arXiv subjects

Adrien Merlini

Publications and source records attributed to Adrien Merlini.

At least 19 recordsLinked to original sources

arXiv2026

Volume-Surface-Wire Integral Equations for EEG Source Imaging

In EEG source imaging, the precision of the imaging of the brain activity depends on the accuracy of the forward head model which is, in turn, affected by the representation of tissue conductivity, including anisotropic compartments such as the skull and white matter. We introduce an EEG source imaging framework based on a hybrid volume-surface-wire integral formulation, enabling tract-aware modeling of anisotropic white matter contributions without a full volumetric head discretization. The proposed framework is assessed on a realistic MRI-derived head anatomy, showing close agreement with FEM reference solutions while avoiding full-volume meshing and providing a natural representation of white matter fiber tracts. We then integrate this model into a real-time pipeline that couples EEG processing with immersive visualization, enabling synchronized inspection of scalp potentials, cortical sources, and white matter fiber tracts-related quantities. The results support the proposed formulation as a tract-aware alternative to canonical formulations, combining robust localization performance with interactive visualization capabilities.

physics.comp-ph↗

arXiv2026

Spectral Filtering of 3D Integral Operators Using Modified Green's Functions

Several recent contributions have analyzed and illustrated the effectiveness of operator filtering, both in terms of regularization and compression, when handling dense matrices arising from the discretization of integral operators, e.g. the single-layer operator. Previous works have introduced different filtering strategies, ranging from Laplacian-based filters to analytically derived ones, with the goal of improving the computational efficiency of iterative and direct solvers for integral equations in the two-dimensional space, like the 2D Electric Field Integral Equation (EFIE). In this work, we propose a filtering strategy based on the spectral truncation of the kernels of integral operators associated with the 3D EFIE. The approach relies on an appropriate spectral representation of the Green's function obtained via the spherical Hankel transform, which provides an analytical foundation for the proposed approach. Finally, we provide semi-analytical and numerical evidence of the impact of this filtering technique on the spectral properties of continuous integral operators and of their discretization through boundary elements, both for the static and dynamic cases.

cs.CE↗

arXiv2026

A Numerical Approach to Operator Filtering within the Adaptive Integral Method for Electromagnetic Integral Equations

Operator filtering allows for the regularization and compression of dense integral operators, effectively mitigating the memory and computational costs associated with iterative solvers. Previous works introduced filters that leverage the analytical spectral truncation of kernels for operators of the 2D Electric Field Integral Equation (EFIE). In this contribution, we will demonstrate how to obtain filtered kernels in a discrete numerical form within the framework of an Adaptive Integral Method (AIM), yielding results entirely comparable to analytical filters. By operating directly on the discrete operator representations, the proposed strategy ensures a native and robust compatibility with fast solver schemes that analytical formulations often lack. The effectiveness of the proposed approach will be demonstrated through numerical results, including its application to the Calder\'on preconditioned EFIE.

cs.CE↗

arXiv2026

Majorization-Minimization Networks for Inverse Problems: An Application to EEG Imaging

Inverse problems are often ill-posed and require optimization schemes with strong stability and convergence guarantees. While learning-based approaches such as deep unrolling and meta-learning achieve strong empirical performance, they typically lack explicit control over descent and curvature, limiting robustness. We propose a learned Majorization-Minimization (MM) framework for inverse problems within a bilevel optimization setting. Instead of learning a full optimizer, we learn a structured curvature majorant that governs each MM step while preserving classical MM descent guarantees. The majorant is parameterized by a lightweight recurrent neural network and explicitly constrained to satisfy valid MM conditions. For cosine-similarity losses, we derive explicit curvature bounds yielding diagonal majorants. When analytic bounds are unavailable, we rely on efficient Hessian-vector product-based spectral estimation to automatically upper-bound local curvature without forming the Hessian explicitly. Experiments on EEG source imaging demonstrate improved accuracy, stability, and cross-dataset generalization over deep-unrolled and meta-learning baselines.

eess.SP↗

arXiv2025

A High-Order Discretization Scheme for Surface Integral Equations for Analyzing the Electroencephalography Forward Problem

