An adaptive multimesh rational approximation scheme for the spectral fractional Laplacian
Alex Bespalov, Raphaël Bulle · 2025
Read An adaptive multimesh rational approximation scheme for the spectral fractional Laplacian by Alex Bespalov, Raphaël Bulle (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-2504-03408.txt.
Introduction
Alex Bespalov and Raphael Bulle design an adaptive solver for a fractional diffusion operator in a 2025 study. The spectral fractional Laplacian models processes with long-range effects, but it has no local formula, so computers approximate it through a rational function trick. Solve many ordinary diffusion problems with different parameters and join the answers.
Current schemes solve every one of those problems on the same mesh. Their multi-mesh version tailors a mesh to each subproblem, refining only where that piece needs it. Tailoring cuts wasted work, which matters because fractional problems already cost far more than classic ones.
Catalog record
| CREATOR | Alex Bespalov, Raphaël Bulle |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 73 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2504-03408
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
An adaptive multimesh rational approximation scheme for the spectral fractional Laplacian Alex Bespalov, Raphaël Bulle Abstract We propose a novel multimesh rational approximation scheme for the numerical solution of the (homogeneous) Dirichlet problem for the spectral fractional Laplacian. The scheme combines a rational approximation of the function λ ↦ λ − s \lambda\mapsto\lambda^{-s} with a family of finite element discretizations of parameter-dependent, non-fractional partial differential equations (PDEs). The key idea that underpins the proposed scheme is that each parametric PDE is numerically solved on an individually tailored finite element mesh. This is in contrast to existing single-mesh approaches that employ the same finite element mesh across all parametric PDEs. We develop an a posteriori error estimation strategy for the proposed rational approximation scheme and design an adaptive multimesh refinement algorithm. Numerical experiments demonstrate that our adaptive multimesh approach achieves faster convergence rates than uniform mesh refinement and yields significant reductions in computational costs—both in terms of the overall number of degrees of freedom and the actual runtime—when compared to the corresponding adaptive algorithm in a single-mesh setting. ℓ keywords finite element method, rational approximation, a posteriori error analysis, adaptive refinement methods, fractional partial differential equations, spectral fractional Laplacian † † runningheads: Adaptive multimesh rational approximation scheme / A. Bespalov, R. Bulle MSC 65N15, 65N30, 65N50, 35R11 1 Introduction Fractional partial differential equations (FPDEs) arise in a wide range of applications including anomalous diffusion, porous media, phase separation in fluids, discontinuous deformations (fractures), and spatial statistics (see [ 9 , 20 , 26 ] and the references therein). They have become a powerful mathematical tool in these application areas due to their ability to model nonlocal phenomena and processes that are characterized by interactions at a distance. The development of effective bespoke numerical methods and algorithms for FPDEs is also of significant interest, as the standard computational techniques applied to them generally incur much greater computational costs than if applied to classical (non-fractional) PDE-based models. In this work, we consider one particular fractional derivative operator, the fractional Laplacian (-\Delta)^{s} ( 0<s<1 ), that has received the most attention from both the mathematical analysis and numerical analysis communities. Among many non-equivalent definitions of the fractional Laplacian on a bounded domain \Omega , our focus here is on the spectral fractional Laplacian (we give a precise definition in section 2 ). For a given s\in(0,1) and f\in L^{2}(\Omega) , we consider the following boundary value problem for the spectral fractional Laplacian: find u:\Omega\to\mathbb{R} such that (-\Delta)^{s}u=f\quad\text{in }\Omega,\qquad u=0\quad\text{on }\partial\Omega. (1) This problem can be discretized directly using, e.g., the finite element method (FEM). However, due to the nonlocal nature of the fractional Laplacian, such a discretization would lead to a dense linear system, see, e.g., [ 26 ] . The resulting linear system can become computationally intractable if the number of degrees of freedom is very large, which is often the case when solving 