Make KroneckerProductLazyTensor call each component's root decomposition #1394
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Outdated suggestions cannot be applied.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.
For PSD tensors, we can write the root decomposition of a Kronecker product,
M = A \otimes B = LL' \otimes NN' = (L \otimes N)(L' \otimes N')
for general root decompositions rather than just the
symeig
case which is implemented currently.This allows for faster and less memory intensive factorizations of very large Kronecker products as it gives the components the ability to use Lanczos.
Note: I also had to slightly modify the batched KP lazy tensor test case to get lanczos to converge here due to the repeated eigenvalue.