Skip to content

Inverse-Modeling/EB_genGK

Repository files navigation

Empirical Bayes based on genGK to estimate hyperparameters

This repository contains MATLAB files for an empirical Bayes method to estimate hyperparameters using an approach based on the generalized Golub-Kahan (genGK) bidiagonalization. The codes accompany the paper:

"Efficient iterative methods for hyperparameter estimation in large-scale linear inverse problems"

  • Hall-Hooper, Saibaba, Chung, and Miller, ACOM, 2024

Project Description

We implement an empirical Bayes (EB) method to estimate hyperparameters 
that maximize the marginal posterior, i.e., the probability density of 
the hyperparameters conditioned on the data.

For problems where the computation of the square root and inverse of 
prior covariance matrices are not feasible, we use an approach based on
the generalized Golub-Kahan bidiagonalization to approximate the 
marginal posterior and seek hyperparameters that minimize the 
approximate marginal posterior. 

Installation

Software language

   MATLAB 9.14 (R2023a)

Requirements

Requirements

These codes require the following packages:

     Regularization Tools package: Hansen. Regularization tools: A
         package for analysis and solution of discrete ill-posed 
         problems. Numerical Algorithms, 1994.
         
     genHyBR: generalized hybrid iterative methods
         by Julianne Chung and Arvind K. Saibaba
         https://github.com/juliannechung/genHyBR

How to Use

See Contents.m

Contributors

    Khalil A. Hall-Hooper
    Department of Mathematics, North Carolina State University

    Arvind K. Saibaba, 
    Department of Mathematics, North Carolina State University
    
    Julianne Chung, 
    Department of Mathematics, Emory University
    
    Scot M. Miller, 
    Department of Environmental Health and Engineering, Johns Hopkins University

Licensing

If you use this codes, you must cite the original authors:

   [1] Hall-Hooper et al. "Efficient iterative methods for 
        hyperparameter estimation in large-scale linear 
        inverse problems". ACOM, 2024.

MIT

Acknowledgement

This work was partially supported by the National Science Foundation under grants DMS-2208294, DMS-2341843, DMS-2026830, and DMS-2026835. Any opinions, findings, conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation.

About

No description, website, or topics provided.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors

Languages