Restarting a Hessian calculation

Message

Vlad · #1 Post by **Vlad** » Wed Apr 21, 2010 1:39 am

Hello!
The longest queue time available on my cluster does not allow me to finish the vibrational frequencies calculation. I talked to my administrators, and they gave me a longer queue, but the computation time seems to be a moving target, as different steps take a different time to converge. I did a search on the subject of restarting such calculations, and the only post that came up was from 2006 by Alex Hofmann: http://cms.mpi.univie.ac.at/vasp-forum/ ... .php?4.740

Has the situation changed since then? It seems that there's no reason why a finite difference Hessian calculation should not be restartable. Are there any solutions out there to this problem? I am using VASP 4.6.32.

Thank you-
Vladimir

graeme · #2 Post by **graeme** » Wed Apr 21, 2010 4:46 am

You can use this older version of the dynamical matrix in vasp to combine separate finite difference calculations into one Hessian:
http://theory.cm.utexas.edu/vtsttools/dynmat/

Another program which can combine finite difference information from multiple calculations is phonopy:
http://phonopy.sourceforge.net/

Vlad · #3 Post by **Vlad** » Thu Apr 22, 2010 5:17 am

Thank you. I am talking to the administrators about applying the VTSTtools patch.

laurent.pedesseau · #4 Post by **laurent.pedesseau** » Wed May 05, 2010 7:31 am

Hi,

Is there a way of restarting this kind of job when using VASP5.2?

Thank you.