Skip to content

Reference implementation of an automatic approach to regularisation for linear(ised) inverse problems, described in Valentine & Sambridge (2020) doi:10.1093/gji/ggy303

License

Notifications You must be signed in to change notification settings

valentineap/optimal-regularisation

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

9 Commits
 
 
 
 
 
 

Repository files navigation

optimal-regularisation

This project provides a reference implementation of the algorithms described in:

Valentine, A.P. and Sambridge, M., Optimal regularisation for a class of linear inverse problem, Geophysical Journal International, 215(2), pp.1003--1021, 2018.

Specifically, these algorithms are intended to solve linear regression problems of the form

minimize_m ||Gm - d||^2 +||D(eps) m||^2

where D(eps) is a regularisation operator that depends on one or more tuneable parameters, eps. We use a hierarchical Bayesian approach to select the 'optimal' choice of eps. In addition to a general-purpose algorithm, we provide one that is substantially more efficient for the common case where the regularisation operator is of Tikhonov form. For full details, please refer to the paper.

About

Reference implementation of an automatic approach to regularisation for linear(ised) inverse problems, described in Valentine & Sambridge (2020) doi:10.1093/gji/ggy303

Resources

License

Stars

Watchers

Forks

Packages

No packages published

Languages