PATHSAMPLE: Difference between revisions

From Docswiki
Jump to navigation Jump to search
import>Dw34
(New page: Example setup for BLJ60 - DJW 1/2/11 In this example the objective was to link a crystal structure to a minimum from the amorphous region. OPTIM was first used to create min.data.info fi...)
 
import>Csw34
(Replacing page with '= Example input = * BLJ60 example input')
Line 1: Line 1:
Example setup for BLJ60 - DJW 1/2/11
= Example input =


* [[BLJ60 example input]]
In this example the objective was to link a crystal
structure to a minimum from the amorphous region. OPTIM was first used to
create min.data.info files using odata files such as:

DUMPDATA
BFGSSTEPS 5000
BFGSMIN
ENDHESS
MAXBFGS 0.2 2.0
UPDATES 100 100
BFGSCONV 0.000001
BINARY 48 1.5 0.5 0.8 0.88
PARAMS 3.587037905 3.587037905 3.587037905 1.79
NORESET
POINTS
LS -11.46001238000000 5.188783135000000 10.47622879000000
LS -2.955568051000000 1.530923988000000 8.223564292000001
etc.

This produced min.data.info.finish and min.data.info.start, which were copied
to a clean directory and read into min.data and points.min using READMIN with
this pathdata file:

NATOMS 60
COPYFILES perm.allow
BULK 3.587037905 3.587037905 3.587037905
PERMDIST
SYSTEM LS
TEMPERATURE 0.71
CONNECTIONS 1
SEED 1
PERTURB 0.40
ETOL 1.0D-7
ITOL 0.01D0
DIRECTION AB
EXEC /home/wales/bin/OPTIM
TRIPLES
ADDTRIPLES
STARTTRIPLES
READMIN min.data.info.start
CYCLES 0
CPUS 1

A single OPTIM connect run was then started for DIJINITCONT. This initial
cycle is necessary before tryin multiple cycles, because there is only one
possible connection attempt with only two minima. The relevant pathdata lines were:
DIJINITCONT EXP
CYCLES 100
JOBSPERNODE 1
PAIRLIST 1
with a qsub file:
#!/bin/csh
#PBS -N BLJ60.1.3.37
#PBS -q h8
#PBS -j oe
#PBS -l nodes=1:ppn=8

cd $PBS_O_WORKDIR
cat $PBS_NODEFILE >& output
wc output > nodes.info
cat $PBS_NODEFILE >> nodes.info
echo $USER >> nodes.info
echo $PBS_O_WORKDIR >> nodes.info

/home/wales/bin/PATHSAMPLE >>& output

echo
qstat -f ${PBS_JOBID}@volkhan
echo

A subsequent run with CYCLES 100 on 8 cores found an initial path in five cycles,
which took a few minutes wall clock time.

Revision as of 15:42, 1 February 2011

Example input