[ase-users] Pulling out data from NEB

[Jesper Toft Rasmussen]

Hi Jens Jørgen,

Sorry for not replying earlier, but now I have finally tested what you said. It is indeed the forces which are different for the two cases.
I have included some MatLAB plots showing the force components for all 25 atoms in my system. (Only a single image in my neb has been shown here, by the same trend is apparent in all images).

The deviation plot is the abs difference between the forces obtained by the GUI and my script. As you can see, it is quite high for certain atoms.
These are exactly my constrained atoms: the corner carbon atoms and a single hydrogen atom.

So, if I understand this correctly: My script sets the forces to zero for constrained atoms, but the GUI does not?
How should I fix this so that I obtain the correct forces to input in my NEB calculation?

Best regards


Jesper Toft Rasmussen

PhD student

DTU Nanotech



Technical University of Denmark

[Description: http://www.dtu.dk/images/DTU_email_logo_01.gif]

Department of Micro- and Nanotechnology

Ørsteds Plads

Building 344

DK - 2800 Kgs. Lyngby

Denmark

Direct +45 45256341

jestr at nanotech.dtu.dk<mailto:jestr at nanotech.dtu.dk>

www.nanotech.dtu.dk<http://www.nanotech.dtu.dk/>





From: Jens Jørgen Mortensen
Sent: 14. januar 2013 21:10
To: Jesper Toft Rasmussen
Cc: 'ase-users at listserv.fysik.dtu.dk'
Subject: Re: [ase-users] Pulling out data from NEB

Den 14-01-2013 11:29, Jesper Toft Rasmussen skrev:
Hi Jens Jørgen,

Of course - stupid mistake to forget a minimal working example.
I have a small script that reads as follows:

from ase.io import read,write
from ase.neb import fit
slice = read('neb.traj at -11:')
write('neb_last_step.traj',slice)
print [i.get_potential_energy() for i in slice]
s, E, Sfit, Efit, lines = fit(slice)

Where neb.traj is obtained from another run,

neb = NEB(images)
neb.interpolate()
qn = FIRE(neb,logfile='optimize.log',trajectory='neb.traj')
qn.run(fmax=0.01)

After all calculations I plot s,E,Sfit,Efit, and lines in MatLAB to get the previously included figures.
>From what I can see from the source code, this is also the way the GUI works,
which is why I have spent a long time scratching my head..

I have made a local workaround by modifying the gui/neb.py file to output s,E,Sfit,Efit and lines to txt files,
which I then use In my further analysis. But an actual solution would be nice.

The energies in the two plots look similar, but the slopes of the small lines are wrong and those are calculated from the forces.  Try to print the forces from the fit0() function.  Are they the same when using the GUI and your script?

Are your atoms constrained in any way?

Jens Jørgen



Thank you in advance,


Jesper Toft Rasmussen

PhD student

DTU Nanotech



Technical University of Denmark

[http://www.dtu.dk/images/DTU_email_logo_01.gif]

Department of Micro- and Nanotechnology

Ørsteds Plads

Building 344

DK - 2800 Kgs. Lyngby

Denmark

Direct +45 45255694

jestr at nanotech.dtu.dk<mailto:jestr at nanotech.dtu.dk>

www.nanotech.dtu.dk<http://www.nanotech.dtu.dk/>





From: Jens Jørgen Mortensen
Sent: 10. januar 2013 16:50
To: Jesper Toft Rasmussen
Cc: 'ase-users at listserv.fysik.dtu.dk<mailto:ase-users at listserv.fysik.dtu.dk>'
Subject: Re: [ase-users] Pulling out data from NEB

Den 02-01-2013 14:29, Jesper Toft Rasmussen skrev:
Dear all ASE users,

I have set up a system, where I want to perform nudged elastic band calculations using the ASE interface. After having created the initial/final images and interpolating, I come to the task of analyzing the data.

By using the ASE GUI, I have tested the NEB tool, but since I need to further fiddle with the data, I need some way of extracting these data sets. One way is to use the fit/fit0 routines (found from the source code). However, when I do this I clearly don't get the same result as the GUI provides. I have added images as proof. It is clear that the tangents are completely off.

Can anyone tell me why this is the case? And what I should do to get out the correct data?

Can you show us how you used fit/fit0 ?

Jens Jørgen



 Best regards,


Jesper Toft Rasmussen

PhD student

DTU Nanotech



Technical University of Denmark

[http://www.dtu.dk/images/DTU_email_logo_01.gif]

Department of Micro- and Nanotechnology

Ørsteds Plads

Building 344

DK - 2800 Kgs. Lyngby

Denmark

Direct +45 45255694

jestr at nanotech.dtu.dk<mailto:jestr at nanotech.dtu.dk>

www.nanotech.dtu.dk<http://www.nanotech.dtu.dk/>










_______________________________________________

ase-users mailing list

ase-users at listserv.fysik.dtu.dk<mailto:ase-users at listserv.fysik.dtu.dk>

https://listserv.fysik.dtu.dk/mailman/listinfo/ase-users


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.fysik.dtu.dk/pipermail/ase-users/attachments/20130211/605fa205/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.gif
Type: image/gif
Size: 1055 bytes
Desc: image001.gif
URL: <http://listserv.fysik.dtu.dk/pipermail/ase-users/attachments/20130211/605fa205/attachment.gif>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: force.png
Type: image/png
Size: 15183 bytes
Desc: force.png
URL: <http://listserv.fysik.dtu.dk/pipermail/ase-users/attachments/20130211/605fa205/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: force_deviation.png
Type: image/png
Size: 14895 bytes
Desc: force_deviation.png
URL: <http://listserv.fysik.dtu.dk/pipermail/ase-users/attachments/20130211/605fa205/attachment-0001.png>