[Pw_forum] R: Re: "vc_relax" + "nscf" calculation

[Paolo Giannozzi]

On Thu, 2012-03-01 at 10:11 +0100, giacsport at libero.it wrote:

> After the "vc-relax" (ibrav=2) calculation, an "scf" calculation 
> with ibrav=0 and starting from the optimized cell parameter (alat)
> from the "vc-relax" calculation. Then on top of it the "nscf" one.

I don't think it is no longer needed to perform an intermediate scf
calculation. Just get the final optimized cell parameters and
positions (the latter will be read from data file anyway, but
the input cell parameters determine the real- and reciprocal 
space grid). You should have the same FFT grid parameters 
(nr1, nr1s etc.: they are reprinted on output) that in the final 
scf run of the vc-relax calculation (may differ from those in
the initial run). Works properly only on recent releases 
(4.3.2 and later)

P.
-- 
Paolo Giannozzi, IOM-Democritos and University of Udine, Italy