arXiv subjects

Bernard Haasdonk

Publications and source records attributed to Bernard Haasdonk.

At least 19 recordsLinked to original sources

arXiv2026

On Symmetric Kernel Collocation for Nonlinear PDEs

This paper considers kernel-based approximation methods for nonlinear partial differential equations. To this end, the problem is formulated as an optimal-recovery generalized interpolation problem, that is, as an optimization problem in an RKHS with nonlinear functional constraints. This formulation provides the basis for a convergence analysis carried out directly in the RKHS and extends existing results by relaxing the uniqueness assumption on the PDE solution. In the nonunique case, the limiting object is characterized as a minimum-norm solution. Furthermore, a residual-greedy strategy for adaptive collocation point selection is proposed, and convergence of the resulting sequence of generalized interpolants is established. Numerical experiments for a stationary nonlinear heat equation illustrate the method and indicate that residual-greedy point selection can lead to markedly smaller PDE residuals than point sets selected according to fill-distance criteria.

math.NA↗

arXiv2026

Symplecticity-preserving prediction of parameter-dependent Hamiltonian dynamics by Generalized Kernel Interpolation

We extend the kernel-based symplectic predictor of [1] to a parameter-augmented setting in which the learned flow-map surrogate depends not only on the state, but also on additional variables such as physical parameters and macro time-step sizes. The method uses a product kernel ansatz on a parameter and macro step augmented domain and constructs the prediction through an implicit symplectic-Euler-type update. Hence, for every fixed admissible parameter and time-step instance, the resulting large-step predictor is symplectic by construction. The training problem is formulated as gradient Hermite--Birkhoff interpolation in a reproducing kernel Hilbert space. Efficient surrogates are obtained by greedy center selection. We show that the convergence analysis from the non-augmented setting carries over to the product-kernel framework and derive corresponding prediction error bounds. Numerical experiments for a pendulum with varying length and time-step size and for a parameter-dependent discretized wave equation illustrate the accuracy and structure-preserving behavior of the proposed approach.

math.NA↗

arXiv2026

Solving Approximation Tasks with Greedy Deep Kernel Methods

Kernel methods are versatile tools for function approximation and surrogate modeling. In particular, greedy techniques offer computational efficiency and reliability through inherent sparsity and provable convergence. Inspired by the success of deep neural networks and structured deep kernel networks, we consider deep, multilayer kernels for greedy approximation. This multilayer structure, consisting of linear kernel layers and optimizable kernel activation function layers in an alternating fashion, increases the expressiveness of the kernels and thus of the resulting approximants. Compared to standard kernels, deep kernels are able to adapt kernel intrinsic shape parameters automatically, incorporate transformations of the input space and induce a data-dependent reproducing kernel Hilbert space. For this, deep kernels need to be pretrained using a specifically tailored optimization objective. In this work, we not only introduce deep kernel greedy models, but also present numerical investigations and comparisons with neural networks, which clearly show the advantages in terms of approximation accuracies. As applications we consider the approximation of model problems, the prediction of breakthrough curves for reactive flow through porous media and the approximation of solutions for parameterized ordinary differential equation systems.

math.NA↗

arXiv2026

Refined rates of convergence for target-data dependent greedy generalized interpolation with Sobolev kernels

Greedy methods have recently been successfully applied to generalized kernel interpolation, or the recovery of a function from data stemming from the evaluation of linear functionals, including the approximation of solutions of linear PDEs by symmetric collocation. When applied to kernels generating Sobolev spaces as their native Hilbert spaces, some of these greedy methods can provide the same error guarantee of generalized interpolation on quasi-uniform points. More importantly, certain target-data-adaptive methods even give a dimension- and smoothness-independent improvement in the speed of convergence over quasi-uniform points, thus offering advantages for high-dimensional problems. These convergence rates however contain a spurious logarithmic term that limits this beneficial effect. The goal of this note is to remove this factor, and this is possible by using estimates on metric entropy numbers.

math.NA↗

arXiv2026

Symplecticity-Preserving Prediction of Hamiltonian Dynamics by Generalized Kernel Interpolation

In this work, a kernel-based surrogate for integrating Hamiltonian dynamics that is symplectic by construction and tailored to large prediction horizons is proposed. The method learns a scalar potential whose gradient enters a symplectic-Euler update, yielding a discrete flow map that exactly preserves the canonical symplectic structure. Training is formulated as a gradient Hermite--Birkhoff interpolation problem in a reproducing kernel Hilbert space, providing a systematic framework for existence, uniqueness, and error control. Algorithmically, the symplectic kernel predictor is combined with structure-preserving model order reduction, enabling efficient treatment of high-dimensional discretized PDEs. Numerical tests for a pendulum, a nonlinear spring--mass chain, and a semi-discrete wave equation show nearly algebraic greedy convergence and long-time trajectory errors reduce by two to three orders of magnitude compared to an implicit midpoint baseline at the same macro time step.

