NEB/ neb.dat (wrong energies) + INTERNAL ERROR

[gvinios]

Hi everyone,

I'm new to VASP and have been trying to do NEB calculation. I was able to go over all steps, but ended up with weird energies. I'll start simple.. please have a look at what I found when used nebbarrier.pl

0 0.000000 0.000000 -0.002801 0
1 0.703065 -447.045935 -1.209352 1
2 1.409051 -446.307600 -0.537018 2
3 2.118795 -446.485345 0.750331 3
4 2.831517 -446.559945 -0.855753 4
5 3.548049 -446.098089 0.000641 5
6 4.072363 -446.443866 1.166645 6
7 4.597088 -447.099679 1.098122 7
8 5.122262 -447.409455 0.002567 8

Third column should be energy. These values match what I see on the OSZICAR files (very last line). However, one calculation before this (another system, another calculation), it didn't match and made much more sense. Those energies were positive and perfectly gave me a reasonable energy profile (same parameters used in both).

Does anyone know why this energies are not right? Is there any other way to get the energy barrier from NEB other than through the nebbarrier script?

Thank you in advance!
Rgds

[graeme]

If you put the OUTCAR from the calculation of the first minimum in the 00 directory, things should look better.

[gvinios]

Thanks a lot for considering replying this thread, prof.

So, I have inserted the OUTCAR from the standard relaxation of the starting point already. Did you mean it or something else?

Taking the opportunity, is that a manual way of extracting the individual energies to find the energy barrier?

Thanks in advance

[graeme]

ok, then something else is wrong - if you post the calculation, I will take a look

[gvinios]

Hi Prof,

I went through many threads, and I found one comment of yours that pointed me to what the problem could be.

Basically, the initial position was somehow relaxed with a different PREC setting compared to the final position. For the initial position, I accidentally used PREC = Accurate, while for the final position, I used PREC = Normal. Once I used 'Normal' for both and replaced the OUTCAR file in both the first and last folders in NEB, everything worked out.

Thank you so much for maintaining this forum!
I learn a lot from your comments :)

[graeme]

yup, that sounds right - good catch.

[gvinios]

Hey prof.

Returning after several successful NEB calculations :) and extensive efforts to resolve the issue shown in attachment -Please take a look. I found this thread https://www.vasp.at/forum/viewtopic.php?f=3&t=10370 where the links suggested by the admin are no longer available. Despite not being able to figure it out yet, I'd say it might be a well known error. Could you please shed some light on it for me?

Thank you in advance!

[graeme]

I don't know this error, but I can see that it is inside the electronic structure optimization. Also, while it does report this as an error, it seems to have continued the calculation, so you might be ok. If it continues to cause trouble, I would suggest trying a different IALGO optimizer, and/or change the mixing values (AMIX, BMIX, ... ).

[gvinios]

Thank you once again for your willingness to help, Professor.

I will report what happened and how the problem was resolved, with the intention of helping others.

* I wasn't using the mixing values (AMIX, BMIX, etc).

* IALGO = 38 yields
internal ERROR SETYLM_AUG: 63 -66 1353 7

*IALGO = 48 yields
internal ERROR SETYLM_AUG: 62 -66 1354 14

* Changing IBRION = 3 to IBRION = 1 worked with no IALGO tag set (i.e., default value) ✓

* Worth meanting that I do not set IOPT for my calculations as I let it follows what IBRION specifies. Also, the cluster I am using to run the calc does not accept IOPT = 3 at all - I don't know why and neither do the cluster staff.

* Last point, all trials exhibited the exactly the same behavior - finished the ionic relaxation at DAV: 19, gave the error message (as showed above), and proceed to the next ionic step getting stuck at DAV: 7.

Hope that is helpful to someone!