[DFTB-Plus-User] Question about hessian generation to use in modes

Reinaldo Pis Diez reinaldo.pisdiez at gmail.com
Tue Feb 8 16:12:22 CET 2011

Dear dftb+/modes users,

I was running the modes utility to get the frequencies of some 
aggregates. There is no problem with that utility but I've seen a 
non-expected behavior in the prior runs with dftb+ needed to create 
the hessian.

Firstly, I run dftb+ with the ConjugateGradient option to the Driver 
keyword to optimize the geometry of the isolated aggregate. It took 
about 100 cycles to optimize with a MaxForceComponent of 0.01 a.u., 
no too strict but the calculation is part of a set of jobs on 
several aggregates with increasing size and convergence becomes 
difficult for large sizes.

Secondly, I prepare dftb_in.hsd with the SecondDerivatives option to 
Driver to generate hessian.out. I append the geometry that was 
optimized in the previous step, using the *.gen file. But, to my 
surprise, the job starts with a geometry optimization again before 
generating the hessian. It took about 30 cycles now, and the problem 
here is that I cannot get the re-optimized geometry as the 
OutputPrefix option is not valid in SecondDerivatives.

As a consequence of this the hessian and the optimzed geometry I'm 
using for modes do not correspond to the same dftb+ run.

Any hints will be appreciated.



PS: Sorry for the long mail!

More information about the DFTB-Plus-User mailing list