[CP2K:3088] hfx_communication: distribute_ks_matrix

[Urban Borštnik]

Hello,

It's hard to tell without a complete code listing.

Cheers,
Urban

On Mon, 2011-02-07 at 07:29 -0800, Andrey wrote:
> Dear CP2K developers,
> 
> The program crashes in the hfx_energy_potential.F, during the work of
> integrate_four_center(...) routine
> (line: 1888)
> ...
>        CALL dbcsr_printmat(full_ks,SIZE(full_ks,1),1)
> The full_ks matrix, taken from the qs_env%x_data%full_ks_alpha ... is
> initialized.
> 
>        CALL distribute_ks_matrix(para_env, full_ks, ks_matrix(ispin)
> %matrix, shm_number_of_p_entries, &
>                                  shm_is_assoc_atomic_block,&
>                                  shm_block_offset, natom,
> last_sgf_global, kind_of, basis_parameter, &
>                                  off_diag_fac=0.5_dp,error=error)
> ...
>      CALL cp_dbcsr_print (ks_matrix(ispin)%matrix,
> matlab_format=.TRUE., variable_name="ks_matrix(ispin)%matrix",
> error=error)
> ...
> 
> After passing the distribute_ks_matrix(...)(without errors), the
> full_ks matrix must be distributed to the ks_matrix(ispin)%matrix.
> However, it does not happen. The output variable ks_matrix(ispin)
> %matrix remains uninitialized and that results to the critical error.
> 
> It would be great if you could give me some advice regarding the
> problem
> 
> Best wishes,
> Andrey
>