math.NA↗

arXiv2025

Escaping the native space of Sobolev kernels by interpolation

Classical convergence analysis for kernel interpolation typically assumes that the target function $f$ lies in the reproducing kernel Hilbert space $\mathcal{H}_k\!\left(Ω\right)$ induced by a kernel on a domain $Ω\subset\mathbb{R}^N$. For many applications, however, this assumption is overly restrictive. We develop a general framework for analyzing the convergence of kernel interpolation {beyond the native space}. Let $A(Ω)$ and $B(Ω)$ be Banach spaces with continuous embeddings $\mathcal{H}_k\!\left(Ω\right) \hookrightarrow A(Ω)\hookrightarrow B(Ω)$, assume point evaluation is continuous on $A(Ω)$, and that $\mathcal{H}_k\!\left(Ω\right)$ is dense in $A(Ω)$. For a nested sequence of node sets $(X_n)_{n\ge1}\subsetΩ$ with $\bigcup_n X_n$ dense, we characterize convergence of the kernel interpolants in the $B(Ω)$-norm for all target functions in $A(Ω)$ via the uniform boundedness of the interpolation operators $Π^{\,n}_{A,B}:A(Ω)\to B(Ω)$. This yields a necessary and sufficient condition under which kernel interpolation extends beyond $\mathcal{H}_k\!\left(Ω\right)$. Specializing to Sobolev kernels of order $τ>N/2$ on bounded Lipschitz domains, we show that every $f \in C(\overlineΩ)$ can be approximated in the $L^2(Ω)$-norm by interpolation using quasi-uniform nested centers. Moreover, for a subclass of Sobolev kernels (including integer-order Matérn kernels), we prove that the Lebesgue constant is uniformly bounded on $[a,b]\subset\mathbb{R}$ under quasi-uniform centers; within our framework this implies supremum norm convergence of the interpolants for every target functions $f \in C([a,b])$.

math.NA↗

arXiv2025

Recovery of the optimal control value function in reproducing kernel Hilbert spaces from verification conditions

Approximating the optimal value function $v^*$ for infinite-horizon, nonlinear, autonomous optimal control problems is both challenging and essential for synthesizing real-time optimal feedback. We develop an abstract optimal recovery framework in reproducing kernel Hilbert spaces (RKHS) for reconstructing unknown target functions from mixed equality and inequality functional constraints. Within this framework, the approximation of $v^*$ is cast as a collocation-type problem derived from verification conditions for optimality -- most prominently, the Hamilton-Jacobi-Bellman (HJB) equation -- that uniquely characterizes $v^*$. As the set of collocation points becomes dense in the ambient domain $Ω$, we establish convergence of the RKHS approximants to $v^*$: globally on $Ω$ in the RKHS norm when $v^*$ is analytic, and locally (in a neighborhood of the origin) in the RKHS norm when $v^*$ is bounded from above and below by quadratic functions. Furthermore, we show that a practical numerical realization of the abstract scheme reduces to the classical policy iteration algorithm. Numerical experiments support the effectiveness of the proposed approach.

math.OC↗

arXiv2025

Convergence Rates for Realizations of Gaussian Random Variables

This paper investigates the approximation of Gaussian random variables in Banach spaces, focusing on the high-probability bounds for the approximation of Gaussian random variables using finitely many observations. We derive non-asymptotic error bounds for the approximation of a Gaussian process $ X $ by its conditional expectation, given finitely many linear functionals. Specifically, we quantify the difference between the covariance of $ X $ and its finite-dimensional approximation, establishing a direct relationship between the quality of the covariance approximation and the convergence of the process in the Banach space norm. Our approach avoids the reliance on spectral methods or eigenfunction expansions commonly used in Hilbert space settings, and instead uses finite, linear observations. This makes our result particularly suitable for practical applications in nonparametric statistics, machine learning, and Bayesian inference.

math.ST↗

arXiv2025

On the Convergence of the Policy Iteration for Infinite-Horizon Nonlinear Optimal Control Problems

