[QE-users] cell_dofree = 'volume'

[Lorenzo Paulatto]

> Is there any technical reasons for this?
> I was supposing that simply changing only celldm(1) should work.

Dear Carlo,
in a sense you are right, but the BFGS algorithm relaxes the 9 Cartesian 
cell components, it is not aware of ibrav, this requires some minor 
coding ans probably not a huge deal but has to be done. Also, in order 
to estimate convergence you have to be able to say when the pressure is 
minimized in the constrained relaxation, i.e. you have to project the 
pressure on ibrav, which requires some ibrav-dependent algebra.

The advantage of the damp-w algorithm, is that it only uses energy to 
decide convergence. Its disadvantage, is that it only uses energy to 
estimate convergence.

What you can always do, is a few constant-cell relax at different values 
of ibrav, then fit with ev.x


cheers

-- 
Lorenzo Paulatto - Paris