a problem in NEB calculation

Message

enthusiastic · #1 Post by **enthusiastic** » Fri Nov 20, 2009 7:16 am

This is my first time of running NEB calculation
I didn't know the suitable parameters.so I read from the vasp manu, and write the following INCAR
SYSTEM = NEB
PREC = Normal ! standard precision

!ENMAX = 400 ! maximum of ENMAX for B and N
ALGO = Fast ! electronic relaxation

!NELM = 40
ISMEAR = 0 ; SIGMA = 0.1 ! Gaussian smearing, -5 for tetrahedron method
! IDIPOL = 4 !full dipole correction
ICHAIN = 0

!ISPIN = 2 magnetic system
IBRION = 3
ISIF= 2

IMAGES = 4
SPRING = -5

NSW = 800

IBRION = 2
!POTIM = 0.2 default is 0.5

LREAL = Auto ! real space projection for a large system
! remove this for an accurate calculation
! EDIFF = 0.1E-04 1.E-04 later, electronic convergence criteria
EDIFFG = -0.04 ! later, criteria for ionic relaxation
********************
and I set 00 01 02 03 04 05 six folders under the direction.
however, it stopped without any hints.
I paste the output files asbelow:

**
running on 8 nodes
each image running on 2 nodes
ERROR: the following files does not exist 03/POSCAR
distr: one band on 1 nodes, 2 groups
vasp.4.6.28 25Jul05 complex
01/POSCAR found : 3 types and 34 ions
LDA part: xc-table for Pade appr. of Perdew
00/POSCAR found : 3 types and 34 ions
05/POSCAR found : 3 types and 34 ions
POSCAR, INCAR and KPOINTS ok, starting setup
WARNING: wrap around errors must be expected
FFT: planning ... 4
reading WAVECAR
WARNING: random wavefunctions but no delay for mixing, default for NELMDL
entering main loop
N E dE d eps ncg rms rms(c)
DAV: 1 0.118120353361E+04 0.11812E+04 -0.89943E+04 7496 0.128E+03
DAV: 2 -0.256730400418E+03 -0.14379E+04 -0.13918E+04 9728 0.364E+02
DAV: 3 -0.365047061435E+03 -0.10832E+03 -0.10744E+03 8352 0.106E+02
DAV: 4 -0.368067013619E+03 -0.30200E+01 -0.30120E+01 9800 0.182E+01
DAV: 5 -0.368114452036E+03 -0.47438E-01 -0.47369E-01 8664 0.221E+00 0.500E+01
RMM: 6 -0.301747823872E+03 0.66367E+02 -0.40589E+02 7498 0.493E+01 0.186E+01
RMM: 7 -0.296024503580E+03 0.57233E+01 -0.15215E+01 7510 0.115E+01 0.887E+00
RMM: 8 -0.294890551646E+03 0.11340E+01 -0.27635E+00 7470 0.437E+00 0.481E+00
RMM: 9 -0.294573435392E+03 0.31712E+00 -0.13732E+00 7442 0.334E+00 0.226E+00
RMM: 10 -0.294488186253E+03 0.85249E-01 -0.17701E-01 8110 0.996E-01 0.901E-01
RMM: 11 -0.294479552844E+03 0.86334E-02 -0.85405E-02 7963 0.646E-01 0.379E-01
RMM: 12 -0.294479196099E+03 0.35674E-03 -0.14632E-02 7676 0.289E-01 0.223E-01
RMM: 13 -0.294479547422E+03 -0.35132E-03 -0.44563E-03 7544 0.174E-01 0.782E-02
RMM: 14 -0.294479622198E+03 -0.74776E-04 -0.69448E-04 6173 0.578E-02
******

I checked all of the folders, and found only two of them ran and stopped at the same time.

So can you help me?

and another question:
when the jobs are done , where can I find the results of the diffusion path?and Can I got the results of the active energy of diffusion?

Thanks alot.

boris · #2 Post by **boris** » Fri Nov 20, 2009 9:05 am

Hi

In the folders 00 and 06, you have to put the initial and final POSCAR file, respectively.

What POSCAR files did you put in the folders 01 02 03 and 04 ?

