[QE-users] Enthalpy report in vc-relax runs

[Lorenzo Monacelli]

Hi,

The final enthalpy is not computed with the same wavefunction as the 
final energy.

This is because the number of wave-function is chosen by the original 
cutoff according to the original volume, and kept fixed during the whole 
relax.

However, since you are changing the volume, the number of wavefunction 
that satisfy the cutoff condition changes. This is why at the end of the 
vc-relax pw automatically does an extra scf calculation, redetermining 
the number of wf to use by the original cutoff at the final volume.

This is why energy / forces and stress tensor of the last calculation 
differs a bit from the last relaxation step (even if the cell and the 
atoms are the same).


The enthalpy printed is computed in the last step of vc-relax, while the 
energy you are taking is computed in the last scf calculation.

Therefore the correct result of the enthalpy is the one you performed 
manually adding the PV factor to the energy (note that this difference 
should go to zero as the cutoff -> infinity).

Bests,

LM


On 24/06/2018 12:04, Print S. wrote:
> Dear users,
>
> I have done vc-relax jobs using pw.x binary v.6.0 (svn rev. 13079). It 
> completed  with this information in the final step
>
> ...
> !    total energy              =   -2893.96255976 Ry
> ...
>
>           total   stress  (Ry/bohr**3) (kbar)     P=    5.50
> ...
>
>      Final enthalpy =   -2893.4953806269 Ry
> Begin final coordinates
>      new unit-cell volume =  12495.34754 a.u.^3 (  1851.61945 Ang^3 )
>
> ...
>
> So I got the final enthalpy as reported by pw.x (call this H_reported) 
> and I calculate the enthalpy from total energy using
>
> H = E+PV
> (called H_calc)
>
>
> It turned out that H_reported and H_calc are not equal.
>
> H_reported = -2893.4953806269 Ry = -39367.97664 eV
> H_calc = E+PV=-39368.05688 eV
>
> H_reported - H_calc = -0.08024 eV
>
> So my questions are,
>
> 1) how does the pw.x determine final enthalpy in its report?
> 2) Is it theoretically correct to determine H=E+P*V directly from E, P 
> and V reported in pw.x output. (Or should I determine P(V) by fitting 
> the equation of state to E-V curve then derive H from that?)
> 3) If both the H_reported and H_calc are theoretically correct, why do 
> the values differ?
>
> My problem is not in this single case, it is that I did several relax 
> jobs and the discrepancies in H_reported and H_calc are not systematic.
>
> Thank you,
> Vichawan Sakulsupich
>
> MSc, Chulalongkorn University, THAILAND
>
>
> -- 
> Vichawan Sakulsupich [Print]
>
>
> _______________________________________________
> users mailing list
> users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20180624/6d885f6d/attachment.html>