Elena Peña Tapia

One more comment, about the failed unit test, you will see that there is a complaint about something related to kernels, this is because your Jupiter notebook metadata is specifying...

> Why would this not be in `neural networks` (same comment for the kernel PR but in kernels of course) and subclass NeuralNetwork. I would think we would place this...

This issue has already been discussed with @Zoufalc and @ebermot, and is connected to #393.

I think that this could be a good first issue for me, but it might take more than one sprint with my current bandwidth. Would you like me to work...

Changing this issue into an epic for better tracking, as it will be divided into 2. - [ ] #575

The idea behind this comment is to explain the change in direction that you will see in my refactoring PR (#722). As discussed with @mrossinek and @woodsp-ibm, we will change...

> I note that `estimate_observables` now returns the variances and shots if obtained from metadata (and sets these to 0.0 otherwise). Should I add this for the main eigenvalue too?...

> It seems that there are some fidelity tests for mismatched left and right circuits that are testing the sampler usage by fidelity to presumably ensure its catching the expected...

Well, you already fixed the CI but decided to remove the test anyway to prevent future situations like this. It's a bit trivial, so if you could quickly review it,...

The simulator seed can be fixed via `set_options`: `backend.set_options(seed_simulator=seed)`, same as with any fake backend. I get that the `seed` argument can be confusing, maybe we can add a clarifying...