Added seniority_zero core functions in new subpackage

[obriente]

This is a first commit (13 files + 13 test files) for the code used in the seniority-zero / error mitigation experiment published in https://arxiv.org/abs/2210.10799 .

[obriente]

@mpharrigan any idea what's going on with the checks here?

[mpharrigan]

I think it's an anti-abuse measure for first-time contributors: https://docs.github.com/en/actions/managing-workflow-runs/approving-workflow-runs-from-public-forks

I've pressed the button to let the CI run.

Please update the title of this PR to provide a self-contained one-line description of the change

[mpharrigan]

Also, when you push new changes make sure to also ping me with a comment as pushing changes doesn't necessarily email me

[obriente]

@mpharrigan - I think I fixed all you comments above, thanks for going over the PR. I guess you might need to approve the workflow again to check pytest? (Currently it passes on my machine)

[obriente]

@mpharrigan - sorry to be a pain, can you approve the workflow again?

[mpharrigan]

That's really annoying. I've never seen it behave this way before

[mpharrigan]

@fdmalone can you give this a review

[mpharrigan]

Looks like you need additional requirements. ReCirq provides a basic set of default dependencies in https://github.com/quantumlib/ReCirq/blob/master/requirements.txt.

Please add a file in your subfolder named extra-requirements.txt. You can follow e.g. https://github.com/quantumlib/ReCirq/blob/master/recirq/hfvqe/extra-requirements.txt for an example.

You will also need to add the directory name to the list of subpackages with extra requirements in dev_tools/write_ci_requirements.py here: https://github.com/quantumlib/ReCirq/blob/master/dev_tools/write-ci-requirements.py#L77-L79

I think that should be sufficient and please don't hesitate to ping me with any questions. I tried to make it so it won't wait for approval to run the CI checks going forward

[obriente]

@mpharrigan - I'm now getting a test failing in qaoa because scipy.optimize expected a 1D array input. The test passes on my machine, however. Looking at the code, it seems that the issue is you have a list of 2 1x1 ndarrays and this is throwing an error? I'm trying to avoid this with np.concatenate, let's see if that works.

[fdmalone]

LGTM. I imagine the scipy issue is probably a version update thing. concatenating seems fine, but possibly it should be pulled out into a separate PR since it touches the qaoa experiment? @mpharrigan what do you think?

[mpharrigan]

Since it's a small change and I've reviewed it and it's already in this PR let's just sneak it in.