The ABINIT Users Mailing List ( CLOSED )

["Erik Ylvisaker" <eylvisaker@physics.ucdavis.edu>]

Yaroslav,

This behavior is not unusual -- it is typical to see the total energy 
oscillate as the k-point mesh is increased.  You should however see that the 
oscillations become smaller and smaller as convergence is increased, which 
appears to be happening in your data.  The k-point mesh is not a variational 
parameter (as ecut is) so the total energy can go up or down as the mesh is 
changed.

~Erik Ylvisaker

----- Original Message ----- 
From: <yshtogun@cas.usf.edu>
To: <forum@abinit.org>
Sent: Sunday, June 25, 2006 3:38 PM
Subject: [abinit-forum] ngkpt - convergence ?


> (Dobrogo Dnya!)
> Dear Abinit Users?
> I tried to investigate convergence according to different k-points mesh.
>  I used next k-points mesh:
>   (2 k-points) - 1 1 4
>   (3 k-points) - 1 1 6
>   (4 k-points) - 1 1 8
>   (5 k-points) - 1 1 10
>   (6 k-points) - 1 1 12
>   (8 k-points) - 1 1 16
> thre are my result:
> k              Etot (Ha)                       Etot (eV)
> 2 -144.780880485148 -3939.68820712992
> 3 -144.788626129349 -3939.89897683245
> 4 -144.780626293347 -3939.68129021906
> 5 -144.784690801234 -3939.79189110600
> 6 -144.784609401319 -3939.78967610162
> 8 -144.784449706748 -3939.78533059123
>
>  I got strange behavior for 2, 3, 4 k- points. It just jumps up and down.
>  Why does it jump? I only changed ngkpt in input files.
>
>  Thank you in advance!
>    Yaroslav!