A unified framework for multiscale spectral generalized FEMs and low-rank approximations to multiscale PDEs

Chupeng Ma

2024-03-15

Abstract:This work presents an abstract framework for the design, implementation, and analysis of the multiscale spectral generalized finite element method (MS-GFEM), a particular numerical multiscale method originally proposed in [I. Babuska and R. Lipton, Multiscale Model.\;\,Simul., 9 (2011), pp.~373--406]. MS-GFEM is a partition of unity method employing optimal local approximation spaces constructed from local spectral problems. We establish a general local approximation theory demonstrating exponential convergence with respect to local degrees of freedom under certain assumptions, with explicit dependence on key problem parameters. Our framework applies to a broad class of multiscale PDEs with $L^{\infty}$-coefficients in both continuous and discrete, finite element settings, including highly indefinite problems (convection-dominated diffusion, as well as the high-frequency Helmholtz, Maxwell and elastic wave equations with impedance boundary conditions), and higher-order problems. Notably, we prove a local convergence rate of $O(e^{-cn^{1/d}})$ for MS-GFEM for all these problems, improving upon the $O(e^{-cn^{1/(d+1)}})$ rate shown by Babuska and Lipton. Moreover, based on the abstract local approximation theory for MS-GFEM, we establish a unified framework for showing low-rank approximations to multiscale PDEs. This framework applies to the aforementioned problems, proving that the associated Green's functions admit an $O(|\log\epsilon|^{d})$-term separable approximation on well-separated domains with error $\epsilon>0$. Our analysis improves and generalizes the result in [M. Bebendorf and W. Hackbusch, Numerische Mathematik, 95 (2003), pp.~1-28] where an $O(|\log\epsilon|^{d+1})$-term separable approximation was proved for Poisson-type problems.

Numerical Analysis

What problem does this paper attempt to address?

### What problems does this paper attempt to solve?

This paper aims to solve the computational challenges in multiscale partial differential equations (PDEs). Specifically, the paper focuses on multiscale PDEs with heterogeneous coefficients that oscillate on possibly non - separated scales, which pose computational difficulties for traditional numerical techniques.

To solve these problems, the paper proposes a unified framework for the multiscale spectral generalized finite element method (MS - GFEM). MS - GFEM is a special numerical multiscale method, originally proposed by Babuska and Lipton. By introducing an abstract local approximation theory, this paper not only improves the convergence rate of MS - GFEM, but also establishes a unified framework to prove the low - rank approximation of multiscale PDEs.

### Main contributions

1. **Improved convergence rate**: The paper proves that for a series of multiscale PDEs (including highly indefinite problems such as convection - dominated diffusion, high - frequency Helmholtz equations, etc.), the local convergence rate of MS - GFEM is \(O(e^{-cn^{1/d}})\), which is better than the previous result \(O(e^{-cn^{1/(d + 1)}})\).

2. **Unified framework**: Based on the abstract local approximation theory of MS - GFEM, the paper establishes a unified framework for proving the low - rank approximation of multiscale PDEs. This framework is applicable to a wide range of problems and proves that the related Green functions have \(O(|\log\epsilon|^d)\)-term separable approximations on well - separated domains.

3. **Low - rank approximation**: The paper reveals the close connection between the approximation mechanism in the structured inverse method and MS - GFEM, provides a strict theoretical basis, and clarifies the relationships between these methods.

4. **Wide applicability**: This framework is applicable not only to continuous and discrete finite element settings, but also to high - order problems and various highly indefinite problems.

### Key techniques and methods involved

- **Multiscale spectral generalized finite element method (MS - GFEM)**: Constructs the optimal local approximation space by solving local eigenvalue problems.
- **Low - rank approximation**: Utilizes the low - rank property of Green functions for matrix compression and fast algorithm design.
- **Caccioppoli inequality and weak approximation property**: Serve as two basic conditions to ensure the exponential decay of local errors.

In conclusion, this paper provides a powerful theoretical framework that can efficiently and accurately solve the computational problems in multiscale PDEs and provides a solid foundation for further research and development.