I am attempting to accurately calculate the DOS for a certain configuration of sodium and niobate. I performed a structural relaxation successfully, copied the CONTCAR to POSCAR, and edited the INCAR file for a single step, SCF calculation. At first I was receiving the infamous ZHEGV error after 1 to 5 electronic iterations, which I attempted to resolve in numerous ways. I changed ALGO from Fast to Normal in order to avoid using the RMM-DIIS algorithm. I deleted the old WAVECAR file. I changed the number of processors I used, etc. None of these solutions worked, individually or together.
Now, instead of terminating VASP early and printing a ZHEGV error, the computer simply runs for the entire allotted wall time without doing a single electronic step, prints empty output files, and terminates regularly with no error output. Anyone have any ideas?
Relevant files -
INCAR:
PREC=Accurate
ALGO=Fast
EDIFF=1E-4
ENCUT=500 eV
NPAR=8
NSIM=8
IBRION=-1
ISIF=2
LORBIT=11
NEDOS=5000
ISMEAR=-5
KPOINTS:
Automatic generation
0
G
2 2 2
0. 0. 0.
Single Step Calculation Failure, no error output
Moderators: Global Moderator, Moderator
-
- Newbie
- Posts: 2
- Joined: Wed Jul 09, 2014 4:23 pm
- License Nr.: 5-55
Single Step Calculation Failure, no error output
Last edited by Lithium on Wed Jul 09, 2014 4:42 pm, edited 1 time in total.
-
- Newbie
- Posts: 2
- Joined: Wed Jul 09, 2014 4:23 pm
- License Nr.: 5-55
Single Step Calculation Failure, no error output
PROBLEM RESOLVED:
It turns out that there was no error, it's just that VASP was unable to complete a single electronic optimization step within the time limit given. A more efficient parallelization using NCORE=24 allowed VASP to finish the entire job in less than two hours. I recommend NCORE=24 for anyone using NERSC's Hopper (Cray) machine.
It turns out that there was no error, it's just that VASP was unable to complete a single electronic optimization step within the time limit given. A more efficient parallelization using NCORE=24 allowed VASP to finish the entire job in less than two hours. I recommend NCORE=24 for anyone using NERSC's Hopper (Cray) machine.
Last edited by Lithium on Thu Jul 10, 2014 5:58 pm, edited 1 time in total.