[AMBER] [NEB] Convergence criteria

From: Madhur Aggarwal via AMBER <amber.ambermd.org>
Date: Wed, 21 Sep 2022 13:23:32 +0000

 Dear amber community,

I am performing NEB calculations on a protein system in explicit water solvent to capture the minimum energy path for the conformational transition of my protein between two stable states by following the simulated annealing protocol as mentioned on the amber website (http://ambermd.org/tutorials/advanced/tutorial5_amber11/index.htm).

I have obtained some results and I want to do a convergence check to make sure my path is actually converged. I was thinking of the following 2 metrics as convergence criteria :

1. Calculating the RMSD of each image w.r.t itself in immediate previous frame and doing this for all frames in the final 5COOL2 step. If the images stop moving much, this RMSD value would decrease for each image as the 5COOL2 run progresses.

2. From the 5COOL2 log files, I can extract the sum of potential energy of all images from each frame. I expect to see the sum of potential energy decreasing in the 5COOL2 step as we move towards the minimum energy path. If this sum stops varying much after certain number of frames, this means the system has converged to a minimum energy path.

I was wondering if there are any standard metrics people use for proving convergence for NEB runs or if anyone could suggest a more refined/better approach of defining the above 2 criteria?

Thanks and regards,

Madhur Aggarwal

Undergraduate researcher,
IIIT Hyderabad, India

AMBER mailing list
Received on Wed Sep 21 2022 - 06:30:04 PDT
Custom Search