BKSF mapper fails to map H2 operator from driver

Environment

• Qiskit Nature version: main branch (did not try release)
• Python version: 3.6
• Operating system: Linux

What is happening?

In #432 I was asked had I tried with BKSF as I had added Parity and BK mappers into the test_excited_states_solvers.py test case when fixing the referenced issue and observing issues around two qubit reduction and symmetries. I did not try BKSF, you can see my comment in the referenced PR, but in trying it fails in the ground state computation. Now the bksf test file includes an pre-defined operator that is commented as H2 - in this case its built from the driver and here it fails with the following exception.

<snip>
line 480, in _add_edges_for_term
    _type = _interaction_type(n_number, n_raise, n_lower)
  File "...qiskit_nature/mappers/second_quantization/bksf.py", line 421, in _interaction_type
    raise ValueError("unexpected number of number operators: ", n_number)
ValueError: ('unexpected number of number operators: ', 0)

How can we reproduce the issue?

Based on the

from qiskit import BasicAer
from qiskit.utils import algorithm_globals, QuantumInstance
from qiskit_nature.drivers import UnitsType
from qiskit_nature.drivers.second_quantization import PySCFDriver
from qiskit_nature.mappers.second_quantization import BravyiKitaevSuperFastMapper, BravyiKitaevMapper
from qiskit_nature.converters.second_quantization import QubitConverter
from qiskit_nature.problems.second_quantization import ElectronicStructureProblem
from qiskit_nature.algorithms import  GroundStateEigensolver, VQEUCCFactory
 


algorithm_globals.random_seed = 8
driver = PySCFDriver(
    atom="H .0 .0 .0; H .0 .0 0.75",
    unit=UnitsType.ANGSTROM,
    charge=0,
    spin=0,
    basis="sto3g",
)
es_problem = ElectronicStructureProblem(driver)
quantum_instance = QuantumInstance(
    BasicAer.get_backend("statevector_simulator"),
    seed_transpiler=90,
    seed_simulator=12,
)


#converter = QubitConverter(BravyiKitaevMapper())
converter = QubitConverter(BravyiKitaevSuperFastMapper())
solver = VQEUCCFactory(quantum_instance)
gsc = GroundStateEigensolver(converter, solver)

result = gsc.solve(es_problem)
print(result)

What should happen?

If the code above is switched to say the BravyiKitaevMapper (not the super fast one) then it works and prints the result as below. I would expect the same with BKSF

=== GROUND STATE ENERGY ===
 
* Electronic ground state energy (Hartree): -1.842686681847
  - computed part:      -1.842686681847
~ Nuclear repulsion energy (Hartree): 0.70556961456
> Total ground state energy (Hartree): -1.137117067287
 
=== MEASURED OBSERVABLES ===
 
  0:  # Particles: 2.000 S: 0.000 S^2: 0.000 M: 0.000
 
=== DIPOLE MOMENTS ===
 
~ Nuclear dipole moment (a.u.): [0.0  0.0  1.41729459]
 
  0: 
  * Electronic dipole moment (a.u.): [0.0  0.0  1.41729468]
    - computed part:      [0.0  0.0  1.41729468]
  > Dipole moment (a.u.): [0.0  0.0  -0.00000009]  Total: 0.00000009
                 (debye): [0.0  0.0  -0.00000024]  Total: 0.00000024

Any suggestions?

No response

[mrossinek]

This appears to boil down to the fact that the BKSFMapper is unable to handle all-identity operators like IIII. @jlapeyre could you take a look at this one?

[jlapeyre]

@jlapeyre is that something you can add as part of #451?

Yes, I pushed a commit to #451 doing this.