CyberSec Research

This paper presents a generalized energy-based modeling framework extending recent formulations tailored for differential-algebraic equations. The proposed structure, inspired by the port-Hamiltonian formalism, ensures passivity, preserves the power balance, and facilitates the consistent interconnection of subsystems. A particular focus is put on low-frequency power applications in electrical engineering. Stranded, solid, and foil conductor models are investigated in the context of the eddy current problem. Each conductor model is shown to fit into the generalized energy-based structure, which allows their structure-preserving coupling with electrical circuits described by modified nodal analysis. Theoretical developments are validated through a numerical simulation of an oscillator circuit, demonstrating energy conservation in lossless scenarios and controlled dissipation when eddy currents are present.

Magnetic Particle Imaging (MPI) is a tomographic imaging modality capable of real-time, high-sensitivity mapping of superparamagnetic iron oxide nanoparticles. Model-based image reconstruction provides an alternative to conventional methods that rely on a measured system matrix, eliminating the need for laborious calibration measurements. Nevertheless, model-based approaches must account for the complexities of the imaging chain to maintain high image quality. A recently proposed direct reconstruction method leverages weighted Chebyshev polynomials in the frequency domain, removing the need for a simulated system matrix. However, the underlying model neglects key physical effects, such as nanoparticle anisotropy, leading to distortions in reconstructed images. To mitigate these artifacts, an adapted direct Chebyshev reconstruction (DCR) method incorporates a spatially variant deconvolution step, significantly improving reconstruction accuracy at the cost of increased computational demands. In this work, we evaluate the adapted DCR on six experimental phantoms, demonstrating enhanced reconstruction quality in real measurements and achieving image fidelity comparable to or exceeding that of simulated system matrix reconstruction. Furthermore, we introduce an efficient approximation for the spatially variable deconvolution, reducing both runtime and memory consumption while maintaining accuracy. This method achieves computational complexity of O(N log N ), making it particularly beneficial for high-resolution and three-dimensional imaging. Our results highlight the potential of the adapted DCR approach for improving model-based MPI reconstruction in practical applications.

Physics-Informed Neural Networks (PINNs) have emerged as a powerful framework for solving partial differential equations (PDEs). However, their performance heavily relies on the strategy used to select training points. Conventional adaptive sampling methods, such as residual-based refinement, often require multi-round sampling and repeated retraining of PINNs, leading to computational inefficiency due to redundant points and costly gradient computations-particularly in high-dimensional or high-order derivative scenarios. To address these limitations, we propose RL-PINNs, a reinforcement learning(RL)-driven adaptive sampling framework that enables efficient training with only a single round of sampling. Our approach formulates adaptive sampling as a Markov decision process, where an RL agent dynamically selects optimal training points by maximizing a long-term utility metric. Critically, we replace gradient-dependent residual metrics with a computationally efficient function variation as the reward signal, eliminating the overhead of derivative calculations. Furthermore, we employ a delayed reward mechanism to prioritize long-term training stability over short-term gains. Extensive experiments across diverse PDE benchmarks, including low-regular, nonlinear, high-dimensional, and high-order problems, demonstrate that RL-PINNs significantly outperforms existing residual-driven adaptive methods in accuracy. Notably, RL-PINNs achieve this with negligible sampling overhead, making them scalable to high-dimensional and high-order problems.

This paper is concerned with non-uniform fully-mixed FEMs for dynamic coupled Stokes-Darcy model with the well-known Beavers-Joseph-Saffman (BJS) interface condition. In particular, a decoupled algorithm with the lowest-order mixed non-uniform FE approximations (MINI for the Stokes equation and RT0-DG0 for the Darcy equation) and the classical Nitsche-type penalty is studied. The method with the combined approximation of different orders is commonly used in practical simulations. However, the optimal error analysis of methods with non-uniform approximations for the coupled Stokes-Darcy flow model has remained challenging, although the analysis for uniform approximations has been well done. The key question is how the lower-order approximation to the Darcy flow influences the accuracy of the Stokes solution through the interface condition. In this paper, we prove that the decoupled algorithm provides a truly optimal convergence rate in L^2-norm in spatial direction: O(h^2) for Stokes velocity and O(h) for Darcy flow in the coupled Stokes-Darcy model. This implies that the lower-order approximation to the Darcy flow does not pollute the accuracy of numerical velocity for Stokes flow. The analysis presented in this paper is based on a well-designed Stokes-Darcy Ritz projection and given for a dynamic coupled model. The optimal error estimate holds for more general combined approximations and more general coupled models, including the corresponding model of steady-state Stokes-Darcy flows and the model of coupled dynamic Stokes and steady-state Darcy flows. Numerical results confirm our theoretical analysis and show that the decoupled algorithm is efficient.

