This hotfix branch fixes a bug where the flag --eig-io-parity-inflate yes (more specifically, the underlying variable set in the eigensolver parameters structure) wasn't being respected when saving vectors from single-parity operators.
As part of fixing this bug, I extended the io_test unit tests to explicitly check parity inflation for both even and odd vectors. The test to save an inflated single-parity field, load the resulting full field, and make sure the even or odd subset of the loaded full field (as appropriate) matches the saved field.
This bug was originally reported by @leonhostetler, who has independently verified that this fix addresses the issue he was hitting in his workflow --- thank you for catching it!
(FYI @maddyscientist --- I clang-formated this along the way, so there's no need for a subsequent clang-format commit.)
This hotfix branch fixes a bug where the flag
--eig-io-parity-inflate yes
(more specifically, the underlying variable set in the eigensolver parameters structure) wasn't being respected when saving vectors from single-parity operators.As part of fixing this bug, I extended the
io_test
unit tests to explicitly check parity inflation for both even and odd vectors. The test to save an inflated single-parity field, load the resulting full field, and make sure the even or odd subset of the loaded full field (as appropriate) matches the saved field.This bug was originally reported by @leonhostetler, who has independently verified that this fix addresses the issue he was hitting in his workflow --- thank you for catching it!
(FYI @maddyscientist --- I
clang-format
ed this along the way, so there's no need for a subsequentclang-format
commit.)