ABINIT Discussion Forums

Posted: **Fri Jan 18, 2013 5:30 pm**

Dear all,

I am doing Ecut conference study on 80 atom supercell through 13 dataset. The program runs fine for the first dataset. When it start to read the previous WFK it fails and exit. I have build the code --with-mpi-IO. The following is the end of the log file:

-P-0000 hdr_check: WARNING -
-P-0000 Restart of self-consistent calculation need translated wavefunctions.
-P-0000 Indeed, critical differences between current calculation and
-P-0000 restart file have been detected in:
-P-0000 * the plane-wave cutoff
-P-0000 ================================================================================
-inwffil : will read wavefunctions from disk file ecut_xo_DS1_WFK
<Jan 18 07:35:47.563123> BE_MPI (ERROR): The error message in the job record is as follows:
<Jan 18 07:35:47.563233> BE_MPI (ERROR): "killed with signal 11"

Many thanks,

Posted: **Fri Jan 18, 2013 11:25 pm**

Hi,

Sadly, it is not possible to restart a calculation from a density, potential or wavefunction file that was obtained using a different cutoff. More precisely (this is a comment extracted directly form the relevant part of the code, e.g. the subroutine hdr_check) :

Code: Select all

!!****f* ABINIT/hdr_check
!! (...)
!!           self-consistent case : direct restarts
!!           ======================================
!!
!! A direct restart will be allowed provided the following quantities in
!! old and new calculations are the same:
!!
!!   (A) the primitive vectors                             (tprim)
!!   (B) the plane-wave cutoff                             (tecut)
!!   (C) nkpt, kpt(3,nkpt), wtk(nkpt)                      (tkpt)
!!   (D) istwfk(nkpt), the format of wavefunctions         (twfk)
!!   (E) nspinor, the scalar or spinor wf characteristics  (tspinor)
!! For PAW calculations:
!!   (F) the use of PAW method                             (tpaw)
!!   (G) the number of lmn elements for the paw basis      (tlmn)
!!   (H) the energy cutoff for the double (fine) grid      (tdg)
!! For WVL calculations:
!!   (I) the number of wavelets differs                    (twvl)
!!   (J) the space-grid size differs                       (tgrid)

The solution is simple but costs a little extra in CPU time : don't read the wavefunctions from the previous dataset.

Good luck!

Posted: **Sat Jan 19, 2013 12:27 am**

Thank you very much jolafc - and sorry for this stuipd question then - but the only way to know was to read the code!

ABINIT Discussion Forums

failure in reading previous WFK

failure in reading previous WFK

Re: failure in reading previous WFK

Re: failure in reading previous WFK