Stéphane Lanteri, Guillaume Leroy‌, Ines Revol [LAAS,‌ Toulouse, France].

There‌‌ is significant recent interest in designing ultra-thin crystalline‌ silicon solar cells with‌ active layer thickness of‌‌ a few micrometers. Efficient light absorption in such‌ thin films requires both‌ broadband antireflection coatings and‌‌ effective light trapping techniques, which often have different‌ design considerations. In collaboration‌ with physicists from the‌‌ Sunlit team at C2N (Centre for Nanosciences and‌ Nanotechnology) in Campus Paris-Saclay)‌ and the Fraunhofer-Institut für‌‌ Solare Energiesysteme ISE in Freiburg, Germany, we exploit‌ statistical learning methods for‌ the inverse design of‌‌ material nanostructuring with the‌ goal of optimizing light trapping properties of ultraphin‌ solar cells. This objective is challenging because the‌ underlying electromagnetic wave problems exhibit multiple resonances, while‌ the geometrical settings are non-trivial. Such multi-resonant solar‌ cell structures are attractive for maximizing light absorption‌ for the full solar light spectrum as illustrated‌ in Fig. 10. This year, we have‌ started a new study on the design of‌ a tandem solar cell for which we propose‌ to consider a multi-objective optimization setting to address‌ simultaneously the minimization of two absorber layers and‌ the maximization of the short-circuit current in those‌ layers. Very promising results have been obtained that‌ will be soon subimitted for publication in a‌ an appropriate venue.

Figure 10.a — Figure 10: Optimization of‌ light absorption in a solar cell based on‌ a pyramidal grating. Top figures: geometrical model. Bottom:‌ absorption spectra.

Figure 10.b — Figure 10: Optimization of‌ light absorption in a solar cell based on‌ a pyramidal grating. Top figures: geometrical model. Bottom:‌ absorption spectra.

8.7.3 Plasmonic sensing with nanocubes

Participants:‌ Antoine Moreau [Institut Pascal, Clermont-Ferrand, France], Stéphane‌ Lanteri, Guillaume Leroy, Claire Scheid.‌

The propagation of light in a slit between‌ metals is known to give rise to guided‌ modes. When the slit is of nanometric size,‌ plasmonic effects must be taken into account, since‌ most of the mode propagates inside the metal.‌ Indeed, light experiences an important slowing-down in the‌ slit, the resulting mode being called gap-plasmon. Hence,‌ a metallic structure presenting a nanometric slit can‌ act as a light trap, i.e. light will‌ accumulate in a reduced space and lead to‌ very intense, localized fields. We study the generation‌ of gap plasmons by various configurations of silver‌ nanocubes separated from a gold substrate by a‌ dielectric layer, thus forming a narrow slit under‌ the cube. When excited from above, this configuration‌ is able to support gap-plasmon modes which, once‌ trapped, will keep bouncing back and forth inside‌ the cavity. We exploit statistical learning methods for‌ the goal-oriented inverse design of cube size, dielectric‌ and gold layer thickness, as well as gap‌ size between cubes in a dimer configuration (see‌ Fig. 11). This study is conducted in collaboration with Antoine Moreau‌ at Institut Pascal (CNRS).‌ Starting in January 2024,‌‌ we will continue this study in the context‌ of the ANR SWAG-P‌ project, which is coordinated‌‌ by Antoine Moreau from Institut Pascal, Clermont-Ferrand, France.‌

Figure 11.a — Figure 11‌: Optimization of a‌‌ plasmonic nanocube dimer setting for the generation of‌ Fano resonances.

Figure 11.b — Figure 11‌: Optimization of a‌‌ plasmonic nanocube dimer setting for the generation of‌ Fano resonances.

8.7.4 Multiple‌ scattering in random media‌‌

Participants: Stéphane Descombes, Stéphane Lanteri, Guillaume‌ Leroy, Cédric Legrand‌, Gian Luca Lippi‌‌ [INPHYNI laboratory, Nice].

Fluorescence signals emitted by‌ probes, used to characterize‌ the expression of biological‌‌ markers in tissues or cells, can be very‌ hard to detect due‌ to a small amount‌‌ of molecules of interest (proteins, nucleic sequences), to‌ specific genes expressed at‌ the cellular level, or‌‌ to the limited number of cells expressing these‌ markers in an organ‌ or a tissue. Access‌‌ to information coming from weaker emitters can only‌ come from strengthening the‌ signal, since electronic post-amplification‌‌ raises the noise floor as well. Molecule-specific biochemical‌ processes are being developed‌ for this purpose, and‌‌ a new mechanism based on the simultaneous action‌ of stimulated emission and‌ multiple scattering induced by‌‌ nanoparticles suspended in the sample has been recently‌ demonstrated to effectively amplify‌ weak fluorescence signals. A‌‌ precise assessment of the signal fluorescence amplification that‌ can be achieved by‌ such a scattering medium‌‌ requires an electromagnetic wave propagation modeling approach capable‌ of accurately and efficiently‌ coping with multiple space‌‌ and time scales, as well as with non-trivial‌ geometrical features (shape and‌ topological organization of scatterers‌‌ in the medium). In the context of a‌ collaboration with physicists from‌ the Institut de Physique‌‌ de Nice INPHYNI (Gian Luca Lippi from the‌ complex photonic systems and‌ materials group), we initiated‌‌ this year a study on the simultaneous action‌ of stimulated emission and‌ multiple scattering by randomly‌‌ distributed nanospheres in a bulk medium (see Fig.‌ 12). From the‌ numerical modeling point of‌‌ view, our short term goal is to develop‌ a time-domain numerical methodology‌ for the simulation of‌‌ random lasing in a‌ gain medium.

Figure 12.a — Figure 12: Multiple‌ scattering by randomly distributed nanospheres in a bulk‌ medium.

8.7.5 Nonlinear wavefront shaping with optical metasurfaces‌

Participants: Giuseppe Leo [CNRS-MPQ, Université Paris Cité, France]‌, Jean-Michel Gerard [PHELIQS, CEA and Université Grenoble‌ Alpes, France], Mahmoud Elsawy, Stéphane Lanteri‌, Francisco Teixeira Orlandini.

In recent years,‌ the control of sub-wavelength light-matter interactions has enabled‌ the observation of new linear optical phenomena, further‌ establishing a new class of ultra-thin devices for‌ real-world applications. To extend metasurface functionality and implement‌ nonlinear manipulations, the scientific community has considered optical‌ metasurfaces for harmonic emission field control. However, the‌ performance of nonlinear metasurfaces is still modest. Flat‌ optics have also shown their potential in the‌ nonlinear optics with wavefront shaping in far-harmonic fields.‌ There is currently a related effort by the‌ nonlinear nanophotonics community to seek higher conversion efficiencies‌ across narrow resonances of the quasi-bound states of‌ the continuum (qBIC) associated with strong near-field coupling‌ and nonlocal resonance modes. However, the overall performance‌ is relatively weak as most of the current‌ studies ignore the strong near-field coupling between neighboring‌ cells. In this collaboration, we exploit the specific‌ advantages of 'thin' resonators that behave like phased-array‌ antennas, unlike photonic crystals with strong localized energies,‌ to develop highly efficient nonlinear metasurfaces for nonlinear‌ wavefront shaping. Within this strategy, we improve the‌ performance of nonlinear wavefront shaping by inverse design‌ optimization of both meta-atoms and meta-molecules. In addition,‌ we consider long-term design options for nonlinear metasurfaces‌ based on perfect nonlocal responses and long-range near-field‌ coupling, and rigorous computational methods to address nonlinearities‌ in terms of nonlocal configurations. Our preliminary results‌ (see Fig. 13) show the ability to‌ improve the second harmonic generation signal by almost‌ a factor of seven. Fabrication and characterization of‌ the structure is currently underway, and the results‌ will be published in a prominent journal. Since‌ October 2024, we continue this study in the‌ context of the ANR NO-RESTRAIN project, which is‌ coordinated by Giuseppe Leo from the MPQ (Matériaux‌ et Phénomènes Quantiques) at Université Paris Cité, France.‌

Figure‌ 13: Optimized second-harmonic generation (SHG) from an‌ asymmetric nanochair (inset). The left column represent the‌ comparison of SHG response as a function of‌ the pump wavelength for three different designs (classical‌ denote the design without optimization). The right column‌ refers to the field profiles at the two‌ frequencies $2 ω$ and $ω$ for the optimal‌ design.

