[molpro-user] error running MRCI in parallel

[William Chua]

Dear Molpro Users,

I came across this error while running an MRCI calculation in a parallel machine:

 USED MEMORY IN CIPRO:              400     18220
 FREE MEMORY IN CIPRO:        649999500 649981680
 ? Error
 ?  Inconsistent memory
 ? The problem occurs in check_address

 GA ERROR fehler on processor   0

My input file goes this way:

  ***,
  memory,650,m

  file,2,test.wfu;
  punch,test.pun,rewind;
  status,2,clear;

  basis=6-31g(d)

  [I'm sorry I'm not at liberty to reveal details about the molecule]

  {gprint,civector;}
 
  {multi;start,2143.2;occ,34;closed,24;wf,58,1,0;
  state,3;
  cpmcscf,nacm,2.1,3.1,accu=1.0d-07,record=5100.1
  cpmcscf,grad,2.1,spin=0.0,accu=1.0d-07,record=5101.1
  cpmcscf,grad,3.1,spin=0.0,accu=1.0d-07,record=5102.1}
 
  {force
  samc,5100.1
  conical,6000.1}
 
  {force
  samc,5101.1
  conical,6000.1}
 
  {force
  samc,5102.1
  conical,6000.1}
 
  {optg,startcmd=multi,stepmax=0.005,maxit=4;inactive,d1,d2,d3,d4,d5,d6,d7,d8,d9,d4_2,d5_2,d6_2,d7_2,d8_2,d9_2}

 reforb=2140.2
 refci=6000.2
 
 {multi;start,reforb;occ,34;closed,24;wf,58,1,0;state,3;
 orbital,reforb}

 {put,molden,test.cas.molf,new}

 {ci;orbital,reforb;occ,34;core,18;closed,24;
 wf,58,1,0;
 state,3;
 save,refci}
  ---

I ran my job on a Primergy RX200S3 machine using 2 nodes with 4 cores per node. The machine has 1.4 GB memory/core. Previous unsuccessful rubs indicate that I needed at least 1000 MW memory for the MRCI.

The header goes this way:

                                         ***  PROGRAM SYSTEM MOLPRO  ***
                         Copyright, University College Cardiff Consultants Limited, 2004

                                    Version 2006.1 linked 24 Mar 2008 16:05:03


 **********************************************************************************************************************************
 LABEL * conical intersection search
 Linux-2.6.9-42.ELsmp/bk-b0180(x86_64) 64 bit mpp version (ifort9.1/mkl_em64t)           DATE:   8-Oct-08         TIME: 12:38:55
 **********************************************************************************************************************************

 Patch level:      152
 **********************************************************************************************************************************

 Variable memory set to  650000000 words,  buffer space   230000 words

The submission script goes as follows:
#!/bin/csh -f
# @$-q queuename
# @$-lP 2
# @$-lp 4
# @$-lm 1.4gb
# @$-eo
# @$-oi
#
cd $QSUB_WORKDIR
set MOLPRO=/usr/local/molpro2006/bin/molpro
set INPUT=test.com
setenv TCGRSH /usr/bin/plesh
#

###########################################
cat $QSUB_NODEINF > hostlist

foreach node (`cat hostlist`)
  echo $USER\:$node\:1 >> tmpfile
end

cat tmpfile | awk '{printf("%s,",$1)}' |sed -n 's/,$/\n/gp' > hostparam

rm tmpfile
###########################################

set WORK=/work/${LOGNAME}/tmp.$$
if ( ! -d ${WORK} ) mkdir -p ${WORK}

${MOLPRO} -d ${WORK} -I ${WORK} -N `cat hostparam` ${INPUT}

if ( -d ${WORK} ) /bin/rm -rf ${WORK}

exit 0

What caused this error and how do I fix it?

Best regards,

William Chua