[Pw_forum] variable cell MD target pressure

[Davide Ceresoli]

On 10/24/2013 10:14 AM, luisen wrote:
>
> Dear Davide,
>
> Thank you very much for your rapid answer. I have just one quick thought about
> the consistency between the pressure obtained from several NVT runs
> and the result obtained directly from the stress tensor.
>
> Did you use the "soft" cutoff method? (ecfixed, qcutz, q2sigma stuff)
I didn't. What are good values for ecfixed, qcutz and q2sigma?

> Did you include in the calculation of the derivative the kinetic energy?
Yes.

> Do you think the entropic term of the Helmholtz free energy may have
> an important impact?
No idea.


> I have another question. In liquids it doesn't really make sense to change the
> shape of the simulation cell. It is only the volume that matters. Moreover, for
> certain magnitudes it is more convenient to have cubic
> cells because of the increased degeneracy in modules of k-vectors in reciprocal
> space, which helps with statistics. However I could not find
> a switch in the input parameters that changes the 3 axes in the same amount, so
> as to preserve the shape of the box. It is possible to keep it
> orthorhombic, but not cubic. Is there any way to do that without modifying the
> code ?
There should be cell_dofree='volume', but it's not implemented. Actually
one would simply need the Andersen barostat (Allen-Tildeslesy page 232),
in addition to Parrinello-Rahman and similar methods.

Anyone willing to implement it?

> Yours,
>
>      Luis Enrique
>
Best,
     Davide