[DFTB-Plus-User] DFTB+ Geometry Relaxation external socket

Ben Hourahine benjamin.hourahine at strath.ac.uk
Wed May 2 03:06:12 CEST 2018


Hello James,


did you get the structure relaxed? Depending on your system, limiting
the size of the lattice optimization steps may help:


Driver = ConjugateGradient {

   LatticeOpt = Yes

.

.

.

  MaxLatticeStep = 1E-3

}



The optimisation methods are to the nearest local minima, so yes the
starting configuration can effect both the speed to converge and the
structure.

Periodic optimization of lattice and atomic coordinates definitely needs
to be improved in the future. Internally they are doing an iterative
version of what you propose:

optimise lattice {
  optimise atoms at a fixed lattice {
  }
}

If the lattice step is large, this means the atom optimisation inside
that cycle can have trouble.

Regards

Ben

On 25/04/18 12:43, James A Charles wrote:
>
> Hi,
>
>
> Thank you very much for the response. I have tried the LBFGS and
> indeed that is the one I have had the most luck with. However, It
> still takes around a 1000 iterations and I was hoping that I could
> reduce that number. With the LBFGS algorithm, do you know how
> important the initial guess is? Is the final result dependent on the
> initial guess? 
>
>
> I have also tried larger systems with a solute embedded in the solvent
> and that is much more difficult to converge. I apply what I think is
> the standard procedure for geometry relaxation of first relaxing
> internal coordinates keeping the lattice vectors constant and then
> allowing cell relaxation. 
>
>
> Thanks,
>
> James 
>
> ------------------------------------------------------------------------
> *From:* DFTB-Plus-User
> <dftb-plus-user-bounces at mailman.zfn.uni-bremen.de> on behalf of Ben
> Hourahine <benjamin.hourahine at strath.ac.uk>
> *Sent:* Tuesday, April 24, 2018 3:44:00 AM
> *To:* dftb-plus-user at mailman.zfn.uni-bremen.de
> *Subject:* Re: [DFTB-Plus-User] DFTB+ Geometry Relaxation external socket
>  
>
> Hello James,
>
>
> apologies about the delay in replying. The socket interface is for use
> with the external I-Pi code (https://github.com/i-pi/i-pi), which can
> relax geometries, but is better known for advanced molecular dynamics
> (various forms of path integral, replica exchange, ...). They provide
> a few example scripts for DFTB+ (both the dftb_in.hsd and their own
> input.xml files). We are currently working on a section for the
> recipes document that shows this in use.
>
>
> DFTB+ can also be used via the ASE software
> (https://wiki.fysik.dtu.dk/ase/), which doesn't use the socket
> interface, but instead via disc files.
>
>
> For the geometry drivers built into DFTB+, did you get a chance to try
> the LBFGS optimiser as this might be better for your case?
>
>
> Regards
>
>
> Ben
>
>
> On 17/04/18 02:24, James A Charles wrote:
>>
>> Hello,
>>
>>
>> I have been trying to relax ice crystals with DFTB+ and have tried
>> the drivers included in DFTB+ without much luck. I was reading
>> through the documentation and noticed Socket{} as a driver. What is
>> the purpose of this? and is there an example of which external tools
>> this works with? 
>>
>>
>> Thanks,
>>
>> James 
>>
>>
>>
>> _______________________________________________
>> 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
>
> -- 
>       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 <mailto: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
> 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

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

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20180502/87e105b1/attachment.html>


More information about the DFTB-Plus-User mailing list