kpoint grid commensurate with lattice
Posted: Tue Nov 10, 2009 11:35 pm
While using vasp 5.2, I noticed an error that I've never seen before: my kpoint grid is not commensurate with the lattice. I'm using automatic kpoint generation, so my kpoint file may look like this:
My POSCAR file looks like this:
Here is the error.
When I use automatic kpoint generation, isn't the grid automatically reduced with respect to the symmetry of the supercell? So what does this error mean?
Second, when I calculate dielectric constants (LEPSILON=.TRUE.), the program loop to calculate epsilon after the SCF calculation sometimes needs to change the kpoint grid--why is this? Furthermore, it cannot change the kpoint grid when I set NPAR--this makes the calculation very slow--why should a quick kpoint meshing require the entire calculation to use the default NPAR value?
Thanks for your insights.
Code: Select all
Automatic 7x7x7
0
Gamma
7Â 7Â 7Â
0. 0. 0.
Code: Select all
7.6086174048,   0.0016548302,   0.0000000000
-3.8065974586,  -6.5879342971,   0.0000000000
0.0000000000,   0.0000000000, -12.4458399153
32
Direct
0.000Â 0.000Â 0.000
0.167Â 0.167Â 0.167
0.000Â 0.500Â 0.500
0.167Â 0.667Â 0.667
0.500Â 0.000Â 0.000
0.667Â 0.167Â 0.167
0.500Â 0.500Â 0.500
0.667Â 0.667Â 0.667
0.000Â 0.000Â 0.000
0.167Â 0.167Â 0.167
0.000Â 0.500Â 0.500
0.167Â 0.667Â 0.667
0.500Â 0.000Â 0.000
0.667Â 0.167Â 0.167
0.500Â 0.500Â 0.500
0.667Â 0.667Â 0.667
0.167Â 0.167Â 0.167
0.000Â 0.000Â 0.000
0.500Â 0.000Â 0.000
0.000Â 0.500Â 0.500
0.500Â 0.500Â 0.500
0.000Â 0.000Â 0.000
0.500Â 0.000Â 0.000
0.000Â 0.500Â 0.500
0.500Â 0.500Â 0.500
0.667Â 0.167Â 0.167
0.167Â 0.667Â 0.667
0.667Â 0.667Â 0.667
0.667Â 0.667Â 0.667
0.167Â 0.667Â 0.667
0.167Â 0.167Â 0.167
0.667Â 0.167Â 0.167
Here is the error.
Code: Select all
 -----------------------------------------------------------------------------Â
|Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â |
|           W    W    AA    RRRRR   N    N  II  N    N   GGGG   !!!           |
|           W    W   A  A   R    R  NN   N  II  NN   N  G    G  !!!           |
|           W    W  A    A  R    R  N N  N  II  N N  N  G       !!!           |
|           W WW W  AAAAAA  RRRRR   N  N N  II  N  N N  G  GGG   !            |
|Â Â Â Â Â Â Â Â Â Â Â WWÂ Â WWÂ Â AÂ Â Â Â AÂ Â RÂ Â Â RÂ Â Â NÂ Â Â NNÂ Â IIÂ Â NÂ Â Â NNÂ Â GÂ Â Â Â GÂ Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â |
|           W    W  A    A  R    R  N    N  II  N    N   GGGG   !!!           |
|Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â |
|      Your generating k-point grid is not commensurate to the symmetry       |
|      of the lattice.  This can cause   slow convergence with respect        |
|      to k-points for HF type calculations                                   |
|      suggested SOLUTIONS:                                                   |
|       ) if not already the case, use automatic k-point generation           |
|       ) shift your grid to Gamma (G) (e.g. required for hex or fcc lattice) |
|Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â |
 ----------------------------------------------------------------------------- Second, when I calculate dielectric constants (LEPSILON=.TRUE.), the program loop to calculate epsilon after the SCF calculation sometimes needs to change the kpoint grid--why is this? Furthermore, it cannot change the kpoint grid when I set NPAR--this makes the calculation very slow--why should a quick kpoint meshing require the entire calculation to use the default NPAR value?
Thanks for your insights.