Controlled GlobalPhase is ZGate #1113
Conversation
| exponent: float = np.real_if_close(np.log(self.coefficient) / (np.pi * 1.0j)).item() # type: ignore | ||
| bloq = ZPowGate(exponent=exponent, eps=self.eps) |
There was a problem hiding this comment.
Please handle the case when self.coefficient is symbolic
There was a problem hiding this comment.
Should we also have an override where we return a ZGate instead of ZPowGate if exponent is 1 ?
There was a problem hiding this comment.
That's a good question, and I think it's part of a broader issue. How do we handle cases where you can express the same "thing" (unitary) via different bloq+attribute combinations? I think the answer is to keep things as non-magical as possible when constructing bloqs and handle overlap in the bloq analysis functions. Happy to discuss more.
For this PR in particular: since ZPowGate(exponent=1) is a valid bloq, I'll leave as-is.
I'll handle symbolic coefficient
| return super().get_ctrl_system(ctrl_spec=ctrl_spec) | ||
|
|
||
| # Otherwise, it's a ZPowGate | ||
| exponent: float = np.real_if_close(np.log(self.coefficient) / (np.pi * 1.0j)).item() # type: ignore |
There was a problem hiding this comment.
this wouldn't work when self.coefficient is symbolic. Maybe we should update this to self.exponent to match the other rotation gates?
There was a problem hiding this comment.
Thanks.
Did you want to change the semantics to just store the exponent directly as the attribute or just change the name?
There was a problem hiding this comment.
the former, then we can safely pass self.exponent to ZPowGate. But would require updating all call sites. Based on a preliminary scan, most of them are 1j, -1j, -1 and would simplify SU2:
I'm happy to take this PR if you'd like.
|
can be closed: this change was merged #1117 |
|
Thank you! |
The single-bit control of a global phase is
ZGate.More basic gates #1109