convergence of total energy with respect to cutoff and no of kpoints

[Borninger]

Hi,

I'm a VASP newby and I have a problem. I'm doing some GGA calculation for Au bulk. First of all I calculated the EOS. After that I varied ENCUT from 150 to 900 eV. Taking 600 eV I took the Monkhorst-Pack (upper line from 4 to 20 even and 5 to 19 odd, while in the lower line all was 0) for convergence with respect to the kpoints.

My problem is, that everything converges but from the wrong side. I thought the total energy gets minimized, but in my calculation it gets maximized.

Has anyone an idea what the cause of the problem might be?

Thanks for the replies.

<span class='smallblacktext'>[ Edited ]</span>

[admin]

concerning the increase of the total energy when the
-- cutoffs are increased: if the effect is small, it may be due to the kinetic energy corrections added (please have a look at the accuracy.pdf file of the vasp-workshop)
-- k-point meshes are increased (at constant ENCUT): this really sounds strange,
please check wheter really all input parameters (except the number of k-points) were constant in these tests

[Borninger]

Thank you,

I've repeated the calculations and in case of the cutoff I started with a reaaly small cutoff energy. So if I watched coarse enough, it all converged from the right side. It seems to be the same with the kpoints. In the mean time I have asked some experienced VASP user, who told me that small deviations can occur but are not important as long as there is convergence at all.

[dnumb]

Hi there,

I am also a sort of a vasp newby and have a similar problem when trying to do single-point energy convergence calculations of a mineral structure (muscovite composed of Al, H, K, O and Si corresponding to a total of 84 atoms) with respect to ENCUT.


I am using:
*VASP 4.6;
*PAW_GGA Al 05Jan2001, PAW_GGA H 07Jul1998, PAW_GGA K_sv 04May1998, PAW_GGA O 05Jan2001 and PAW_GGA Si 05Jan2001;
*KPOINT file:
Automatic mesh
0
M
1 1 1
0 0 0

*INCAR file given for me to use was:
SYSTEM = muscovite-2M1

Startparameter for this Run:
NWRITE = 2
LPETIM=F write-flag & timer
ISTART = 0 job : 0-new 1-cont 2-samecut

Electronic Relaxation 1
PREC = HIGH
ENMAX = 200.00 eV
NELM = 60
EDIFF = 1E-05 stopping-criterion for ELM
BMIX = 3.00
INIWAV = 1 random initial wavefunctions
NPAR = 2

Ionic Relaxation
EDIFFG = -0.01 stopping-criterion for IOM
NSW = 0 number of steps for IOM
NBLOCK = 1; KBLOCK = 40 inner block; outer block
IBRION = 2 ionic relax: 0-MD 1-quasi-New 2-CG
ISIF = 2 stress and relaxation
IWAVPR = 1 prediction: 0-non 1-charg 2-wave 3-comb
ISYM = 1 0-nonsym 1-usesym
LCORR = F Harris-correction to forces
ISMEAR = 0; SIGMA = 0.01


From these files (conditions) and just varying the Ecut the Energies obtained were:
Ecut F= (energy in eV, correct?)
200 -542.6230600
300 -549.8089800
400 -542.2314600
500 -541.1163100
600 -541.3176400

I thought that you should always get a lower energy as you improve your basis-set, or am I doing/saying something stupid :S ?

Thanks in advance for your time,


David

[alex]

Hi there,

in VASP there is some correction for (low) cutoff energies. You'll find it by
grep "kinetic energy error for atom=" OUTCAR
Sum up and remove from the total energy. The correction should be larger for small cutoffs.
The pure total energies should show the expected behaviour.

cheers,

alex

[dnumb]

Hi Alex,

But if there is a greater correction, I checked my results and you are right about this, for smaller cutoffs then the behaviour will be augmented, no?

In my case this was translated in and logarithmic-like increase :S

ENCUT F
200 -551.0760600
300 -551.0665800
400 -542.3617600
500 -541.1432100
600 -541.3358400

Cheers


David

[dnumb]

Hi there,

It appears that my thread was blocked while the admin was confirming my license affiliation. But I am back now so I would like to ask if anybody could please get back to me on this

Cheers


David
<span class='smallblacktext'>[ Edited Thu Feb 18 2010, 03:40PM ]</span>

[dnumb]

Can anybody please help me with this problem, as I am stuck with this and this influences every result that I can ever get and have now started to doubt if there is any problem with my VASP :S