Replies: 2 comments 2 replies
-
@klapo Thanks for investigating this! So unless I'm mistaken, I believe this is working as intended? The real components shouldn't be being added to the imaginary components due to the multiplication by
Though I might be misunderstanding the issue. Do you have any examples/test cases that illustrate your issues? |
Beta Was this translation helpful? Give feedback.
-
@sichinaga Thank you for the explanation! Based on a couple quick tests (although very possibly wrong) the No need to apology about the delay! Otherwise I'll need to apologize about my delay as well 😓 |
Beta Was this translation helpful? Give feedback.
-
@sichinaga I am a little unsure if I am misunderstanding the eigenvalue sorting options for bopdmd.
The documentation of
BOPDMDOperator
states the eigenvalue sorting follows this logic: "Eigenvalues will be sorted by real part and then by imaginary part to break ties ifeig_sort="real"
, by imaginary part and then by real part to break ties ifeig_sort="imag"
, or magnitude ifeig_sort="abs"
.The options "abs" and "real" behave as intended. But, the order back from
eig_sort="imag"
doesn't behave as I would have guessed.The code for sorting by the imaginary components is:
The result of this logic is that real components are added to the imaginary components. I get modes with a large real eigenvalue but a small imaginary eigenvalue (i.e., low frequency) being misplaced in the sorting relative to my expectations. I am unsure if I am misunderstanding the intent of the documentation or if this is a bug. Any feedback?
Beta Was this translation helpful? Give feedback.
All reactions