problem in HSE calculation!

Message

fzcsydl · #1 Post by **fzcsydl** » Tue May 25, 2010 11:00 pm

Hi all,

When I do HSE calculation, I always get error like:
ROT: 52 -0.512664845560E+05 0.16667E+04
ROT: 53 -0.529331275721E+05 -0.16666E+04
ROT: 54 -0.512664880920E+05 0.16666E+04
ROT: 55 -0.529331222301E+05 -0.16666E+04
ROT: 56 -0.512664904293E+05 0.16666E+04
ROT: 57 -0.529331186988E+05 -0.16666E+04
ROT: 58 -0.512664919747E+05 0.16666E+04
ROT: 59 -0.529331163645E+05 -0.16666E+04
ROT: 60 -0.512664929960E+05 0.16666E+04
gam=****** g(H,U,f)= 0.256E+01 0.170E+02 0.270E+04 ort(H,U,f) = 0.233E+01-0.105E+02-0.190E+04
LAPACK: Routine ZTRTRI failed! 1
LAPACK: Routine ZTRTRI failed! 1
LAPACK: Routine ZTRTRI failed! 1
LAPACK: Routine ZTRTRI failed! 1

What does "LAPACK: Routine ZTRTRI failed!" mean?
how to resolve this problem?

Thanks a lot!

panda · #2 Post by **panda** » Wed May 26, 2010 2:30 am

I have had trouble with HSE calculations in the past, but last week finally found some information on the forums after performing a search on the bug reports section of the forum - apparently you have to set your stack size to unlimited via the command ulimit -u unlimited to allow the arrays to behave appropriately. You are getting a LAPACK error because you are using LAPACK as your linear algebra dll and because your stack size has a cap then when the routine ZTRRTRI is called from the VASP code you get an error. I am using the mkl and my error (before setting the stack size to unlimited) was just "stacking fault" lol that is always what my errors are - not very helpful, eh?

fzcsydl · #3 Post by **fzcsydl** » Wed May 26, 2010 10:59 pm

Thanks.
I am not sure whether this is the problem or maybe this is a problem with my model system. BTW, what do you mean by stack size limitation? It is about memory or what else?