Skip to content

Avoid dense matrices in chol_cap_mat for ConstantDiagLinearOperator A #120

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Osburg wants to merge 1 commit into
base: main
Choose a base branch
from
Open

Avoid dense matrices in chol_cap_mat for ConstantDiagLinearOperator A #120

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 5 additions & 1 deletion linear_operator/operators/low_rank_root_added_diag_linear_operator.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -40,7 +40,11 @@ def chol_cap_mat(self):
V = self._linear_op.root.mT
C = ConstantDiagLinearOperator(torch.ones(*V.batch_shape, 1, device=V.device, dtype=V.dtype), V.shape[-2])

cap_mat = to_dense(C + V.matmul(A_inv.matmul(U)))
if isinstance(self._diag_tensor, ConstantDiagLinearOperator):
sigma_inv = A_inv.diag_values[0]
cap_mat = to_dense(C + sigma_inv * V.matmul(U))
else:
cap_mat = to_dense(C + V.matmul(A_inv.matmul(U)))
Copy link
Collaborator

@kayween kayween Jan 13, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Note that A_inv here is a DiagLinearOperator. Thus, A_inv.matmul(U) basically scales each row of U by corresponding entries in A_inv.

chol_cap_mat = psd_safe_cholesky(cap_mat)

return chol_cap_mat
Expand Down