[DFTB-Plus-User] Queries about the SCCTolerance and MaxSCCIterations

[Bálint Aradi]

Dear Ranajit,

welcome in the community!

Based on what you write, I'd think, that the MD resulted after a few 
steps in an unphysical system, which is hard or can not be converged.

Have a look first at the resulting geometry and also on its electronic 
structure. Are they reasonable at all. If not, probably your system 
started from a state with big forces (or big velocities, if you 
initialized the velocities by hand) and combined with a longish time 
step resulted in an unreasonable geometry.

The SccTolerance can be indeed made more coarse (resulting in a worse 
energy conservation during an NVE MD), but it still won't able to cope 
with very unphysical geometries. One workaround is to set 
ConvergentForcesOnly = No, which would continue the MD, even if the SCC 
was not converged for a given geometry. However, the forces it uses then 
may be absolutely unphysical, so use it with care.

Finally, in case the geometry is reasonable, and SCC still does not 
converge, you may have ended up in a metallic / partially filled state. 
In that case, giving some electronic temperature (Filling) can help to 
get the system converged.

Best regards,

Bálint



On 26.08.21 12:16, rsaha at icredd.hokudai.ac.jp wrote:
> Hello All,
> 
> I am using DFTB+ for a few months. I am using this software optimisation
> as well as MD. I have a few queries about the SCC iterations.
> 
> For optimisation, I have used SCCTolerance = 1e-6 and MaxSCCIterations =
> 500. The geometry and SCC converged well for optimisation process.
> For MD, these same parameters have been used and it worked well this far,
>   though the time taken is very large. Recently, I am facing a problem in
> generating the tragectory. The SCC is not converging after few steps in
> the MD.
> 
> After searching in the web, I found (https://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/2020/003310.html)
>   a solution as lowering the SCCTolerance and adding ForceEvaluation.
> Currently, I am doing MD with, SCCTolerance = 1e-3 and MaxSCCIterations
> = 200 and also ForceEvaluation = Dynamics (in Driver section).
> 
> (1) I do not know, is this parameters are correct to use for MD?
> (2) What values of the SCCTolerance and MaxSCCIterations can be used?
> How much lower values can be used? How to get the idea of the
> SCCTolerance and MaxSCCIterations that will be better for MD (or
> optimisation)?
> 
> If anyone can comment or have any suggestion, please write to me. I will
> be happy to hear from you.
> 
> Thank you for your time.
> 
> Regards
> Ranajit
> 
> 
> _______________________________________________
> DFTB-Plus-User mailing list
> DFTB-Plus-User at mailman.zfn.uni-bremen.de
> https://mailman.zfn.uni-bremen.de/cgi-bin/mailman/listinfo/dftb-plus-user
> 


-- 
Dr. Bálint Aradi
Bremen Center for Computational Materials Science, University of Bremen
http://www.bccms.uni-bremen.de/cms/people/b-aradi/


-------------- next part --------------
A non-text attachment was scrubbed...
Name: OpenPGP_signature
Type: application/pgp-signature
Size: 840 bytes
Desc: OpenPGP digital signature
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20210914/24602e1a/attachment.sig>