[DFTB-Plus-User] About Python Interface with DFTB

[Bálint Aradi]

Dear Junmian Zhu,

> I am trying to look for a Python interface with DFTB+ (or other
> versions of DFTB ). I have been using Atomic Simulation Environment
> (ASE), which interacts with DFTB by file IO and therefore is quite
> slow, which does not fit our purpose. Are there any other ways to
> directly interact with DFTB? For instance, are there a library like
> PySCF that can directly calculate the DFTB functional?

In case, you want to use DFTB+ as an energy server you have at the
moment various options:

* ASE-like wrapping, with file I/O. As you point it out, not very efficient.

* Communicating via sockets: You run DFTB+ which listens on a socket.
You can send geometries via the socket and receive the energies and
forces on that socket. So far, we have implemented the I-Py protocol for
the socket communication, but it should be straightforward to implement
any other protocol, in case you need further quantitites.

* We have an experimental branch with DFTB+ as library (branch 'mmapi'
in my personal fork). Currently, it has only a Fortran API. The
C/Python wrapper around it are planned of course, but had not been
implemented yet. (So you would have to invest some time in that.)

I think, currently your most efficient way would be to try the socket
communication. If that turnes out to be too slow for some reasons, you
can still invest the time into the Python bindings.

  Best regards,

  Bálint

-- 
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: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: OpenPGP digital signature
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20181031/8f014034/attachment.sig>