How is LUMO occupancy different from zero in XTB calculation?

I run some calculations using xTB. From one of the outputs I got:

*** convergence criteria satisfied after 182 iterations ***

#    Occupation            Energy/Eh            Energy/eV
-------------------------------------------------------------
1        2.0000           -0.6285969             -17.1050
...           ...                  ...                  ...
254        1.7813           -0.3250849              -8.8460
255        1.7322           -0.3248659              -8.8401
256        1.4992           -0.3241340              -8.8201
257        1.3002           -0.3236807              -8.8078
258        1.1554           -0.3233899              -8.7999
259        1.1298           -0.3233404              -8.7985
260        0.9584           -0.3230132              -8.7896 (HOMO)
261        0.9131           -0.3229267              -8.7873 (LUMO)
262        0.8611           -0.3228267              -8.7846
263        0.7655           -0.3226383              -8.7794
264        0.6860           -0.3224747              -8.7750
265        0.6470           -0.3223914              -8.7727
...           ...                  ...                  ...
516                         17.7730979             483.6306
-------------------------------------------------------------


Question: how can be the occupancy of molecular orbitals equal/above the LUMO greater than zero?

• Why does your LUMO have so high occupancy (almost close to 1)? I think probably your system has static correlation, or maybe the SCF did not converge to the correct solution (182 iterations is too high I feel). Sep 14 at 9:50
• @SRMaiti, about the occupancy, this is my point. Turning off the Fermi smearing, return the integer occupancies: HOMO and below, 2, and LUMO and above, 0. About the number of iterations, my system is a boron-nitride (120 atoms) + 4 nikel atoms nanotube and the calculations were run with double the accuracy as default.
– Camps
Sep 14 at 12:56