[QE-users] Tetragonal vc-relax, is breaking the symmetry

[Lorenzo Paulatto]

> I am performing a vc-relax calculation over tetragonal crystal, for that 
> a put: ibrav=6 ; celldm(1)=16.64 ; celldm(3)=1.44 ; cell_dofree= xyz.
> 
> I believed if i was doing the vc-relax with these parameters, the 
> calculation would keep fixed a=b, and the tetragonal symmetry remains. 
Well, not necessarily, cell_dofree="xyz", in your case, only guarantees 
that the axis remain orthogonal. What you need is cell_dofree="volume", 
I think it does not work for non-cubic systems (and I have the 
impression is it not enforced when using the default BFGS algorithm)

However, your case is quite simple,  you could try to modify cell_force, 
at line 995 of Modules/cell_base.f90 to do something like

     IF( isotropic ) THEN
       !
       ! Isotropic force on the cell
       !
       fiso = (fcell(1,1)+fcell(2,2))/2.0_DP
       do i=1,2
           fcell(i,i)=fiso
       end do
     END IF

I.e. instead of averaging the cell force over all three directions, you 
only average over the first two. This of course will break the usual 
cell_dofree="volume", you can do it properly by adding another 
cell_dofree and another switch, but I think it is worth doing a quick 
test first

cheers


-- 
Lorenzo Paulatto - Paris