[Pw_forum] Advanced error report about vc-relax of pw.x

[Paolo Giannozzi]

On Tue, 2011-04-12 at 14:34 +0800, GAO Zhe wrote:

>      from sym_rho_init_shell : error #         2
>      lone vector

this is a consequence of non-local interactions between recent
changes to the way G-vectors are computed and the final scf step,
performed in variable-cell optimization only, with G-vectors set 
for the final cell (and not for the initial cell as it is done during
the optimization). Apparently in this last step some G-vectors are
"lost" and the algorithm that collects G-vector shells complains.
 
For the time being, waiting for a better solution, increase the
FFT grid to nr1=32, nr2=32, nr3=32. Also note that you can get the 
final coordinates and lattice parameters from output and perform
the final scf calculation (as it used to be done some time ago)
if you want. The last step is done only to make it clear what
the result will look like with the final cell.

P.
-- 
Paolo Giannozzi, IOM-Democritos and University of Udine, Italy