calculate PARCHG, erro reading WAVCAR

Message

bossten · #1 Post by **bossten** » Mon Dec 12, 2005 2:34 am

when I was calculating PARCHG, some erros appeared like this:
found WAVECAR, reading the header
nup: number of bands has changed, file:1160 present:1168
POSCAR, INCAR and KPOINTS ok, starting setup
FFT: planning ... 1
reading WAVECAR
WARNING: random wavefunctions but no delay for mixing, default for NELMDL

when I set NBANDS in INCAR , the same erro can't be avoided, how should I do? thanks

bossten · #2 Post by **bossten** » Mon Dec 12, 2005 10:31 am

is it the reason that when doing the calculation in different number of nodes (for example, WAVECAR is abtained on 2 nodes, and now calculate the PARCHG on 4 nodes), NPAR will be changed and accordingly NBANDS is also changed? let me check it

superyoyo · #3 Post by **superyoyo** » Wed Dec 14, 2005 4:08 am

Hi
I never calculate PARCHG on different nodes and it go well.
Probably , your INCAR file must contain lines
LPARD =.TRUE. (you know it) and do no relaxing (NSW=0)
I think when calculate PARCHG VASP does not re-calculate any thing except read WAVECAR and get PARCHG. So you can calculate on one node. (not waste your time).

^_^

Good Luck

#4 Post by **admin** » Fri Dec 16, 2005 2:31 pm

please make sure that you use the same number of bands in the follow-up run (by setting NBANDS) as in the run by which WAVECAR was written.
NBANDS is a integer multiple of the number of nodes, therefore it may change slightly if you do a calculation on a different
number of nodes, even if you use exactly the same input. I therefore also suggest not to change the number of nodes for the follow-up run.

bossten · #5 Post by **bossten** » Fri Dec 16, 2005 3:13 pm

it's very grateful of you, when i used the same number of nodes, vasp read the wavecar correctly.the feeling is great when you find the problem and get rid of it. thanks