get_qcs_objects_for_notebook needs cleanup and bugfixes

[tanujkhattar]

Description of the issue
get_qcs_objects_for_notebook was added in #4286 and is super useful to abstract out the authentication logic for accessing QCS. However, I think there are a few rough edges -- from a user perspective, I haven't dug into the technical details yet.

1. In the getting started with QCS tutorial, the object is used for authentication and verification that the authentication went through (device_sampler.is_signed_in) but after that, there's no usage of the device_sampler object. This doesn't look like the intended usage of an object specifically has a device and sampler objects ?
2. It looks like sqrt_iswap gateset string is hardcoded deep inside the get_qcs_objects_for_notebook method. This also looks like a bug and something that shouldn't be intended?

   Cirq/cirq-google/cirq_google/engine/qcs_notebook.py

   Line 119 in 6562a05

   sampler = get_engine_sampler(processor_id, gate_set_name="sqrt_iswap")

Cirq version
0.13.0

[tanujkhattar]

The call to get_engine_device(processor_id) in get_qcs_objects_for_notebook is wrong as it doesn't pass any gateset info / project_id parameters and hence returns an empty device which cannot be used for validation.

[mpharrigan]

(1) seems fine

(2) -- I think the goal is to remove gateset as a required arg for get_engine_sampler. it's weird to have to pick your target gateset instead of getting a device and it telling you its supported gateset(s)

(3) this seems like a bug in get_engine_device. we shouldn't be returning empty devices

[dstrain115]

Did we fix this already in another issue? If not, this seems like some doc cleanup that should likely be addressed if possible before 1.0. Tentatively assigning before 1.0 to this issue.

[maffoo]

This is partly fixed by #5045, which cleans up the logic including removing hard-coded sqrt_iswap. We could potentially refactor things a bit to address the fact that the start.ipynb notebook doesn't actually use the sampler or device returned by the helper function, but rather just checks that we are authenticated and then connects to quantum engine later by calling cirq_google.get_engine(). So we could perhaps split out the authentication stuff into a separate helper function to be called in situations like this.

[maffoo]

I think we can address some of the shortcomings of get_qcs_objects_for_notebook in #5240. @verult, should I assign this issue to you?

[verult]

Unassigning myself for now in case anyone can get to this before I can! Otherwise happy to take on this when I get some spare cycles.

[mpharrigan]

@verult what's the status of #5240?

[verult]

#5240 should ideally be replaced by calling a refactored version of get_qcs_objects_for_notebook which would expose

• the project_id used for signing in and
• the processor_id

in addition to (or maybe in place of) existing return values.

[dstrain115]

I'm taking this on in #5759

[dstrain115]

Processor_id and project_id are now returned in #5759 and the function is cleaned up.

I have a follow up in #5766 that attempts authentication only if virtual=False and tests this out in cirq_google notebooks to enable better testing and will close this after that PR is in.

[dstrain115]

@wcourtney @verult I think we can close this now. Do you agree?