Support ActOnStabilizerCHFormArgs in the common gates and testing #3203
Conversation
There was a problem hiding this comment.
Hey @smitsanghavi, thanks for submitting this! I started reviewing it! I will need to dig deeper in the next couple of days but I found some nits to cleanup in the meantime.
smitsanghavi
changed the title
| assert all( | ||
| gate._act_on_(args) # type: ignore | ||
| for gate in [ZPowGate(), HPowGate()]) | ||
| state.omega *= (1 + 1j) / (2**0.5) # type: ignore |
There was a problem hiding this comment.
Why is the type ignore needed?
There was a problem hiding this comment.
mypy assumes state.omega is an int since it was initialized by an int. It complains when I try to assign a complex value to it. Maybe adding + 0j to the initialization will fix it. Let me know if you think that's better.
There was a problem hiding this comment.
Left a bunch of comments!
Also, I'm a bit worried about the fact that phases just go unnoticed...
In assert_all_implemented_act_on_effects_match_unitary shouldn't we use
np.testing.assert_allclose(np.reshape(
stabilizer_ch_form.state_vector(), protocols.qid_shape(qubits)),
state_vector,
atol=1e-07, err_msg="inconsistent stabilizer ch form")
instead of
testing.assert_allclose_up_to_global_phase(np.reshape(
stabilizer_ch_form.state_vector(), protocols.qid_shape(qubits)),
state_vector,
atol=1e-07)
?
One of the benefits of the CH-form is that it is phase sensitive. Why are we checking unitary consistency in a phase insensitive way?
That makes sense. There were some tests adding extra global phases that were breaking it previously. Added support for |
balopat
added
automerge
CirqBot
added
the
front_of_queue_automerge
balopat
removed
automerge
Also some minor bug fixes and refactoring in StabilizerStateCHForm, ActOnCliffordTableauArgs and consistent_act_on.
#2948 #2423