Number of G-vector changed in star.

[hyeonseo_park]

I am not sure if this is a bug or my mistake
I was trying to calculate phonon dispersion for a 2x2x2 supercell of TiSe2.
I used IBRION=6 (finite difference method with symmetry)
I obtained this in log file:
-----------------------------------------------------------------------------
| _ ____ _ _ _____ _ |
| | | | _ \ | | | | / ____| | | |
| | | | |_) | | | | | | | __ | | |
| |_| | _ < | | | | | | |_ | |_| |
| _ | |_) | | |__| | | |__| | _ |
| (_) |____/ \____/ \_____| (_) |
| |
| internal error in: mkpoints_change.F at line: 786 |
| |
| internal error in GENERATE_KPOINTS_TRANS: number of G-vector |
| changed in star 7741 7740 |
| |
| If you are not a developer, you should not encounter this problem. |
| Please submit a bug report. |
| |
-----------------------------------------------------------------------------




------------------------------------------------------------------------------------------------------------------------
INCAR:
SYSTEM = TiSe2

#start parameters
NWRITE = 1
PREC = Accurate #precision mode

#electronic optimization
ENCUT = 350 #cutoff energy
EDIFF = 1.0e-8 #breakout condition for SC loop
NELMIN = 6 #minimum number of electronic SCF steps
NELM = 100
ALGO = Normal

#ionic relaxation
IBRION = 6 #relaxation method

#DOS-related
ISMEAR = 0 #determines how the partial occupancies are set for each orbial
SIGMA = 0.005
#Write flags
LWAVE = F
#Exchange correlation treatment
GGA = PE
#VdW interaction
IVDW = 11
#performance optimization
NCORE = 1 #set to 1 for Hessians

------------------------------------------------------------------------------------------------------------------------
KPOINTS:

Automatic mesh
0
Gamma
10 10 5
0. 0. 0.
------------------------------------------------------------------------------------------------------------------------
POSCAR:

Ti1 Se2
1.0
7.0799999237 0.0000000000 0.0000000000
-3.5400007783 6.1314593213 0.0000000000
0.0000000000 0.0000000000 12.0139999390
Ti Se
8 16
Direct
0.000000000 0.000000000 0.000000000
0.000000000 0.000000000 0.500000000
0.000000000 0.500000000 0.000000000
0.000000000 0.500000000 0.500000000
0.500000000 0.000000000 0.000000000
0.500000000 0.000000000 0.500000000
0.500000000 0.500000000 0.000000000
0.500000000 0.500000000 0.500000000
0.166666672 0.333333343 0.128322378
0.166666672 0.333333343 0.628322363
0.166666672 0.833333373 0.128322378
0.166666672 0.833333373 0.628322363
0.666666687 0.333333343 0.128322378
0.666666687 0.333333343 0.628322363
0.666666687 0.833333373 0.128322378
0.666666687 0.833333373 0.628322363
0.333333343 0.166666672 0.371677637
0.333333343 0.166666672 0.871677637
0.333333343 0.666666687 0.371677637
0.333333343 0.666666687 0.871677637
0.833333373 0.166666672 0.371677637
0.833333373 0.166666672 0.871677637
0.833333373 0.666666687 0.371677637
0.833333373 0.666666687 0.871677637

------------------------------------------------------------------------------------------------------------------------

There are several features of my system.
1) I extended lattice parameter to fit the experimental values. This might introduce stress on my system, and the POSCAR is only relaxed with respect to ionic displacement (not cell displacement).
2. This issue was resolved when I decreased SYMPREC (1e-5 -> 1e-7). This slightly broke the symmetry, which increased the total amount of calculation.
3. I constructed supercell using VESTA software.

[alexey.tal]

Hi,

It is an issue which occurs when we rotate the orbitals from a k-point in IBZ to the full BZ. We have not yet implemented a proper mechanism to solve it.
But there is a number of things one could try to avoid this problem. As you have done already, you can reduce the symmetry either by SYMPREC or ISYM=0, but that makes the calculation more heavy.
Perhaps a better way would be to symmetrize the structure first. You can find a python script here.