Kjf/noise learning

[kaelynj]

trafficstars

Closes #1830

[qiskit-bot]

One or more of the following people are relevant to this code:

• @abbycross
• @javabster
• @lerongil

[review-notebook-app[bot]]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[kaelynj]

I want to add a link to the NoiseLearnerOptions api ref page, but will need to do that later once the runtime 0.28 api refs are built.

[javabster]

Overall this looks good, just 2 things:

• number of qubits in the code examples is very small, any way we can scale it up a bit?
• doc is missing a "next steps" section at the bottom

[SamFerracin]

• number of qubits in the code examples is very small, any way we can scale it up a bit?

If this is what we want, then I suggest looking for the largest ring in the backend that we are using, and placing entangling gates on each pair. This would result in one "many qubits" noise-learning experiment, as opposed to many "few qubit" experiments with the GHZ circuits that would take forever.

For example, on ibm_bangkok one could run the following:

ring = [0, 1, 4, 7, 10, 12, 15, 18, 21, 23, 24, 25, 22, 19, 16, 14, 11, 8, 5, 3]
layer = list(zip(ring[::2], ring[1::2]))

circuit = QuantumCircuit(27)
for pair in layer:
    circuit.cx(*pair)

noise_learner.run(circuit)

[kaelynj]

Overall this looks good, just 2 things:

* number of qubits in the code examples is very small, any way we can scale it up a bit?

* doc is missing a "next steps" section at the bottom

Just finished addressing these! This PR is ready for another review.

[kaelynj]

Finished this last round of edits. @javabster or @jyu00 do either of you have any more comments/suggestions?