Empty Spheres in VASP6 with spin polarized calculations in parallel

[roberto.robles]

Dear VASP team,

I believe I have found a bug regarding calculations using Empty Spheres. When doing a parallel calculation for a spin polarized system using VASP 6.4.1 the DOS extracted from DOSCAR shows a wrong value for the spin down channel. The attached example is for a non magnetic system (just a Au surface) so up and down channels should be equal. As can be seen in the attached images, the result is correct for a serial calculation, but not for a parallel one. In addition the result is correct for VASP 5.4.4 both in serial and in parallel.

In addition DOS is wrong not only for the empty spheres, but also for as many atoms as empty spheres you have included. In this example, there are 4 atoms and one empty sphere, so the results for the empty sphere and the last atom are wrong.

Best regards,

Roberto

[merzuk.kaltak]

Dear Roberto,

Empty spheres are deprecated and not supported by vasp-6 anymore.
We suggest using LOCPROJ that is much more versatile.
LOCPROJ allows to project orbitals on any site or any set of coordinates with defined orbital character.

[roberto.robles]

Thank you for the answer.

Could you provide an example for LOCPROJ which gives "exactly" the same projection as the Empty Spheres method?

For empty spheres the corresponding DOS was simply written to the DOSCAR file. Is there a provided tool to extract DOS from PROJCAR/LOCPROJ or we should write our own program?

Best,

Roberto