Skip to Content.
Sympa Menu

forum - A problem about high ecut in SCF calculation

forum@abinit.org

Subject: The ABINIT Users Mailing List ( CLOSED )

List archive

A problem about high ecut in SCF calculation


Chronological Thread 
  • From: hyello25@gmail.com
  • To: forum@abinit.org
  • Subject: A problem about high ecut in SCF calculation
  • Date: Tue, 8 Jan 2008 10:09:58 +0100

Dear Abinit users:
When I do SCF calculations for supercell BZO/BTO, if ecut<=160Ha,it is
OK. However, when I use ecut>=170Ha, it gives a error message like this:
ITER STEP NUMBER 1
vtorho : nnsclo_now= 2, note that nnsclo,dbl_nnsclo,istep= 0 0 1
-P-0000 leave_test : synchronization done...
vtorho: loop on k-points and spins done in parallel
/opt/mpich/bin/mpirun.ch_shmem: line 91: 16625 Segmentation fault
/opt/etsf/abinit/5.4/bin/abinip

I test different pps, Teter-extended-norm-conserving and HGH, for both
two, this problem appears. Then I use different K-point mesh, this problem
appears too.

Also, When I do optimization of cell geometry(optcell=1) for cubic
ferromagnetic LaMnO3 with HGH pps, when ecut >= 130Ha, this problem emerge
again! And the erroe message gives:
ETOT 1 -158.14103404663 -1.58E+02 1.01E+00 1.91E+06 0.0E+00 0.00E+00
0.318
scprqt: <Vxc>= -4.2423491E-01 hartree
/opt/mpich/bin/mpirun.ch_shmem: line 91: 3691 Segmentation fault
/opt/etsf/abinit/5.4/bin/abinip

I do the calculation using the version of abinit 5.4.4, operated on
RHEL4.
I don't know what is wrong with it and what does "Segmentation fault"
mean.....
Can you give me a hand?
PS: the file /opt/mpich/bin/mpirun.ch_shmem is given below and I mark the
line 91. Also I give the input file of BZO/BTO supercell and LaMnO3.
Best Regards,
Chen Zhu
Jan. 07. 2008

-----------------------------------------------------------------------------------------
The file /opt/mpich/bin/mpirun.ch_shmem:

#! /bin/sh
exitstatus=1
if [ "$argsset" = "" ] ; then
. $MPIRUN_HOME/mpirun.args
argsset=1
fi
# ch_shmem device
#
# Pass to mpirun that this is a shared memory device. Since there
# may be several, try using the default value.
# This may not be what we want....
MPIRUN_DEVICE=$default_device
export MPIRUN_DEVICE
#
#
# Handle debuggers
#
#startpgm="eval"
MPICH_NP=$np
export MPICH_NP
# If MPICH_NP doesn't work, use -np $np
startpgm="$progname $cmdLineArgs"

if [ "$debugger" != "" ] ; then
if [ "$debugger" = "totalview" ] ; then
if [ "$TOTALVIEW" = "" ] ; then
TOTALVIEW="totalview"
fi
# We need to put -mpichtv into an environment variable
startpgm="$TOTALVIEW $progname -a $cmdLineArgs -mpichtv"
else
# There should be a way to specify a debugger startup method, for
# special vendor-specific debuggers (like SGI's cvd) etc.
# gdb or dbx-like debuggers
dbgfile=$PWD_TRIAL/PId$$
echo "Test" > $dbgfile
if [ -s $dbgfile ] ; then
/bin/rm -f $dbgfile
else
dbgfile=$HOME/PId$$
fi
# If gdb has an equivalent, the documentation is well hidden.
# - .gdbinit - and the documentation IS well hidden. Fortunately,
# the .gbdinit file is read; gdb provides the -nx (instead of -x)
# command if you don't want to read the default init file.
if [ "$debugger" = "dbx" -a -r $HOME/.dbxinit ] ; then
cat $HOME/.dbxinit > $dbgfile
fi
# Regretably, xxgdb does NOT provide this argument (even though it
# claims to).
# if [ "$debugger" = "xxgdb" -a -r $HOME/.gdbinit ] ; then
# cat $HOME/.gdbini > $dbgfile
# fi
# handle USR1 pass for gdb; add an initial break
if [ "$debugger" = "gdb" -o "$debugger" = "xxgdb" -o \
"$debugger" = "ddd" ] ; then
echo "handle SIGUSR1 nostop pass" >> $dbgfile
echo "br MPI_Init" >> $dbgfile
#echo "set args $cmdLineArgs -np $np" >> $dbgfile
echo "set args $cmdLineArgs" >> $dbgfile
echo "run" >> $dbgfile
else
echo "ignore USR1" >> $dbgfile
# This may not work, depending on when the symbol table is read.
# An alternative is to use the startup file to load the
# program with "debug $progname" ,followed with this, followed
# with run command args.
echo "debug $progname" >> $dbgfile
echo "stop in MPI_Init" >> $dbgfile
# echo "run $cmdLineArgs -np $np" >> $dbgfile
echo "run $cmdLineArgs" >> $dbgfile
fi
# cat >> $dbgfile <<.
## Use this command to start your programm:
#run $cmdLineArgs -p4pg $p4pgfile
#.
commandfile=`echo $commandfile | sed -e s@%f@$dbgfile@g`
startpgm="$debugger $commandfile $progname"
fi
fi