9 Bilateral contracts and grants with industry‌

9.1 Bilateral contracts with industry

Simulation of photonic pigments

Participants: Alexis Gobé‌, Stéphane Lanteri.‌

Duration: Oct 2024 -‌‌ Mar 2026
Local coordinator: Stéphane Lanteri
Participants: Valérie‌ Alard [LVMH], Nicolas Benoot‌ [LVMH]
To manufacture a‌‌ colored material, one normally use a dye or‌ pigment. But another approach‌ to producing color is‌‌ to fabricate a nanostructure that reflects or scatters‌ light so that waves‌ of certain frequencies can‌‌ interfere constructively. These nanostructured materials are said to‌ have structural colors. Unlike‌ traditional colors, which comes‌‌ from light-absorbing dyes or pigments that absorb light,‌ structural colors can be‌ made resistant to fading.‌‌ In this context, it is desirable to obtain‌ a structural color that‌ is independent of angle,‌‌ i.e., the color is the same regardless of‌ the orientation of the‌ material, and whatever the‌‌ angle between the light source and the eyes.‌ There are many structurally‌ colored materials that, like‌‌ an opal stone, are iridescent, which means that‌ the color changes depending‌ on viewing angle and‌‌ orientation. The reason for this is that the‌ nanostructure of these materials‌ is well-ordered (or crystalline),‌‌ as in photonic crystals. To manufacture materials whose‌ color is independent of‌ angle, we need to‌‌ create disordered nanostructures. These materials are called photonic‌ glasses. The aim is‌ to study how the‌‌ optical properties of these glasses are linked to‌ their structure and the‌ particles. This type of‌‌ study is primarily based on experiment. In this‌ project, we relied on‌ numerical modeling to study‌‌ the optical properties of photonic pigments.

Metasurfaces for‌ the visible, mid-infrared and‌ long infrared

Participants: Mahmoud‌‌ Elsawy, Arthur Gouinguenet, Stéphane Lanteri.‌

Duration: Sep 2024 -‌ March 2026
Local coordinator:‌‌ Stéphane Lanteri
Participants: Michel Jegouzo [Safran E&D], Pascal‌ Junique [Safran E&D], Emmanuel‌ Kling [Safran E&D]
We‌‌ initiated this year a novel collaboration with Safran‌ Electronics & Defense in‌ Eragny for the design‌‌ of several variants of metasurfaces or the visible,‌ mid-infrared and long infrared‌ spectra. In this context,‌‌ we explore various shapes of meta-atoms for the‌ definition of metadeflectors and‌ metalenses driven by precise‌‌ operational performances.

10 Partnerships and cooperations

10.1 International‌ initiatives

10.1.1 Associate Teams‌ in the framework of‌‌ an Inria International Lab or in the framework‌ of an Inria International‌ Program

DNN4Photonics

Title:
Deep‌‌ Neural Networks for the design of photonic devices‌
Partner Institution(s):
Laboratory of‌ Applied and Computational Electromagnetics‌‌ (LEMAC), Universidade Estadual de Campinas, Brazil
Coordinators:
Stéphane‌ Lanteri and Hugo Enrique‌ Hernandez Figueroa (LEMAC laboratory)‌‌
Date/Duration:
Jan 2024 to Dec 2026
Additionnal info/keywords:‌
nanophotonics, Deep Learning
Summary.‌
In the context of‌‌ this partnership with researchers form the LEMAC laboratory‌ at Universidade Estadual de‌ Campinas, we aim at‌‌ studying and developing disruptive approaches based on Deep‌ Neural Networks for modeling‌ and shaping the interactions‌‌ of optical waves (or light waves) with matter‌ when the latter is‌ structured at the subwavelength‌‌ scale. Electronic devices that exploit these interactions in‌ their design are called‌ photonic devices. We propose‌‌ to investigate data-driven Deep‌ Learning methodologies for two main objectives: (1) devising‌ fast and reliable surrogates of light wave propagation‌ in the general case of three-dimentional spatial domains‌ with complex scattering objects and, (2) devising inverse‌ design strategies of photonic devices with unprecedented properties.‌

Other international visits to the team

Haogang Cai‌

Status
associate professor
Institution of origin:
Tech4Health Institute,‌ NYU School of Medicine; Biomedical Engineering, NYU School‌ of Engineering
Country:
USA
Dates:
Mar 30 to‌ Apr 16
Context of the visit:
collaboration on‌ the design of metasurfaces for biosensing and participation‌ to the ICON-W workshop at Université Côte d'Azur‌
Mobility program/type of mobility:
research stay

10.2 National‌ initiatives

10.2.1 ANR projects

NumPEx (Digital for Exascale)‌

Participants: Daria Hrebenshchykova, Stéphane Lanteri, Victor‌ Michel-Dansac [MACARON project-team, Centre Inria de l'Université de‌ Lorraine].

Type: PEPR (Priority Research Program and‌ Equipment)
Coordinator of the Exa-MA project: Université de‌ Strasbourg
Partners: CEA, École Polytechnique, Inria, Sorbonne Université,‌ Université de Strasbourg
Inria contact: Stéphane Lanteri
Duration:‌ Jan 2023 to Mar 2028
The team is‌ involved in the Exa-MA (Methods and Algorithms for‌ Exascale) project of PEPR NumPEx, and more particularly‌ in the work-package dedicated to model order reduction‌ and SciML. The Exa-MA project aims to push‌ the frontiers of exascale computing by developing cutting-edge‌ numerical methods, algorithms, and software libraries. The PhD‌ thesis of Daria Hrebenshchykova is funded by this‌ project.

SWEET (Sub-WavelEngth Electro-optic sysTems)

Participants: Henri Camon‌ [CNRS-LAAS], Jean-Yves Duboz [CNRS-CRHEA], Mahmoud Elsawy‌, Olivier Gauthier-Lafaye [CNRS-LAAS], Samira Khadir [CNRS-CRHEA]‌, Stéphane Lanteri, Daniel Turover [NAPA Technologies]‌.

Type: ANR
Duration: Jan 2023 to Dec‌ 2026
Coordinator: CNRS-LAAS (Toulouse)
Partners: Inria (ATLANTIS project-team),‌ CNRS-CRHEA (Sophia Atipolis), NAPA Technologies (Archamps)
Inria contact:‌ Stéphane Lanteri
Beam steering is a key enabling‌ photonic technology that would improve the performance of‌ light detection and ranging modules (LiDAR). A typical‌ LiDAR component consists of a light source for‌ illumination, a light modulating device to scan the‌ scene and finally a fast detection system to‌ recover the optical information received from the scene.‌ The operation principle of conventional LiDARs relies on‌ the Time-of-Flight (ToF) measurement, where a pulsed laser‌ directed toward a distant reflective object measures the‌ propagation round-trip time (ToF) of light pulses propagating‌ from the laser to the scene and back‌ to the detection module. The LiDAR sector is‌ currently ongoing important research and development efforts to‌ enable real-time sensing of the distance of fast-moving‌ objects, with applications in robotics, autonomous vehicles and‌ future augmented reality devices. Dynamic beam steering with‌ competitive performances requires the deflection of a light‌ beam along any arbitrary direction to spatially scan‌ a large angular field-of-view (FoV) with high speed‌ and high efficiency. SWEET addresses several drawbacks of‌ current LiDAR technologies by proposing innovative ultrafast beam‌ steering systems, their combinations and integration into demonstrator.‌ We propose to realize an ultrafast 2D beam‌ steering system using innovative transparent tunable metasurface using LC and III-nitride materials.‌ Our motivation is to‌ develop a generic technology‌‌ to meet the market needs of LiDAR applications‌ in terms of operation‌ speed, FoV, angular resolution,‌‌ manufacturability. As indicators, we consider the most stringent‌ requirements for LiDAR integration‌ in the automotive industry.‌‌ In this context, the general objective of our‌ contribution to the SWEET‌ project is to design‌‌ active metasurface components for dynamic beam steering.

SWAG-P‌ (Sensing With A Gap-Plasmon)‌

Participants: David Duche [CNRS-IM2NP]‌‌, Frederic Dumur [CNRS-ICR], Stéphane Lanteri,‌ Olivie Margeat [CNRS-IM2NP],‌ Antoine Moreau [CNRS-Institut Pascal]‌‌, Carmen Ruiz-Herrero [CNRS-IM2NP], Claire Scheid,‌ Beniamino Sciacca [CNRS-CINaM].‌

