Add block Lanczos quadrature - #193
Open
rileyjmurray wants to merge 1 commit into
Open
rileyjmurray wants to merge 1 commit into
rileyjmurray wants to merge 1 commit into
Conversation
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
BlockLanczosQFAapproximatesB^T f(A) Busing the block Lanczos recurrence, avoiding reconstruction of the large matrixf(A)B. It supports fixed depth, a heuristic window stop, and a Gauss–Radau bracket check with explicit certification status. The bracket interpretation depends on the function and spectral assumptions; the heuristic stop supplies no certificate.This extracts the block quadrature component from #138. It is based on #192 and reuses the block recurrence introduced in #191. The optional banded eigensolver is in #195.
Validation. All 537 core Release tests passed locally with Spack Clang 19.1.3 on macOS. Tests compare with independent dense matrix functions and scalar quadrature, and exercise depth caps, failed positive-definiteness checks, stopping controls, and workspace reuse.