My Community

Hi,
I was hoping to use the low scaling fully self-consistent QSGW routine in VASP 6(.1.1), via the ALGO = QPGWR tag (wiki/index.php/Practical_guide_to_GW_ca ... ons:_QPGWR), but everytime I try to do so, VASP doesn't seem to recognise this tag from the INCAR, and just performs a normal PBE calculation.

This was my INCAR:
NBANDS = 128
ISMEAR = 0
SIGMA = 0.05
ALGO = QPGWR
LOPTICS = .TRUE.
LPEAD = .TRUE.
NELM = 12
EDIFF = 1E-8
NOMEGA = 24

Note that I attempted to rerun with MAXMEM and NTAUPAR set in the INCAR, without LOPTICS and LPEAD, without NBANDS and EDIFF, without NOMEGA, with and without the WAVECAR and WAVEDER files from a previous run present in the calculation directory, none of which worked.
I also tried using the exact same INCAR given on the wiki here: wiki/index.php/Practical_guide_to_GW_ca ... lculations, but got the same result.

KPOINTS:
Automatic mesh
0
Gamma
8 8 8
0. 0. 0.

The system is Li Co O2, with GW POTCARs. I have also tried with a Sn-Sb-I-S system, which also didn't work.

Is there something that I am doing wrong, or is this routine not fully implemented in VASP6 yet? I couldn't find an example case in the testsuite.
Any help with this would be fantastic.

From double checking the Forum guidelines, I realise I need to also add a zip with the jobscript, OUTCAR and stdout, so have attached that plus the input files.

Dear Sean,

ALGO=QPGWR is an experimental feature and shouldn't have been be documented on our wiki. I have changed this.
Thank you for pointing this out.

QPGWR is an experimental feature that is not available in 6.1 yet.
Note, QPGW0R is experimental (and available in 6.1) and is primarily used to determine partially self-consistent Brueckner orbitals (and their eigen-energies).
To determine Quasi-Particle energies within the low-scaling GW algorithm we propose to use ALGO = GW[0]R.

We will update the wiki page (and this thread) as soon as ALGO=QPGWR and QPGW0R is robust and published.
Sorry for the inconvenience.

Hi Merzuk,

Thanks for the reply!

That's unfortunate, I'll have to make do with the other routines available.
Looking forward to seeing ALGO = QPGWR in future VASP 6 versions.
Cheers!
Seán