#
# Build a suitable command for the whole thing
#
if [ $just_testing = 1 ] ; then
doitall="echo"
else
doitall="eval"
fi
#
$doitall $startpgm #### line 91#####
exitstatus=$?
exit $exitstatus
-------------------------------------------------------------
The input file for BZO/BTO(with Teter-extended-noem-conserving pps given on
abinit.org website):

iscf 5
toldfe 1.0d-10

#Definition of the k-point grids
kptopt 1
nshiftk 1
shiftk 0.5 0.5 0.5
ngkpt 6 6 3


#Definition of the unit cell
acell 7.6881721615E+00 7.6881721615E+00 1.5229661071E+01
rprim 1.0 0.0 0.0
0.0 1.0 0.0
0.0 0.0 1.0

#atoms types
ntypat 4
znucl 56 40 22 8
natom 10
typat 1 1 2 3 4 4 4 4 4 4
xred 0.0000000000E+00 0.0000000000E+00 -9.4442195375E-03
0.0000000000E+00 0.0000000000E+00 5.1272810994E-01
5.0000000000E-01 5.0000000000E-01 2.5165554212E-01
5.0000000000E-01 5.0000000000E-01 7.5161181164E-01
5.0000000000E-01 5.0000000000E-01 -1.1128669085E-02
5.0000000000E-01 0.0000000000E+00 2.5093260203E-01
0.0000000000E+00 5.0000000000E-01 2.5093260203E-01
5.0000000000E-01 5.0000000000E-01 5.1361827983E-01
5.0000000000E-01 0.0000000000E+00 7.5124697051E-01
0.0000000000E+00 5.0000000000E-01 7.5124697051E-01

ecut 170
ixc 3
nband 48
nstep 200
diemac 12.0

--------------------------------------------------------------
The input file for LaMnO3(with HGH pps given on abinit.org website):

toldfe 1.0d-10
kptopt 1
nshiftk 1
shiftk 0.5 0.5 0.5

#Definition of the unit cell
acell 3*3.880 angstrom
rprim 1 0 0
0 1 0
0 0 1

#Definition of the atom types
ntypat 3
znucl 57 8 25

occopt 7
tsmear 0.005

#Definition of the atoms
natom 5
typat 1 2 2 2 3
xred
0.0 0.0 0.0
1/2 1/2 0.0
1/2 0.0 1/2
0.0 1/2 1/2
1/2 1/2 1/2

#Definition of the planewave basis set
ecut 120

nspinor 1
nsppol 2
nspden 2
spinat 0.0 0.0 0.0
0.0 0.0 0.0
0.0 0.0 0.0
0.0 0.0 0.0
0.0 0.0 4.0

ixc 1
nband 53
#Definition of the SCF procedure
nstep 100
diemac 1.0d6


!---------------------------!

G.R. Qian
Physics Dept. NJU CN
!---------------------------!


  • A problem about high ecut in SCF calculation, hyello25, 01/08/2008

Archive powered by MHonArc 2.6.16.

Top of Page