UCC and fully occupied MOs (Fix for issue #1277)

[julenl]

Summary

Improved the functionality of UCC a little bit, to work if only one of the spin registers is fully occupied.

Details and comments

The configuration check looked for whether any number of alpha or beta orbitals (length of the spin register) was greater or equals than the number of particles.
Obviously, there cannot be more particles than spin orbitals, and in order to handle this case, I updated the previous any(n >= ... to any(n > ....
For handling case where both alpha and beta are the same as the number of particles (i.e. (2, 2) for the He2 molecule), I created a new raise condition, notifying that the function is still not implemented.
Implementing this option will probably involve dealing with the empty excitation list at line 339.
For the case where either alpha or beta is equals to the number of particles (i.e. H-He molecule), it should just pass through.

[coveralls]

Pull Request Test Coverage Report for Build 7057038901

• 2 of 5 (40.0%) changed or added relevant lines in 1 file are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage decreased (-0.02%) to 86.753%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
qiskit_nature/second_q/circuit/library/ansatzes/ucc.py	2	5	40.0%

Totals
Change from base Build 7020358022:	-0.02%
Covered Lines:	8769
Relevant Lines:	10108

---

💛 - Coveralls

[mrossinek]

Thanks, this looks like a good start. Can you please also do the following changes?

• add a unittest which asserts that the behavior works as intended
• verify whether the docstrings on the section of which errors are being raised needs to be updated
• add a release note (see here for instructions)

And finally:

• please also check all of the above for the HartreeFock initial state and add a unittest if it already works (just to be certain that it won't break in the future)

[julenl]

So, @mrossinek

• Unit test for a system equivalent to the "single H" molecule with spin register (1, 0) added
• Docstrings do not need update, because the raised error corresponds to an internal method
• Release note added
• The HartreeFock one does not have this issue, because it directly checks for "n spatial orbitals" > "particles in spin register"

[mrossinek]

Thanks, this looks good to me 👍

---

On a side note: I noticed that none of the errors which _check_configuration (and _check_ucc_configuration) might raise when raise_on_failure=True are documented.. I guess this does not show up, because these are all part of internal methods, which get called under the hood in Qiskit's BlueprintCircuit (which is a base-class of these circuits incl. HartreeFock).

We should probably at some point look into addressing this documentation-wise.