dflemin3 / approxposterior

A Python package for approximate Bayesian inference and optimization using Gaussian processes
https://dflemin3.github.io/approxposterior/
MIT License
41 stars 9 forks source link

MAINT: merging quality of life changes from dev to master #47

Closed dflemin3 closed 5 years ago

dflemin3 commented 5 years ago

This PR adds various quality-of-life and stability improvements to approxposterior by addressing the following issues: #44 #41 #38. Additionally, this PR sets the stage for user-defined white_noise objects to add terms to the diagonal of the GP covariance matrix for numerical stability. This PR is preparing for future changes to better initialize GP hyperparameter optimization, including fitting for the GP amplitude which has so far not been included because of numerical stability issues.