Crashing Structure & "hit a member that was already found in another star"

[Greekman]

Hello,

i have again a problem in one binary 2D-Alloy structure A1B5:

POSCAR:
-Vol
3.0 -1.0 0.0
2.0 1.0 0.0
0 0 6.192497648827324473
1 4
C
0 0 0 A
1 0 0 B
2 0 0 B
3 0 0 B
4 0 0 B

with KPOINTS:
Gamma
7 7 1
0 0 0


the Problem is that
1) in my OUTPUT-FILE the Sentence
"hit a member that was already found in another star"
is written really alot! and the outputfile is than around 10GB!!
i checked in the Forum than i can ignore it ... but it's really almost the whole outputfile!
2) after the first 7-10 electronic steps VASPs stops or crashed!

I am calculating on a Linux-based Cluster each time with 32 CPUs

Does anybody know what could be the problem?

thank you!

[admin]

the easiset way is to set ISYM=0 to avoid this error.

[sven]

I have also encountered this error, but only when I used a geometry according to the following lattice vectors

5.8380000000000000 0.0000000000000000 0.0000000000000000
-1.4595000000000000 2.5279281536467763 0.0000000000000000
0.0000000000000000 0.0000000000000000 53.6633861430581049

The absolute lengths of the vectors are of course irrelevant, but |a1|=2|a2|, where a1 (a2) is the first (second) lattice vector.

If NGX != 2 NGY, the "hit a member" error arises.

So the solution in my case is either to explicitly set NGX and NGY in the INCAR file, so that NGX = 2 NGY or to set PREC = High or PREC = Accurate, in which case the relation NGX = 2 NGY is fulfilled by the default choice of NGX and NGY done by Vasp.

Maybe the solution to your problem can be found along the same lines, i.e. a careful choice of NGX and NGY.
<span class='smallblacktext'>[ Edited Fri May 16 2008, 01:48AM ]</span>

[admin]

please have a look whether it helps to set SYMPREC to the precision which has to be accounted for the read lattice vectors of the Bravais matrix. The default value is 10-5 this might not be enough (within the numerical accuracy) if you give the Bravais matrix with an accuracy of 10-16

[sven]

No, I have tried varying SYMPREC from 1E-3 down to 1E-16. From 1E-16 and lower, I got another error:

VERY BAD NEWS! internal error in subroutine SGRCON:
Found some non-integer element in rotation matrix 1

This error makes Vasp exit, so I have not been able to try smaller values of SYMPREC. However, none of the SYMPREC settings above 1E-16 resolved the "hit a member" error. The only thing that seems to help is to always let NGX = 2 NGY.