Type: ANR
Duration: Jan‌‌ 2024 to Dec 2027
Coordinator: CNRS-Institut Pascal (Clermont-Ferrand)‌
Partners: Inria (ATLANTIS project-team),‌ CNRS-IM2NP (Marseille), CNRS-CINaM (Marseille),‌‌ CNRS-ICR (Marseille)
Inria contact: Claire Scheid
Point of‌ Care (POC) tests are‌ expected to continue to‌‌ be increasingly relied on in the coming years.‌ They have been central‌ to the strategy for‌‌ combating the COVID-19 pandemic and are expected to‌ play an increasingly important‌ role in the future‌‌ for fighting other epidemics, particularly in developing countries‌ such as India or‌ China. The World Health‌‌ Organization (WHO) has defined the ideal POC test,‌ emphasizing the importance of‌ Affordable, Sensitive, Specific, User-‌‌ friendly, Rapid/Robust, Equipment-free, and Deliverable (ASSURED) diagnostic methods.‌ Among all the techniques‌ used to design biosensors,‌‌ optical techniques are booming, thanks to numerous innovations‌ in nanophotonics and plasmonics.‌ However, not so many‌‌ innovations meet the ASSURED criteria. To be deliverable,‌ a device has to‌ be stable for months.‌‌ It is important to use as little materials‌ as possible, because sensors‌ have to be disposed‌‌ of after use. Miniaturization is attractive in this‌ perspective, especially since it‌ is associated with lower‌‌ detection limits. Plasmonic resonators have an edge here‌ because of their large‌ interaction cross-section: their response‌‌ is thus easier to measure. Regarding all the‌ ASSURED criteria, plasmonic sensors‌ at the end of‌‌ a fiber offer very interesting characteristics. The goal‌ of the SWAG-P project‌ is to explore the‌‌ potential of a new class of plasmonic resonators‌ with unique properties, called‌ gap-plasmon resonators, for the‌‌ detection of biological molecules of interest. These resonators‌ are optical “patch antennas”‌ that can be fabricated‌‌ simply by depositing metal nanocubes of a few‌ tens of nanometers on‌ a dielectric film, deposited‌‌ on a very thin layer of functionalizable gold.‌ We want to fabricate‌ a biosensor based on‌‌ an optical fiber to interrogate gap-plasmon resonators through‌ the thin metallic film,‌ from within the fiber.‌‌ Such a biosensor can potentially meet all the‌ criteria of the ASSURED‌ framework.

NO-RESTRAIN (NOnlinear metasuRfacE‌‌ on Silicon photodiode foR infrAred detectIoN)

Participants: Jean-Michel‌ Gerard [PHELIQS, CEA and‌ Université Grenoble Alpes, France]‌‌, Mahmoud Elsawy, Giuseppe Leo [CNRS-MPQ, Université‌ Paris Cité, France],‌ Stéphane Lanteri.

Type:‌‌ ANR
Duration: Oct 2024 to Sep 2027
Coordinator:‌ CNRS-MPQ (Grenoble)
Partners: Inria‌ (ATLANTIS project-team), CNRS-MPQ, Université‌‌ Paris Cité (Paris), PHELIQS,‌ CEA and Université Grenoble Alpes (Grenoble)
Inria contact:‌ Mahmoud Elsawy
In the last years, metasurfaces have‌ revolutionized the field of optics, with the promise‌ of replacing bulky optical systems and providing new‌ functionalities by nanostructured thin films. Flat optics also‌ showed its potential in the nonlinear regime, mostly‌ in III-V semiconductors, and in the NOMOS project‌ MPQ and PHELIQS recently achieved second harmonic generation‌ with phase-front control in a nonlinear metasurface (NLMS).‌ Today the NO-RESTRAIN project aims to harness the‌ full potential of NLMSs by tackling an old‌ problem of nonlinear optics: the upconversion of infrared‌ radiation into the silicon absorption band via sum‌ frequency generation (SFG). By heterogeneous integration of a‌ high-efficiency NLMS, a linear metalens and a single‌ photon avalanche photodiode (SPAD), the project will demonstrate‌ a miniature device allowing for ultrafast detection of‌ infrared signals with wavelength beyond the fast-detection limit‌ of InGaAs APDs. To this end, we will‌ address a triple challenge in design, forefront nanofabrication,‌ and ultrafast characterization. Based on the complementary competences‌ of partners INRIA , MPQ and PHELIQS in‌ design, nanofabrication and nanophotonic measurements, we will both‌ push forward fundamental research and go beyond purely‌ academic interest. The NO-RESTRAIN device will provide the‌ breakthrough of ultrafast detection of infrared radiation at‌ 300 K, spurring the transition of the young‌ NLMS research field from fundamental research into a‌ set of high-impact applied technologies.

DNN4Photonics (Deep Neural‌ Networks for the design of photonic devices)

Participants:‌ Arthur Clini de Souza, Mahmoud Elsawy,‌ Stéphane Lanteri, Huynh Thanh Phuong Lê.‌

Type: ANR-FAPESP
Duration: Jan 2025 to Jun 2028‌
Coordinator: Inria (ATLANTIS project-team)
Partners: Laboratory of Applied‌ and Computational Electromagnetics (LEMAC), Universidade Estadual de Campinas‌ (Campinas, Brazil), Solnil (Marseille)
Inria contact: Stéphane Lanteri‌
The team is coordinating the DNN4Photonics2 project.‌ In DNN4Photonics, our primary objective is to develop‌ innovative methodologies relying on Deep Neural Networks (DNN)‌ for the modeling and optimization of diverse metasurface‌ configurations, with a particular emphasis on large-scale structures.‌ In DNN4Photonics, our research thrust revolves around exploring‌ data-driven DL techniques to develop rapid and dependable‌ surrogates capable of emulating the simulation of three-dimensional‌ spatial domains involving complex scattering objects. Moreover, a‌ key focus is to find an intelligent and‌ efficient formulation for DNN-based inverse design strategies (e.g.,‌ retrieving the structural morphology of a metasurface provided‌ a target function) tailored for large-scale photonic devices.‌ This encompasses not only enhancing the efficiency and‌ reliability of large-scale metasurface modeling but also strategically‌ tailoring the designs for applications demanding high precision,‌ such as the optimization of RGB metalenses at‌ visible frequency for achromatic, micrometer- and millimeter-sized devices.‌

MAXINET (Physics informed neural networks for harmonic Maxell‌ equations)

Participants: Stéphane Descombes, Mahmoud Elsawy,‌ Stéphane Lanteri, Alexandre Pugin.

Type: ANR-ASTRID‌
Duration: Jan 2025 to Dec 2028
Coordinator: Inria‌ (ATLANTIS project-team)
Partners: Thales Research & Technology (Palaiseau)‌
Inria contact: Stéphane Descombes
The team is coordinating the MAXINET project, which‌ aims to develop a‌ new generation of artificial‌‌ intelligence (AI)-enhanced digital engineering tools for the simulation‌ time-harmonic electromagnetic wave propagation‌ and the design of‌‌ radio-frequency components and systems. These tools will be‌ based on the Physics-Informed‌ Neural Network (PINN) and‌‌ Neural Operator (NO) concepts, which are modern concepts‌ in Scientific Machine Learning‌ (SciML) research. MAXINET proposes‌‌ to explore these concepts by focusing on the‌ scientific obstacles identified for‌ their effective application to‌‌ the specificities of time-harmonic electromagnetic wave propagation problems‌ including non-trivial boundary conditions,‌ complex geometries and multi-material‌‌ domains (heterogeneous media).

10.2.2 DGA/AID RAPID project

AEROCOM‌ (Ultra-flat and low-cost antennas)‌

Participants: Ayoub Bellouch,‌‌ Guillaume Bouchet [NANOE], Guillaume de Calan [NANOE]‌, Mahmoud Elsawy,‌ Van Hoang [Thales Research‌‌ & Technology], Guillaume Leroy, Stéphane Lanteri‌, Julien Sourice [NANOE]‌, Erika Vandelle [Thales‌‌ Research & Technology].

Type: RAPID
Duration: Jan‌ 2023 to May 2026‌
Coordinator: NANOE (Palaiseau)
Partners:‌‌ Inria (ATLANTIS project-team), Thales Research & Technology (Palaiseau)‌
Inria contact: Stéphane Lanteri‌
The development of agile,‌‌ ultra-flat and low-cost Ka-band antennas is a major‌ challenge to enable Internet‌ accessibility in mobility, in‌‌ particular on board of public land and air‌ transport (trains, buses, airliners),‌ and to secure communication‌‌ servers (for combat aircraft, military vehicles, etc.). A‌ possible antenna architecture to‌ address this challenge is‌‌ composed of a radiating source and a deflection‌ system consisting of two‌ deflectors. The compactness and‌‌ the moderate cost of the de-pointing system could‌ be obtained thanks to‌ the sub-wavelength structuring technique,‌‌ and to the shaping by additive manufacturing. Indeed,‌ the sub-wavelength patterning technique‌ has recently shown the‌‌ possibility to realize antenna components much thinner than‌ a homogeneous bulk material,‌ with equivalent or even‌‌ better radio frequency performance. In this context, the‌ general objective of our‌ contribution to the AEROCOM‌‌ project is to develop an advanced numerical methodology‌ for the virtual design‌ of subwavelength structured deflectors‌‌ and their cascading to achieve an ultra-flat Ka-band‌ antenna system consisting of‌ two such metadeflectors.

11‌‌ Dissemination

Participants: Stéphane Descombes, Mahmoud Elsawy,‌ Stéphane Lanteri, Claire‌ Scheid.

