FORCES and GRID in Supercells

Queries about input and output files, running specific calculations, etc.


Moderators: Global Moderator, Moderator

Post Reply
Message
Author
DukeSilver
Newbie
Newbie
Posts: 1
Joined: Thu Jun 23, 2016 4:35 pm

FORCES and GRID in Supercells

#1 Post by DukeSilver » Tue Jun 28, 2016 6:56 am

I am trying to calculate the phonon dispersion curve for a 2D material. Once I have properly optimized the primitive structure until forces are completely negligible, I created a supercell. However, when I computed the forces for this undistorted supercell, there were huge forces.

One of the things I have tried is increasing the grid NGX,Y,Z manually in the input. But, when I check the OUTPUT it seems that VASP is overriding the grid I have selected. Below follows the relevant parts of my output:

max r-space proj IRMAX = 1 max aug-charges IRDMAX= 423666

dimension x,y,z NGX = 180 NGY = 180 NGZ = 200

dimension x,y,z NGXF= 360 NGYF= 360 NGZF= 400

support grid NGXF= 720 NGYF= 720 NGZF= 800


(...)

I would recommend the setting:

dimension x,y,z NGX = 129 NGY = 129 NGZ = 175

(...)

NNGXGis ok and might be reduce to 130

NGY is ok and might be reduce to 130

NGZ is ok and might be reduce to 174

Broyden mixing: mesh for mixing (old mesh)

NGX = 65 NGY = 65 NGZ = 87

(NGX =360 NGY =360 NGZ =400)

gives a total of 367575 points

It seems that VASP is not using the grid I define in the input.
I wonder how to force to get the grid I want and solve the problem with the forces in the supercell cell

admin
Administrator
Administrator
Posts: 2921
Joined: Tue Aug 03, 2004 8:18 am
License Nr.: 458

Re: FORCES and GRID in Supercells

#2 Post by admin » Tue Jun 28, 2016 1:38 pm

Check if there is no mistake in producing the supercell.
For primitive cell and the supercell you should get
similar forces (k-point sampling should be proportionally
decreased).

Post Reply