van der Waals correction, problem with GGA=RE INCAR file option

Message

lukasz.rzeznik · #1 Post by **lukasz.rzeznik** » Mon Sep 16, 2013 9:03 am

I am new to VASP and DFT.

I would like to calculate relaxation of C2H6 single molecule including vdW correction.

When I used in my INCAR only the option LVDW = .TRUE. I managed to do that.

But from the site http://cms.mpi.univie.ac.at/vasp/vasp/v ... et_al.html I see that some more advanced approaches are also possible. Those options are connected with change the POTCAR's PP to RE by setting in INCAR GGA = RE. And here is my problem. I dont have such POTCAR potentials.

When I try to do this with the POTCAR I have, I see the remark: "kernel differs, recalculating vdW kernel".

I have vdw_kernel.bindat and checked that it is the right file (for small endian clusters, the option "hexdump -d vdw_kernel.bindat | head" gives nd=35"

My INCAR file:

general:
System = Ethane
ISTART = 2

IBRION = 2
NSW = 120
EDIFFG = -1.0e-06
EDIFF = 1.0E-06

GGA = PE
PREC = normal
ENCUT = 450
ISF = 3
NELMIN = 5
IALGO = 38
ISMEAR = 0; SIGMA = 0.1
LREAL = .FALSE.
ADDGRID = .TRUE.
LWAVE = .FALSE.
LCHARG = .FALSE.
LUSE_VDW= .TRUE.
AGGAC = 0.0000

If I use GGA=RE instead of GGA=PE the system warns me that I specify xc-type that is not found in the POTCAR and starts recalculating the kernel.

I put the vdw_kernel.bindat file in the folder from which I run the calculations. When I change the name of the file I have "No kernel file found".

hexdump -d vdw_kernel.bindat | head
0000000 00004 00000 00035 00000 00004 00000 00280 00000
0000010 00000 00000 00000 00000 22192 16991 01694 16335
0000020 23344 10783 14900 16352 40148 25625 31358 16361
0000030 39486 01975 51830 16369 14628 26219 20170 16375
0000040 03016 00250 21813 16381 54142 40746 62680 16385
0000050 20140 34521 36010 16389 46910 30277 31029 16393
0000060 44136 10529 49735 16397 34467 44667 14386 16401
0000070 39276 25085 50799 16403 04362 50115 37102 16406
0000080 35748 49022 40253 16409 57984 08625 61804 16412
0000090 58045 21214 18956 16416 51366 04120 17983 16418

Where I make a mistake?

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

lukasz.rzeznik · #2 Post by **lukasz.rzeznik** » Mon Sep 16, 2013 11:54 am

If I remove the line GGA = PE from INCAR file and use potpaw_LDA in the POTCAR I can run the calculations with the comment "You have switched on vdW-DFT", however the vdw_kernel.bindat is not mentioned and probably not used in this case (when I change its name the calculations run with the same comment).

kelum · #3 Post by **kelum** » Mon Sep 16, 2013 3:09 pm

Hi,
for the vdW-DF calculations you can use PBE POTCARs, there are no special revPBE or vdW POTCARs.
What is the exact statement before the "kernel differs" message?
VASP should in that case write out "nmesh" and "nd" which should have the same value. What is it in your case?
j

lukasz.rzeznik · #4 Post by **lukasz.rzeznik** » Mon Sep 16, 2013 4:50 pm

Hi,
nmesh is much larger.
When I use POTCAR like below:

Code: Select all

PAW_PBE?C_h?06Feb2004
???4.00000000000000?????
?parameters?from?PSCTR?are:
???VRHFIN?=C:?s2p2
???LEXCH??=?PE
???EATOM??=???147.1560?eV,???10.8157?Ry
???TITEL??=?PAW_PBE?C_h?06Feb2004?

and remove the GGA = 91 line from INCAR I got the comment like that:

Code: Select all

?-----------------------------------------------------------------------------
|?????????????????????????????????????????????????????????????????????????????|
|??ADVICE?TO?THIS?USER?RUNNING?'VASP/VAMP'???(HEAR?YOUR?MASTER'S?VOICE?...):??|
|?????????????????????????????????????????????????????????????????????????????|
|??????You?have?switched?on?vdW-DFT.??????????????????????????????????????????|
|??????This?routine?was?written?and?supplied?by?Jiri?Klimes.??????????????????|
|??????We?recommed?that?you?carefully?read?and?cite?the?following?????????????|
|??????publication????????????????????????????????????????????????????????????|
|??????J.?Klimes,?D.R.?Bowler,?A.?Michelides??????????????????????????????????|
|????????J.?Phys.:?Cond?Matt.?22?022201?(2010)????????????????????????????????|
|??????J.?Klimes,?D.R.?Bowler,?A.?Michelides??????????????????????????????????|
|????????Phys.?Rev.?B.?83,?195131?(2011)??????????????????????????????????????|
|??????and?references?therein.????????????????????????????????????????????????|
|?????????????????????????????????????????????????????????????????????????????|
?-----------------------------------------------------------------------------

LDA?part:?xc-table?for?Pade?appr.?of?Perdew
POSCAR,?INCAR?and?KPOINTS?ok,?starting?setup
WARNING:?small?aliasing?(wrap?around)?errors?must?be?expected
FFT:?planning?...(?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_1?_)
WAVECAR?not?read
WARNING:?random?wavefunctions?but?no?delay?for?mixing,?default?for?NELMDL
?_entering?_main?_loop
?_?_?_?_?_?_?_N?_?_?_?_?_?_?_E?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_dE?_?_?_?_?_?_?_?_?_?_?_?_?_d?_eps?_?_?_?_?_?_?_ncg?_?_?_?_?_rms?_?_?_?_?_?_?_?_?_?_rms(c)
?_nmesh?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_4841498
?_nd?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_?_35
?_kernel?_differs,?_recalculating?_vdW?_kernel
?_it?_is?_strongly?_recommened?_to?_download?_the?_file?_vdw_kernel.bindat
?_(it?_is?_included?_in?_the?_source?_directory?_of?_vasp.5)
?_the?_calculation?_of?_the?_kernel?_might?_take?_days?_...
?_?_?_0.5071049521164697?_?_?_?_?_?_?_?_0.3735438613710874
^C?_

<span class='smallblacktext'>[ Edited Tue Sep 17 2013, 08:56AM ]</span>

kelum · #5 Post by **kelum** » Tue Sep 17, 2013 10:17 am

Hi,
the warnings about different POTCARs can be usually ignored.
Maybe your default integer size is not 4 bytes, what compiler are you using?
Do you have access to the code? If so could you change the
integer :: nmesh, IOstatus
line in the code to
integer*4 :: nmesh, IOstatus
and try to run?

j

lukasz.rzeznik · #6 Post by **lukasz.rzeznik** » Tue Sep 17, 2013 11:18 am

I asked system administrator to do this. He will do it tomorrow. I will write how the program works with this change in the code.

lukasz.rzeznik · #7 Post by **lukasz.rzeznik** » Thu Sep 26, 2013 11:36 am

It seems that everything is fine after this change in the code. However, now I do not see the message about nmesh and nd, just 3 warnings (about NPAR, publications citation, different xc functional requested in INCAR file vs one exists in POTCAR file) and the calculations begin. I have done some relaxation of ethane molecule by using several xc-functionals (all vdW corrections mentioned in Vasp manual+LDA+GGA) and the results have sense.

kelum · #8 Post by **kelum** » Thu Sep 26, 2013 9:22 pm

Good, the nd is written out only when it's some problem with it.
IIRC the numerical grid used for interpolation should be written out instead at that point.

jik