The ABINIT Users Mailing List ( CLOSED )

["张�s" <zhangting1980323@gmail.com>]

Dear all:

     When I run multistep abinip jobs, there are errors like that in log file:
------------------------------

--------------------------------------------------------------------------------------------
 wfsinp: loop on k-points and spins done in parallel
 pareigocc : MPI_ALLREDUCE
-P-0000  leave_test : synchronization done...
 wfsinp: loop on k-points done in parallel
 newkpt : in paral mode, only node 0 describes in the main output file the copying of wfs.
-P-0000  leave_test : synchronization done...
 newkpt: loop on k-points done in parallel
 pareigocc : MPI_ALLREDUCE


 loper3 : eigenq array
 loper3 : total number of electrons, from k and k+q
  fully or partially occupied states are    8.000000E+01 and    7.999968E+01.
 Initialisation of the first-order wave-functions :
  ireadwf=   0
 newkpt : in paral mode, only node 0 describes in the main output file the copying of wfs.
-P-0000  wfconv :    52 bands set=0 with npw=  53738, for ikpt=   1
-P-0000  wfconv :    52 bands set=0 with npw=  53710, for ikpt=   2
-P-0000  wfconv :    52 bands set=0 with npw=  53738, for ikpt=   3
-P-0000  wfconv :    52 bands set=0 with npw=  53738, for ikpt=   4
-P-0000  leave_test : synchronization done...
 newkpt: loop on k-points done in parallel
 pareigocc : MPI_ALLREDUCE


     iter   2DEtotal(Ha)       deltaE(Ha) residm    vres2

 getcut: wavevector=  0.0000  0.0000  0.0000  ngfft= 180 180  30
         ecut(hartree)=     45.000   => boxcut(ratio)=   2.10286

 ITER STEP NUMBER     1
-P-0000  leave_test : synchronization done...
 vtorho3: loop on k-points and spins done in parallel
 vtorho3 : MPI_ALLREDUCE, buffer of size16250936 bytes
----------------------------------------------------------------------------------------------------------------------------

    And then, the job fails. The job is to get phonon 1st order  perturbation wave functions, at some random q-vector. I've checked some other situations, and found that even with just 1 abinip (without using mpirun command), the same problem still occurs. And I've recompiled the abinip with another version of mpich, which is MPICH2, the problem remains also.
    Can someone tell me what's the reason of breaking out? Thanks!


                                                                                             Zhang Ting
                                                                                             Peking Univ
                                                                                             Feb,16th,2007