11.1 Promoting‌‌ scientific activities

11.1.1 Scientific events: organization

Member of‌ the organizing committees

Stéphane‌ Lanteri was a member‌‌ of the organization comittee of the "ICON -‌ Nonlinear cell photonics" thematic‌ semester of Université Côte‌‌ d'Azur. The overarching ICON project was to consolidate‌ ongoing research initiatives at‌ Université Côte d'Azur, encompassing‌‌ various aspects of photonics and life sciences, from‌ experimental and applied methods‌ to numerical modeling. Additionally,‌‌ the project has enriched these endeavors by integrating‌ external insights and expertise,‌ fostering collaborations with researchers‌‌ from France and beyond. In this context, the‌ ICON-W workshop has been‌ organized at Université Côte‌‌ d'Azur on April 2-5, 2025.

11.1.2 Invited talks‌

"Advanced numerical design methodologies‌ for next-generation metasurface architectures",‌‌ Mahmoud Elsawy , General Assembly of GDR Ondes,‌ October 29-30, 2025, Besançon,‌ France
"Trimer metasurfaces for‌‌ highly sensitive biomedical sensors",‌ Mahmoud Elsawy , PIERS 2025 - PhotonIcs &‌ Electromagnetics Research Symposium, May 4-8, 2025, Abu Dhabi,‌ United Arab Emirates
"Numerical modelling in nanoplasmonics", Claire‌ Scheid , Workshop in honor of the 60th‌ birthday of Patrick Ciarlet, June 17, 2025, Institut‌ Henri Poincaré, Paris, France
"Inverse design of nanophotonic‌ devices using high-order fullwave solvers and statistical learning‌ global optimization algorithms" Stéphane Lanteri , AES 2025‌ - 11th International Conference on Antennas and Electromagnetic‌ Systems, July 1-4, 2025, Tangier, Morocco

11.1.3 Leadership‌ within the scientific community

Since July 2023, Claire‌ Scheid is member of the Scientific Committee of‌ the "Maison de la simulation et des interactions"‌ of Université Côte d'Azur
Since October 2024, Claire‌ Scheid is an elected CNU (Comité National des‌ Universités) member of the 26th section (applied mathematics)‌
Since June 2024, Stéphane Lanteri is a member‌ of the Scientific Council of the 3iA Côte‌ d'Azur
Since January 2023, Mahmoud Elsawy is a‌ member of the Scientific Council of the academia‌ of Complex System (Physics section) of Université Côte‌ d'Azur Idex

11.1.4 Research administration

Mahmoud Elsawy is‌ a member of the ANR evaluation committee CE24‌ in 2025 and 2026
Claire Scheid is responsible‌ for the membership management of the applied and‌ industrial mathematics french association (SMAI) since January 2016‌
Claire Scheid is responsible (on the math side)‌ for the training programme "Double Licence Mathématique-Physique" (L1‌ to L3) of Université Côte d'Azur, since September‌ 2024
Stéphane Descombes is responsible of the second‌ year of the master Mathematics for Data Sciences‌ from Université Côte d'Azur since September 2024
Stéphane‌ Descombes is responsible of the second year of‌ the double diploma Université Côte d'Azur - EDHEC‌ Business School since September 2023
Stéphane Lanteri is‌ a member of the Project-team Committee's Bureau of‌ the Inria research center at Université Côte d'Azur‌ since January 2022
Stéphane Lanteri is a member‌ of the Project-team Committee's Bureau of the Inria‌ research center at Université Côte d'Azur since January‌ 2022
Since September 2022, Stéphane Lanteri is the‌ Deputy Head of Science of the Inria Research‌ Center at Université Côte d'Azur

11.2 Teaching -‌ Supervision - Juries - Educational and pedagogical outreach‌

11.2.1 Teaching

Mahmoud Elsawy , Resolution of linear‌ and nonlinear systems, L3, 12 h, Université Côte‌ d'Azur
Mahmoud Elsawy , Machine Learning in Python,‌ MAM5, 12 h, Polytech Nice Sophia, Université Côte‌ d'Azur
Claire Scheid , Option modélisation, M2 Agrégation,‌ 45 h, Université Côte d'Azur
Claire Scheid ,‌ Approximation Numérique des fonctions, intégrales et équations différentielles,‌ L3, 36 h, Université Côte d'Azur
Claire Scheid‌ , Complément d'algèbre linéaire, L2, 20 h, Université‌ Côte d'Azur
Claire Scheid , Equations différentielles ordinaires,‌ L3, 11 h, Université Côte d'Azur
Stéphane Descombes‌ , Scientific Machine Learning, MAM5, 15 h, Polytech‌ Nice Sophia, Université Côte d'Azur
Stéphane Descombes ,‌ Approximation Numérique des fonctions, intégrales et équations différentielles,‌ L3, 28 h, Université Côte d'Azur
Stéphane Descombes‌ , Introduction aux équations aux dérivées partielles, M1, 30 h, Université Côte‌ d'Azur.
Stéphane Descombes ,‌ Calcul scientifique, M1, 30‌‌ h, Université Côte d'Azur
Stéphane Descombes , Système‌ dynamique et analyse numérique,‌ L3, 40 h, Université‌‌ Côte d'Azur
Stéphane Lanteri , Scientific Machine Learning,‌ MAM5, 20 h, Polytech‌ Nice Sophia, Université Côte‌‌ d'Azur

11.2.2 Supervision

PhD in progress: Arthur Clini‌ De Souza. (Cifre PhD‌ grant with Solnil, Marseille),‌‌ Deep neural networks for the design of large-scale‌ photonic devices for active‌ beam steering, co-supervised by‌‌ Mahmoud Elsawy and Stéphane Lanteri .
PhD in‌ progress: Daria Hrebenshchykova (PEPR‌ NumPEx PhD fellowship), Building‌‌ physics-based multilevel substitution models from neural networks. Application‌ to electromagnetic wave propagation,‌ co-supervised by Victor Michel-Dansac‌‌ (MACARON project-team, Centre Inria de l'Université de Lorraine),‌ Victorita Dolean (Eindhoven University‌ of Technology, The Netherlands)‌‌ and Stéphane Lanteri .
PhD in progress: Enzo‌ Isnard (Cifre PhD grant‌ with Thales Research &‌‌ Technology, Palaiseau), Modeling and optimization of freeform optical‌ metasurfaces integrated into an‌ imaging system, co-supervised by‌‌ Mahmoud Elsawy and Stéphane Lanteri .
PhD in‌ progress: Carlotta Filippin (ANR‌ SWAG-P PhD fellowship), Reduced-order‌‌ modeling and global optimization for the robust design‌ of Gap Plasmon Resonators,‌ co-supervised by Claire Scheid‌‌ , Antoine Moreau (Institut Pascal, Université Clermont-Auvergne) and‌ Stéphane Lanteri .
PhD‌ in progress: Roman Gelly‌‌ (PhD fellowship from Université Côte d'Azur), Advanced numerical‌ modeling of time-modulated metasurfaces,‌ co-supervised by Mahmoud Elsawy‌‌ and Stéphane Lanteri .
PhD defended: Thibault Laufroy‌ (PhD fellowship from Université‌ Côte d'Azur), High order‌‌ finite element type solvers for thermoplamonics, co-supervised by‌ Yves d'Angelo and Claire‌ Scheid . Defended in‌‌ December 2025.
PhD in progress: Cédric Legrand (DGA-Inria‌ PhD fellowship), High order‌ finite element type solvers‌‌ for modeling gain media, co-supervised by Stéphane Descombes‌ and Stéphane Lanteri .‌
PhD in progress: Huynh‌‌ Thanh Phuong Lê (ANR-FAPEST DNN4Photonics fellowship), Deep Learning‌ methods for the design‌ of large-scale photonic devices,‌‌ co-supervised by Mahmoud Elsawy and Stéphane Lanteri .‌
PhD in progress: Florentin‌ Proust, Novel finite element‌‌ methods for dealing with high frequency wave propagation‌ problems, co-supervised by Maxime‌ Ingremeau (Université Côte d'Azur,‌‌ LJAD) and Théophile Chaumont-Frelet .
PhD in progress:‌ Alexandre Pugin (DGA-Inria PhD‌ fellowship), Physically informed neural‌‌ networks for building surrogate models in numerical electromagnetics,‌ co-supervised by Stéphane Descombes‌ and Stéphane Lanteri .‌‌
Postdoc (until Sep 2025): Ayoub Bellouch (AEROCOM RAPID‌ project), Advanced computational design‌ of cascaded metadeflectors for‌‌ an ultra-flat antenna system, co-supervised by Mahmoud Elsawy‌ and Stéphane Lanteri .‌
Postdoc in progress: Alemayehu‌‌ Getahun Kumela (Inria Exploratory Action META4PIQ), Metasurfaces for‌ quantum information processing, supervised‌ by Mahmoud Elsawy .‌‌
Postdoc in progress: Francisco Teixeira Orlandini (ANR NO-RESTRAIN‌ project), Computational design of‌ active and nonlinear optical‌‌ metasurfaces, co-supervised by Mahmoud Elsawy and Stéphane Lanteri‌ .

