Restarting a CI-NEB

Vasp transition state theory tools

Moderator: moderators

Post Reply
s.zarrini
Posts: 9
Joined: Tue Aug 19, 2014 5:24 pm

Restarting a CI-NEB

Post by s.zarrini »

++++++++++++++++++++++++++++++++++++++++++++++
Dear VTSTTools users,

It is easy to restart a usual relaxation (ISIF =2) crashed by time limit (cp CONTCAR POSCAR and then start from scratch), however, I was wondering if it is possible to restart a CI-NEB calculation which has stopped just because of time limit? If so, how please? The CI-NEB has done by quick-min optimizer (IBRION=3, POTIM=0, IOPT=1).

The second question is concerning to "LSCALAPACK", I know for NEB the LSCALAPACK must be "FALSE" as VASP manual states, but would you please let me know if it should be "FALSE" for CI-NEB as well?
Any helpful comment would be highly appreciated.

Cheers,
Salman Zarrini
++++++++++++++++++++++++++++++++++++++++++++++
graeme
Site Admin
Posts: 2035
Joined: Tue Apr 26, 2005 4:25 am
Contact:

Re: Restarting a CI-NEB

Post by graeme »

You can restart an NEB calculation in the same way as a regular calculation: copy the CONTCAR files to POSCAR and restart. We have the "vfin.pl dir_to_which_the_old_calculation_is_moved" which will do this automatically (for regular, dimer, and NEB calculations).

I don't remember any restrictions using SCALAPACK with NEB, but I will ask around about this.
chill
Posts: 96
Joined: Tue Jul 28, 2009 9:04 pm

Re: Restarting a CI-NEB

Post by chill »

There isn't any issue with using SCALAPACK and the VTSTCODE NEB as far as I am aware. Please report any issue you may find.
s.zarrini
Posts: 9
Joined: Tue Aug 19, 2014 5:24 pm

Re: Restarting a CI-NEB

Post by s.zarrini »

[quote="chill"]There isn't any issue with using SCALAPACK and the VTSTCODE NEB as far as I am aware. Please report any issue you may find.[/quote]

++++++++++++++++++++++++++++++++++++++++++++++
Thank you Graeme and Chill for your answer, in VASP manual, page No. 86 some statements about "LscaLAPACK
-tag" can be found while several images are run at the same time by setting IMAGES=X, if I understood well.
In following I have one more question about my CI-NEB calculation of water molecule dissociation on Ni(111) surface where 6 of images have reasonable electronic and ionic convergence behaviors while one of them does not behave well, and because of that all the CI-NEB is slowing down as other 6 images have to be waiting for the other one. The noticeably point is magnetization in every ionic step of every image where in well behaved images it is as what I expected (around 10 as it was indicated by MAGMOM in INCAR), but for the another one the magnetization is suppressed and oscillating around 5, leading to having highly number of electronic step in this image while in the others the number of electronic steps in every ionic step is not that mach.
The CI-NEB directory of my calculation has been enclosed, any useful point would be appreciated in advance.

Cheers,
Salman Zarrini
++++++++++++++++++++++++++++++++++++++++++++++
Attachments
ci-oh-h-n.tar.gz
CI-NEB calculation water molecule dissociation on Ni(111) surface
(2.88 MiB) Downloaded 665 times
Post Reply