Report Approximate Truncation Error from Apply Function#232
Report Approximate Truncation Error from Apply Function#232mtfishman merged 16 commits intoITensor:mainfrom
Conversation
|
@JoeyT1994 take a look at the callback design in |
|
@mtfishman Oh yeah that's quite a nice way to do it. I have updated the code to use a callback for the |
|
@mtfishman I can't say I understand the reason for the failing tests. I am not seeing anything locally, it seems to be erroring here when running |
|
Looks good, thanks! I have no idea what's going on with those tests, I've seen that issue randomly but only in this package. Maybe one of the tests uses a lot of memory? |
|
Yeah maybe next time we see it I can try to remove the test that causes it |
2 participants
This small PR allows passing a
truncation_error!reference to theapplyfunction in the case of the simple update routine. This will be filled in with thetruncerrfrom the corresponding call tofactorize_svdallowing one to keep track of the approximate error induced by the simple update routine.Note that this is not possible to infer solely from the
singular_values!reference object due to the fact that the spectrum is not normalized byfactorize_svd!.Tests are updated in
apply_test.jlto make sure the functionality is working.