11.2.3 Juries

Claire‌ Scheid was a member‌‌ of the HDR committee of Erell Jamelot: Contributions‌ to the analysis of‌ finite element methods for‌‌ models in electromagnetism, neutronics and fluid dynamics, CEA‌ Saclay, Institut Polytechnique de‌ Paris, September 2025.
Stéphane‌‌ Lanteri was a reviewer‌ and a member of the jury of the‌ PhD thesis of Julien Besset: A model order‌ reduction strategy for parametrized PDEs: a new paradigm‌ for efficient subsurface imaging, Université de Pau et‌ des Pays de l'Adour, June 2025.
Stéphane Lanteri‌ was a member of the jury of the‌ PhD thesis of Adrien Talatizi: Simulation of ultrasonic‌ wave propagation in polycrystalline materials, Université Paris Sciences‌ et Lettres, June 2025.

12 Scientific production

12.1‌ Major publications

1 articleE.Emmanuel Agullo,‌ L.Luc Giraud, A.Alexis Gobé,‌ M.Matthieu Kuhn, S.Stephane Lanteri and‌ L.Ludovic Moya. High order HDG method‌ and domain decomposition solvers for frequency‐domain electromagnetics.‌International Journal of Numerical Modelling: Electronic Networks, Devices‌ and FieldsOctober 2019HAL DOI back to‌ text back to text
2 miscT.Théophile‌ Chaumont-Frelet. Asymptotically constant-free and polynomial-degree-robust a posteriori‌ estimates for space discretizations of the wave equation‌.April 2022HAL
3 miscT.Théophile‌ Chaumont-Frelet, V.Victorita Dolean and M.Maxime‌ Ingremeau. Efficient approximation of high-frequency Helmholtz solutions‌ by Gaussian coherent states.August 2022HAL‌
4 unpublishedT.Théophile Chaumont-Frelet, A.Alexandre‌ Ern and M.Martin Vohralík. On the‌ derivation of guaranteed and p-robust a posteriori error‌ estimates for the Helmholtz equation.August 2020‌, working paper or preprintHAL
5 report‌T.Théophile Chaumont-Frelet and M.Maxime Ingremeau.‌ Decay of coefficients and approximation rates in Gabor‌ Gaussian frames.Inria2023HAL
6 article‌A.Arthur Clini de Souza, S.Stéphane‌ Lanteri, H. E.Hugo Enirique Hernández-Figueroa,‌ M.Marco Abbarchi, D.David Grosso,‌ B.Badre Kerzabi and M.Mahmoud Elsawy.‌ Back-propagation optimization and multi-valued artificial neural networks for‌ highly vivid structural color filter metasurfaces.Scientific‌ Reports131December 2023, 21352HAL‌DOI back to textback to text
7‌ articleS.Stéphane Descombes, S.Stéphane Lanteri‌ and L.Ludovic Moya. Locally implicit discontinuous‌ Galerkin time domain method for electromagnetic wave propagation‌ in dispersive media applied to numerical dosimetry in‌ biological tissues.SIAM Journal on Scientific Computing‌3852016, A2611-A2633HAL DOI
8‌ articleS.Stéphane Descombes, S.Stéphane Lanteri‌ and L.Ludovic Moya. Temporal convergence analysis‌ of a locally implicit discontinuous galerkin time domain‌ method for electromagnetic wave propagation in dispersive media‌.Journal of Computational and Applied Mathematics316‌May 2017, 122--132HAL DOI
9 article‌M.Mahmoud Elsawy, M.Mickael Binois,‌ R.Régis Duvigneau, S.Stéphane Lanteri and‌ P.Patrice Genevet. Optimization of metasurfaces under‌ geometrical uncertainty using statistical learning.Optics Express‌29192021, 29887HAL DOI back‌ to text back to text
10 articleM.‌ M.Mahmoud M R Elsawy, A.Anthony‌ Gourdin, M.Mickael Binois, R.Régis‌ Duvigneau, D.Didier Felbacq, S.Samira Khadir, P.Patrice‌ Genevet and S.Stéphane‌ Lanteri. Multiobjective statistical‌‌ learning optimization of RGB metalens.ACS photonics‌88July 2021‌, 2498–2508HAL DOI‌‌back to text back to text
11 article‌M. M.Mahmoud M‌ R Elsawy, S.‌‌Stéphane Lanteri, R.Régis Duvigneau, G.‌Gauthier Brière, M.‌ S.Mohamed Sabry Mohamed‌‌ and P.Patrice Genevet. Global optimization of‌ metasurface designs using statistical‌ learning methods.Scientific‌‌ Reports91November 2019HAL DOI back‌ to text back to‌ text
12 articleM.‌‌Mahmoud Elsawy, S.Stéphane Lanteri, R.‌Régis Duvigneau, J.‌Jonathan Fan and P.‌‌Patrice Genevet. Numerical optimization methods for metasurfaces‌.Laser and Photonics‌ Reviews1410October‌‌ 2020, 1900445HALDOI back to text‌
13 articleX.-F.Xiao-Feng‌ He, L.Liang‌‌ Li, S.Stéphane Lanteri and K.Kun‌ Li. Model order‌ reduction for parameterized electromagnetic‌‌ problems using matrix decomposition and deep neural networks‌.Journal of Computational‌ and Applied Mathematics431‌‌October 2023, 115271HAL DOI back to‌ text
14 articleX.-F.‌Xiao-Feng He, L.‌‌Liang Li, S.Stéphane Lanteri and K.‌Kun Li. Reduced‌ Order Modeling for Parameterized‌‌ Electromagnetic Simulation Based on Tensor Decomposition.IEEE‌ Journal of Multiscale and‌ Multiphysics Computational Techniques8‌‌August 2023, 296-305HAL DOI
15 article‌S.Stephane Lanteri,‌ C.Claire Scheid and‌‌ J.Jonathan Viquerat. Analysis of a Generalized‌ Dispersive Model Coupled to‌ a DGTD Method with‌‌ Application to Nanophotonics.SIAM Journal on Scientific‌ Computing393January‌ 2017, A831 -‌‌ A859HAL DOI back to text back to‌ text
16 articleK.‌Kun Li, T.-Z.‌‌Ting-Zhu Huang, L.Liang Li and S.‌Stéphane Lanteri. A‌ reduced-order discontinuous Galerkin method‌‌ based on a Krylov subspace technique in nanophotonics‌.Applied Mathematics and‌ Computation358October 2019‌‌, 128-145HAL DOIback to text
17‌ articleK.Kun Li‌, T.-Z.Ting-Zhu Huang‌‌, L.Liang Li and S.Stéphane Lanteri‌. Non-intrusive reduced-order modeling‌ of parameterized electromagnetic scattering‌‌ problems using cubic spline interpolation.Journal of‌ Scientific Computing872‌May 2021HAL DOI‌‌back to text back to text
18 article‌K.Kun Li,‌ T.-Z.Ting-Zhu Huang,‌‌ L.Liang Li and S.Stéphane Lanteri.‌ POD-based model order reduction‌ with an adaptive snapshot‌‌ selection for a discontinuous Galerkin approximation of the‌ time-domain Maxwell's equations.‌Journal of Computational Physics‌‌396November 2019, 106-128HAL DOI back‌ to text
19 article‌K.Kun Li,‌‌ T.Ting‐zhu Huang, L.Liang Li and‌ S.Stéphane Lanteri.‌ Simulation of the interaction‌‌ of light with 3‐D metallic nanostructures using a‌ proper orthogonal decomposition‐Galerkin reduced‐order‌ discontinuous Galerkin time‐domain method‌‌.Numerical Methods for Partial Differential EquationsSeptember‌ 2022HAL DOI
20‌ articleK.Kun Li‌‌, T.-Z.Ting-Zhu Huang‌, L.Liang Li, Y.Ying Zhao‌ and S.Stéphane Lanteri. A non-intrusive model‌ order reduction approach for parameterized time-domain Maxwell's equations‌.Discrete and Continuous Dynamical Systems - Series‌ B281March 2022, 449-473HAL‌DOI
21 articleL.Liang Li, S.‌Stéphane Lanteri and R.Ronan Perrussel. A‌ hybridizable discontinuous Galerkin method combined to a Schwarz‌ algorithm for the solution of 3d time-harmonic Maxwell's‌ equations.Journal of Computational Physics256January‌ 2014, 563-581HALDOI back to text‌
22 articleK.Kun Li, Y.Yixin‌ Li, L.Liang Li and S.Stéphane‌ Lanteri. Surrogate modeling of time-domain electromagnetic wave‌ propagation via dynamic mode decomposition and radial basis‌ function.Journal of Computational Physics491October‌ 2023, 112354HALDOI back to text‌
23 articleN.Nikolai Schmitt, C.Claire‌ Scheid, S.Stéphane Lanteri, A.Antoine‌ Moreau and J.Jonathan Viquerat. A DGTD‌ method for the numerical modeling of the interaction‌ of light with nanometer scale metallic structures taking‌ into account non-local dispersion effects.Journal of‌ Computational Physics316July 2016HAL DOI
24‌ articleN.Nikolai Schmitt, C.Claire Scheid‌, J.Jonathan Viquerat and S.Stéphane Lanteri‌. Simulation of three-dimensional nanoscale light interaction with‌ spatially dispersive metals using a high order curvilinear‌ DGTD method.Journal of Computational Physics373‌November 2018, 210-229HAL back to text‌

