Likelihood-informed dimension reduction for nonlinear inverse problems
arXiv:1403.4680 · doi:10.1088/0266-5611/30/11/114015
Abstract
The intrinsic dimensionality of an inverse problem is affected by prior information, the accuracy and number of observations, and the smoothing properties of the forward operator. From a Bayesian perspective, changes from the prior to the posterior may, in many problems, be confined to a relatively low-dimensional subspace of the parameter space. We present a dimension reduction approach that defines and identifies such a subspace, called the "likelihood-informed subspace" (LIS), by characterizing the relative influences of the prior and the likelihood over the support of the posterior distribution. This identification enables new and more efficient computational methods for Bayesian inference with nonlinear forward models and Gaussian priors. In particular, we approximate the posterior distribution as the product of a lower-dimensional posterior defined on the LIS and the prior distribution marginalized onto the complementary subspace. Markov chain Monte Carlo sampling can then proceed in lower dimensions, with significant gains in computational efficiency. We also introduce a Rao-Blackwellization strategy that de-randomizes Monte Carlo estimates of posterior expectations for additional variance reduction. We demonstrate the efficiency of our methods using two numerical examples: inference of permeability in a groundwater system governed by an elliptic PDE, and an atmospheric remote sensing problem based on Global Ozone Monitoring System (GOMOS) observations.
References in corpus (2)
Cited by in corpus (52)
- Quantifying and Reducing Model-Form Uncertainties in Reynolds-Averaged Navier-Stokes Simulations: A Data-Driven, Physics-Based Bayesian Approach
- Dimension-independent likelihood-informed MCMC
- An introduction to sampling via measure transport
- Geometric MCMC for Infinite-Dimensional Inverse Problems
- Inverse modeling of nonisothermal multiphase poromechanics using physics-informed neural networks
- Accelerating MCMC with active subspaces
- Scalable posterior approximations for large-scale Bayesian inverse problems via likelihood-informed parameter and state reduction
- A Stein variational Newton method
- Bayesian inference of heterogeneous epidemic models: Application to COVID-19 spread accounting for long-term care facilities
- Survey of multifidelity methods in uncertainty propagation, inference, and optimization
- Deep composition of tensor-trains using squared inverse Rosenblatt transports
- Bayesian, frequentist, and information geometric approaches to parametric uncertainty quantification of classical empirical interatomic potentials
- A multiscale strategy for Bayesian inference using transport maps
- Randomized Truncated SVD Levenberg-Marquardt Approach to Geothermal Natural State and History Matching
- Adaptive Dimension Reduction to Accelerate Infinite-Dimensional Geometric Markov Chain Monte Carlo
- Strategic model reduction by analysing model sloppiness: a case study in coral calcification
- Data-Free Likelihood-Informed Dimension Reduction of Bayesian Inverse Problems
- Greedy inference with structure-exploiting lazy maps
- Accelerated dimension-independent adaptive Metropolis
- Prior normalization for certified likelihood-informed subspace detection of Bayesian inverse problems
- Inference via low-dimensional couplings
- hIPPYlib: An Extensible Software Framework for Large-Scale Inverse Problems Governed by PDEs; Part I: Deterministic Inversion and Linearized Bayesian Inference
- Importance Sampling: Intrinsic Dimension and Computational Cost
- A low-rank ensemble Kalman filter for elliptic observations
- A Data-Scalable Randomized Misfit Approach for Solving Large-Scale PDE-Constrained Inverse Problems
- Stochastic Subspace Descent
- Replica Exchange for Non-Convex Optimization
- Multilevel Dimension-Independent Likelihood-Informed MCMC for Large-Scale Inverse Problems
- High-dimensional Nonlinear Bayesian Inference of Poroelastic Fields from Pressure Data
- Ensemble sampler for infinite-dimensional inverse problems
- Bayesian non-parametric inference for -coalescents: consistency and a parametric method
- An adaptive independence sampler MCMC algorithm for infinite dimensional Bayesian inferences
- Generalized bounds for active subspaces
- Scalable optimization-based sampling on function space
- A randomized maximum a posterior method for posterior sampling of high dimensional nonlinear Bayesian inverse problems
- Analysis of sloppiness in model simulations: unveiling parameter uncertainty when mathematical models are fitted to data
- Multilevel Sequential Monte Carlo with Dimension-Independent Likelihood-Informed Proposals
- A stochastic subspace approach to gradient-free optimization in high dimensions
- A probabilistic framework for approximating functions in active subspaces
- Likelihood informed dimension reduction for inverse problems in remote sensing of atmospheric constituent profiles
- Reduced Modeling of Unknown Trajectories
- Optimal and scalable methods to approximate the solutions of large-scale Bayesian problems: Theory and application to atmospheric inversions and data assimilation
- Iterative importance sampling algorithms for parameter estimation
- Goal-oriented optimal approximations of Bayesian linear inverse problems
- On Unifying Randomized Methods For Inverse Problems
- Optimal low-rank posterior mean and distribution approximation in linear Gaussian inverse problems on Hilbert spaces
- Sharp detection of low-dimensional structure in probability measures via dimensional logarithmic Sobolev inequalities
- Bayesian inverse regression for dimension reduction with small datasets
- Uncertainty quantification for ptychography using normalizing flows
- Reduced Order Modeling using Shallow ReLU Networks with Grassmann Layers
- Localization for MCMC: sampling high-dimensional posterior distributions with local structure
- Spatial localization for nonlinear dynamical stochastic models for excitable media