Abstract:We construct finite-dimensional approximations of solution spaces of divergence form operators with $L^\infty$-coefficients. Our method does not rely on concepts of ergodicity or scale-separation, but on the property that the solution space of these operators is compactly embedded in $H^1$ if source terms are in the unit ball of $L^2$ instead of the unit ball of $H^{-1}$. Approximation spaces are generated by solving elliptic PDEs on localized sub-domains with source terms corresponding to approximation bases for $H^2$. The $H^1$-error estimates show that $\mathcal{O}(h^{-d})$-dimensional spaces with basis elements localized to sub-domains of diameter $\mathcal{O}(h^\alpha \ln \frac{1}{h})$ (with $\alpha \in [1/2,1)$) result in an $\mathcal{O}(h^{2-2\alpha})$ accuracy for elliptic, parabolic and hyperbolic problems. For high-contrast media, the accuracy of the method is preserved provided that localized sub-domains contain buffer zones of width $\mathcal{O}(h^\alpha \ln \frac{1}{h})$ where the contrast of the medium remains bounded. The proposed method can naturally be generalized to vectorial equations (such as elasto-dynamics).

What problem does this paper attempt to address?

The problem that this paper attempts to solve is how to construct a finite - dimensional approximation space to approximate the solution space of divergence form operators with non - separable scales and high - contrast media. Specifically, the author proposes a method that does not rely on scale - separation or ergodicity assumptions, but is based on the compact embedding property of the solution space in the $H^1$ space. ### Core problems of the paper 1. **Constructing a finite - dimensional approximation space**: How to construct a finite - dimensional space to approximate the solution space of divergence - type operators, especially for cases with complex coefficients (such as high - contrast media). 2. **Handling non - separable scales**: Traditional multiscale methods usually rely on the scale - separation assumption, but this paper aims to handle cases without obvious scale separation. 3. **High - contrast media**: How to maintain the accuracy of the approximation method when the media contrast is very high. ### Key points of the method - **Compact embedding property**: Utilize the compact embedding property of the solution space in $H^1$, rather than relying on scale - separation or ergodicity assumptions. - **Localized basis functions**: Generate localized basis functions by solving elliptic partial differential equations (PDEs) on local regions. These basis functions can effectively capture the main features of the solution. - **Error estimate**: Provide a strict error estimate, indicating that the proposed localized method can achieve high precision when dealing with elliptic, parabolic, and hyperbolic problems. ### Main conclusions - For problems with non - separable scales and high - contrast media, basis functions can be generated by solving elliptic PDEs on local sub - domains, thereby constructing an effective finite - dimensional approximation space. - This method is not only applicable to static problems (such as elliptic problems), but also applicable to time - dependent problems (such as parabolic and hyperbolic problems), and can also maintain high precision in high - contrast media. ### Formula summary The key formulas involved in the paper include: - The form of the divergence - type operator: \[ - \text{div}(a(x)\nabla u(x)) = g(x),\quad x\in\Omega \] where $a(x)$ is a symmetric and uniformly elliptic coefficient matrix. - Compact embedding condition: \[ \lambda_{\min}(a)|\xi|^2\leq\xi^T a(x)\xi\leq\lambda_{\max}(a)|\xi|^2 \] - Construction of localized basis functions: \[ h^{- 2\alpha}\psi_i-\text{div}(a\nabla\psi_i)=\Delta\phi_i\quad\text{in }B(x_i,C_1h^\alpha\ln(1 / h))\cap\Omega \] - Error estimate: \[ \|u - u_h\|_{H^1_0(\Omega)}\leq \begin{cases} Ch&\text{if }\alpha\in(0,\frac{1}{2}]\\ Ch^{2 - 2\alpha}&\text{if }\alpha\in[\frac{1}{2},1) \end{cases} \] Through these methods and formulas, the author successfully solves the problem of constructing an efficient finite - dimensional approximation space under complex media conditions.

Localized bases for finite dimensional homogenization approximations with non-separated scales and high-contrast

Polyharmonic homogenization, rough polyharmonic splines and sparse super-localization

Hybrid Localized Spectral Decomposition for multiscale problems

Localized Harmonic Characteristic Basis Functions for Multiscale Finite Element Methods

Adaptive Local (AL) Basis for Elliptic Problems with $L^\infty$-Coefficients

Homogenization of nondivergence-form elliptic equations with discontinuous coefficients and finite element approximation of the homogenized problem

Local Convergence of Inexact Methods under the Hölder Condition

Homogenization of nonlocal convolution type operators: Approximation for the resolvent with corrector

Super-localized orthogonal decomposition for convection-dominated diffusion problems

Hierarchical Super-Localized Orthogonal Decomposition Method

Localized Orthogonal Decomposition Method with $H^1$ Interpolation for Multiscale Elliptic Problem

Sharp error estimates for spatial-temporal finite difference approximations to fractional sub-diffusion equation without regularity assumption on the exact solution

Computational multiscale methods for nondivergence-form elliptic partial differential equations

Homogenization of non-local energies on disconnected sets

Local decay rates of best-approximation errors using vector-valued finite elements for fields with low regularity and integrable curl or divergence

High-frequency homogenization of multidimensional hyperbolic equations

Uniform Calderón-Zygmund estimates in multiscale elliptic homogenization

Local well-posedness for dispersive equations with bounded data

A parabolic local problem with exponential decay of the resonance error for numerical homogenization

Localized subspace iteration methods for elliptic multiscale problems