12.2 Publications of the year

International journals

25‌ articleC.-S.Chen-Song Gan, L.Liang Li‌, S.Stéphane Lanteri, K.Kun Li‌, X.-F.Xiao-Feng He and B.Bin Li‌. Reduced-order modeling for parameterized electromagnetic simulation based‌ on free-form deformation.IEEE Antennas and Wireless‌ Propagation Letters247July 2025, 1595-1599‌HAL DOI back to text
26 articleE.‌Enzo Isnard, S.Sébastien Héron, S.‌Stéphane Lanteri and M.Mahmoud Elsawy. Hybrid‌ model to simulate optical systems combining metasurfaces and‌ classical refractive elements.Optics ExpressDecember 2025‌HAL DOI back to text

International peer-reviewed conferences‌

27 inproceedingsA.Ayoub Bellouch, M.Mahmoud‌ Elsawy, S.Stéphane Lanteri, E.Erika‌ Vandelle and T. Q.Thi Quynh Van Hoang‌. Optimization strategy to design rotating nonlocal meta-deflectors‌ for Risley-Prism antenna systems.IEEE XploreEuCAP‌ 2025 - 19th European Conference on Antennas and‌ Propagation2025 19th European Conference on Antennas and‌ Propagation (EuCAP)Stockholm, FranceIEEEMarch 2025,‌ 1-5HAL DOI back to text
28 inproceedings‌A.Ayoub Bellouch, M.Mahmoud Elsawy,‌ G.Guillaume Leroy, M.Mickael Binois,‌ R.Régis Duvigneau and S.Stéphane Lanteri.‌ Multi-Fidelity Bayesian Optimization of Metasurface Designs.Technical‌ Programme International Symposium on Electromagnetic Theory URSI EMTS‌ 2025 / IEEE XploreEMTS 2025 - 25th‌ URSI International Symposium on Electromagnetic TheoryBologna, Italy‌June 2025HAL back to text
29 inproceedingsT.Théophile Chaumont-Frelet,‌ V.Victorita Dolean,‌ M.Maxime Ingremeau and‌‌ F.Florentin Proust. A Galerkin method with‌ microlocalised shape functions to‌ solve high-frequency Helmholtz problems‌‌.Book of abstracsConference on Mathematics of‌ Wave Phenomena 2025Karlsruhe,‌ GermanyFebruary 2025HAL‌‌back to text
30 inproceedingsV.Victorita Dolean‌, D.Daria Hrebenshchykova‌, S.Stéphane Lanteri‌‌ and V.Victor Michel-Dansac. Neural network-driven domain‌ decomposition for efficient solutions‌ to the Helmholtz equation‌‌.DD 2025 - XXIX International Conference on‌ Domain Decomposition Methods in‌ Science and EngineeringMilano,‌‌ ItalyDecember 2025HALback to text
31‌ inproceedingsM.Mahmoud Elsawy‌ and R.Roman Gelly‌‌. Numerical modeling of time-modulated metasurfaces with a‌ Discontinuous Galerkin method.‌SpringerlinkOWTNM 2025 -‌‌ 31st International Workshop on Optical Wave & Waveguide‌ Theory and Numerical Modelling‌Optical and Quantum Electronics‌‌Nottingham, United KingdomApril 2025HAL back to‌ text
32 inproceedingsM.‌Mahmoud Elsawy, H.‌‌Hao Wang, A.Arash Nemati, S.‌Stéphane Lanteri and H.‌Haogang Cai. Trimer‌‌ Metasurfaces for Highly Sensitive Biomedical Sensors.IEEE‌ XplorePIERS 2025 -‌ Progress in Electromagnetics Research‌‌ Symposium2025 Photonics & Electromagnetics Research Symposium -‌ Spring (PIERS-Spring)Abu Dhabi,‌ United Arab EmiratesMay‌‌ 2025HAL back to text
33 inproceedingsT.‌T. Hoang, E.‌E. Vandelle, A.‌‌Ayoub Bellouch, M.Mahmoud Elsawy, S.‌Stéphane Lanteri and B.‌B. Loiseaux. Low-profile‌‌ Risley scanner using dielectric nonlocal metasurfaces for Ka-band‌ SATCOM applications.IEEE‌ XploreEuCAP 2025 -‌‌ 19th European Conference on Antennas and Propagation2025‌ 19th European Conference on‌ Antennas and Propagation (EuCAP)‌‌Stockholm, FranceIEEEMarch 2025, 1-5HAL‌DOI back to text‌
34 inproceedingsC.Cédric‌‌ Legrand, S.Stéphane Lanteri and S.Stéphane‌ Descombes. Semiclassical numerical‌ modeling of gain materials‌‌ with a high order discontinuous Galerkin time-domain solver‌.ENUMATH 2023 -‌ European Conference on Numerical‌‌ Mathematics and Advanced Applications154Lecture Notes in‌ Computational Science and Engineering‌Lisbon, PortugalSpringer Nature‌‌ SwitzerlandApril 2025, 115-123HAL DOI

Conferences‌ without proceedings

35 inproceedings‌M.Mahmoud Elsawy,‌‌ A.Arthur Clini de Souza, E.Enzo‌ Isnard, R.Roman‌ Gelly and S.Stéphane‌‌ Lanteri. Advanced numerical design methodologies for next-generation‌ metasurface architectures.GDR‌ Ondes 2025 - 11ème‌‌ Conférence Plénière du GDR ONDES SUPMICROTECH-ENSMMBesançon, France‌October 2025HAL back‌ to text
36 inproceedings‌‌C.Carlotta Filippin, S.Stephane Lanteri,‌ C.Claire Scheid,‌ F.F Pichi and‌‌ M.M Strazzullo. Graph-based nonlinear reduced-order modeling‌ for time-domain electromagnetics.‌MORTech 2025 - 7th‌‌ International Workshop on Model Order Reduction TechniquesZaragoza,‌ SpainNovember 2025HAL‌back to text
37‌‌ inproceedingsE.Enzo Isnard, S.Sébastien Héron‌, M.Mahmoud Elsawy‌ and S.Stéphane Lanteri‌‌. Optimization of imaging systems containing metasurfaces using‌ a ray-wave model.‌Metamaterials 2025 -19th International‌‌ Congress on Artificial Materials‌ for Novel Wave PhenomenaAmsterdam, NetherlandsSeptember 2025‌HAL back to text
38 inproceedingsP. E.‌Paloma E S Pellegrini, F. T.Francisco‌ T Orlandini, S. V.Silvia V G‌ Nista, S.Stéphane Lanteri, H. E.‌Hugo Enrique Hernandez-Figueroa and S.Stanislav Moshkalev.‌ Thermal scanning probe lithography for fabrication of perforated‌ metallic films.Integrated Photonics Research, Silicon and‌ NanophotonicsMarseille, FranceOptica Publishing Group2025,‌ JM3B.3HAL DOI

Scientific book chapters

39 inbook‌M.Mahmoud Elsawy, A.Alexi Gobé,‌ G.Guillaume Leroy, S.Stéphane Lanteri and‌ C.Claire Scheid. Advanced computational design of‌ complex nanostructured photonic devices using high order discontinuous‌ Galerkin methods and statistical learning global optimization.‌LNCSE-146Emerging Technologies in Computational Sciences for Industry,‌ Sustainability and InnovationLecture Notes in Computational Science‌ and EngineeringSpringer Nature SwitzerlandNovember 2025,‌ 223-239HAL DOI back to text

Edition (books,‌ proceedings, special issue of a journal)

40 proceedings‌CEMRACS 2023 - Scientific machine learning.CEMRACS‌ 202381Luminy (CIRM, Centre International de Rencontres‌ Mathématiques), FranceEDP SciencesOctober 2025, 1-1‌HAL DOI

