Scaled Vecchia approximation for fast computer-model emulation
From MaRDI portal
Abstract: Many scientific phenomena are studied using computer experiments consisting of multiple runs of a computer model while varying the input settings. Gaussian processes (GPs) are a popular tool for the analysis of computer experiments, enabling interpolation between input settings, but direct GP inference is computationally infeasible for large datasets. We adapt and extend a powerful class of GP methods from spatial statistics to enable the scalable analysis and emulation of large computer experiments. Specifically, we apply Vecchia's ordered conditional approximation in a transformed input space, with each input scaled according to how strongly it relates to the computer-model response. The scaling is learned from the data, by estimating parameters in the GP covariance function using Fisher scoring. Our methods are highly scalable, enabling estimation, joint prediction and simulation in near-linear time in the number of model runs. In several numerical examples, our approach substantially outperformed existing methods.
Recommendations
- A fast and calibrated computer model emulator: an empirical Bayes approach
- Multiresolution functional ANOVA for large-scale, many-input computer experiments
- Massively parallel approximate Gaussian process regression
- Locally induced Gaussian processes for large-scale simulation experiments
- Computer emulation with nonstationary Gaussian processes
Cites work
- 2010 Rietz lecture: When does the screening effect hold?
- A case study competition among methods for analyzing large spatial data
- A class of multi-resolution approximations for large spatial datasets
- A general framework for Vecchia approximations of Gaussian processes
- A unifying view of sparse approximate Gaussian process regression
- Approximating Likelihoods for Large Spatial Data Sets
- Assessing probabilistic forecasts of multivariate quantities, with an application to ensemble predictions of surface winds
- Bayesian calibration of computer models. (With discussion)
- Bayesian Design and Analysis of Computer Experiments: Use of Derivatives in Surface Prediction
- Combining Field Data and Computer Simulations for Calibration and Prediction
- Compression, inversion, and approximate PCA of dense kernel matrices at near-linear computational complexity
- Design and analysis of computer experiments. With comments and a rejoinder by the authors
- Efficient Algorithms for Bayesian Nearest Neighbor Gaussian Processes
- Efficient emulators of computer experiments using compactly supported correlation functions, with an application to cosmology
- Efficient global optimization of expensive black-box functions
- Emulating satellite drag from large simulation experiments
- Fast prediction of deterministic functions using sparse grid experimental designs
- Gaussian Predictive Process Models for Large Spatial Data Sets
- Gaussian process learning via Fisher scoring of Vecchia's approximation
- scientific article; zbMATH DE number 43985 (Why is no real title available?)
- Improving the performance of predictive process modeling for large datasets
- Interpolation of spatial data. Some theory for kriging
- Nonseparable dynamic nearest neighbor Gaussian process models for large spatio-temporal data with an application to particulate matter analysis
- Sparse Cholesky Factorization by Kullback--Leibler Minimization
- Structured discrepancy in Bayesian model calibration for ChemCam on the Mars Curiosity rover
- Uncertainty quantification using the nearest neighbor Gaussian process
- Vecchia-Laplace approximations of generalized Gaussian processes for big non-Gaussian spatial data
Cited in
(22)- Multiresolution functional ANOVA for large-scale, many-input computer experiments
- Locally induced Gaussian processes for large-scale simulation experiments
- Large-scale local surrogate modeling of stochastic simulation experiments
- Shift and scale coupling methods for perfect simulation
- Massively parallel approximate Gaussian process regression
- Accurate emulators for large-scale computer experiments
- A sequential split-and-conquer approach for the analysis of big dependent data in computer experiments
- Emulating satellite drag from large simulation experiments
- Correlation-based sparse inverse Cholesky factorization for fast Gaussian-process inference
- Sensitivity Prewarping for Local Surrogate Modeling
- Towards Improved Heliosphere Sky Map Estimation with Theseus
- Hierarchical shrinkage Gaussian processes: applications to computer code emulation and dynamical system recovery
- ProSpar-GP: Scalable Gaussian Process Modeling with Massive Nonstationary Datasets
- The inverse Kalman filter
- Block Vecchia Approximation for Scalable and Efficient Gaussian Process Computations
- Sparse inverse Cholesky factorization of dense kernel matrices by greedy conditional selection
- Enhancing approximate modular Bayesian inference by emulating the conditional posterior
- Estimating basis functions in massive fields under the spatial mixed effects model
- Synthesizing data products, mathematical models, and observations for lake temperature forecasting
- Locally anisotropic nonstationary covariance functions on the sphere
- Linear-Cost Vecchia Approximation of Multivariate Normal Probabilities
- Vecchia-approximated Deep Gaussian Processes for Computer Experiments
This page was built for publication: Scaled Vecchia approximation for fast computer-model emulation
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5097836)
