Is there any way to create non-overlapping images for NEB?

Vasp transition state theory tools

Moderator: moderators

Post Reply
nokimann
Posts: 16
Joined: Mon Sep 10, 2018 6:27 pm

Is there any way to create non-overlapping images for NEB?

Post by nokimann » Sat Nov 17, 2018 9:21 am

Hi, VTSTTools

I'm trying to create images from trans to cis transition.
But some of the structures created in the image seem to cause a warning message.

| The distance between some ions is very small |
| please check the nearest neighbor list in the OUTCAR file |
| I HOPE YOU KNOW, WHAT YOU ARE DOING |

Is there any way to create images with non-overlapping of atoms automatically?

I saw a topic 'NEB initial image construction,' but I need more detail explanations.
I did "nebmake.pl POSCAR_trans POSCAR_cis 8", and attached just my initial(trans) and final(cis) POSCARs.

1. To create images with non-overlapping atoms, should I fix them manually?
2. In the TOPIC: NEB initial image construction, there is a tip that separate initial & final to initial, transition state & final.
How can I find the transition state? Is this correct that
Step 1: nebmake.pl POSCAR_trans POSCAR_cis 2
Step 2: And then, run NEB
Step 3: By using from the last results as the transition states, create more images
nebmake.pl POSCAR_trans POSCAR_TS1 2
nebmake.pl POSCAR_TS1 POSCAR_TS2 2
nebmake.pl POSCAR_TS2 POSCAR_cis 2
Step 4: Run NEB again with all eight images
But even in these steps, suppose they are correct steps, sometimes it causes overlapping images.

As getting complex initial and final images, it's hard to modify structures manually.

Best,
Noki
Attachments
POSCARs.tar.gz
(2.57 KiB) Downloaded 235 times

graeme
Site Admin
Posts: 1860
Joined: Tue Apr 26, 2005 4:25 am
Location: University of Texas at Austin
Contact:

Re: Is there any way to create non-overlapping images for NEB?

Post by graeme » Sat Nov 17, 2018 2:50 pm

We have a little script called pushapart.pl which will push apart any pairs of atoms in a POSCAR file to a specified distance.

A better solution is to use the IDPP method, from the Jónsson group, which is implemented in ASE. This method uses a model hamiltonian to give a sensible initial reaction pathway, even for rotations of molecules which cause problems with linear interpolation.

One final trick is to include a guess of your transition state. Then you can create two linear paths, one from the initial to the transition state, and the second from the transition state to the final state. If you stitch these together, they can give a sensible initial path.

nokimann
Posts: 16
Joined: Mon Sep 10, 2018 6:27 pm

Re: Is there any way to create non-overlapping images for NEB?

Post by nokimann » Sat Nov 17, 2018 9:08 pm

Then, I'll need to check ASE.

Thank you, professor.

Post Reply