A Nystrom-based high-order (HO) discretization scheme for surface integral equations (SIEs) for analyzing the electroencephalography (EEG) forward problem is proposed in this work. We use HO surface elements and interpolation functions for the discretization of the interfaces of the head volume and the unknowns on the elements, respectively. The advantage of this work over existing isoparametric HO discretization schemes resides in the fact that the interpolation points are different from the mesh nodes, allowing for the flexible manipulation of the order of the basis functions without regenerating the mesh of the interfaces. Moreover, the interpolation points are chosen from the quadrature rules with the same number of points on the elements simplifying the numerical computation of the surface integrals for the far-interaction case. In this contribution, we extend the implementation of the HO discretization scheme to the double-layer and the adjoint double-layer formulations, as well as to the isolated-skull-approach for the double-layer formulation and to the indirect adjoint double-layer formulation, employed to improve the solution accuracy in case of high conductivity contrast models, which requires the development of different techniques for the singularity treatment. Numerical experiments are presented to demonstrate the accuracy, flexibility, and efficiency of the proposed scheme for the four SIEs for analyzing the EEG forward problem.

math.NA↗

arXiv2025

Limitations of Nyquist Criteria in the Discretization of 2D Electromagnetic Integral Equations at High Frequency: Spectral Insights into Pollution Effects

The use of boundary integral equations in modeling boundary value problems-such as elastic, acoustic, or electromagnetic ones-is well established in the literature and widespread in practical applications. These equations are typically solved numerically using boundary element methods (BEMs), which generally provide accurate and reliable solutions. When the frequency of the wave phenomenon under study increases, the discretization of the problem is typically chosen to maintain a fixed number of unknowns per wavelength. Under these conditions, the BEM over finite-dimensional subspaces of piecewise polynomial basis functions is commonly believed to provide a bounded solution accuracy. If proven, this would constitute a significant advantage of the BEM with respect to finite element and finite difference time domain methods, which, in contrast, are affected by numerical pollution. In this work, we conduct a rigorous spectral analysis of some of the most commonly used boundary integral operators and examine the impact of the BEM discretization on the solution accuracy of widely used integral equations modeling two-dimensional electromagnetic scattering from a perfectly electrically conducting cylinder. We consider both ill-conditioned and well-conditioned equations, the latter being characterized by solution operators bounded independently of frequency. Our analysis, which is capable of tracking the effects of BEM discretization on compositions and sums of different operators, reveals a form of pollution that affects, in different measures, equations of both kinds. After elucidating the mechanism by which the BEM discretization impacts accuracy, we propose a solution strategy that can cure the pollution problem thus evidenced. The defining strength of the proposed theoretical model lies in its capacity to deliver deep insight into the root causes of the phenomenon.

cs.CE↗

arXiv2024

On the Computation of Square Roots and Inverse Square Roots of Gram Matrices for Surface Integral Equations in Electromagnetics

Surface integral equations (SIEs)-based boundary element methods are widely used for analyzing electromagnetic scattering scenarii. However, after discretization of SIEs, the spectrum and eigenvectors of the boundary element matrices are not usually representative of the spectrum and eigenfunctions of the underlying surface integral operators, which can be problematic for methods that rely heavily on spectral properties. To address this issue, we delineate some efficient algorithms that allow for the computation of matrix square roots and inverse square roots of the Gram matrices corresponding to the discretization scheme, which can be used for revealing the spectrum of standard electromagnetic integral operators. The algorithms, which are based on properly chosen expansions of the square root and inverse square root functions, are quite effective when applied to several of the most relevant Gram matrices used for boundary element discretizations in electromagnetics. Tables containing different sets of expansion coefficients are provided along with comparative numerical experiments that evidence advantages and disadvantages of the different approaches. In addition, to demonstrate the spectrum-revealing properties of the proposed techniques, they are applied to the discretization of the problem of scattering by a sphere for which the analytic spectrum is known.

math.NA↗

arXiv2023

Calder\'on Strategies for the Convolution Quadrature Time Domain Electric Field Integral Equation

