Sparse Gaussian processes for solving nonlinear PDEs
From MaRDI portal
Abstract: This article proposes an efficient numerical method for solving nonlinear partial differential equations (PDEs) based on sparse Gaussian processes (SGPs). Gaussian processes (GPs) have been extensively studied for solving PDEs by formulating the problem of finding a reproducing kernel Hilbert space (RKHS) to approximate a PDE solution. The approximated solution lies in the span of base functions generated by evaluating derivatives of different orders of kernels at sample points. However, the RKHS specified by GPs can result in an expensive computational burden due to the cubic computation order of the matrix inverse. Therefore, we conjecture that a solution exists on a ``condensed" subspace that can achieve similar approximation performance, and we propose a SGP-based method to reformulate the optimization problem in the ``condensed" subspace. This significantly reduces the computation burden while retaining desirable accuracy. The paper rigorously formulates this problem and provides error analysis and numerical experiments to demonstrate the effectiveness of this method. The numerical experiments show that the SGP method uses fewer than half the uniform samples as inducing points and achieves comparable accuracy to the GP method using the same number of uniform samples, resulting in a significant reduction in computational cost. Our contributions include formulating the nonlinear PDE problem as an optimization problem on a ``condensed" subspace of RKHS using SGP, as well as providing an existence proof and rigorous error analysis. Furthermore, our method can be viewed as an extension of the GP method to account for general positive semi-definite kernels.
Recommendations
- Numerical Gaussian processes for time-dependent and nonlinear partial differential equations
- Solving and learning nonlinear PDEs with Gaussian processes
- Operator learning with Gaussian processes
- Bayesian numerical methods for nonlinear partial differential equations
- Active learning based sampling for high-dimensional nonlinear partial differential equations
Cites work
- A unifying view of sparse approximate Gaussian process regression
- Bayesian numerical homogenization
- Bayesian numerical methods for nonlinear partial differential equations
- Gaussian processes for machine learning.
- Geometry on probability spaces
- Improved Bounds for the Nyström Method With Application to Kernel Classification
- Mean field games: numerical methods
- Multigrid with Rough Coefficients and Multiresolution Operator Decomposition from Hierarchical Information Games
- Numerical approximation of partial differential equations
- Numerical Gaussian processes for time-dependent and nonlinear partial differential equations
- Numerical methods for mean field games based on Gaussian processes and Fourier features
- On the Nyström method for approximating a gram matrix for improved kernel-based learning
- Operator-adapted wavelets, fast solvers, and numerical homogenization. From a game theoretic approach to numerical approximation and algorithm design
- PAC-Bayesian Generalisation Error Bounds for Gaussian Process Classification
- Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
- Proximal methods for stationary mean field games with local couplings
- Shannon sampling. II: Connections to learning theory
- Solving and learning nonlinear PDEs with Gaussian processes
- Sparse Cholesky Factorization by Kullback--Leibler Minimization
- Sparse spectrum Gaussian process regression
- Variational Fourier features for Gaussian processes
- Weak adversarial networks for high-dimensional partial differential equations
Cited in
(14)- Solving and learning nonlinear PDEs with Gaussian processes
- Gaussian process learning of nonlinear dynamics
- Error analysis of kernel/GP methods for nonlinear and parametric PDEs
- The ADMM-PINNs algorithmic framework for nonsmooth PDE-constrained optimization: a deep learning approach
- Operator learning with Gaussian processes
- Learning surrogate potential mean field games via Gaussian processes: a data-driven approach to ill-posed inverse problems
- A Gaussian process framework for solving forward and inverse problems involving nonlinear partial differential equations
- Sparse Cholesky factorization for solving nonlinear PDEs via Gaussian processes
- H^ -optimal control via game-theoretic differential dynamic programming and Gaussian processes
- Consistency of variational inference for Besov priors in non-linear inverse problems
- A unified perspective of Gaussian process approximation for differential equations
- Random feature method solving second-order nonlinear integrable systems
- Temporal categorical modeling using latent Gaussian processes
- Solving functional PDEs with Gaussian processes and applications to functional renormalization group equations
This page was built for publication: Sparse Gaussian processes for solving nonlinear PDEs
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6173368)
