How to use "neb2dim.pl"

[yamaks]

Dear all,

I am wondering how to use the "neb2dim.pl" script.
I understand the usage of "neb2dim.pl" as follows;

1. To run CI-NEB calculation.
2. To execute "vfin.pl"
3. To enter the newly-created directory (and to execute "nebresults.pl")
4. In the directory, to execute "neb2dim.pl"

However, in these procedures, the inputs of dimer calculation seem generated from POSCARs of the CI-NEB calculation. I am not sure this is the correct and reasonable way.
Is there any way or script to generate dimer inputs using the results of CI-NEB calculations?

I am sorry if similar topics already exist.
Thank you very much in advance,
All the best

[graeme]

It sounds like you are doing things right. This should generate an initial configuration approximated at the saddle point of the NEB calculation, with an initial mode along the reaction coordinate. If you have done that and see a problem, please attach the directory with the calculation and let me know what the issues is.

[yamaks]

Sorry, this might be my comprehension problem.
Is it OK to use POSCARs of the NEB calculations rather than CONTCARs? I thought I had to adopt CONTCARs to consider the results of the NEB calculations.

[graeme]

You want to use the CONTCARs. Oh goodness, I think you may have found a 20 year old bug in our scripts! Wow, ok, I'll look into this today, but yes, it does appear that our neb2dim script uses the POSCAR files and it should use the CONTCAR files. Well, that's pretty embarrassing, but really good that you noticed this!

[yamaks]

Thank you for your reply.

The neb2dim script is so helpful. I, my teacher, and my friend (who noticed this problem) always use this. The teacher says using POSCARs generated by neb2dim.pl shows bad convergence and replacing them with the highest-energy CONTCARs of the NEB calculation often greatly shortens the Dimer calculation time. We found this kind of thing is due to the bug, according to your reply. Thank you for your information. I will let him know.

I would appreciate it if you fix the bug at your convenience. (If you are not busy. The fact that no one has noticed for 20 years implies this bug might be not so a big problem. :)

Sincerely,