[Pw_forum] Out of range subscript for igk_l2g_kdip

Paolo Giannozzi giannozz at democritos.it
Tue Sep 30 22:24:39 CEST 2008


On Sep 30, 2008, at 21:49 , Brad Malone wrote:

> The error comes about because in the subroutine write_this_wfc the  
> subroutine write_wfc is called, which has as one of its arguments  
> igk_l2g_kdip(:, ik-iks+1). iks is the first index of the first k  
> point in a particular pool. However, ik in this routine runs over  
> ALL the kpoints, thus indexing negative values (e.g., ik goes from  
> 1-28 while iks goes from 17-18). If debugging flags are not turned  
> on the program doesn't crash and I haven't tested to see what  
> difference in might make in the output. But turning debugging flags  
> on catches this error and exits the program.

it is a known problem, or more exactly, it is known not to be a  
problem in practise.
The subscript goes out of bound only on those processors that do not  
use the
information contained in the out-of-bound array. It is ugly but  
harmless. Thank
you anyway: we'll think how to fix it. I know a simple solution that  
would simplify
a lot the horrible mess of collecting the wavefunctions with pools of  
processors,
but it works well only in presence of parallel file systems

Paolo
---
Paolo Giannozzi, Dept of Physics, University of Udine
via delle Scienze 208, 33100 Udine, Italy
Phone +39-0432-558216, fax +39-0432-558222





More information about the Pw_forum mailing list