Reports & preprints

41 miscA.‌Ayoub Bellouch, M.Mahmoud Elsawy, S.‌Stéphane Lanteri, E.Erika Vandelle and T.‌ Q.Thi Quynh Van Hoang. Nonlocal metasurfaces‌ based on advanced Bayesian learning for satellite communications‌.March 2025HAL
42 reportR.Roman‌ Gelly and M.Mahmoud Elsawy. Numerical modeling‌ of wave propagation in passive and time-modulated media‌.RR-9569Centre Inria d'Université Côte d'AzurJanuary‌ 2025, 49HALback to text
43‌ reportC.Cédric Legrand, S.Stéphane Lanteri‌ and S.Stéphane Descombes. Semiclassical numerical modeling‌ of gain materials with a high-order Implicit-Explicit Discontinuous‌ Galerkin time-domain solver.RR-9579INRIA Sophia Antipolis‌March 2025HAL back to text
44 report‌A.Alexandre Pugin, S.Stéphane Lanteri,‌ S.Stéphane Descombes and M.Mahmoud Elsawy.‌ Application of Physics-Informed Neural Networks to Maxwell equations‌.RR-9588Centre Inria d'Université Côte d'AzurMay‌ 2025HAL back to text
45 miscY.‌Yves d'Angelo, T.Thibault Laufroy and C.‌Claire Scheid. Numerical approximation of Thermoplasmonics effects‌ in a Discontinuous Galerkin framework.November 2025‌HAL back to text

Other scientific publications

46‌ inproceedingsA.Arthur Clini de Souza, S.‌Stephane Lanteri, H.Hugo Enrique Hernandez-Figueroa,‌ M.Marco Abbarchi, B.Badre Kerzabi,‌ P. E.Paloma Elias da Silva Pellegrini,‌ S. V.Silvia Vaz Guerra Nista and M.‌Mahmoud Elsawy. Deep learning methodolody for designing‌ large scale matelenses in reflection.NANOP 2025‌ - Nanophotonics and Micro/Nano Optics International Conference 2025‌Paris, FranceOctober 2025HAL
47 inproceedingsC.‌Carlotta Filippin, M.Maria Strazzullo, S.‌Stéphane Lanteri and F.Federico Pichi. Nonlinear‌ reduced-order modeling for time-domain electromagnetics.EMS-TAG-SciML 2025‌ - Annual Meeting of EMS activity group on Scientific Machine LearningMilan,‌ ItalyMarch 2025HAL‌
48 inproceedingsA.Alemayehu‌‌ Getahun Kumela, S.Stephane Lanteri and M.‌Mahmoud Elsawy. Advanced‌ nonlinear metasurface architecture for‌‌ entangled photon pair generation.NANOP 2025: Functional‌ NanophotonicsParis, FranceOctober‌ 2025HAL back to‌‌ text
49 inproceedingsA.Alexandre Pugin, S.‌Stéphane Lanteri and S.‌Stéphane Descombes. Application‌‌ of Physics-Informed Neural Networks to Maxwell equations.‌Numerical Analysis School 2025‌ - Solving partial differential‌‌ equations in fields physics faster with physics-based machine‌ learningPalaiseau, FranceJune‌ 2025HAL

12.3 Cited‌‌ publications

50 articleE.E. Agullo, L.‌L. Giraud and Y.‌Y.F. Jing. Block‌‌ GMRES method with ineact breackdowns and deflated restarting‌.SIAM J. Matrix‌ Anal. Appl.354‌‌2014, 1625--1261back to text
51 book‌G.G. Baffou.‌ Thermoplasmonics.Cambridge University‌‌ Press2017back to text
52 articleT.‌T. Chaumont-Frelet, V.‌V. Dolean and M.‌‌M. Ingremeau. Efficient approximation of high-frequency Helmholtz‌ solutions by Gaussian coherent‌ states.Numer. Math.‌‌1562024, 1385–1426back to text back‌ to text
53 article‌X.X. Chen,‌‌ H.H.R. Park, M.M. Pelton,‌ X.X. Piao,‌ N.N.C. Lindquist,‌‌ H.H. Im, Y.Y.. Kim,‌ J.J.S. Ahn,‌ K.K.J. Ahn,‌‌ N.N. Park, D.D.S. Kim and‌ S.S.H. Oh.‌ Atomic layer lithography of‌‌ wafer-scale nanogap arrays for extreme confinement of electromagnetic‌ waves.Nature Comm.‌42013, 2361‌‌back to text
54 articleH.H.T. Chen‌, A.A.J. Taylor‌ and N.N. Yu‌‌. A review of metasurfaces: physics and applications‌.Rep. Progr. Phys.‌7972016,‌‌ 076401back to text
55 articleC.C.‌ Cirac\`i, R.R.T.‌ Hill, J.J.J‌‌ Mock, Y.Y. Urzhumov, A.A.I.‌ Fernández-Domínguez, S.S.A.‌ Maier, J.J.B.‌‌ Pendry, A.A. Chilkoti and D.D.R.‌ Smith. Probing the‌ ultimate limits of plasmonic‌‌ enhancement.Science33760982012, 1072--1074‌back to text
56‌ articleF.F. Ding‌‌, A.A. Pors and S.S.I. Bozhevolnyi‌. Gradient metasurfaces: a‌ review of fundamentals and‌‌ applications.Rep. Progr. Phys.8122018‌, 026401back to‌ text
57 articleR.‌‌R. Duvigneau and P.P. Chandrashekar. Kriging-based‌ optimization applied to flow‌ control.Int. J.‌‌ Num. Fluids692012, 1701--1714back to‌ text
58 articleR.‌R. Duvigneau, J.‌‌J. Labroquère and E.E. Guilmineau. Comparison‌ of turbulence closures for‌ optimized active control.‌‌Comput. Fluids1242016, 67--77back to‌ text
59 bookS.‌S.V. Gaponenko. Introduction‌‌ to nanophotonics.Cambridge University Press2010back‌ to text
60 article‌P.P. Genevet and‌‌ F.F. Capasso. Holographic optical metasurfaces: a‌ review of current progress‌.Rep. Progr. Phys.‌‌7822015,‌ 024401back to text
61 articleY.Y.‌ He, M.M. Razi, C.C.‌ Forestiere, L. D.L. Dal Negro and‌ R.R.M. Kirby. Uncertainty quantification guided robust‌ design for nanoparticles morphology.Comp. Meth. Appl.‌ Mech. Engrg.893362018, 578--593back‌ to text
62 bookS.S.A. Maier.‌ Plasmonics - Fundamentals and applications.Springer2007‌back to text back to text
63 book‌V.V.V. Mitin, V.V.A. Kochelap and‌ M.M.A. Stroscio. Introduction to nanoelectronics.‌Cambridge University Press2012back to text
64‌ articleS.S. Molesky, Z.Z. Lin‌, A.A.Y. Piggott, W.W. Jin‌, J.J. Vuckovic and A.A.W. Rodriguez‌. Inverse design in nanophotonics.Nat. Phot.‌122018, 659--670back to text
65‌ articleF.F. Pichi, B.B. Moya‌ and J. H.J.S. Hesthaven a. A‌ graph convolutional autoencoder approach to model order reduction‌ for parametrized PDEs.J. Comput. Phys.501‌2024, 112762back to text
66 article‌G.G. Pichon, E.E. Darve,‌ M.M. Faverge, P.P. Ramet and‌ J.J. Roman. Sparse supernodal solver using‌ block low-rank compression: design performance and analysis.‌J. Comp. Sci.272018, 255--270back‌ to text
67 articleM.M. Raissi,‌ P.P. Perdikaris and G.G.E. Karniadakis.‌ Physics-informed neural networks: a deep learning framework for‌ solving forward and inverse problems involving nonlinear partial‌ differential equations.J. Comput. Phys.3782019‌, 686--707back to text back to text‌
68 articleM.M. Sacher, F.F.‌ Hauville, R.R. Duvigneau, O. L.‌O. Le Mâitre, N.N. Aubib and‌ M.M. Durand. Efficient optimization procedure in‌ non-linear fluid-structure interaction problem: application to mainsail trimming‌ in upwind conditions.J. Fluids and Struct.‌692017, 209--231back to text
69‌ phdthesisN.N. Schmitt. High order simulation‌ and calibration strategies for spatial dispersion models in‌ nanophotonics.University of Nice-Sophia Antipolis2018back‌ to text
70 articleK.K. Yao,‌ R.R. Unni and Y.Y. Zheng.‌ Intelligent nanophotonics: merging photonics and artificial intelligence at‌ the nanoscale.Nanophot.832019,‌ Open AccessURL: https://doi.org/10.1515/nanoph-2018-0183back to text