Policy iteration (PI) is a widely used algorithm for synthesizing optimal feedback control policies across many engineering and scientific applications. When PI is deployed on infinite-horizon, nonlinear, autonomous optimal-control problems, however, a number of significant theoretical challenges emerge - particularly when the computational state space is restricted to a bounded domain. In this paper, we investigate these challenges and show that the viability of PI in this setting hinges on the existence, uniqueness, and regularity of solutions to the Generalized Hamilton-Jacobi-Bellman (GHJB) equation solved at each iteration. To ensure a well-posed iterative scheme, the GHJB solution must possess sufficient smoothness, and the domain on which the GHJB equation is solved must remain forward-invariant under the closed-loop dynamics induced by the current policy. Although fundamental to the method's convergence, previous studies have largely overlooked these aspects. This paper closes that gap by introducing a constructive procedure that guarantees forward invariance of the computational domain throughout the entire PI sequence and by establishing sufficient conditions under which a suitably regular GHJB solution exists at every iteration. Numerical results are presented for a grid-based implementation of PI to support the theoretical findings.

math.OC↗

arXiv2025

Kernel-based Greedy Approximation of Parametric Elliptic Boundary Value Problems

We recently introduced a scale of kernel-based greedy schemes for approximating the solutions of elliptic boundary value problems. The procedure is based on a generalized interpolation framework in reproducing kernel Hilbert spaces and was coined PDE-$β$-greedy procedure, where the parameter $β\geq 0$ is used in a greedy selection criterion and steers the degree of function adaptivity. Algebraic convergence rates have been obtained for Sobolev-space kernels and solutions of finite smoothness. We now report a result of exponential convergence rates for the case of infinitely smooth kernels and solutions. We furthermore extend the approximation scheme to the case of parametric PDEs by the use of state-parameter product kernels. In the surrogate modelling context, the resulting approach can be interpreted as an a priori model reduction approach, as no solution snapshots need to be precomputed. Numerical results show the efficiency of the approximation procedure for problems which occur as challenges for other parametric MOR procedures: non-affine geometry parametrizations, moving sources or high-dimensional domains.

math.NA↗

arXiv2025

A trust-region framework for optimization using Hermite kernel surrogate models

In this work, we present a trust-region optimization framework that employs Hermite kernel surrogate models. The method targets optimization problems with computationally demanding objective functions, for which direct optimization is often impractical due to expensive function evaluations. To address these challenges, we leverage a trust-region strategy, where the objective function is approximated by an efficient surrogate model within a local neighborhood of the current iterate. In particular, we construct the surrogate using Hermite kernel interpolation and define the trust-region based on bounds for the interpolation error. As mesh-free techniques, kernel-based methods are naturally suited for medium- to high-dimensional problems. Furthermore, the Hermite formulation incorporates gradient information, enabling precise gradient estimates that are crucial for many optimization algorithms. We prove that the proposed algorithm converges to a stationary point, and we demonstrate its effectiveness through numerical experiments, which illustrate the convergence behavior as well as the efficiency gains compared to direct optimization.

math.NA↗

arXiv2025

Analysis of Structured Deep Kernel Networks

In this paper, we leverage a recent deep kernel representer theorem to connect kernel based learning and (deep) neural networks in order to understand their interplay. In particular, we show that the use of special types of kernels yields models reminiscent of neural networks that are founded in the same theoretical framework of classical kernel methods, while benefiting from the computational advantages of deep neural networks. Especially the introduced Structured Deep Kernel Networks (SDKNs) can be viewed as neural networks (NNs) with optimizable activation functions obeying a representer theorem. This link allows us to analyze also NNs within the framework of kernel networks. We prove analytic properties of the SDKNs which show their universal approximation properties in three different asymptotic regimes of unbounded number of centers, width and depth. Especially in the case of unbounded depth, more accurate constructions can be achieved using fewer layers compared to corresponding constructions for ReLU neural networks. This is made possible by leveraging properties of kernel approximation.

cs.LG↗

arXiv2025

Convergence Analysis of a Greedy Algorithm for Conditioning Gaussian Random Variables

In the context of Gaussian conditioning, greedy algorithms iteratively select the most informative measurements, given an observed Gaussian random variable. However, the convergence analysis for conditioning Gaussian random variables remains an open problem. We adress this by introducing an operator $M$ that allows us to transfer convergence rates of the observed Gaussian random variable approximation onto the conditional Gaussian random variable. Furthermore we apply greedy methods from approximation theory to obtain convergence rates. These greedy methods have already demonstrated optimal convergence rates within the setting of kernel based function approximation. In this paper, we establish an upper bound on the convergence rates concerning the norm of the approximation error of the conditional covariance operator.

math.ST↗

arXiv2025

Adaptive meshfree approximation for linear elliptic partial differential equations with PDE-greedy kernel methods

