The ABINIT Users Mailing List ( CLOSED )

[delaire@caltech.edu]

Dear Abinit users,

I am trying to run a parallel calculation with abinip to compute phonon 
response functions.
However, the parallel calculation systematically hangs on some parts of the 
response function calculation (ie, for certain phonon q-points).

Specifically, if I use a kpt grid that is the same as the qpt grid, then the 
full RF calculation works fine (ie, I let it calculate phonons on a 
symmetry-reduced 4x4x4 qpt grid, and I use a 4x4x4 grid for ngkt, see input 
file below).
But if I try to increase to number of kpt in the electronic grid to some 
multiple of the 4x4x4 qpt grid used for phonons, then the computation hangs: 
it appears to run with all cpus at 100%, but nothing gets output anymore.
For example, if I set ngkpt to 8x8x8, then this systematically happens for 
the perturbation in my dataset 4, qpt = 0.5 0.0 0.0.

Also, note that the paral_tests are passed succesfully with our Abinit 
installation.

Your comments on this would be helpful.

On another point, I am not sure when the RF parallelization should be turned 
on (paral_rf  1)? Could you please advise?

Thanks and regards,
Olivier


####################################################
# pure V : computation of the phonon spectrum

  #ndtset  3
  ndtset  9
#Set 1 : ground state self-consistency

  getwfk1   0            # Cancel default
  kptopt1   3            # Automatic generation of k points, no symmetry
                         
  tolvrs1   1.0d-18      # SCF stopping criterion (modify default)
  prtden1   1
    nqpt1   0            # Cancel default
  rfphon1   0            # Cancel default
 
#Q vectors for all datasets

#Complete set of symmetry-inequivalent qpt chosen to be commensurate
# with kpt mesh so that only one set of GS wave functions is needed.
#Generated automatically by running GS calculation with kptopt=1,
# nshift=0, shiftk=0 0 0 (to include gamma) and taking output kpt set
# file as qpt set. Set nstep=1 so only one iteration runs.

     nqpt   1            # One qpt for each dataset (only 0 or 1 allowed)
                         # This is the default for all datasets and must
                         #  be explicitly turned off for dataset 1.

#### grid corresonding to nqpt 4 4 4 reduced by symmetry:

  qpt2                       0       0       0
  qpt3                    0.25       0       0
  qpt4                     0.5       0       0
  qpt5                    0.25    0.25       0
  qpt6                    0.25    0.25    0.25
  qpt7                   -0.25    0.25    0.25
  qpt8                     0.5     0.5    0.25
  qpt9                     0.5     0.5     0.5

#Set 2: phonon calc at Gamma
   kptopt2  2          # k-point set reduced only by time-reversal symetry

#Sets 3-9 : Finite-wave-vector phonon calculations (defaults for all datasets)

   getwfk   1          # Use GS wave functions from dataset1
   kptopt   3          # Need full k-point set for finite-Q response
   rfphon   1          # Do phonon response
  rfatpol   1 1        # Treat displacements of all atoms
    rfdir   1 1 1      # Do all directions (symmetry will be used)
   tolvrs   1.0d-8     # This default is active for sets 2-9

#######################################################################
#Common input variables

#Definition of the unit cell
  acell   3*5.8074        # 
  rprim  -0.5  0.5  0.5   # BCC primitive vectors (to be scaled by acell)
          0.5 -0.5  0.5 
          0.5  0.5 -0.5

#Definition of the atom types
   ntypat   1         # There are two types of atom
    znucl   23        # The keyword "znucl" refers to the atomic number of 
the 
                      # possible type(s) of atom. The pseudopotential(s) 
                      # mentioned in the "files" file must correspond
                      # to the type(s) of atom.

#Definition of the atoms
    natom   1         # There are two atoms
    typat   1         # The first is of type 1 (Al), the second is of type 2 
(As).
                       
     xred   0.0  0.0  0.0

#Gives the number of band, explicitely (do not take the default)
#    nband   4    check the importance of this 

#Exchange-correlation functional

      ixc   11             # GGA-PBE

#Definition of the planewave basis set

     ecut   40.0           # Maximal kinetic energy cut-off, in Hartree

#Definition of the k-point grid
  ngkpt   4  4  4 

# ngkpt   8  8  8   # <<< use this for higher accuracy 
         
  nshiftk   2              # Use one copy of grid only (default)
  shiftk   0.25  0.25  0.25   # shift vectors to apply to the grid
          -0.25 -0.25 -0.25

#  nshiftk    1
#  shiftk    0 0 0


#Definition of occupation numbers
  occopt 4      # "cold smearing" option for occupation of levels
  tsmear 0.005  
  nband  10

#Definition of the SCF procedure
  iscf   5          # Self-consistent calculation, using algorithm 5
  nstep  100         # Maximal number of SCF cycles

# parallelization of the response function calculation:
#paral_rf  1