Skip to content

merge v4 into master: release 4.0.2 - #669

Merged
mostofi merged 1678 commits into
masterfrom
develop
Aug 27, 2026
Merged

merge v4 into master: release 4.0.2#669
mostofi merged 1678 commits into
masterfrom
develop

Conversation

@JeromeCCP9

Copy link
Copy Markdown
Collaborator

No description provided.

JeromeCCP9 and others added 30 commits February 2, 2026 16:52
specific to transl_inv_full case.  Change history too complex to
explicity revert
previous test sources were generated with the sorted bk vectors, which
change was since regressed
wann_phases() in Stengel Spalding case needs to be tested

new benchmark required for testw90_knbo3_higher_stengel_spaldin test

this commit causes combination of keywords to be an error
Signed-off-by: Cristian Le <cristian.le@mpsd.mpg.de>
From Seung-Ju Hong  (sjhong6230): see issue #586 and PR #584

Test testw90_knbo3_higher_stengel_spaldin reinstated with new reference data
replace fortran binary spn test data with portable ascii
…etry - as they are now only used with the site symmetry path
i.e. incorrect behaviour when ifx combined with (GCC compiled) Openblas

fix replaces zdotc function (return result passing being the problem) with zgemv subroutine
avoid Intel/GCC BLAS ABI incompatibility by avoiding blas functions
JeromeCCP9 and others added 22 commits July 28, 2026 10:33
all k-points assigned to rank 0, serialising the calculation instead of
curent duplication.

corrects a bug that non-root ranks did not have a copy of the default dist-k
distribution array.
The projectability-windows banner again reports the disentangled window
as [dis_proj_min, 1.0], as on develop: frozen high-projectability states
are part of the disentanglement window (PR #650 review). The
testw90_graphene_pdwf benchmark is restored to the develop version and
the testw90_graphene_dis_proj_auto benchmark regenerated.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
- dis_proj_min/dis_proj_max no longer have default values: they are
  either set explicitly or determined automatically
- explicit thresholds disable the (defaulted) automatic thresholding;
  combining them with an explicit dis_proj_auto = .true. is an input
  error
- dis_froz_proj with automatic thresholding disabled requires both
  thresholds; providing only one of them is an input error
- dis_proj_auto without dis_froz_proj is no longer an error (it simply
  has no effect)
- dis_proj_auto_classes is renamed to dis_proj_auto_num_classes, and
  the local flags recording which projectability keywords were provided
  are given descriptive names
- documentation, parameter tables and CHANGELOG updated; the CHANGELOG
  entry is shortened to match the style of existing entries

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The looser omegaI/omegaOD tolerances were the only inline tolerance
override in jobconfig; they now live in userconfig as a named program
section, as for the other tests that need non-default tolerances.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Automatic thresholds for projectability disentanglement (`dis_proj_auto`)
"or later" was missing in the license update description
Update CHANGELOG.md with correction to license update
@JeromeCCP9
JeromeCCP9 requested a review from mostofi August 24, 2026 08:22
@JeromeCCP9

Copy link
Copy Markdown
Collaborator Author

Let me add #663 also--these are simply corrections and belong in this release.

@mostofi
mostofi merged commit 3d69500 into master Aug 27, 2026
11 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.