A B-spline-Heaviside collocation method for solving Fredholm integral equations with piecewise Hölder

Maria Capcelea, Titu Capcelea · 2025

Read A B-spline-Heaviside collocation method for solving Fredholm integral equations with piecewise Hölder by Maria Capcelea, Titu Capcelea (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-06590.txt.

Introduction

Maria Capcelea and Titu Capcelea solve Fredholm integral equations whose data jump at known points along a smooth closed curve. Because the smooth integral part stays continuous, the solution keeps exactly the jumps of the data. Ordinary methods smear those jumps or invent new ones, wrecking accuracy near the breaks.

Their method pairs one global periodic spline with Heaviside step functions carrying precisely the prescribed jumps. The scheme turns triangular: jump sizes come first, then a modest collocation system yields the smooth remainder. Stability and convergence rates are proven for spline orders two through four, quadrature effects are bounded, and numerical tests match the predicted accuracy against established rivals built for jumps.

Catalog record

CREATORMaria Capcelea, Titu Capcelea
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE46 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-06590

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A B-spline-Heaviside collocation method for solving Fredholm integral equations with piecewise Hölder-continuous right-hand sides Thanks: Submitted to the editors DATE. Maria Capcelea, Titu Capcelea Abstract This work presents a collocation method for solving linear Fredholm integral equations of the second kind defined on a closed contour in the complex plane. The right-hand side of the equation is a piecewise continuous function that may have a finite number of jump discontinuities and is known numerically at discrete points on the contour. The proposed approach employs a combination of B-spline functions and Heaviside step functions to ensure accurate approximation near discontinuity points and smooth behavior elsewhere on the contour. Convergence in the norm of piecewise Hölder spaces is established, together with explicit error estimates. Numerical results illustrate the effectiveness and convergence rate of the method. keywords collocation, Fredholm integral equation, closed contour, b-spline, piecewise Hölder spaces, convergence rate Funding. This work was funded by the † † runningheads: B-spline-Heaviside collocation method for integral equations / M. Capcelea and T. Capcelea MSC 45B05, 65R20, 65D15 1 Introduction and problem formulation Fredholm integral equations of the second kind arise in many practical applications, including boundary integral formulations in potential theory, elasticity, electromagnetism, inverse problems, and transport phenomena in heterogeneous media. Such equations frequently involve kernels or right-hand sides that are piecewise continuous, particularly when the underlying physical model exhibits discontinuities in boundary data, coefficients, or source terms. Fredholm integral equations with piecewise continuous right-hand sides naturally occur in several important contexts: • Boundary value problems in potential theory and elasticity, where boundary data or loading conditions exhibit jumps - for instance, at interfaces between materials with different physical properties or along crack edges. These situations lead to integral equations whose right-hand sides are piecewise continuous functions [ 8 ] . • Inverse problems or parameter identification in heterogeneous media, where the piecewise-defined coefficients or source terms induce discontinuities in the data of the corresponding integral equations [ 14 ] . • Transport or radiative‐transfer models in composite or layered materials, which often reduce to Fredholm equations with piecewise-defined source functions or solutions that are piecewise continuous across interface boundaries [ 12 ] . These examples motivate the development of numerical methods capable of handling low-regularity data. Let \Gamma be a closed, piecewise smooth contour that forms the boundary of the simply connected domain \Omega^{+}\subset{\mathbb{C}} , and let the point z=0\in\Omega^{+} . Consider the Riemann function z=\psi(w) , which provides a conformal map from the exterior D^{-} of the unit circle \Gamma_{0}:=\{\,w\in{\mathbb{C}}:|w|=1\,\} onto the exterior of the contour \Gamma , denoted by \Omega^{-} , such that \psi(\infty)=\infty and \psi^{\prime}(\infty)>0 . The function \psi(w) maps the circle \Gamma_{0} onto the contour \Gamma . Consequently, the points on the contour \Gamma can be described using the function \psi(w) . Fix a finite set of points (the jump set) {\mathcal{D}}=\left\{t^{d}_{1},...,t^{d}_{n_{d}}\right\}\subset\Gamma . Let 0<\alpha\leq 1 . Denote by PH_{\alpha}\left(\Gamma,{\mathcal{D}}\right) the Banach space of all functions on \Gamma that satisfy a Hölder condition with exponent \alpha , except possibly at the points of {\mathcal{D}} , where the functions are left-continuous and may have jump discontinuities of the first kind. Let \mathcal{A}(\Gamma\setminus\mathcal{D}) denote the finite family of open, connected arcs of \Gamma\setminus\mathcal{D} between successive points of \mathcal{D} . On each connected