[Pw_forum] geometry optimization

[Lorenzo Paulatto]

In data 27 ottobre 2009 alle ore 15:00:25, mohnish pandey  
<mohnish.iitk at gmail.com> ha scritto:
> I can fix this problem by preserving the symmetry and
> getting eqb. lattice constan.?

Dear Mohnish,
the code actually do preserve the symmetry, it is enforced actively when  
computing the forces and cannot be broken during the relax procedure.  
What's happening is that the system has a lower symmetry than you think  
since the very beginning.

There are two possible reasons:
1. you have not specified the initial atomic positions with sufficient  
precision, the code consider an atom to be in a high symmetry position if  
it is less than 1.d-6 away from it (I think in units of alat=celldm(0),  
but I would have to check)
2. the symmetry of your system has a free parameter you where not aware  
of, e.g. in a Wurzite structure the distance between alternate planes can  
change in a specific way without breaking symmetry.

These two problems have two corresponding solutions:
1. specify more digits, or (with QE 4.1 and next) specify the position as  
a fraction, e.g.
   Ga 1/3 2/3 0
note that xcrysden cannot yet interpret an input file written like this  
and it is still a bit experimental
2. lock the atoms in position, you can do that adding three zeroes after  
the atom's coordinate in the input, e.g.:
  Ga  0.3333333 0.6666667 0.000000  0 0 0
have a look at the manual (Doc/INPUT_PW.txt) for details on details on  
this notation

best regards



-- 
Lorenzo Paulatto
SISSA  &  DEMOCRITOS (Trieste)
phone: +39 040 3787 511
skype: paulatz
www:   http://people.sissa.it/~paulatto/

     *** save italian brains ***
  http://saveitalianbrains.wordpress.com/