3D problems and/or using a very fine computational mesh. In addition, for small values of s , the solution to ( 1 ) exhibits boundary layers, calling for approximations that employ adaptive and/or anisotropic mesh refinement, see, e.g., [ 18 , 6 ] . Several alternative discretization strategies for problems involving the fractional Laplacian (and applicable to more general nonlocal problems) are available in the literature. Among these strategies we mention the methods that employ the Dirichlet-to-Neumann maps to reformulate ( 1 ) in 2D as a three-dimensional non-fractional problem (see [ 27 , 5 ] ), reduced basis methods that construct a discrete space of small dimension where the solution u is sought (see [ 3 ] ), the semi-group method that transforms the elliptic problem ( 1 ) into a non-fractional parabolic problem (see [ 19 ] ), and the methods that employ rational approximations of the function \lambda\mapsto\lambda^{-s} in order to reformulate a fractional PDE as a set of independent non-fractional parameter-dependent PDEs (see [ 13 , 14 , 23 , 31 , 15 ] ). In particular, the strategies combining rational approximations of \lambda^{-s} and finite element discretizations of non-fractional parametric PDEs have become popular. In these strategies, called rational approximation schemes , instead of assembling and solving one dense linear system, the FEM is used to assemble several independent sparse linear systems that can be solved in parallel. The approximate solution to problem ( 1 ) is then obtained as a linear combination of those independently generated finite element approximations. The existing rational approximation schemes employ the same finite element space (associated with a single finite element mesh on \Omega ) across all non-fractional parametric PDE problems; we refer to this approach as a single-mesh scheme. A posteriori error estimation and adaptive solution techniques are fairly recent topics in the numerical analysis of FPDEs. The following a posteriori error estimation strategies have been developed to guide adaptive mesh refinement in the FEM-based methods for fractional problems: residual-based error estimators [ 1 , 22 ] , a gradient recovery based error estimator [ 33 ] , and hierarchical error estimators [ 18 , 15 ] . In this work, we propose a novel multimesh framework for rational approximation schemes where the underlying meshes (and hence FEM approximations) can be different for different non-fractional parametric problems. Within this framework, we develop an a posteriori error estimation strategy that drives the adaptive multimesh refinement algorithm for the numerical solution of problem ( 1 ). We emphasize that our multimesh construction, the error estimation strategy, and the adaptive algorithm are generic, in the sense that they can be used with any rational approximation of the function \lambda^{-s} (see § 3.1 below). Our practical implementation, however, employs a specific rational approximation—the one proposed by Bonito and Pasciak in [ 13 ] —but can be easily adapted to other approximations such as, e.g., the BURA-based schemes [ 23 ] . Numerical experiments demonstrate two main advantages of our adaptive multimesh framework: faster convergence rates (particularly, for smaller values of s\in(0,1) ) than the rates for uniform mesh refinement (see [ 13 ] ) and significant reductions in computational costs—measured by the overall number of degrees of freedom and the actual computational time—when compared to the adaptive algorithm proposed in [ 15 ] in a single-mesh setting. The paper is organized as follows. In section 2 , we define the spectral fractional Laplacian operator (-\Delta)^{s} in ( 1 ). Section 3 introduces a multimesh rational approximation scheme for problem ( 1 ), including the underlying rational approximation of the function \lambda^{-s} , finite element discretizations as well as the concept of union mesh that plays a key role in computing the fully discrete solution and in a posteriori error analysis. In section 4 , we discuss local hierarchical error indicators and propose two a posteriori estimates of the overall discretization error in the L^{2} -norm. An adaptive multimesh refinement algorithm is formulated in section 5 . The effectiveness of two error estimation strategies and the performance of the proposed adaptive algorithm are assessed in a series of numerical experiments presented in section 6 . 2 The spectral fractional Laplacian Let \Omega\subset\mathbb{R}^{d} ( d=2,3 ) be a connected bounded domain with polygonal/polyhedral boundary \partial\Omega and let \omega be any open connected subset