We consider meshless approximation for solutions of boundary value problems (BVPs) of elliptic Partial Differential Equations (PDEs) via symmetric kernel collocation. We discuss the importance of the choice of the collocation points, in particular by using greedy kernel methods. We introduce a scale of PDE-greedy selection criteria that generalizes existing techniques, such as the PDE-$P$-greedy and the PDE-$f$-greedy rules for collocation point selection. For these greedy selection criteria we provide bounds on the approximation error in terms of the number of greedily selected points and analyze the corresponding convergence rates. This is achieved by a novel analysis of Kolmogorov widths of special sets of BVP point-evaluation functionals. Especially, we prove that target-data dependent algorithms that make use of the right hand side functions of the BVP exhibit faster convergence rates than the target-data independent PDE-$P$-greedy. The convergence rate of the PDE-$f$-greedy possesses a dimension independent rate, which makes it amenable to mitigate the curse of dimensionality. The advantages of these greedy algorithms are highlighted by numerical examples.

math.NA↗

arXiv2024

Data-driven identification of latent port-Hamiltonian systems

Conventional physics-based modeling techniques involve high effort, e.g., time and expert knowledge, while data-driven methods often lack interpretability, structure, and sometimes reliability. To mitigate this, we present a data-driven system identification framework that derives models in the port-Hamiltonian (pH) formulation. This formulation is suitable for multi-physical systems while guaranteeing the useful system theoretical properties of passivity and stability. Our framework combines linear and nonlinear reduction with structured, physics-motivated system identification. In this process, high-dimensional state data obtained from possibly nonlinear systems serves as input for an autoencoder, which then performs two tasks: (i) nonlinearly transforming and (ii) reducing this data onto a low-dimensional latent space. In this space, a linear pH system, that satisfies the pH properties per construction, is parameterized by the weights of a neural network. The mathematical requirements are met by defining the pH matrices through Cholesky factorizations. The neural networks that define the coordinate transformation and the pH system are identified in a joint optimization process to match the dynamics observed in the data while defining a linear pH system in the latent space. The learned, low-dimensional pH system can describe even nonlinear systems and is rapidly computable due to its small size. The method is exemplified by a parametric mass-spring-damper and a nonlinear pendulum example, as well as the high-dimensional model of a disc brake with linear thermoelastic behavior.

math.DS↗

arXiv2024

Greedy Kernel Methods for Approximating Breakthrough Curves for Reactive Flow from 3D Porous Geometry Data

We address the challenging application of 3D pore scale reactive flow under varying geometry parameters. The task is to predict time-dependent integral quantities, i.e., breakthrough curves, from the given geometries. As the 3D reactive flow simulation is highly complex and computationally expensive, we are interested in data-based surrogates that can give a rapid prediction of the target quantities of interest. This setting is an example of an application with scarce data, i.e., only having available few data samples, while the input and output dimensions are high. In this scarce data setting, standard machine learning methods are likely to ail. Therefore, we resort to greedy kernel approximation schemes that have shown to be efficient meshless approximation techniques for multivariate functions. We demonstrate that such methods can efficiently be used in the high-dimensional input/output case under scarce data. Especially, we show that the vectorial kernel orthogonal greedy approximation (VKOGA) procedure with a data-adapted two-layer kernel yields excellent predictors for learning from 3D geometry voxel data via both morphological descriptors or principal component analysis.

math.NA↗

arXiv2024

Error Analysis of Randomized Symplectic Model Order Reduction for Hamiltonian systems

Solving high-dimensional dynamical systems in multi-query or real-time applications requires efficient surrogate modelling techniques, as e.g., achieved via model order reduction (MOR). If these systems are Hamiltonian systems their physical structure should be preserved during the reduction, which can be ensured by applying symplectic basis generation techniques such as the complex SVD (cSVD). Recently, randomized symplectic methods such as the randomized complex singular value decomposition (rcSVD) have been developed for a more efficient computation of symplectic bases that preserve the Hamiltonian structure during MOR. In the current paper, we present two error bounds for the rcSVD basis depending on the choice of hyperparameters and show that with a proper choice of hyperparameters, the projection error of rcSVD is at most a constant factor worse than the projection error of cSVD. We provide numerical experiments that demonstrate the efficiency of randomized symplectic basis generation and compare the bounds numerically.

math.NA↗

arXiv2024

Model reduction on manifolds: A differential geometric framework

Using nonlinear projections and preserving structure in model order reduction (MOR) are currently active research fields. In this paper, we provide a novel differential geometric framework for model reduction on smooth manifolds, which emphasizes the geometric nature of the objects involved. The crucial ingredient is the construction of an embedding for the low-dimensional submanifold and a compatible reduction map, for which we discuss several options. Our general framework allows capturing and generalizing several existing MOR techniques, such as structure preservation for Lagrangian- or Hamiltonian dynamics, and using nonlinear projections that are, for instance, relevant in transport-dominated problems. The joint abstraction can be used to derive shared theoretical properties for different methods, such as an exact reproduction result. To connect our framework to existing work in the field, we demonstrate that various techniques for data-driven construction of nonlinear projections can be included in our framework.

math.NA↗