In this work, we introduce new integral formulations based on the convolution quadrature method for the time-domain modeling of perfectly electrically conducting scatterers that overcome some of the most critical issues of the standard schemes based on the electric field integral equation (EFIE). The standard time-domain EFIE-based approaches typically yield matrices that become increasingly ill-conditioned as the time-step or the mesh discretization density increase and suffer from the well-known DC instability. This work presents solutions to these issues that are based both on new Calder\'on strategies and quasi-Helmholtz projectors regularizations. In addition, to ensure an efficient computation of the marching-on-in-time, the proposed schemes leverage properties of the Z-transform -- involved in the convolution quadrature discretization scheme -- when computing the stabilized operators. The two resulting formulations compare favorably with standard, well-established schemes. The properties and practical relevance of these new formulations will be showcased through relevant numerical examples that include canonical geometries and more complex structures.

math.NA↗

arXiv2023

Linear-in-Complexity Computational Strategies for Modeling and Dosimetry at TeraHertz

This work presents a fast direct solver strategy allowing full-wave modeling and dosimetry at terahertz (THz) frequencies. The novel scheme leverages a preconditioned combined field integral equation together with a regularizer for its elliptic spectrum to enable its compression into a non-hierarchical skeleton, invertible in quasi-linear complexity. Numerical results will show the effectiveness of the new scheme in a realistic skin modeling scenario.

math.NA↗

arXiv2023

High-order quasi-Helmholtz Projectors: Definition, Analyses, Algorithms

The accuracy of the electric field integral equation (EFIE) can be substantially improved using high-order discretizations. However, this equation suffers from ill-conditioning and deleterious numerical effects in the low-frequency regime, often jeopardizing its solution. This can be fixed using quasi-Helmholtz decompositions, in which the source and testing elements are separated into their solenoidal and non-solenoidal contributions, then rescaled in order to avoid both the low-frequency conditioning breakdown and the loss of numerical accuracy. However, standard quasi-Helmholtz decompositions require handling discretized differential operators that often worsen the mesh-refinement ill-conditioning and require the finding of the topological cycles of the geometry, which can be expensive when modeling complex scatterers, especially in high-order. This paper solves these drawbacks by presenting the first extension of the quasi-Helmholtz projectors to high-order discretizations and their application to the stabilization of the EFIE when discretized with high-order basis functions. Our strategy will not require the identification of the cycles and will provide constant condition numbers for decreasing frequencies. Theoretical considerations will be accompanied by numerical results showing the effectiveness of our method in complex scenarios.

math.NA↗

arXiv2023

On some Operator Filtering Strategies Based on Suitably Modified Green's Functions

Recent contributions showed the benefits of operator filtering for both preconditioning and fast solution strategies. While previous contributions leveraged laplacian-based filters, in this work we introduce and study a different approach leveraging the truncation of appropriately chosen spectral representations of operators' kernels. In this contribution, the technique is applied to the operators of the 2D TE- and TM-electric field integral equations (EFIE). We explore two different spectral representations for the 2D Green's function that lead to two distinct types of filtering of the EFIE operators. Numerical results corroborate the effectiveness of the newly proposed approaches, also in the Calder\'on preconditioned EFIE

math.NA↗

arXiv2022

Laplacian Filtered Loop-Star Decompositions and Quasi-Helmholtz Laplacian Filters: Definitions, Analysis, and Efficient Algorithms

Quasi-Helmholtz decompositions are fundamental tools in integral equation modeling of electromagnetic problems because of their ability of rescaling solenoidal and non-solenoidal components of solutions, operator matrices, and radiated fields. These tools are however incapable, per se, of modifying the refinement-dependent spectral behavior of the different operators and often need to be combined with other preconditioning strategies. This paper introduces the new concept of filtered quasi-Helmholtz decompositions proposing them in two incarnations: the filtered Loop-Star functions and the quasi-Helmholtz Laplacian filters. Because they are capable of manipulating large parts of the operators' spectra, new families of preconditioners and fast solvers can be derived from these new tools. A first application to the case of the frequency and h-refinement preconditioning of the electric field integral equation is presented together with numerical results showing the practical effectiveness of the newly proposed decompositions.

math.NA↗

arXiv2022

On a Calder\'on preconditioner for the symmetric formulation of the electroencephalography forward problem without barycentric refinements

