Abstract:There has been a surge of interest in uncertainty quantification for parametric partial differential equations (PDEs) with Gevrey regular inputs. The Gevrey class contains functions that are infinitely smooth with a growth condition on the higher-order partial derivatives, but which are nonetheless not analytic in general. Recent studies by Chernov and Le (Comput. Math. Appl., 2024, and SIAM J. Numer. Anal., 2024) as well as Harbrecht, Schmidlin, and Schwab (Math. Models Methods Appl. Sci., 2024) analyze the setting wherein the input random field is assumed to be uniformly bounded with respect to the uncertain parameters. In this paper, we relax this assumption and allow for parameter-dependent bounds. The parametric inputs are modeled as generalized Gaussian random variables, and we analyze the application of quasi-Monte Carlo (QMC) integration to assess the PDE response statistics using randomly shifted rank-1 lattice rules. In addition to the QMC error analysis, we also consider the dimension truncation and finite element errors in this setting.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to apply the Quasi - Monte Carlo (QMC) method for uncertainty quantification in partial differential equations (PDEs) with generalized Gaussian input uncertainty. Specifically, the paper focuses on the evaluation of response statistics of parameterized PDEs under Gevrey - regular input random fields. Compared with previous studies, this paper relaxes the uniformly bounded assumption on the input random field, allows parameter - dependent boundaries, and considers the generalized β - Gaussian distribution as the probability measure of the input random variable. In addition, the paper also analyzes QMC integration error, dimension truncation error, and finite - element error. ### Background and Motivation Uncertainty Quantification (UQ) has become increasingly important for the modeling and simulation of partial differential equations with random coefficients. These equations can be used to describe problems in various scientific and engineering disciplines, such as materials science, fluid mechanics, and environmental science, etc. The key to UQ is how to statistically characterize the behavior of the solution given the statistical properties of the input random field. However, such problems are usually computationally expensive, so it is particularly important to study efficient numerical algorithms such as sparse grids and QMC methods. ### Main Contributions 1. **Relaxed Assumptions**: The paper relaxes the uniformly bounded assumption on the input random field and allows parameter - dependent boundaries. This makes the model more flexible and able to handle a wider range of practical problems. 2. **Generalized β - Gaussian Distribution**: The paper considers the generalized β - Gaussian distribution as the probability measure of the input random variable, which is more general than the traditional Gaussian distribution and can better describe the uncertainty in practice. 3. **Application of QMC Method**: The paper analyzes in detail the application of the QMC method in high - dimensional integration, especially for elliptic PDEs. Through the randomly shifted rank - 1 lattice - point rule, the paper provides an effective method to approximate high - dimensional integrals. 4. **Error Analysis**: The paper not only analyzes the QMC integration error, but also considers the dimension truncation error and the finite - element error, and provides strict convergence rate estimates. ### Mathematical Model The model problem considered in the paper is an elliptic PDE with a random diffusion coefficient: \[-\nabla\cdot(a(x,\omega)\nabla u(x,\omega)) = f(x),\quad x\in D,\] \[u(x,\omega) = 0,\quad x\in\partial D,\] where \(a(x,\omega)\) is a log - normal random field. For numerical approximation, the input random field is represented as a finite sum of parameters: \[a_s(x,y)=a_0(x)\exp\left(\sum_{j = 1}^s\sqrt{\lambda_j}\psi_j(x)y_j\right),\] where \(y_j\sim N(0,1)\). ### Key Results 1. **Parameter Regularity**: The paper proves the parameter regularity of the PDE solution under the Gevrey - regularity assumption. 2. **QMC Error Bounds**: By applying the QMC method, the paper derives strict bounds on the integration error, and combined with the dimension truncation error and the finite - element error, gives an estimate of the total error. 3. **Numerical Experiments**: The paper verifies the theoretical results through numerical experiments and shows the effectiveness of the QMC method. ### Conclusion The paper expands the application range of the QMC method in uncertainty quantification by relaxing the assumptions on the input random field. In particular, the generalized β - Gaussian distribution and the parameter - dependent boundary conditions proposed in the paper provide new tools for dealing with complex uncertainties in practical problems. These results are of great significance for improving the efficiency and accuracy of numerical simulations.

Quasi-Monte Carlo for partial differential equations with generalized Gaussian input uncertainty

Parabolic PDE-constrained optimal control under uncertainty with entropic risk measure using quasi-Monte Carlo integration

Quasi-Monte Carlo integration for feedback control under uncertainty

Quasi-Monte Carlo and discontinuous Galerkin

Density estimation for elliptic PDE with random input by preintegration and quasi-Monte Carlo methods

Generalized dimension truncation error analysis for high-dimensional numerical integration: lognormal setting and beyond

Random time step probabilistic methods for uncertainty quantification in chaotic and geometric numerical integration

QMC integration for lognormal-parametric, elliptic PDEs: local supports and product weights

Variational Monte Carlo - Bridging Concepts of Machine Learning and High Dimensional Partial Differential Equations

The Gevrey class implicit mapping theorem with application to UQ of semilinear elliptic PDEs

Uncertainty quantification for generalized Langevin dynamics

Leveraging Viscous Hamilton–Jacobi PDEs for Uncertainty Quantification in Scientific Machine Learning

Quasi-Monte Carlo for unbounded integrands with importance sampling

Uncertainty quantification for random domains using periodic random variables

Leveraging viscous Hamilton-Jacobi PDEs for uncertainty quantification in scientific machine learning

An introduction to uncertainty quantification for kinetic equations and related problems

Higher order QMC Galerkin discretization for parametric operator equations

Nonasymptotic Convergence Rate of Quasi-Monte Carlo: Applications to Linear Elliptic PDEs with Lognormal Coefficients and Importance Samplings

PDE-constrained Gaussian process surrogate modeling with uncertain data locations