You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Make all GeminiOneZoneModels use correlated noise (#459)
Implements #448.
The test on the noisy GHZ fidelity now fails. On the one hand, of
course, since the fidelities did change. On the other hand, however, we
now somehow get a singular matrix for two qubits which produces a
corresponding `np.LinAlg` warning and results in a fidelity > 1. FWIW,
checking manually you can see that the matrices are pretty close, so the
issue really seems to be with the way the fidelity is calculated.
What's also kind of fishy: the fidelity for three qubits increased
(slightly). Before, we had: 0.9518, now it's 0.957. @tcochran-quera@lamq317 did you see something similar when comparing the one-zone model
with and without correlated noise or is there something wrong here?
You can now also pass in the correlated rates either as dictionary or as
array
```python
model = GeminiOneZoneModel(cz_paired_correlated_rates = np.array(...))
model = GeminiOneZoneModel(cz_paired_error_probabilities = {"II": 0.9, "XX": ...})
```
Not sure about the naming though.
---------
Co-authored-by: Luis Martinez <[email protected]>
Co-authored-by: Tyler A Cochran <[email protected]>
0 commit comments