English

Reduced Sum Implementation of the BURA Method for Spectral Fractional Diffusion Problems

Numerical Analysis 2021-05-20 v1 Numerical Analysis

Abstract

The numerical solution of spectral fractional diffusion problems in the form Aαu=f{\mathcal A}^\alpha u = f is studied, where A\mathcal A is a selfadjoint elliptic operator in a bounded domain ΩRd\Omega\subset {\mathbb R}^d, and α(0,1]\alpha \in (0,1]. The finite difference approximation of the problem leads to the system Aαu=f{\mathbb A}^\alpha {\mathbf u} = {\mathbf f}, where A{\mathbb A} is a sparse, symmetric and positive definite (SPD) matrix, and Aα{\mathbb A}^\alpha is defined by its spectral decomposition. In the case of finite element approximation, A{\mathbb A} is SPD with respect to the dot product associated with the mass matrix. The BURA method is introduced by the best uniform rational approximation of degree kk of tαt^{\alpha} in [0,1][0,1], denoted by rα,kr_{\alpha,k}. Then the approximation uku{\bf u}_k\approx {\bf u} has the form uk=c0f+i=1kci(Ad~iI)1f{\bf u}_k = c_0 {\mathbf f} +\sum_{i=1}^k c_i({\mathbb A} - {\widetilde{d}}_i {\mathbb I})^{-1}{\mathbf f}, d~i<0{\widetilde{d}}_i<0, thus requiring the solving of kk auxiliary linear systems with sparse SPD matrices. The BURA method has almost optimal computational complexity, assuming that an optimal PCG iterative solution method is applied to the involved auxiliary linear systems. The presented analysis shows that the absolute values of first %d~i{\widetilde{d}}_i {d~i}i=1k\left\{{\widetilde{d}}_i\right\}_{i=1}^{k'} can be extremely large. In such a case the condition number of Ad~iI{\mathbb A} - {\widetilde{d}}_i {\mathbb I} is practically equal to one. Obviously, such systems do not need preconditioning. The next question is if we can replace their solution by directly multiplying f{\mathbf f} with ci/d~i-c_i/{\widetilde{d}}_i. Comparative analysis of numerical results is presented as a proof-of-concept for the proposed RS-BURA method.

Keywords

Cite

@article{arxiv.2105.09048,
  title  = {Reduced Sum Implementation of the BURA Method for Spectral Fractional Diffusion Problems},
  author = {Stanislav Harizanov and Nikola Kosturski and Ivan Lirkov and Svetozar Margenov and Yavor Vutov},
  journal= {arXiv preprint arXiv:2105.09048},
  year   = {2021}
}

Comments

8 pages, 4 figures

R2 v1 2026-06-24T02:15:26.369Z