[DFTB-Plus-User] R: Phonon Mode Calculations

alessandro.pecchia alessandro.pecchia at ismn.cnr.it
Tue Jul 25 17:18:07 CEST 2017

Dear Lachlan,
you need sk-files with repulsive part included otherwise you do not get the right forces. Then in the input you set SecondDerivatives as diver. The Hessian is currently computed with finite differences.The 'modes' code uses a LAPACK eigensolver for real symmetric matrices, if I remember correctly. Good relaxed structures and electronic bandstructures do not necessarily implay accurate vibrations as that's related to the curvature around the minimum, harder to get in many cases. It depends if you can live with about 10-20% accurate frequencies. Do you need phonon dispersions ?
Kind Regards,
Alessandro Pecchia

Inviato dal mio dispositivo Samsung

-------- Messaggio originale --------
Da: Lachlan Oberg <u5805496 at anu.edu.au> 
Data: 24/07/17  07:31  (GMT+01:00) 
A: dftb-plus-user at mailman.zfn.uni-bremen.de 
Oggetto: [DFTB-Plus-User] Phonon Mode Calculations 


I have two questions regarding the MODES algorithm used to calculate phonon modes. Firstly, what method does it use to calculate the eigenmodes given the Hessian? Secondly, what part of the .skf files is required to determine the Hessian? Are the integral
 tables sufficient, or do I explicitly require .skf files whose geometry can be optimized (i.e., they contain the
repulsive spline interaction)? For example, does a faithful reproduction of the conduction/valence bands necessarily mean
a faithful reproduction of the modes?

Kind regards
Lachlan Oberg

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.zfn.uni-bremen.de/pipermail/dftb-plus-user/attachments/20170725/24f2b888/attachment.html>

More information about the DFTB-Plus-User mailing list