Physics information aided kriging using stochastic simulation models
From MaRDI portal
Publication:5015299
DOI10.1137/20M1331585zbMATH Open1483.65019arXiv1809.03461OpenAlexW3214056652MaRDI QIDQ5015299FDOQ5015299
Authors: Xiu Yang, Guzel Tartakovsky, Alexandre M. Tartakovsky 
Publication date: 7 December 2021
Published in: SIAM Journal on Scientific Computing (Search for Journal in Brave)
Abstract: In this work, we propose a new Gaussian process regression (GPR) method: physics information aided Kriging (PhIK). In the standard data-driven Kriging, the unknown function of interest is usually treated as a Gaussian process with assumed stationary covariance with hyperparameters estimated from data. In PhIK, we compute the mean and covariance function from realizations of available stochastic models, e.g., from realizations of governing stochastic partial differential equations solutions. Such constructed Gaussian process generally is non-stationary, and does not assume a specific form of the covariance function. Our approach avoids the optimization step in data-driven GPR methods to identify the hyperparameters. More importantly, we prove that the physical constraints in the form of a deterministic linear operator are guaranteed in the resulting prediction. We also provide an error estimate in preserving the physical constraints when errors are included in the stochastic model realizations. To reduce the computational cost of obtaining stochastic model realizations, we propose a multilevel Monte Carlo estimate of the mean and covariance functions. Further, we present an active learning algorithm that guides the selection of additional observation locations. The efficiency and accuracy of PhIK are demonstrated for reconstructing a partially known modified Branin function, studying a three-dimensional heat transfer problem and learning a conservative tracer distribution from sparse concentration measurements.
Full work available at URL: https://arxiv.org/abs/1809.03461
Recommendations
- APIK: Active Physics-Informed Kriging Model with Partial Differential Equations
- Physics-informed cokriging: a Gaussian-process-regression-based multifidelity method for data-model convergence
- Intrinsic Kriging and prior information
- Kriging: methods and applications
- Gradient Extrapolated Stochastic Kriging
Monte Carlo methods (65C05) Learning and adaptive systems in artificial intelligence (68T05) Gaussian processes (60G15) Numerical solutions to stochastic differential and integral equations (65C30)
Cites Work
- Efficient global optimization of expensive black-box functions
- Gaussian processes for machine learning (GPML) toolbox
- Design and analysis of computer experiments. With comments and a rejoinder by the authors
- Gaussian processes for machine learning.
- An explicit link between Gaussian fields and Gaussian Markov random fields: the stochastic partial differential equation approach
- Multigrid with Rough Coefficients and Multiresolution Operator Decomposition from Hierarchical Information Games
- A unifying view of sparse approximate Gaussian process regression
- Probabilistic numerics and uncertainty in computations
- Title not available (Why is that?)
- Title not available (Why is that?)
- Title not available (Why is that?)
- Multilevel Monte Carlo Path Simulation
- Multilevel Monte Carlo methods and applications to elliptic PDEs with random coefficients
- Bayesian numerical homogenization
- Multi-level Monte Carlo finite element method for elliptic PDEs with stochastic coefficients
- Average-case analysis of numerical problems
- Title not available (Why is that?)
- Fast and Exact Simulation of Stationary Gaussian Processes through Circulant Embedding of the Covariance Matrix
- High-Order Collocation Methods for Differential Equations with Random Inputs
- A reproducing kernel Hilbert space approach to functional linear regression
- Convergence analysis of multilevel Monte Carlo variance estimators and application for random obstacle problems
- Improved multilevel Monte Carlo convergence using the Milstein scheme
- Predicting the output from a complex computer code when fast approximations are available
- Gaussian process for nonstationary time series prediction
- Support vector machine active learning with applications to text classification
- Adaptive ANOVA decomposition of stochastic incompressible and compressible flows
- Near-optimal sensor placements in Gaussian processes: theory, efficient algorithms and empirical studies
- Reweighted \(\ell_1\) minimization method for stochastic elliptic differential equations
- Advanced Lectures on Machine Learning
- Estimation of arbitrary order central statistical moments by the multilevel Monte Carlo method
- A hybrid anchored-ANOVA - POD/Kriging method for uncertainty quantification in unsteady high-fidelity CFD simulations
- Multifidelity information fusion algorithms for high-dimensional systems and massive data sets
- Additive covariance kernels for high-dimensional Gaussian process modeling
- Multi-fidelity Gaussian process regression for prediction of random fields
- Survey of multifidelity methods in uncertainty propagation, inference, and optimization
- Machine learning of linear differential equations using Gaussian processes
- Numerical Gaussian processes for time-dependent and nonlinear partial differential equations
- Multi-fidelity stochastic collocation method for computation of statistical moments
- Immiscible front evolution in randomly heterogeneous porous media
- Physics-informed cokriging: a Gaussian-process-regression-based multifidelity method for data-model convergence
- When Bifidelity Meets CoKriging: An Efficient Physics-Informed MultiFidelity Method
- A Comparison of Closures for Stochastic Advection-Diffusion Equations
Cited In (5)
- Physics-informed cokriging: a Gaussian-process-regression-based multifidelity method for data-model convergence
- Physics-informed machine learning method with space-time Karhunen-Loève expansions for forward and inverse partial differential equations
- Learning with partition of unity-based kriging estimators
- APIK: Active Physics-Informed Kriging Model with Partial Differential Equations
- Physics-Based Kriging Surrogates for a Class of Finite Element Codes
Uses Software
This page was built for publication: Physics information aided kriging using stochastic simulation models
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5015299)