Approximating a manifold-valued function from samples of input-output pairs consists of modeling the relationship between an input from a vector space and an output on a Riemannian manifold. We propose a function approximation method that leverages and unifies two prior techniques: (i) approximating a pullback to the tangent space, and (ii) the Riemannian moving least squares method. The core idea of the new scheme is to combine pullbacks to multiple tangent spaces with a weighted Fr\'echet mean. The effectiveness of this approach is illustrated with numerical experiments on model problems from parametric model order reduction.

We propose a characterization of a $p$-Laplace higher eigenvalue based on the inverse iteration method with balancing the Rayleigh quotients of the positive and negative parts of solutions to consecutive $p$-Poisson equations. The approach relies on the second eigenvalue's minimax properties, but the actual limiting eigenvalue depends on the choice of initial function. The well-posedness and convergence of the iterative scheme are proved. Moreover, we provide the corresponding numerical computations. As auxiliary results, which also have an independent interest, we provide several properties of certain $p$-Poisson problems.

We construct an efficient primal-dual forward-backward (PDFB) splitting method for computing a class of minimizing movement schemes with nonlinear mobility transport distances, and apply it to computing Wasserstein-like gradient flows. This approach introduces a novel saddle point formulation for the minimizing movement schemes, leveraging a support function form from the Benamou-Brenier dynamical formulation of optimal transport. The resulting framework allows for flexible computation of Wasserstein-like gradient flows by solving the corresponding saddle point problem at the fully discrete level, and can be easily extended to handle general nonlinear mobilities. We also provide a detailed convergence analysis of the PDFB splitting method, along with practical remarks on its implementation and application. The effectiveness of the method is demonstrated through several challenging numerical examples.

In this paper, we study the numerical simulation of stochastic differential equations (SDEs) on the special orthogonal Lie group $\text{SO}(n)$. We propose a geometry-preserving numerical scheme based on the stochastic tangent space parametrization (S-TaSP) method for state-dependent multiplicative SDEs on $\text{SO}(n)$. The convergence analysis of the S-TaSP scheme establishes a strong convergence order of $\mathcal{O}(\delta^{\frac{1-\epsilon}{2}})$, which matches the convergence order of the previous stochastic Lie Euler-Maruyama scheme while avoiding the computational cost of the exponential map. Numerical simulation illustrates the theoretical results.

Stochastic differential equations (SDEs) on Riemannian manifolds have numerous applications in system identification and control. However, geometry-preserving numerical methods for simulating Riemannian SDEs remain relatively underdeveloped. In this paper, we propose the Exponential Euler-Maruyama (Exp-EM) scheme for approximating solutions of SDEs on Riemannian manifolds. The Exp-EM scheme is both geometry-preserving and computationally tractable. We establish a strong convergence rate of $\mathcal{O}(\delta^{\frac{1 - \epsilon}{2}})$ for the Exp-EM scheme, which extends previous results obtained for specific manifolds to a more general setting. Numerical simulations are provided to illustrate our theoretical findings.

The existence of explicit symplectic integrators for general nonseparable Hamiltonian systems is an open and important problem in both numerical analysis and computing in science and engineering, as explicit integrators are usually more efficient than the implicit integrators of the same order of accuracy. Up to now, all responses to this problem are negative. That is, there exist explicit symplectic integrators only for some special nonseparable Hamiltonian systems, whereas the universal design involving explicit symplectic integrators for general nonseparable Hamiltonian systems has not yet been studied sufficiently. In this paper, we present a constructive proof for the existence of explicit symplectic integrators for general nonseparable Hamiltonian systems via finding explicit symplectic mappings under which the special submanifold of the extended phase space is invariant. It turns out that the proposed explicit integrators are symplectic in both the extended phase space and the original phase space. Moreover, on the basis of the global modified Hamiltonians of the proposed integrators, the backward error analysis is made via a parameter relaxation and restriction technique to show the linear growth of global errors and the near-preservation of first integrals. In particular, the effective estimated time interval is nearly the same as classical implicit symplectic integrators when applied to (near-) integrable Hamiltonian systems. Numerical experiments with a completely integrable nonseparable Hamiltonian and a nonintegrable nonseparable Hamiltonian illustrate the good long-term behavior and high efficiency of the explicit symplectic integrators proposed and analyzed in this paper.

We look for traveling waves of the semi-discrete conservation law $4\dot u_j +u_{j+1}^2-u_{j-1}^2 = 0$, using variational principles related to concepts of ``hidden convexity'' appearing in recent studies of various PDE (partial differential equations). We analyze and numerically compute with two variational formulations related to dual convex optimization problems constrained by either the differential-difference equation (DDE) or nonlinear integral equation (NIE) that wave profiles should satisfy. We prove existence theorems conditional on the existence of extrema that satisfy a strict convexity criterion, and numerically exhibit a variety of localized, periodic and non-periodic wave phenomena.

