Accepting circuits as initial states #817
Conversation
andrea-pasquale
requested review from
scarrazza,
stavros11 and
Edoardo-Pedicillo
andrea-pasquale
changed the title
Codecov ReportPatch coverage:
Additional details and impacted files@@ Coverage Diff @@
## measurebasis #817 +/- ##
==============================================
Coverage 100.00% 100.00%
==============================================
Files 97 97
Lines 12930 12953 +23
==============================================
+ Hits 12930 12953 +23
Flags with carried forward coverage won't be shown. Click here to find out more.
Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here. ☔ View full report at Codecov. |
| initial state for circuit with density_matrix {circuit.density_matrix}.""", | ||
| ) | ||
| else: | ||
| return self.execute_circuit(initial_state + circuit, None, nshots) |
There was a problem hiding this comment.
Could you please update the documentation with this sum operation?
There was a problem hiding this comment.
I've added the docstring. Thanks for pointing this out.
andrea-pasquale
changed the title
There was a problem hiding this comment.
Thanks for implementing this, looks good.
I was thinking that if you delegate the check if initial state is circuit in Circuit.execute we would not have to update both the numpy and qibolab backends. However, passing a circuit initial state would not work if one tries to use backend.execute_circuit(...) directly and this would create an assymetry. I think your solution is better.
| @@ -331,6 +331,16 @@ def thermal_error_density_matrix(self, gate, state, nqubits): | |||
| def execute_circuit( | |||
| self, circuit, initial_state=None, nshots=None, return_array=False | |||
| ): | |||
| if isinstance(initial_state, type(circuit)): | |||
| if not initial_state.density_matrix == circuit.density_matrix: | |||
There was a problem hiding this comment.
Looking at Circuit.__init__ another attribute that initial_state and circuit may be different is accelerators. I find it highly unlikely that someone will try to use this feature with multigpu, but maybe for consistency we could add the same check here.
There was a problem hiding this comment.
Thanks for catching this. I've just added a check also for accelerators.
This PR implements the following point:
from #720 and PR #799.
I will open a PR in qibolab to support this feature.
Checklist: