NEB with ISIF=3,4 etc

[xph]

Has the VTSTCODE been updated and complied in the last one month?

[raspberry]

Yes, about 2 weeks ago.

[xph]

Maybe try to add ICHAIN=0

[raspberry]

hi,
thanks for your reply. ICHAIN=0 did not help per se as it is set by default. However, I set the FMAGFLAG=T flag and I can tell that the optimization proceeds by observing changes in the cell volumes for each image. However, I still have a problem with the job because it does not converge (kind of). It stops at the line " CONVERGED based on Magnitude of Force", instead of "reached required accuracy..." line, so I assume it is not done yet. I played with the IALGO flag, changing it from 48 to 38 and then 53. With IALGO=48, there no convergence (electronic) is achieved. With IALGO=38 and IALGO=53, the jobs seem to be converging (although not completely. As described above, it always stops/crashes). What's bizarre is that for IALGO=38 and IALGO=53, the jobs stop at different points and different barriers.
Could you please comment on what could be wrong this time. I would greatly appreciate your help.

Below are the forces and energies for different IALGO runs. In my INCAR file, I set EDIFFG = -2E-03.
===================================================
IALGO=53
0 0.000000 -8.912561 0.000000
1 0.001227 -8.789312 0.123249
2 0.005684 -8.565784 0.346777
3 0.003181 -8.377846 0.534715
4 0.001105 -8.302495 0.610066
5 0.001063 -8.314200 0.598361
6 0.003360 -8.349339 0.563222
7 0.000974 -8.370277 0.542284
8 0.000000 -8.400642 0.511919

IALGO=54
0 0.000000 -8.912561 0.000000
1 0.000298 -8.781206 0.131355
2 0.003514 -8.541879 0.370682
3 0.002053 -8.338729 0.573832
4 0.000704 -8.260878 0.651683
5 0.000676 -8.284221 0.628340
6 0.000846 -8.335030 0.577531
7 0.000194 -8.366890 0.545671
8 0.000000 -8.400642 0.511919

IALGO=38
0 0.000000 -8.912561 0.000000
1 0.001170 -8.776879 0.135682
2 0.006804 -8.529006 0.383555
3 0.003990 -8.317491 0.595070
4 0.001403 -8.238329 0.674232
5 0.001320 -8.268251 0.644310
6 0.004154 -8.327620 0.584941
7 0.000672 -8.365102 0.547459
8 0.000000 -8.400642 0.511919
===========================================

[xph]

I think "FMAGFLAG=T" just means converging based on magnitude of the total force rather than converging based on the max atom force.
IALGO sets the algorithm used for electronic step. As long as the electronic step converges within NELM, IALGO will not affect the ionic steps. But if something is wrong with the electronic step, the force for ionic movement will definitely be messed up.

These results are strange: the distance between the initial and final states is 0 and images are so close. Do you know what's wrong with that? They cannot be the same structure. Could you upload the initial and final POSCARs so that we can look into them?

[raspberry]

hi,
What do you mean by "these results are strange: the distance between the initial and final states is 0 and images are so close"? The four columns I am showing represent images, forces, energies, barrier obtained with the nebef.pl script. They are not the same structure. I am not sure why you think that the distance between the initial and final state is 0. What do you mean?
Thank you.

[xph]

Sorry, I thought what you posted were neb.dat from nebbarrier.pl

[raspberry]

hi,
I am still struggling with my ss-NEB calculation. I ran a regular NEB calculation to verify that the vasp version I am using works fine. With the ss-NEB flags turned on, however, my jobs never converge or it do not run at all. I was playing different flags, but no success.
Could you please post some representative ss-NEB INCAR file that was tested to work well? I would greatly appreciate your help.

[graeme]

On conversely, if you post or email your calculation, we can see the actual problem.
I can't believe how much back-and-forth there has been here!