[Pw_forum] inconsistent number of sticks in nscf calculation

[Paolo Giannozzi]

Hi Hande

> Our initial relaxation run where take an automatic kpoint set of  
> 6x6x1 ran
> smoothly without problems. When we try to run a subsequent nscf  
> calculation
> with manually chosen kpoints however, we get the following error :
>
>      from  fft_dlay_set  : error #         7
>       inconsistent number of sticks

this is a bug, or more exactly, a case of inability to cope with weird
input data. The problem is not present in serial execution, or if k- 
points
are automatically calculated; it is present only in parallel execution
(even on a single processor) and if k-points are given as a list, and if
they are way beyond the first Brillouin Zone. If you give a list of  
"good"
k-points the error disappears. Remember that k-points are by default in
units of 2pi/a, where a = lattice parameter, but you set a=1! this is  
perfectly
legitimate as long as everything is consistent with this defintion,  
in particular
atomic positions and k-points.

The serial and parallel-on-a-processor cases use different FFTs and
different algorithms to distribute the "sticks" (columns in G-space).
Apparently the serial algorithm is more robust and better tolerates
strange k-points in input.

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