Regards.

enthusiastic · #3 Post by **enthusiastic** » Thu Nov 26, 2009 1:44 pm

thank you very much.
I did put the initial andfinal POSCAR files into the corresponding folders.
and in hte folders of 01 02 03 and 04, I just removed the atom for diffusion from the initial position into the assumed diffusion path and then transfered them into the POSCAR.
Indeed, I am not sure whether it is right or not?
Thanks again for your advices.

boris · #4 Post by **boris** » Thu Nov 26, 2009 2:06 pm

Ok, so you built the migration pathway "by hand".

If we take a look at the OUTCAR file, it is written "ERROR: the following files does not exist 03/POSCAR". Are you sure that you have put the intermediate POSCAR file in this directory?

By the way, there are scripts which automatically construct all intermediate POSCAR files using the initial and final POSCAR files.

Finally, some hints:
1. There are two IBRION keywords in your INCAR. Why don't you use IBRION=1? It works pretty well.
2. Be careful to have the volume kept constant in your calculations i.e. the header in the POSCAR file should be identical and ISIF=2 in the INCAR file (it seems ok).
3. Be careful to have the number of processors be a multiple of the number of images (multiple of 4 in your case, also seems ok).

Regards.

Boris.
<span class='smallblacktext'>[ Edited Thu Nov 26 2009, 03:08PM ]</span>

jim_king · #5 Post by **jim_king** » Mon Nov 30, 2009 6:59 am

Hi,
I also want to do the TS search job, but i donot know how to buit the intermediate POSCAR. Can you tell where can i get this script? Thank you very much.
[quote="'smallblacktext'>[ Edited Mon Nov 30 2009, 08:05AM "]</span>

boris · #6 Post by **boris** » Mon Nov 30, 2009 2:54 pm

Dear jin_king

There's a whole package available, associated with the TS theory. You can find all necessary scripts on the following website:
http://theory.cm.utexas.edu/vtsttools/downloads/

Regards.

Boris.

enthusiastic · #7 Post by **enthusiastic** » Wed Dec 02, 2009 12:10 am

Dear boris
Thank you so much.
I have solved the problem by your adivices.
and I have another basic question:
which results are useful?
it seems I got an energy, is that respect with the diffusion active energy of this special path?
Actually, I was expected to got all the results of the images.
This is rather a basic question. But I can't get useful informaions on other literatures.
Thanks again.

boris · #8 Post by **boris** » Wed Dec 02, 2009 8:17 am

If everything went well during the NEB calculation, you will find in every folder (01 to 04) an OUTCAR file, just like a regular VASP calculation.

If you take the final total energy in each of these folders, and you plot it with respect to some reaction coordinates, you will have some kind of migration energy curve.

enthusiastic · #9 Post by **enthusiastic** » Wed Dec 02, 2009 9:55 am

Thanks so much for you reply. I will try to analize my calculation.

enthusiastic · #10 Post by **enthusiastic** » Thu Dec 03, 2009 12:59 am

[quote="boris"]If everything went well during the NEB calculation, you will find in every folder (01 to 04) an OUTCAR file, just like a regular VASP calculation.

If you take the final total energy in each of these folders, and you plot it with respect to some reaction coordinates, you will have some kind of migration energy curve. [/quote]
*****
I have another question. is it possible for me to run small kpoints first and then large KPOInTS to shorten the calculation of NEB?
thanks again.

enthusiastic · #11 Post by **enthusiastic** » Thu Dec 03, 2009 12:59 am

[quote="boris"]If everything went well during the NEB calculation, you will find in every folder (01 to 04) an OUTCAR file, just like a regular VASP calculation.

If you take the final total energy in each of these folders, and you plot it with respect to some reaction coordinates, you will have some kind of migration energy curve. [/quote]
*****
I have another question. is it possible for me to run small kpoints first and then large KPOInTS to shorten the calculation of NEB?
thanks again.

boris · #12 Post by **boris** » Thu Dec 03, 2009 7:47 am

Yes. You perform a first calculation with a low number of kpoints and you set LCHARG=.TRUE. in order to print the CHGCAR file in every folder 01 to 04.
Then once the calculation is complete, you move the CONTCAR to the POSCAR, you set ICHARG=1 in the INCAR file and you increase the number of kpoints.

The calculation should restart with the charge density stored in the CHGCAR file, significantly increasing the convergence.

Boris.