Strange orbital ordering by HSE06

[spakinfo]

Hi,

I've been trying to do a HSE06 calculation for a small-gap semiconductor using a pre-concerged WAVECAR file obtained from a standard PBE calculation. While the new calculation appears to be done normally, the eigenvalues in OUTCAR seem to have strange orderings and occupancies as follows:

k-point 1 : 0.0000 0.0000 0.0000
band No. band energies occupation
1 -28.4905 2.00000
2 -28.2677 2.00000
3 -14.8907 2.00000
4 -13.8988 2.00000
5 -11.3840 2.00000
6 -11.3840 2.00000
7 -10.9497 2.00000
8 -10.8343 2.00000
9 -10.8343 2.00000
10 -9.8975 2.00000
11 -5.2481 2.00000
12 -4.1626 2.00000
13 -3.1802 2.00000
14 -3.0254 2.00000
15 -3.0191 2.00000
16 -3.0191 2.00000
17 -2.6326 2.00000
18 -2.7495 2.00000
19 -2.7495 2.00000
20 -2.6881 2.00000
21 -2.6274 2.00000
22 -2.4548 2.00000
23 0.7923 2.00000
24 1.1315 2.00000
25 1.7648 2.00000
26 1.3554 2.00000
27 1.3554 2.00000
28 2.4325 2.00000
29 2.4325 2.00000
30 4.5692 2.00000
31 4.5692 2.00000
32 5.9843 2.00000
33 7.3134 0.00000
34 6.7823 2.00000
35 6.7823 2.00000
36 8.4250 0.00000
37 9.4722 0.00000
38 9.8530 0.00000
39 10.0606 0.00000
40 10.2480 0.00000

k-point 2 : 0.1429 0.0000 0.0000
band No. band energies occupation
.
.
.
As you can see above, the unoccupied state (#33) with much higher energy lies in between two occupied states (#32 and #34). As far as I know, the format of OUTCAR is such that, the eigenvalues for each kpoint are listed in an increasing order. I guess, the above ordering can cause errors in subsequent DOS and band structure calculations. Does anybody know why I'm getting such an output? For the sake of clarity, I'd like to mention that the preconverged PBE eigenvlaues are aligned as expected in an increasing order.
Below you can see the contents of my INCAR file (for HSE06 calculations)

#----------------------
ISTART=1
ISMEAR=0
SIGMA=0.02
NSW=0
IBRION=-1
ISPIN=1
EDIFF=0.0001
EDIFFG=-0.005
NELM=60
ICHARG=1
LMAXMIX=6
ENCUT=500
#----------------------
LHFCALC=.TRUE.
HFSCREEN=0.2
TIME=0.4
ALGO=Damped
ENCUTFOCK=0
#----------------------

Your comments are much appreciated

<span class='smallblacktext'>[ Edited ]</span>

[admin]

this behaviour shows up because the band-indices are NOT re-ordered during the electronic scf calculation, obviously that state moved up in energy during the scf run. As the occupancy of the state is 0.0 (it shifted above the Fermi level), this state is ignored when putting up the charge densities,... so you need not worry about it.