I'm running DFTP on 280 CPUs with 1024.0GB mem. VASP stop with following error
internal error in: hamil_lrf.F at line: 815
LRF_HAMIL internal error: the vector H(1)-e(1) S(1) |phi(0)> is not |
| orthogonal to |phi(0)> (-0.126, 0.008) 0.056
I have attached my input files
Best regards,
Ziqi
Dear Ziqi Yin,
This seems to be a problem in the electronic optimization algorithm. Have you already tried using a different optimization algorithm. Please try to remove the IALGO tag from your INCAR file and set ALGO = All.
All the best Jonathan
Dear Jonathan,
I have tried to change IALGO to ALGO=All, the error disappears.
but I got a new error return when I tried to run parallel :
I set my INCAR as follows and requested 4 nodes total 112 cpus
SIGMA = 0.10000
EDIFF = 1.00e-06
EDIFFG = -1.00e-04
ENCUT = 550
ALGO = All
ADDGRID = True
GGA = PS
ISPIN = 2
PREC = Accurate
IBRION = 7
ISMEAR = 0
NSW = 1
LWAVE = .FALSE.
LCHARG = .FALSE.
LEPSILON = .TRUE.
LREAL = .False.
NPAR = 4
ERROR message :
VASP internal routines have requested a change of the k-point set.
Unfortunately, this is only possible if NPAR=number of nodes.
Please remove the tag NPAR from the INCAR file and restart the
calculation.
I have tried many different NCORE and NPAR combinations as well as IBRION = 7 or 8, only if I remove NPAR and NCORE the code runs normally. How can I run DFPT parallel?
Many thanks
Yours,
Ziqi
NPAR and NCORE are related, i.e. they control the same thing.
Because of a limitation in the implementation, NCORE/=1 cannot yet be used for DFPT calculations.
We plan to fix this in an upcoming release.
In the meantime, you can run the code without setting NCORE or NPAR in the INCAR.
This will default to band parallelism, which might not be the most efficient.
If you want to use some additional parallelism, we recommend setting k-point parallelism KPAR/=1