Add ROHF GDM#453
Open
v-boqinzhang wants to merge 37 commits into
Open
Conversation
v-boqinzhang
requested review from
ConradJohnston,
nabbelbabbel,
sobolevnrm and
wavefunction91
as code owners
Closed
Contributor
There was a problem hiding this comment.
Pull request overview
This PR enables ROHF (restricted open-shell) support in the SCF GDM/DIIS_GDM convergence path by adding ROHF-specific convergence matrix handling to the SCFAlgorithm base and extending GDM to handle ROHF orbital rotations/gradients.
Changes:
- Allow ROHF runs to use
SCFAlgorithmName::GDMandSCFAlgorithmName::DIIS_GDM(previously blocked). - Centralize ROHF “effective Fock + total density” convergence-matrix caching in
SCFAlgorithmand reuse it from DIIS/convergence checks. - Add ROHF-specific kappa packing, orbital rotation, pseudo-canonical transformations, and gradient computation in GDM.
Reviewed changes
Copilot reviewed 8 out of 8 changed files in this pull request and generated 3 comments.
Show a summary per file
| File | Description |
|---|---|
| cpp/tests/test_scf.cpp | Removes the test that asserted ROHF+GDM is invalid (no replacement added). |
| cpp/src/qdk/chemistry/algorithms/microsoft/scf/src/scf_algorithm/scf_algorithm.cpp | Enables ROHF+GDM/DIIS_GDM in the factory; adds ROHF convergence-matrix caching in base algorithm. |
| cpp/src/qdk/chemistry/algorithms/microsoft/scf/src/scf_algorithm/gdm.cpp | Implements ROHF-specific rotations, gradients, and pseudo-canonical/Hessian logic for GDM line search + iteration. |
| cpp/src/qdk/chemistry/algorithms/microsoft/scf/src/scf_algorithm/diis.h | Removes DIIS-owned ROHF cache API and updates internal helper signature. |
| cpp/src/qdk/chemistry/algorithms/microsoft/scf/src/scf_algorithm/diis.cpp | Switches DIIS ROHF handling to use SCFAlgorithm’s shared ROHF convergence cache. |
| cpp/src/qdk/chemistry/algorithms/microsoft/scf/src/scf/scf_impl.h | Exposes core Hamiltonian accessor and a J/K build helper for use by GDM ROHF gradient code. |
| cpp/src/qdk/chemistry/algorithms/microsoft/scf/src/scf/scf_impl.cpp | Updates trial-density Fock build to treat ROHF like unrestricted; implements build_jk_matrices. |
| cpp/src/qdk/chemistry/algorithms/microsoft/scf/include/qdk/chemistry/scf/core/scf_algorithm.h | Adds ROHF convergence-matrix hook + cache accessors and stores ROHF caches in base class. |
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
📊 Coverage Summary
Detailed Coverage ReportsC++ Coverage DetailsPython Coverage DetailsPybind11 Coverage Details |
Contributor
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 9 out of 9 changed files in this pull request and generated 3 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Contributor
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 11 out of 11 changed files in this pull request and generated 8 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Contributor
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 12 out of 12 changed files in this pull request and generated 3 comments.
💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.
Closed
Contributor
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 12 out of 12 changed files in this pull request and generated 3 comments.
Comments suppressed due to low confidence (2)
cpp/src/qdk/chemistry/algorithms/microsoft/scf/src/scf_algorithm/gdm.cpp:1261
- This calls
ERI::build_JK()from insideGDM::iterate(), butSCFImpl::iterate_()invokes the SCF algorithm only onworld_rank == 0, whileERI::build_JK()performs collective MPI reductions whenworld_size > 1. As a result, ROHF GDM will hang under MPI because the non-root ranks never enter this call. The ROHF GDM path should either coordinate the JK build across all ranks or be rejected when MPI is enabled.
scf_impl.build_jk_matrices(scf_impl.get_density_matrix(), J_ao, K_ao);
cpp/src/qdk/chemistry/algorithms/microsoft/scf/src/scf_algorithm/scf_helper.h:31
- This documentation says callers may pass arbitrary sub-block starts, but
compute_atba_gemm()hardcodes the leading dimensions frommandnand has no way to accept the parent matrix stride. That is only correct for contiguous/full-width blocks; a true interior row-major sub-block would be read with the wrong stride. Please either restrict the contract to contiguous blocks or add leading-dimension parameters.
📊 Coverage Summary
Detailed Coverage ReportsC++ Coverage DetailsPython Coverage DetailsPybind11 Coverage Details |
Contributor
There was a problem hiding this comment.
Pull request overview
Copilot reviewed 12 out of 12 changed files in this pull request and generated 1 comment.
Comments suppressed due to low confidence (1)
cpp/src/qdk/chemistry/algorithms/microsoft/scf/src/scf_algorithm/scf_helper.h:33
- The
compute_atba_gemmdoc says the input pointers may reference sub-block starts, but the implementation hard-codes leading dimensions (e.g.,lda = nfor RowMajor). That only works when the referenced blocks are stored contiguously with those exact strides; for typical Eigen sub-blocks from a larger matrix, the stride is the parent’s column count and the result would be incorrect. Please either (a) tighten the doc to state the required memory layout/stride constraints, or (b) extend the helper to accept explicit leading dimensions (lda/ldb/ldc) so true sub-block pointers are supported.
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
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.
No description provided.