whether to keep PSTRESS tag in INCAR after relaxation

[alpinnovianus]

Hi, I have a question regarding PSTRESS tag in INCAR.

We use PSTRESS tag, e.g. PSTRESS = 100 to give hydrostatic pressure of 100 kB to the lattice during ionic relaxation (ISIF = 3, NSW > 0)
Now, suppose we've already obtained the optimized structure under this pressure (the smaller lattice parameter, and changes in relative atomic positions, if any)

We then use this optimized structure and say we want to compute the density of states (changing ISMEAR to -5, NSW = 0) for example.
Especially, I am considering metaGGA / hybrid functional case, where a self-consistent electronic minimization is still required (since CHG and CHGCAR can't be used in these cases).

In the DOS calculation, should I keep or remove the PSTRESS = 100 tag in the INCAR?
I am worried if I keep the PSTRESS tag, I might be applying double the pressure, since the lattice volume is already minimized in the DOS structure.
But if I am wrong, then removing the PSTRESS tag would result in the electronic minimization during the DOS computation to be performed without any applied pressure, which is not correct either.

Therefore, please advise whether we should keep or remove the PSTRESS = 100 tag in the INCAR for DOS calculations after using the optimized structure obtained under pressure.

[jonathan_lahnsteiner2]

Dear alpinnovianus,

I does not matter. You will receive the same result in both cases.
There is no explicit stress term in the electronic Hamiltonian. The
stress only enters implicitly via the box shape and the ionic positions
within this box. The box shape is kept fixed during the computation of the DOS
and so the PSTRESS tag won't affect your results.

A nice way to test such things is with the vasp tutorials.

wiki/index.php/Fcc_Si_DOS

You can do the calculation once with and once without the PSTRESS tag and check
for yourself that the result will be the same.I hope this answers your question.

All the best Jonathan