[DFTB-Plus-User] SCC does not converge.
Goldman, Nir
goldman14 at llnl.gov
Thu Feb 18 20:33:08 CET 2016
Dear Dawid,
I was able to run a number of geometry optimization steps on the larger system you sent earlier (chromo-0.5.gen) by including the following in your input file:
Filling = Fermi {
Temperature [Kelvin] = 300.0
}
I also changed the SCC convergence to 1e-6, which likely helped speed things up.
Setting the electronic temperature to a non-zero value allows for the Mermin functional to be calculated, which essentially “smears" the electrons to excited states by computing fractional occupancies for the system according to the Fermi-Dirac distribution. This can effect can be very important to systems with molecular ions, for example, which can exhibit thermal electronic excitations even at relatively low temperature.
Hope that helps,
Nir
From: <dftb-plus-user-bounces at mailman.zfn.uni-bremen.de<mailto:dftb-plus-user-bounces at mailman.zfn.uni-bremen.de>> on behalf of Dawid Grabarek <dawid.grabarek at pwr.edu.pl<mailto:dawid.grabarek at pwr.edu.pl>>
Reply-To: User list for DFTB+ related questions <dftb-plus-user at mailman.zfn.uni-bremen.de<mailto:dftb-plus-user at mailman.zfn.uni-bremen.de>>
Date: Thursday, February 18, 2016 at 10:43 AM
To: User list for DFTB+ related questions <dftb-plus-user at mailman.zfn.uni-bremen.de<mailto:dftb-plus-user at mailman.zfn.uni-bremen.de>>
Subject: Re: [DFTB-Plus-User] SCC does not converge.
Dear Ben,
Unfortunately, my happiness came too early... . As the optimization proceeds
my system blows up. All bonds break down and atoms just fly into space... .
I have noticed that all oxygen and sulfur atoms get charges equal ca. +1.5
for the "optimized" structure. In fact, the breakdown of all of my systems
begins just after first optimization step. What would you suggest?
Different initial charges?
Best regards,
Dawid Grabarek
----- oryginalna wiadomość -----
od: Ben Hourahine <benjamin.hourahine at strath.ac.uk<mailto:benjamin.hourahine at strath.ac.uk>>
data: czwartek, luty 18, 2016 16:29
temat: Re: [DFTB-Plus-User] SCC does not converge.
do: dftb-plus-user at mailman.zfn.uni-bremen.de<mailto:dftb-plus-user at mailman.zfn.uni-bremen.de>
> Hello Dawid,
>
> have you tried a finite electronic temperature (this can really help with self consistency)?
> The attached example was at 0 Kelvin, I'd suggest something like room temperature as a first test.
>
> Also the SCC tollerance you are using is quite small (try the default value, unless there is a good reason otherwise).
>
> Regards
>
> Ben
>
>
> On 07/02/16 21:11, Dawid Grabarek wrote:
> Dear DFTB+ Experts,
>
> I want to perform geometry optimization of a system consisting of increasing
> number of amino acid residues.
> I have managed to successfully perform optimization for system up to
> 1870 atoms, using non-SCC DFTB approach.
> However when I want to use SCC-DFTB3 for system larger than 460 atoms,
> the charges do not converge even after a few thousand iterations. The iteration
> does not proceed in fact.
