A multiscale strategy for Bayesian inference using transport maps
From MaRDI portal
Publication:3179325
Abstract: In many inverse problems, model parameters cannot be precisely determined from observational data. Bayesian inference provides a mechanism for capturing the resulting parameter uncertainty, but typically at a high computational cost. This work introduces a multiscale decomposition that exploits conditional independence across scales, when present in certain classes of inverse problems, to decouple Bayesian inference into two stages: (1) a computationally tractable coarse-scale inference problem; and (2) a mapping of the low-dimensional coarse-scale posterior distribution into the original high-dimensional parameter space. This decomposition relies on a characterization of the non-Gaussian joint distribution of coarse- and fine-scale quantities via optimal transport maps. We demonstrate our approach on a sequence of inverse problems arising in subsurface flow, using the multiscale finite element method to discretize the steady state pressure equation. We compare the multiscale strategy with full-dimensional Markov chain Monte Carlo on a problem of moderate dimension (100 parameters) and then use it to infer a conductivity field described by over 10,000 parameters.
Recommendations
- Scalable posterior approximations for large-scale Bayesian inverse problems via likelihood-informed parameter and state reduction
- Bayesian inference with optimal maps
- A Bayesian approach to multiscale inverse problems with on-the-fly scale determination
- Transport map accelerated adaptive importance sampling, and application to inverse problems arising from multiscale stochastic reaction networks
- A Bayesian approach to multiscale inverse problems using the sequential Monte Carlo method
Cites work
- scientific article; zbMATH DE number 2140079 (Why is no real title available?)
- scientific article; zbMATH DE number 1866807 (Why is no real title available?)
- scientific article; zbMATH DE number 1909499 (Why is no real title available?)
- scientific article; zbMATH DE number 2117879 (Why is no real title available?)
- scientific article; zbMATH DE number 3210995 (Why is no real title available?)
- A Bayesian approach to characterizing uncertainty in inverse problems using coarse and fine-scale information
- A Bayesian approach to multiscale inverse problems using the sequential Monte Carlo method
- A Family of Nonparametric Density Estimation Algorithms
- A Newton-CG method for large-scale three-dimensional elastic full-waveform seismic inversion
- A locally conservative variational multiscale method for the simulation of porous media flow with multiscale source terms
- A multiscale strategy for Bayesian inference using transport maps
- Adaptive fully implicit multi-scale finite-volume method for multi-phase flow and transport in heterogeneous porous media
- Analysis of the Hessian for inverse scattering problems. I: Inverse shape scattering of acoustic waves
- Bayesian data analysis.
- Bayesian inference with optimal maps
- Coarse-gradient Langevin algorithms for dynamic data integration and uncertainty quantification
- Computational Methods for Inverse Problems
- Data Assimilation
- Density estimation by dual ascent of the log-likelihood
- Dimension-independent likelihood-informed MCMC
- Dimensionality reduction and polynomial chaos acceleration of Bayesian inference in inverse problems
- Distributed optimization and statistical learning via the alternating direction method of multipliers
- Dynamics of fluids in porous media.
- Existence and uniqueness of monotone measure-preserving maps
- Fine scale uncertainty in parameter estimation for elliptic equations
- Handbook of Markov Chain Monte Carlo
- Heterogeneous multiscale methods: a review
- Inverse Problem Theory and Methods for Model Parameter Estimation
- Inverse problems: a Bayesian perspective
- Langevin diffusions and Metropolis-Hastings algorithms
- Large‐Scale Inverse Problems and Quantification of Uncertainty
- Likelihood-informed dimension reduction for nonlinear inverse problems
- Mixed Multiscale Finite Element Methods for Stochastic Porous Media Flows
- Monte Carlo errors with less errors
- Multi-scale finite-volume method for elliptic problems in subsurface flow simulation.
- Multiscale Finite Element Methods
- Multiscale Methods
- Multiscale modelling and inverse problems
- On Augmented Lagrangian Methods with General Lower-Level Constraints
- On the implementation of an interior-point filter line-search algorithm for large-scale nonlinear programming
- Optimal Transport
- Optimal low-rank approximations of Bayesian linear inverse problems
- Polar factorization and monotone rearrangement of vector‐valued functions
- Preconditioning Markov Chain Monte Carlo Simulations Using Coarse-Scale Models
- Probability Theory
- Statistical and computational inverse problems.
- The variational multiscale method -- a paradigm for computational mechanics
- Transport map accelerated Markov chain Monte Carlo
- Variational Multiscale Analysis: the Fine‐scale Green’s Function, Projection, Optimization, Localization, and Stabilized Methods
- Variational multiscale analysis: the fine-scale Green's function for stochastic partial differential equations
Cited in
(28)- Residual-based error correction for neural operator accelerated Infinite-dimensional Bayesian inverse problems
- Inference via low-dimensional couplings
- Bayesian inference with optimal maps
- Dynamic data-driven Bayesian GMsFEM
- Transport map accelerated adaptive importance sampling, and application to inverse problems arising from multiscale stochastic reaction networks
- Some models are useful, but how do we know which ones? Towards a unified Bayesian model taxonomy
- A Bayesian approach to multiscale inverse problems with on-the-fly scale determination
- Bayesian nonparametric generative modeling of large multivariate non-Gaussian spatial fields
- A Riemann-Stein kernel method
- A Geometric Approach to the Transport of Discontinuous Densities
- Multilevel Monte Carlo for smoothing via transport methods
- Bayesian inference of non-linear multiscale model parameters accelerated by a deep neural network
- A multiscale strategy for Bayesian inference using transport maps
- Inference With Aggregate Data in Probabilistic Graphical Models: An Optimal Transport Approach
- Bayesian inference using intermediate distribution based on coarse multiscale model for time fractional diffusion equations
- Transport map accelerated Markov chain Monte Carlo
- Operator shifting for noisy elliptic systems
- Low-rank tensor reconstruction of concentrated densities with application to Bayesian inversion
- hIPPYlib-MUQ: a Bayesian inference software framework for integration of data with complex predictive models under uncertainty
- Data-driven projection pursuit adaptation of polynomial chaos expansions for dependent high-dimensional parameters
- Gibbs Flow for Approximate Transport with Applications to Bayesian Computation
- A distributed framework for the construction of transport maps
- Beyond the model limit: parameter inference across scales
- Scalable Bayesian Transport Maps for High-Dimensional Non-Gaussian Spatial Fields
- Iterative importance sampling algorithms for parameter estimation
- On Large Lag Smoothing for Hidden Markov Models
- Multiscale sampling for the inverse modeling of partial differential equations
- A Continuation Method in Bayesian Inference
This page was built for publication: A multiscale strategy for Bayesian inference using transport maps
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3179325)
