[ase-users] Activating Climbing after reducing force in STRING/NEB

[James Kermode]

Hi Reza

Yes, just run the optimisation twice: first with a fmax=0.2, and then set neb.climb to True and continue the optmisation with a tighter tolerance, with something like the following:

neb = NEB(images)
opt = LBFGS(neb)
opt.run(fmax=0.2)
neb.climb = True
opt.run(fmax=0.01)

Best wishes,
James.


From: ase-users <ase-users-bounces at listserv.fysik.dtu.dk> on behalf of Reza Behjatmanesh-Ardakani via ase-users <ase-users at listserv.fysik.dtu.dk>
Date: Tuesday, 2 March 2021 at 17:38
To: ase-users <ase-users at listserv.fysik.dtu.dk>
Subject: [ase-users] Activating Climbing after reducing force in STRING/NEB
Hi,
Is it possible to activate climbing (with relaxing only the highest energy image with its two neighbors) after reducing force, for example, to 0.2 eV/A? This good idea has been used in FHI-aims and in ORCA.
So, starting STRING/NEB calculation, only simple calculations are done without considering image climbing, and after reducing force to a certain value, only 3 images are used for finding TS, while fixing other images.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/ase-users/attachments/20210305/15d1b2fc/attachment-0001.htm>