save unnecessary matmul#30
Merged
Merged
Conversation
Signed-off-by: Hao Wu <skyw@nvidia.com>
skyw
force-pushed
the
skyw/optimize_soap
branch
from
990a47a to
e3bf176
Compare
Contributor
Author
|
/ok to test e3bf176 |
| # (i.e. the approximated eigenvectors diagonalize the kronecker factor) | ||
| approx_eigenvalue_matrix = eigenbasis.T @ kronecker_factor @ eigenbasis | ||
| # Update eigenbasis when necessary. Update is skipped only when adaptive update criteria is met. | ||
| if _adaptive_criteria_met( |
Contributor
There was a problem hiding this comment.
_adaptive_criteria_met also extracts the diagonal and computes diagonal norm, in addition to matrix frobenius norm. So I think this will need the full approx_eigenvalue_matrix diagonal for diagonal norm and the kronecker factor for frobenius norm, separately.
Signed-off-by: Hao Wu <skyw@nvidia.com>
Signed-off-by: Hao Wu <skyw@nvidia.com>
Signed-off-by: Hao Wu <skyw@nvidia.com>
Contributor
Author
|
/ok to test 7265ff4 |
mkhona-nvidia
pushed a commit
to mkhona-nvidia/Emerging-Optimizers
that referenced
this pull request
Oct 3, 2025
* save unnecessary matmul Signed-off-by: Hao Wu <skyw@nvidia.com> * simplify criteria logic Signed-off-by: Hao Wu <skyw@nvidia.com> * remove max precondition dim Signed-off-by: Hao Wu <skyw@nvidia.com> Signed-off-by: mikail <mkhona@nvidia.com>
mkhona-nvidia
pushed a commit
to mkhona-nvidia/Emerging-Optimizers
that referenced
this pull request
Oct 3, 2025
* save unnecessary matmul Signed-off-by: Hao Wu <skyw@nvidia.com> * simplify criteria logic Signed-off-by: Hao Wu <skyw@nvidia.com> * remove max precondition dim Signed-off-by: Hao Wu <skyw@nvidia.com> Signed-off-by: mikail <mkhona@nvidia.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
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.
Only diagonal of approx_eigenvalue_matrix is needed for _orthogonal_iteration, added logic to skip the last matrix multiply.
Frobenius norm is unitarily invariant, that is for any orthogonal matrix Q and square matrix A ,$||Q^T A Q||_F = ||A||_F$ . So we can calculate norm on kronecker_factor instead of approx_eigenvalue_matrix.