ATLANTIS - 2025

ATLANTIS - 2025

2025Activity report﻿‌​‌Project-TeamATLANTIS

Keywords​‌﻿﻿

Computer Science and Digital​​﻿﻿ Science

Other Research﻿​﻿﻿ Topics and Application Domains​‌﻿﻿

1 Team members,​​﻿﻿ visitors, external collaborators

Research​​​‌ Scientists

Faculty​‌﻿﻿ Members

Post-Doctoral Fellows

PhD Students

Technical​​﻿﻿ Staff

Interns and Apprentices﻿﻿﻿‌

Administrative​​​‌ Assistant

2 Overall﻿‌​‌ objectives

3﻿​﻿﻿ Research program

3.1 Driving​‌﻿﻿ physical fields

3.2﻿​​﻿ Research agenda

3.2.1​​​‌ Core research topics

3.2.2 Complementary topics

4﻿﻿﻿‌ Application domains

4.1 Nanostructures for sunlight﻿﻿﻿‌ harvesting

4.2 Metasurfaces for light​​​‌ front shaping

4.3 Nanostructuring​​​‌ for THz wave generation﻿​﻿﻿

4.4﻿‌​‌ Plasmonic nanostructures for nanoscale﻿​​﻿ sensing

4.5 Plasmonic​​​‌ nanostructures for photothermal effects﻿​﻿﻿

4.6 Light﻿﻿﻿‌ management by disordered nanostructures﻿‌​‌

5 Social and environmental﻿﻿﻿‌ responsibility

5.1 Impact of﻿‌​‌ research results

6 Highlights of the​​﻿﻿ year

7​​​‌ Latest software developments, platforms,﻿​﻿﻿ open data

7.1 Latest​‌﻿﻿ software developments

7.1.1 DIOGENeS​​﻿﻿

7.1.2 POSEidON​​﻿﻿

7.1.3 Geom4PhotoPigments

8 New results﻿​﻿﻿

8.1 High order methods​‌﻿﻿ for complex problems

8.1.1 Time-domain numerical modeling﻿﻿﻿‌ of gain media

8.1.2﻿​​﻿ Time-domain numerical modeling of​​​‌ plasmonic-based nanoscale heating

8.1.3 Time-domain numerical﻿​﻿﻿ modeling of liquid crystal​‌﻿﻿ materials

8.1.4 Numerical modeling of﻿‌​‌ time-modulated metasurfaces

8.1.5 Efficient approximation of﻿​​﻿ high-frequency Helmholtz problems

8.2 Data-driven reduced-order﻿​﻿﻿ modeling

8.2.1 POD-based​​​‌ ROM methods for parameterized﻿﻿﻿‌ electromagnetic problems

8.2.2 Nonlinear ROM with﻿‌​‌ Graph Convolutional Autoencoder

8.3 Numerical​​﻿﻿ optimization and inverse design​​​‌ approaches

8.4 Deep﻿​﻿﻿ Learning methods

8.4.1 PINNs for the​​﻿﻿ parametric Maxwell equations

8.4.2 Multilevel and distributed﻿​​﻿ PINNs for the Helmholtz​​​‌ equation

8.4.3 Efficient deep learning﻿﻿﻿‌ methodology for large-scale metalens﻿‌​‌

8.5 Discovering​​​‌ novel nanoscale structures

8.5.1 Trimer metasurfaces for﻿​​﻿ highly sensitive biomedical sensors​​​‌

8.5.2 Metasurface for quantum﻿​​﻿ information processing

8.6 Software﻿​﻿﻿ developments in DIOGENeS

8.7​‌﻿﻿ Applications

8.7.1 Modeling of​​﻿﻿ centimeter-scale metasurfaces in imaging​​​‌ systems

8.7.2 Optimization﻿﻿﻿‌ of light trapping in﻿‌​‌ nanostructured solar cells

8.7.3 Plasmonic​​﻿﻿ sensing with nanocubes

8.7.4 Multiple﻿﻿﻿‌ scattering in random media﻿‌​‌

8.7.5 Nonlinear wavefront​​﻿﻿ shaping with optical metasurfaces​​​‌

9 Bilateral contracts﻿​﻿﻿ and grants with industry​‌﻿﻿

9.1 Bilateral contracts with​​﻿﻿ industry

10 Partnerships﻿​​﻿ and cooperations

10.1 International​​​‌ initiatives

10.1.1 Associate Teams﻿﻿﻿‌ in the framework of﻿‌​‌ an Inria International Lab﻿​​﻿ or in the framework​​​‌ of an Inria International﻿﻿﻿‌ Program

DNN4Photonics

Other international visits to​​﻿﻿ the team

Haogang Cai​​​‌

10.2 National​‌﻿﻿ initiatives

10.2.1 ANR projects​​﻿﻿

NumPEx (Digital for Exascale)​​​‌

SWEET (Sub-WavelEngth Electro-optic﻿​﻿﻿ sysTems)

SWAG-P​​​‌ (Sensing With A Gap-Plasmon)﻿﻿﻿‌

NO-RESTRAIN (NOnlinear metasuRfacE﻿‌​‌ on Silicon photodiode foR﻿​​﻿ infrAred detectIoN)

DNN4Photonics (Deep Neural​‌﻿﻿ Networks for the design​​﻿﻿ of photonic devices)

MAXINET (Physics informed neural﻿​﻿﻿ networks for harmonic Maxell​‌﻿﻿ equations)

10.2.2﻿​​﻿ DGA/AID RAPID project

AEROCOM​​​‌ (Ultra-flat and low-cost antennas)﻿﻿﻿‌

11﻿‌​‌ Dissemination

11.1 Promoting﻿‌​‌ scientific activities

2025Activity report‌‌Project-TeamATLANTIS

Keywords‌

Computer Science and Digital Science

Other Research Topics and Application Domains‌

1 Team members, visitors, external collaborators

Research‌ Scientists

Faculty‌ Members

Technical Staff

Interns and Apprentices‌

Administrative‌ Assistant

2 Overall‌‌ objectives

3 Research program

3.1 Driving‌ physical fields

3.2 Research agenda

3.2.1‌ Core research topics

4‌ Application domains

4.1 Nanostructures for sunlight‌ harvesting

4.2 Metasurfaces for light‌ front shaping

4.3 Nanostructuring‌ for THz wave generation

4.4‌‌ Plasmonic nanostructures for nanoscale sensing

4.5 Plasmonic‌ nanostructures for photothermal effects

4.6 Light‌ management by disordered nanostructures‌‌

5 Social and environmental‌ responsibility

5.1 Impact of‌‌ research results

6 Highlights of the year

7‌ Latest software developments, platforms, open data

7.1 Latest‌ software developments

7.1.1 DIOGENeS

7.1.2 POSEidON

8 New results

8.1 High order methods‌ for complex problems

8.1.1 Time-domain numerical modeling‌ of gain media

8.1.2 Time-domain numerical modeling of‌ plasmonic-based nanoscale heating

8.1.3 Time-domain numerical modeling of liquid crystal‌ materials

8.1.4 Numerical modeling of‌‌ time-modulated metasurfaces

8.1.5 Efficient approximation of high-frequency Helmholtz problems

8.2 Data-driven reduced-order modeling

8.2.1 POD-based‌ ROM methods for parameterized‌ electromagnetic problems

8.2.2 Nonlinear ROM with‌‌ Graph Convolutional Autoencoder

8.3 Numerical optimization and inverse design‌ approaches

8.4 Deep Learning methods

8.4.1 PINNs for the parametric Maxwell equations

8.4.2 Multilevel and distributed PINNs for the Helmholtz‌ equation

8.4.3 Efficient deep learning‌ methodology for large-scale metalens‌‌

8.5 Discovering‌ novel nanoscale structures

8.5.1 Trimer metasurfaces for highly sensitive biomedical sensors‌

8.5.2 Metasurface for quantum information processing

8.6 Software developments in DIOGENeS

8.7‌ Applications

8.7.1 Modeling of centimeter-scale metasurfaces in imaging‌ systems

8.7.2 Optimization‌ of light trapping in‌‌ nanostructured solar cells

8.7.3 Plasmonic sensing with nanocubes

8.7.4 Multiple‌ scattering in random media‌‌

8.7.5 Nonlinear wavefront shaping with optical metasurfaces‌

9 Bilateral contracts and grants with industry‌

9.1 Bilateral contracts with industry

10 Partnerships and cooperations

10.1 International‌ initiatives

10.1.1 Associate Teams‌ in the framework of‌‌ an Inria International Lab or in the framework‌ of an Inria International‌ Program

Other international visits to the team

Haogang Cai‌

10.2 National‌ initiatives

10.2.1 ANR projects

NumPEx (Digital for Exascale)‌

SWEET (Sub-WavelEngth Electro-optic sysTems)

SWAG-P‌ (Sensing With A Gap-Plasmon)‌

NO-RESTRAIN (NOnlinear metasuRfacE‌‌ on Silicon photodiode foR infrAred detectIoN)

DNN4Photonics (Deep Neural‌ Networks for the design of photonic devices)

MAXINET (Physics informed neural networks for harmonic Maxell‌ equations)

10.2.2 DGA/AID RAPID project

AEROCOM‌ (Ultra-flat and low-cost antennas)‌

11‌‌ Dissemination

11.1 Promoting‌‌ scientific activities

11.1.1 Scientific events: organization

Member of‌ the organizing committees

11.1.2 Invited talks‌

11.1.3 Leadership‌ within the scientific community

11.1.4 Research administration

11.2 Teaching -‌ Supervision - Juries - Educational and pedagogical outreach‌

12.1‌ Major publications