Phase-transition models are an important family of non-equilibrium continuum traffic flow models, offering properties like replicating complex traffic phenomena, maintaining anisotropy, and promising potentials for accommodating automated vehicles. However, their complex mathematical characteristics such as discontinuous solution domains, pose numerical challenges and limit their exploration in traffic flow theory. This paper focuses on developing a robust and accurate numerical method for phase-transition traffic flow models: We propose a second-order semi-discrete central-upwind scheme specifically designed for discontinuous phase-transition models. This novel scheme incorporates the projection onto appropriate flow domains, ensuring enhanced handling of discontinuities and maintaining physical consistency and accuracy. We demonstrate the efficacy of the proposed scheme through extensive and challenging numerical tests, showcasing their potential to facilitate further research and application in phase-transition traffic flow modeling. The ability of phase-transition models to embed the ``time-gap'' -- a crucial element in automated traffic control -- as a conserved variable aligns seamlessly with the control logic of automated vehicles, presenting significant potential for future applications, and the proposed numerical scheme now substantially facilitates exploring such potentials.

In this note we discuss the numerical solution of the eddy current approximation of the Maxwell equations using the simple Pragmatic Algebraic Model to include hysteresis effects. In addition to the more standard time-stepping approach we propose a space-time finite element method which allows both for parallelization and adaptivity simultaneously in space and time. Numerical experiments confirm both approaches yield the same numerical results.

Many-body interactions arise naturally in the perturbative treatment of classical and quantum many-body systems and play a crucial role in the description of condensed matter systems. In the case of three-body interactions, the Axilrod-Teller-Muto (ATM) potential is highly relevant for the quantitative prediction of material properties. The computation of the resulting energies in d-dimensional lattice systems is challenging, as a high-dimensional lattice sum needs to be evaluated to high precision. This work solves this long-standing issue. We present an efficiently computable representation of many-body lattice sums in terms of singular integrals over products of Epstein zeta functions. For three-body interactions in 3D, this approach reduces the runtime for computing the ATM lattice sum from weeks to minutes. Our approach further extends to a broad class of n-body lattice sums. We demonstrate that the computational cost of our method only increases linearly with n, evading the exponential increase in complexity of direct summation. The evaluation of 51-body interactions on a two-dimensional lattice, corresponding to a 100-dimensional sum, can be performed within seconds on a laptop. We discuss techniques for computing the arising singular integrals and compare the accuracy of our results against computable benchmarks, achieving full precision for exponents greater than the system dimension. Finally, we apply our method to study the stability of a three-dimensional lattice system with Lennard-Jones two-body interactions under the inclusion of an ATM three-body term at finite pressure, finding a transition from the face-centered-cubic to the body-centered-cubic lattice structure with increasing ATM coupling strength. This work establishes the mathematical foundation for an ongoing investigation into the influence of many-body interactions on the stability of matter.

We derive strong Lp convergence rates for the Euler-Maruyama schemes of Levy-driven SDE using a new dynamic cutting (DC) method with a time-dependent jump threshold. In addition, we present results from numerical simulations comparing the DC and Asmussen-Rosinski (AR) approaches. These simulations demonstrate the superior accuracy achieved by the DC method.

In this article, we consider issues surrounding integration when using Neural Networks to solve Partial Differential Equations. We focus on the Deep Ritz Method as it is of practical interest and sensitive to integration errors. We show how both deterministic integration rules as well as biased, stochastic quadrature can lead to erroneous results, whilst high order, unbiased stochastic quadrature rules on integration meshes can significantly improve convergence at an equivalent computational cost. Furthermore, we propose novel stochastic quadrature rules for triangular and tetrahedral elements, offering greater flexibility when designing integration meshes in more complex geometries. We highlight how the variance in the stochastic gradient limits convergence, whilst quadrature rules designed to give similar errors when integrating the loss function may lead to disparate results when employed in a gradient-based optimiser.

We propose a variance reduction method for calculating transport coefficients in molecular dynamics using an importance sampling method via Girsanov's theorem applied to Green--Kubo's formula. We optimize the magnitude of the perturbation applied to the reference dynamics by means of a scalar parameter~$\alpha$ and propose an asymptotic analysis to fully characterize the long-time behavior in order to evaluate the possible variance reduction. Theoretical results corroborated by numerical results show that this method allows for some reduction in variance, although rather modest in most situations.

This paper presents a convergence analysis of evolving surface finite element methods (ESFEM) applied to the original Eyles-King-Styles model of tumour growth. The model consists of a Poisson equation in the bulk, a forced mean curvature flow on the surface, and a coupled velocity law between bulk and surface. Due to the non-trivial bulk-surface coupling, all previous analyses -- which exclusively relied on energy-estimate based approaches -- required an additional regularization term. By adopting the $\widehat{H}^{3/2}$ theory and the multilinear forms, we develop an essentially new theoretical framework that enables the application of PDE regularity theory to stability analysis. Based on this framework, we provide the first rigorous convergence proof for the original model without regularization.