probml / pyprobml

Python code for "Probabilistic Machine learning" book by Kevin Murphy
MIT License
6.45k stars 1.52k forks source link

Add GHKF plots to ekf_vs_ukf.ipynb #1083

Closed petergchang closed 2 years ago

petergchang commented 2 years ago

Description

Added GHKF plots to ekf_vs_ukf.ipynb.

Figure Number

book2 - 8.12

Figures

ghkf_source ghkf_target

Issue

1082

Checklist

review-notebook-app[bot] commented 2 years ago

Check out this pull request on  ReviewNB

See visual diffs & provide feedback on Jupyter Notebooks.


Powered by ReviewNB

murphyk commented 2 years ago

Can you please make helper functions

(m,S,C) = ukf.transform(non_linear_transform, mean_pre, cov_pre)
(m,S,C) = ghkf.transform(non_linear_transform, mean_pre, cov_pre)

that wraps up some of the low level logic (I think you can reuse https://github.com/probml/ssm-jax/blob/main/ssm_jax/ggssm/containers.py#L61). Something like this:

Screen Shot 2022-07-27 at 12 39 49 PM

I realize that you may also want to return the transformed sigma points, for plotting, so these could be optional return args.