Fixed identity operator evaluation and extended the accepted input types in compute_rdms

[alexfleury-sb]

There were several bugs preventing users to use the compute rdms function:
1.

{pauli_string_to_of(term): exp_val for term, exp_val in exp_data.items()}

to

{pauli_string_to_of(term): exp_val for term, exp_val in exp_vals.items()}

(wrong dictionary).

List are unhashable, changed output type of the pauli_string_to_of function to a tuple.

if qterm in exp_vals:
    exp_val = exp_vals[qterm] if qterm else 1.

was failing for the () term. I added a conditional statement.

4. Can now accept dictionaries with string ("IXYZ") or openfermion-like ((1, "X), (2, "Y"), ...) terms.

[alexfleury-sb]

I tested all the possibilities, i.e.:

• exp_data with openfermion-like terms ((0, "X"), (1, "Y"), ...)
• exp_data with string terms ("XY...")
• exp_vals with openfermion-like terms
• exp_vals with string terms

Also, a missing entry in exp_vals raises an error. A missing entry in exp_data only raises an error when no compatible bases are detected.

[ValentinS4t1qbit]

Two things:

• Please change the PR title to something a bit better, that briefly describes what the PR did, for all readers.
• I would like you to add tests addressing the use cases this fix covers. The fact that all tests were passing before this PR and are still passing means that what you did is not reflected in them: we thus need more tests, that illustrate what you fixed / contributed.

[ValentinS4t1qbit]

Thank you !