The ABINIT Users Mailing List ( CLOSED )

[Anglade Pierre-Matthieu <anglade@gmail.com>]

Hi,

It is quite likely possible to help you. However, to proceed, one would need : both of your input file and relevant part of the relaxation output file. 
Do you consider possible that your calculation parameters were unconverged ? Or that your scf/non-scf loops were unconverged ?

regards

PMA

On Mon, Jan 19, 2009 at 6:38 AM, kong <kongl@physics.rutgers.edu> wrote:

Hi,

I relaxed a system using optcell =2 and ionmov=3. After the desired convergence
is achieved, I copied the relaxed lattice parameters( acell/angles) and
atomic positions(xangst)  from the output file and start a new calculation
without
relaxation. All the other input variables are same as before.  I am very
surprised
that the new total energy is about 0.2Ha lower than  the relaxed results,
but the stress tensor is three order of magnitude larger!

I note that ngfft/boxcut(ratio) in the relaxation is different. But using the
same ngfft(which also leads to the same boxcut(ratio)) in the input file does
not improve the consistency.
Would anyone help me out on it? Thanks!

Lingzhu

-- 
Pierre-Matthieu Anglade