We present a Calder\'on preconditioning scheme for the symmetric formulation of the forward electroencephalographic (EEG) problem that cures both the dense discretization and the high-contrast breakdown. Unlike existing Calder\'on schemes presented for the EEG problem, it is refinement-free, that is, the electrostatic integral operators are not discretized with basis functions defined on the barycentrically-refined dual mesh. In fact, in the preconditioner, we reuse the original system matrix thus reducing computational burden. Moreover, the proposed formulation gives rise to a symmetric, positive-definite system of linear equations, which allows the application of the conjugate gradient method, an iterative method that exhibits a smaller computational cost compared to other Krylov subspace methods applicable to non-symmetric problems. Numerical results corroborate the theoretical analysis and attest of the efficacy of the proposed preconditioning technique on both canonical and realistic scenarios.

math.NA↗

arXiv2022

Fast Direct Solvers for Integral Equations at Low-Frequency Based on Operator Filtering

This paper focuses on fast direct solvers for integral equations in the low-to-moderate-frequency regime obtained by leveraging preconditioned first kind or second kind operators regularized with Laplacian filters. The spectral errors arising from boundary element discretizations are properly handled by filtering that, in addition, allows for the use of low-rank representations for the compact perturbations of all operators involved. Numerical results show the effectiveness of the approaches and their effectiveness in the direct solution of integral equations.

math.NA↗

arXiv2022

Stabilized Single Current Inverse Source Formulations Based on Steklov-Poincar\'e Mappings

The inverse source problem in electromagnetics has proved quite relevant for a large class of applications. In antenna diagnostics in particular, Love solutions are often sought at the cost of an increase of the dimension of the linear system to be solved. In this work, instead, we present a reduced-in-size single current formulation of the inverse source problem that obtains one of the Love currents via a stable discretization of the Steklov-Poincar\'e boundary operator leveraging dual functions. The new approach is enriched by theoretical treatments and by a further low-frequency stabilization of the Steklov-Poincar\'e operator based on the quasi-Helmholtz projectors that is the first of its kind in this field. The effectiveness and practical relevance of the new schemes are demonstrated via both theoretical and numerical results.

math.NA↗

arXiv2022

On a Constrained Pseudoinverse for the Electromagnetic Inverse Source Problem

Inverse source approaches have shown their relevance for several applications in the past years. They rely on the solution of an ill-posed problem where near-field/current data is reconstructed starting from far-field (or less informative field) information. Standard strategies, including the physically constrained ones using Love conditions, result in linear systems to be pseudoinverted which are still ill-conditioned due to the lack of information from the evanescent fields. In this work we present a generalized pseudoinverse for those problems that allows the inclusion of extra constraints from the evanescent field space, when available. This is obtained by dropping some of the standard Moore-Penrose (MP) requirements and using the resulting degrees of freedom to obtain a generalized pseudoinverse that shows favorable performance in several cases of practical interest.

math.NA↗

arXiv2022

Brain-Computer Interfaces: Investigating the Transition from Visually Evoked to Purely Imagined Steady-State Potentials

Brain-Computer Interfaces (BCIs) based on Steady State Visually Evoked Potentials (SSVEPs) have proven effective and provide significant accuracy and information-transfer rates. This family of strategies, however, requires external devices that provide the frequency stimuli required by the technique. This limits the scenarios in which they can be applied, especially when compared to other BCI approaches. In this work, we have investigated the possibility of obtaining frequency responses in the EEG output based on the pure visual imagination of SSVEP-eliciting stimuli. Our results show that not only that EEG signals present frequency-specific peaks related to the frequency the user is focusing on, but also that promising classification accuracy can be achieved, paving the way for a robust and reliable visual imagery BCI modality.

eess.SP↗

arXiv2022

On a Fast Solution Strategy for a Surface-Wire Integral Formulation of the Anisotropic Forward Problem in Electroencephalography

This work focuses on a quasi-linear-in-complexity strategy for a hybrid surface-wire integral equation solver for the electroencephalography forward problem. The scheme exploits a block diagonally dominant structure of the wire self block -- that models the neuronal fibers self interactions -- and of the surface self block -- modeling interface potentials. This structure leads to two Neumann iteration schemes further accelerated with adaptive integral methods. The resulting algorithm is linear up to logarithmic factors. Numerical results confirm the performance of the method in biomedically relevant scenarios.

math.NA↗