Cholesky decomposition for ExactGP class #115
Open
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Hi,
Really useful library! I'm using GPs for some large scale problems and wanted to be able to use your code. For exact GPs, it is well known that Cholesky decomposition is a better alternative than the base matrix inversion. I think the NumPyro example, which the ExactGP class borrows from, also reflects this now. Just added this slight change.
Let me know if I missed something. Testing the new
test_gp.py
runs successfully on my end.I'm not sure if other parts of the codebase would also benefit from this.