RDM calculation not stored correctly in MolecularData

[rugantio]

Hello, I ran a simple test with the H2 molecule at bond distance, this is what I get from the OpenFermion hdf5:

import os
from openfermion.hamiltonians import MolecularData

THIS_DIRECTORY = '.'
filename = os.path.join(THIS_DIRECTORY, 'src','openfermion','data','H2_sto-3g_singlet_0.7414')
molecule = MolecularData([('H', (0,0,0)),('H', (0,0,0.7414))],'sto-3g', 1, filename=filename)
molecule.load()

# FCI RDM
fci_rdm = molecule.get_molecular_rdm(use_fci=True)

print(fci_rdm)

the output is:

((0, 1), (0, 0)) 0.9872699847206565
((1, 1), (1, 0)) 0.9872699847206565
((2, 1), (2, 0)) 0.012730015279343554
((3, 1), (3, 0)) 0.012730015279343554
((0, 1), (1, 1), (0, 0), (1, 0)) -0.9872699847206565
((0, 1), (1, 1), (1, 0), (0, 0)) 0.9872699847206565
((0, 1), (1, 1), (2, 0), (3, 0)) 0.11210692213387734
((0, 1), (1, 1), (3, 0), (2, 0)) -0.11210692213387734
((1, 1), (0, 1), (0, 0), (1, 0)) 0.9872699847206565
((1, 1), (0, 1), (1, 0), (0, 0)) -0.9872699847206565
((1, 1), (0, 1), (2, 0), (3, 0)) -0.11210692213387734
((1, 1), (0, 1), (3, 0), (2, 0)) 0.11210692213387734
((2, 1), (3, 1), (0, 0), (1, 0)) 0.11210692213387734
((2, 1), (3, 1), (1, 0), (0, 0)) -0.11210692213387734
((2, 1), (3, 1), (2, 0), (3, 0)) -0.012730015279343554
((2, 1), (3, 1), (3, 0), (2, 0)) 0.012730015279343554
((3, 1), (2, 1), (0, 0), (1, 0)) -0.11210692213387734
((3, 1), (2, 1), (1, 0), (0, 0)) 0.11210692213387734
((3, 1), (2, 1), (2, 0), (3, 0)) 0.012730015279343554
((3, 1), (2, 1), (3, 0), (2, 0)) -0.012730015279343554

Instead form OpenFermion-PySCF, using the method run_pyscf:

from openfermion.hamiltonians import MolecularData
from openfermionpyscf import run_pyscf

molecule = MolecularData([('H', (0,0,0)),('H', (0,0,0.7414))],'sto-3g', 1)
molecule = run_pyscf(molecule,
                     run_scf=1,
                     run_fci=1)

# FCI RDM
fci_rdm = molecule.get_molecular_rdm(use_fci=True)

print(fci_rdm)

the output is:

((0, 1), (0, 0)) 1.9745399697399248
((1, 1), (1, 0)) 0.025460030260075338
((0, 1), (0, 1), (0, 0), (0, 0)) 1.9745399697399248
((0, 1), (0, 1), (1, 0), (1, 0)) -0.22421384296984592
((1, 1), (1, 1), (0, 0), (0, 0)) -0.22421384296984592
((1, 1), (1, 1), (1, 0), (1, 0)) 0.025460030260075338

The tests have been run with both packages updated to the latest git commit.

[babbush]

Interesting bug, we'll look into this. The RDM you are getting in the first case includes the spin degree of freedom, whereas the RDM in the second case does not.

[rugantio]

This issue is being addressed inside OpenFermion, here: https://github.com/quantumlib/OpenFermion/pull/535