smooth arc I\in\mathcal{A}(\Gamma\setminus\mathcal{D}) without discontinuities, we have \;f|_{I}\in H_{\alpha}(I)\; , and the usual Hölder norm is defined by \lVert f\rVert_{H_{\alpha}(I)}:=\lVert f\rVert_{\infty,I}+[f]_{\alpha,I}, where \lVert f\rVert_{\infty,I}=\sup_{t\in I}|f(t)|\qquad\text{and}\qquad[f]_{\alpha,I}:=\sup_{\begin{subarray}{c}t,s\in I\\ t\neq s\end{subarray}}\frac{|f(t)-f(s)|}{|t-s|^{\alpha}} is the Hölder seminorm on I . A standard norm on the piecewise Hölder space PH_{\alpha}(\Gamma,\mathcal{D}) is \lVert f\rVert_{PH_{\alpha}(\Gamma,\mathcal{D})}:=\max_{I\in\mathcal{A}(\Gamma\setminus\mathcal{D})}\lVert f\rVert_{H_{\alpha}(I)}. Let us consider the linear Fredholm integral equation of the second kind: \varphi\left(t\right)-\lambda\int_{\Gamma}{k\left(t,\tau\right)\varphi\left(\tau\right)d\tau}=f\left(t\right),t\in\Gamma, (1) which is defined on the contour \Gamma described above. The kernel function k\left(t,\tau\right) is continuous in both variables, i.e., k\in C\left(\Gamma\times\Gamma\right) . The right-hand side function satisfies f\in PH_{\alpha}\left(\Gamma,{\mathcal{D}}\right) , and \lambda\in{\mathbb{C}} is a constant. Equation ( 1 ) can be rewritten in operator form as A\varphi\equiv\left(I-\lambda K\right)\varphi=f , where K\varphi\left(t\right)=\int_{\Gamma}{k\left(t,\tau\right)\varphi\left(\tau\right)d\tau} . The operator K:\ C\left(\Gamma\right)\to C\left(\Gamma\right) is compact; the same conclusion holds on PH_{\alpha}\left(\Gamma,{\mathcal{D}}\right) , since the compactness argument applies separately on each arc and then extends to the piecewise space. If the operator norm satisfies {\left\|\lambda K\right\|}_{PH_{\alpha}\left(\Gamma,{\mathcal{D}}\right)}<1 (Neumann criterion) or if the homogeneous equation A\varphi=0 admits only the trivial solution \varphi\left(t\right)\equiv 0 (Fredholm alternative), then the operator A=I-\lambda K , which defines the left-hand side of equation ( 1 ), is invertible in PH_{\alpha}\left(\Gamma,{\mathcal{D}}\right) and equation ( 1 ) has a unique solution {\rm\ }\varphi\in PH_{\alpha}\left(\Gamma,{\mathcal{D}}\right) . Considering that the right-hand side f is known numerically at a set of distinct points \left\{t_{j}\right\} on the contour \Gamma , we aim to develop an efficient method for computing a sequence of approximations \varphi_{n} to the solution \varphi , which converge to \varphi as n\to\infty in the norm of PH_{\alpha}\left(\Gamma,{\mathcal{D}}\right) . When the kernel and the right-hand side are smooth, classical numerical methods such as Nyström, polynomial or spline collocation, and Galerkin methods achieve high convergence rates [ 1 ] , [ 9 ] . However, in many problems, the solution \varphi and the right-hand side may be only continuous or Hölder continuous, possibly with finite jump discontinuities along the integration contour. In this low-regularity setting, standard global approximations lose accuracy near discontinuities (may suffer Gibbs-type oscillations), and convergence rates are reduced. For example, for a solution \varphi\in H_{\alpha} with Hölder exponent 0<\alpha\leq 1 , spline- or polynomial-based collocation methods of nominal order m achieve an error of order \mathcal{O}\left(h^{\alpha}\right) ( h is the mesh size) in uniform discretizations, rather than the full \mathcal{O}\left(h^{m}\right) [ 1 ] , reflecting the limited smoothness of the solution. Global or piecewise polynomial approximation is generally ineffective for approximating piecewise continuous functions, except when studying convergence in the norm of Lebesgue spaces L_{p},1<p<\infty . In such cases, it has been shown that the sequence of interpolation polynomials converges to the solution \varphi , except at a countable set of points [ 13 ] , [ 5 ] . To address the challenges defined by discontinuities, several strategies have been proposed: 1. Piecewise approximation spaces: The integration contour is partitioned into segments where the solution and coefficients are continuous or Hölder continuous, and local approximants (e.g., splines) are applied on each continuous segment. The convergence rate on each segment is \mathcal{O}\left(h^{\alpha}\right) in Hölder norms when the solution is only Hölder continuous. Discontinuities are sometimes explicitly represented using step functions [ 11 ] . 2. Enriched bases: The approximation space is augmented with basis functions that reflect the structure of the discontinuities (for example Heaviside step functions or singular basis functions). This allows one to recover higher local convergence rates in the presence of jumps. For example, recent work on collocation with Heaviside-type enrichment show convergence in piecewise Hölder norms [ 4 ] . This approach preserves the local convergence rate \mathcal{O}\left(h^{\alpha}\right) on smooth segments and avoids loss of accuracy near jumps. 3. Corrected quadrature and transformations: The integral equation or quadrature rule is transformed so that the discontinuity or singularity is either removed or ”smoothed” (for example by a one-to-one mapping or singularity subtraction), and then a standard discretisation such as the Nyström method is applied [ 7 ] . This enhances stability and convergence even when the right-hand side or kernel is

Read the complete text