> I attach part of my output file:
>
> 9976 -0.21529827E+03 -0.13123659E+02 0.53106709E+01
> 9977 -0.17459612E+03 0.40702148E+02 0.53210677E+01
> 9978 -0.20103686E+03 -0.26440744E+02 0.53230816E+01
> 9979 -0.24096058E+03 -0.39923717E+02 0.53269557E+01
> 9980 -0.19704089E+03 0.43919692E+02 0.53385778E+01
> 9981 -0.21818849E+03 -0.21147597E+02 0.53345550E+01
> 9982 -0.25301554E+03 -0.34827052E+02 0.53358740E+01
> 9983 -0.27482831E+03 -0.21812768E+02 0.53316452E+01
> 9984 -0.29608303E+03 -0.21254728E+02 0.53454981E+01
> 9985 -0.26626843E+03 0.29814602E+02 0.53557124E+01
> 9986 -0.29473653E+03 -0.28468098E+02 0.53569507E+01
> 9987 -0.25516950E+03 0.39567029E+02 0.53763278E+01
> 9988 -0.28536057E+03 -0.30191072E+02 0.53798582E+01
> 9989 -0.28895556E+03 -0.35949882E+01 0.53806192E+01
> 9990 -0.22896112E+03 0.59994443E+02 0.53836906E+01
> 9991 -0.22485550E+03 0.41056157E+01 0.53812648E+01
> 9992 -0.22164301E+03 0.32124948E+01 0.53759401E+01
> 9993 -0.16929036E+03 0.52352651E+02 0.53873225E+01
> 9994 -0.19352833E+03 -0.24237970E+02 0.53825003E+01
> 9995 -0.16294863E+03 0.30579698E+02 0.53834931E+01
> 9996 -0.11288863E+03 0.50060003E+02 0.53970748E+01
> 9997 -0.10136437E+03 0.11524262E+02 0.53984882E+01
> 9998 -0.10537121E+03 -0.40068449E+01 0.53984460E+01
> 9999 -0.84099319E+02 0.21271891E+02 0.53857706E+01
> 10000 -0.76159647E+02 0.79396717E+01 0.53785016E+01
>
> Total Energy: -66.7997002259 H
> Total Mermin free energy: -66.7997002259 H
> ERROR!
> -> SCC is NOT converged, maximal SCC iterations exceeded
>
> I know that going up to 10 000 is crazy but just wanted to see what will happen.
>
> Anyway, I also attach two input files for systems for which I cannot
> reach the SCC convergence. What is more I also attach a PDF file which contains
> various parameters combination that I used to check whether changing
> parameters and Mixer will allow to reach convergence.
> I did that based on what I read in paper by V. Eyert (Journal of
> Computational Physics, 124, 271-285, 1996).
>
> Best wishes,
>
> Dawid Grabarek
>
>
>
> -------
> Dawid Grabarek, M.Sc.
> Wroclaw University of Technology
> Faculty of Chemistry
> Advanced Materials Modelling and Engineering Group
>
>
> _______________________________________________ > DFTB-Plus-User mailing list > DFTB-Plus-User at mailman.zfn.uni-bremen.de<javascript:main.compose('new',%20't=DFTB-Plus-User at mailman.zfn.uni-bremen.de')> https://mailman.zfn.uni-bremen.de/cgi-bin/mailman/listinfo/dftb-plus-user
> -- > Dr. B. Hourahine, SUPA, Department of Physics, > University of Strathclyde, John Anderson Building, > 107 Rottenrow, Glasgow G4 0NG, UK. > +44 141 548 2325, benjamin.hourahine at strath.ac.uk<javascript:main.compose('new',%20't=benjamin.hourahine at strath.ac.uk')> > > 2013/4 THE Awards Entrepreneurial University of the Year > 2012/13 THE Awards UK University of the Year > > The University of Strathclyde is a charitable body, > registered in Scotland, number SC015263
> _______________________________________________
> DFTB-Plus-User mailing list
> DFTB-Plus-User at mailman.zfn.uni-bremen.de<mailto:DFTB-Plus-User at mailman.zfn.uni-bremen.de>
> https://mailman.zfn.uni-bremen.de/cgi-bin/mailman/listinfo/dftb-
> plus-user
-------
Dawid Grabarek, M.Sc.
Wroclaw University of Technology
Faculty of Chemistry
Advanced Materials Modelling and Engineering Group
--
Nir Goldman
Material Sciences Division, PLS
Lawrence Livermore National Laboratory
P.O. Box 808, L-288
Livermore, CA 94551-0808
Phone: (925) 422-3994
Fax: (925) 422-2382
https://pls.llnl.gov/people/staff-bios/msd/goldman-n
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20160218/856a3a0c/attachment-0001.html>
More information about the DFTB-Plus-User
mailing list