Variational inference with vine copulas: an efficient approach for Bayesian computer model calibration

DOI10.1007/S11222-022-10194-ZzbMATH Open1502.62017arXiv2003.12890OpenAlexW3013828499MaRDI QIDQ2110192FDOQ2110192

Authors: Vojtech Kejzlar, Tapabrata Maiti 

Publication date: 21 December 2022

Published in: Statistics and Computing (Search for Journal in Brave)

Abstract: With the advancements of computer architectures, the use of computational models proliferates to solve complex problems in many scientific applications such as nuclear physics and climate research. However, the potential of such models is often hindered because they tend to be computationally expensive and consequently ill-fitting for uncertainty quantification. Furthermore, they are usually not calibrated with real-time observations. We develop a computationally efficient algorithm based on variational Bayes inference (VBI) for calibration of computer models with Gaussian processes. Unfortunately, the speed and scalability of VBI diminishes when applied to the calibration framework with dependent data. To preserve the efficiency of VBI, we adopt a pairwise decomposition of the data likelihood using vine copulas that separate the information on dependence structure in data from their marginal distributions. We provide both theoretical and empirical evidence for the computational scalability of our methodology and describe all the necessary details for an efficient implementation of the proposed algorithm. We also demonstrate the opportunities given by our method for practitioners on a real data example through calibration of the Liquid Drop Model of nuclear binding energies.

Full work available at URL: https://arxiv.org/abs/2003.12890

Recommendations

• Bayesian Projected Calibration of Computer Models
• Efficient calibration for imperfect computer models
• A fast and calibrated computer model emulator: an empirical Bayes approach
• Calibration of computer models with multivariate output
• Computer Model Calibration Using High-Dimensional Output

Mathematics Subject Classification ID

Computational methods for problems pertaining to statistics (62-08) Characterization and structure theory for multivariate probability distributions; copulas (62H05) Applications of statistics to physics (62P35)

Cites Work

• Selecting and estimating regular vine copulae and application to financial returns
• Automatic differentiation variational inference
• Adaptive subgradient methods for online learning and stochastic optimization
• Scaled Gaussian stochastic process for computer model calibration and prediction
• Graphical models, exponential families, and variational inference
• Bayesian calibration of computer models. (With discussion)
• Truncated regular vines in high dimensions with application to financial data
• Title not available (Why is that?)
• Uncertainty Analysis with High Dimensional Dependence Modelling
• Title not available (Why is that?)
• Vines -- a new graphical model for dependent random variables.
• Title not available (Why is that?)
• A unifying view of sparse approximate Gaussian process regression
• Computer Model Calibration Using High-Dimensional Output
• Title not available (Why is that?)
• A Stochastic Approximation Method
• Bayesian Projected Calibration of Computer Models
• Rao-Blackwellisation of sampling schemes
• Combining Field Data and Computer Simulations for Calibration and Prediction
• Exploratory designs for computational experiments
• An introduction to variational methods for graphical models
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Title not available (Why is that?)
• Combining experimental data and computer simulations, with an application to flyer plate experiments
• High-Dimensional Copula Variational Approximation Through Transformation
• Computer model calibration with confidence and consistency
• Truncation of vine copulas using fit indices
• A fast and calibrated computer model emulator: an empirical Bayes approach
• Stochastic gradient MCMC for state space models
• Frequentist consistency of variational Bayes

Cited In (5)

• Specification of informative prior distributions for multinomial models using vine copulas
• A review on computer model calibration
• A vine-copula based adaptive MCMC sampler for efficient inference of dynamical systems
• Deep Gaussian processes for calibration of computer models (with discussion)
• Bayesian model calibration for block copolymer self-assembly: likelihood-free inference and expected information gain computation via measure transport

Uses Software

• AdaGrad
• RMSprop
• ADVI
• ADADELTA

This page was built for publication: Variational inference with vine copulas: an efficient approach for Bayesian computer model calibration

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2110192)