My Community

Hello,

Following this post (forum/viewtopic.php?t=19126),
I wonder if it’s possible to fix the distance between two atoms during structure optimization with IBRION = 1, 2, or 3.

Since ICONST isn’t currently supported for IBRION = 1 or 2 in VASP, are there plans to implement it in the near future?

Or, is there an alternative to fix interatomic distances during relaxation? (FFF function by Selective dynamics isn’t sufficient for my research.)

Thank you!

Hi,

If you really have only two atoms of a molecule whose length you want to keep fixed, in some cases selective dynamics might work if you think that everything else other than the two atoms will relax around these two atoms. It will probably not be efficient, but it might work. If you want to relax a molecule on a surface, then maybe it will not work.

Currently, there are no plans to support ICONST with IBRION/=0.

You might try and find the ground state using simulated annealing: slowly decrease the temperature during an MD run which likely lands the molecule in the minimum energy configuration.

Yet another option is to use ASE as a relaxation engine in combination with VASP.
You can find documentation about that here
https://wiki.fysik.dtu.dk/ase/ase/calculators/vasp.html
and use the bond length constraint described here
https://wiki.fysik.dtu.dk/ase/ase/constraints.html

Hope this helps.

Hello,

Actually, You can change some lines in main.F file in /src directory of VASP source code and recompile.
Check this post: https://w.vasp.at/forum/viewtopic.php?t=19286
Of course it is not tested so use at Your own risk! I did't try this approach.

Best Regards,
Leszek

@Leszek

Thanks for the suggestion, however, that answer is referring to LATTICE_CONSTRAINTS while the original question refers to the ICONST file.

@henrique_miranda @leszek_nowakowski

Thank you for all the advices. It helps me a lot!

I'm trying to find ground state of point defect structure in amorphous oxide with + and - charge.

For this, I need to fix bond lengths for more than two atoms. (about 2 to 4 pairs of atoms)

I've tried MD annealing at 500K using ICONST and this method seems to be really effective at this moment.

Thanks a lot