Open isaacdevlugt opened 1 month ago
Oh that's good to know thanks 😬 I just suggested to UF they use this feature for easier device-based noise testing, where MCM's wouldn't necessarily be present. (seems like it only affects circuits with multiple results though)
Speaking of, it does make me wonder wether we should think of this feature as an "MCM" feature, rather than just an "execution mode" feature. @josh146
@dime10 just double checking since I missed the iteration planning meeting -- are we in the process of resolving this one this iteration?
Didn't realize that my traceback section was improperly copy-pasted from VS code 😅. Fixed!
@dime10 just double checking since I missed the iteration planning meeting -- are we in the process of resolving this one this iteration?
This is one hasn't been scheduled yet because #929 seems more pressing.
Issue description
Specifying
mcm_method="one-shot"
when no MCMs are present in a qjit'd QNode raises an error when the return type is an iterable.Expected behavior: Expect
mcm_method="one-shot"
to not affect a circuit without any MCMs.Actual behavior:
ValueError: Incompatible shapes for broadcasting: shapes=[(10, 3), (1, 0)]
Reproduces how often: 100%
System information:
Source code and tracebacks
Traceback:
Additional information