EDDDAV Error in exact diagonalization (ALGO=Exact)

[juhL]

Hi everyone,

i'm trying to get the RPA correlation energy for a water molecule according to http://cms.mpi.univie.ac.at/vasp/vasp/g ... rgies.html

In step three I get a Error EDDDAV: Call to ZHEGV failed. Returncode = 1 2 32

Using the gamma-point VASP version everything works out fine up to increasing NBANDS by 1.7, however, using NBANDS*2 (or the complex VASP version) results in the above error. So I think it is a memory problem, but the error message makes me curious, as it usually hints to an instability of the algorithm for a specific system. Any hints/tipps how to get this calculation done? For completeness, i'm using the following Input:

GGA = PE
ENCUT = 415
EDIFF = 1E-08
NBANDS = 19213
ALGO = Exaxt
NELM = 1
LOPTICS= .TRUE.

The WAVECAR of the DFT run is used as well as PBE_GW POTCARs. Calculation runs fine for NBANDS = 9607, 12489, 14410, and 16332

[admin]

ALGO = Exaxt ?

[juhL]

... Sorry to bother, thanks a lot ...