IBZKPT error - revised

Message

bth20 · #1 Post by **bth20** » Fri Oct 28, 2005 9:17 pm

Hi there,

I posted the recent message entitled IBZKPT error where I indicated I'm getting the error message:

VERY BAD NEWS! internal error in subroutine IBZKPT:
NKX>IKPTD 46

I figured out that this is because in VASP, IKPD is the maximum number of k-points allowed (in any one direction I think) and that this is set to 45.

So I tried increasing it and recompiling but then I get the error

VERY BAD NEWS! internal error in subroutine IBZKPT:
Fatal error: unable to match k-point! 0

Can anyone help? Why is the number of k-points allowed so restricted, and how can I get around it so that I can use more k-points?

Thanks

Ben

#2 Post by **admin** » Wed Nov 16, 2005 3:35 pm

the error message shows up if the symmetry of the lattice is incompatible with the chosen k-mesh

krihen · #3 Post by **krihen** » Mon Nov 05, 2007 12:00 pm

Hi,

I'm getting the same error

VERY BAD NEWS! internal error in subroutine IBZKPT:
Fatal error: unable to match k-point! 0

I'm first relaxing a structure (iron or chromium carbide) using ISMEAR = 1, and then evaluating the energy using ISMEAR = -5. All the runs of the first type runs OK, but many (not all) of the runs of the second type crashes with the 'unable to match k-point' error.

The input to the runs are the same, except of course that CONTCAR is copied to POSCAR before start of second run. I tried removing the IBZKPT file before starting second run, to no effect.

So solution seems to be not to use ISMEAR = -5, but how then should I obtain accurate energies ?

-Krister

#4 Post by **admin** » Tue Nov 06, 2007 12:19 pm

please check if the symmetry of the lattice was preserved during the relaxation runs. the KPOINTS file for the second run must of course be compatible with the symmetry with that lattice (in case the symmetry changed)

krihen · #5 Post by **krihen** » Thu Nov 08, 2007 8:28 am

All files are identical in the first and second runs, except of course the POSCAR file, as I already mentioned.

In addition, I erased the WAVECAR and CHG* files, so that Vasp in the second run should treat the structure as "new". Therefore the concept "lattice symmetry changed" does not apply.

The conclusion seems to be that the implementation of BlÃ¶chl's method --- ISMEAR = -5 --- is buggy ??

-Krister

#6 Post by **admin** » Thu Nov 08, 2007 10:41 am

certainly not. the error message is due to the choice of the k-mesh, not to the BZ integration algorithm

krihen · #7 Post by **krihen** » Thu Nov 08, 2007 10:58 am

So what is the problem ?

I'm using a KPOINTS file of the format

------------------------------
// Comment
0
M
nk1 nk2 nk3
0 0 0
------------------------------

where nk1, nk2, nk3 are integers, not necessarily the same.

-Krister

#8 Post by **admin** » Fri Nov 09, 2007 12:49 pm

the ratios of nk1 nk2 and nk3 have to scale roughly like the inverse of the corresponding legths of the lattice parameters of you cell. Mabye that is the reason.

manolo · #9 Post by **manolo** » Thu Dec 20, 2007 7:29 am

Hello,

I also have trouble with the IBZKPT subroutine.

These are the vector of my cell in the POSCAR file:
11.455020 0.000000 -2.811503
0.000000 2.910000 0.000000
0.000000 0.000000 5.621000

and this is the KPOINTS I used

Automatic mesh
0
Monkhorst Pack
1 4 2
0. 0. 0.

I set the points so that nk1, nk2 and nk3 scale like the inverse of the lengths of the lattice vectors. I got this error message:
"VERY BAD NEWS! internal error in subroutine IBZKPT:
Fatal error: unable to match k-point! 0"

I changed the KPOINTS file to
Automatic mesh
0
Monkhorst Pack
2 8 4
0. 0. 0.

And I got a new error message:
"VERY BAD NEWS! internal error in subroutine IBZKPT:
Reciprocal lattice and k-lattice belong to different class of lattices. 96"

Whatever I do for this calculation I always get those two error messages. Please, could you tell us what it means and how to get rid of this error?!!

Thank you very much

Amritendu · #10 Post by **Amritendu** » Mon Oct 25, 2010 11:53 am

Hi,
I encountered the same problem. However, when I changed the Kmesh scheme (Monkhorst-Pack to Gamma) the job ran with same Kpoints.

Amritendu

shaldar · #11 Post by **shaldar** » Tue Dec 14, 2010 1:46 pm

Hi,
I was trying to do shape relaxation (ISIF =4) of a sheet of atoms. The KPOINTS file used is
Automatic mesh
0
M
3 3 1
0. 0. 0.

Initial run starts well, but when re started the run from the latest CONTCAR, I am getting the error "VERY BAD NEWS! internal error in subroutine IBZKPT: "

My Initial lattice vectors are
1.00000000000000
6.5175000060701596 11.2886411488437997 0.0000000000000000
-13.0350000121402001 0.0000000000000750 0.0000000000000040
0.0000000000000070 0.0000000000000050 20.0000000186269986

and at the point of restart it is

6.4938000192778320 11.2447798426244692 0.0013028499758000
-12.9806514016667638 -0.0040115708216802 0.0030206323466521
-0.0046528127588851 0.0049995268147400 20.1656776870495520

could you tell us how to change KPOINTS mesh in this cases ?

Thank you very much in advance

VASP001 · #12 Post by **VASP001** » Fri Sep 09, 2011 11:18 am

Try set the KPOINTS is like this :

Auto-G
0
M
45 1 1

Tong Lijia · #13 Post by **Tong Lijia** » Thu Sep 12, 2013 7:09 am

VERY BAD NEWS! internal error in subroutine IBZKPT: NKPT>NKDIM 35072
My way：reduced K mesh from 49X49X49 to 21X21X21