[ase-users] NEB code

[Peterson, Andrew]

I would check the literature for your field to see what parameters people
are using; they should report this in their publications. For our work, 0.1
eV/A is a bit loose; something like 0.03-0.05 eV/A is more normal.

But really the best practice is to do a convergence test. For example,
"over-converge" several of your calculations, and watch how the barrier
changes with fmax during the calculation. Then you can decide your
tolerance based on energies, not forces.

On Fri, Aug 30, 2019 at 11:05 AM Alaithan, Zainab <
z.alaithan17 at imperial.ac.uk> wrote:

> Thank you very much for the response. I am new to NEB calculation so I am
> also not quite sure about the convergence criteria. is 0.1 ev fmax
> sufficient or is it too high? Do you also think that AutoNEB can speed the
> calculation?
> ------------------------------
> *From:* Peterson, Andrew <Andrew_Peterson at brown.edu>
> *Sent:* 30 August 2019 14:48:16
> *To:* Alaithan, Zainab <z.alaithan17 at imperial.ac.uk>
> *Cc:* Ase-users <ase-users at listserv.fysik.dtu.dk>
> *Subject:* Re: [ase-users] NEB code
>
> It looks to me like you need to run the NEB optimization longer. The
> forces don't always need to go monotonically down, and often go up for some
> time period during an optimization before going down again. You may also
> want to look at the evolution of the shape of the band within ASE GUI (or
> with ase.neb.NEBTools) to see how it evolves to give you a little more
> insight into what is going on.
>
> On Fri, Aug 30, 2019 at 6:13 AM Alaithan, Zainab via ase-users <
> ase-users at listserv.fysik.dtu.dk> wrote:
>
> Thank you for the explanation. I am also facing convergence problem with
> my reaction as you can see in the log file. I am not sure what could be the
> cause for the optimizer to behave in this way. The fmax was going down
> initially as can be seen in neb_1.log until it reached 0.4 in neb_2.log and
> then started to go up again. On the other hand, the energy is going down
> consistently but slowly where the forces start to increase.
>
>
> It seems to me that the energy surface at this region is rugged but the
> curvature is flatter. I am not sure if there is a remedy for this, or
> whether there are alternative convergence criteria or tests that I can use
> to locate the transition state geometry with more accuracy.
>
>
> Regards,
> ------------------------------
> *From:* ase-users-bounces at listserv.fysik.dtu.dk <
> ase-users-bounces at listserv.fysik.dtu.dk> on behalf of Peterson, Andrew
> via ase-users <ase-users at listserv.fysik.dtu.dk>
> *Sent:* 23 August 2019 20:42:53
> *To:* ase-users at listserv.fysik.dtu.dk <ase-users at listserv.fysik.dtu.dk>
> *Subject:* Re: [ase-users] NEB code
>
>
> You probably saw the line
> <https://gitlab.com/ase/ase/blob/master/ase/neb.py#L315> f = forces[i - 1],
> and then f is modified. That first line returns a “view” of the original
> array, so when f is modified so is the original forces array. It's just a
> numpy way of dealing with slices.
>
>
> On Tue, Aug 20, 2019 at 11:37 AM Alaithan, Zainab via ase-users <
> ase-users at listserv.fysik.dtu.dk> wrote:
>
> I am examining NEB code because I am facing some convergence problems when
> I increase the number of images from 3. I noticed that NEB get.forces
> method returns the real forces instead of the modified forces.
>
>
> The modified forces are calculated as 'f'
> but the method returns 'forces' rather than 'f'
>
>
> am I missing something?
>
>
> Thank you,
>
>
> Regards,
> _______________________________________________
> ase-users mailing list
> ase-users at listserv.fysik.dtu.dk
> https://listserv.fysik.dtu.dk/mailman/listinfo/ase-users
>
>
>
> --
> Andrew Peterson
> Associate Professor
> Brown University School of Engineering
> Barus & Holley 247
> 184 Hope Street
> Providence, RI 02912
> (401) 863-2153
> http://brown.edu/go/catalyst
> _______________________________________________
> ase-users mailing list
> ase-users at listserv.fysik.dtu.dk
> https://listserv.fysik.dtu.dk/mailman/listinfo/ase-users
>
>
>
> --
> Andrew Peterson
> Associate Professor
> Brown University School of Engineering
> Barus & Holley 247
> 184 Hope Street
> Providence, RI 02912
> (401) 863-2153
> http://brown.edu/go/catalyst
>


-- 
Andrew Peterson
Associate Professor
Brown University School of Engineering
Barus & Holley 247
184 Hope Street
Providence, RI 02912
(401) 863-2153
http://brown.edu/go/catalyst
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/ase-users/attachments/20190830/4519e08c/attachment.html>