test

#!/bin/sh

export basedir=$PWD

# compile lda code
cd lib/lda-c-dist; make || exit

# compile using ant
cd $basedir; ant -f build.xml || exit

#data = {'med.out.1', 'test.5000.data'}
data='med.out'
if [ "$1" ];then
    data=$1
fi
userdata=$data.`whoami`

echo $userdata

# download test.data if it doesn't exist

if test ! -f /tmp/$data.`whoami`; then
   cd /tmp/
   wget http://www.falsifian.org/a/9xwk/$data --output-document=$userdata
   cd $basedir
fi

out_dir=experiment-$(date -u +%F.%T%Z)
if [ "$2" ];then
    out_dir=$2
fi
mkdir $out_dir || exit
echo Output will go in $out_dir

#0.1,0.3,0.5,0.7,0.9 \
args=\
"-Dplusone.dataFile=/tmp/$userdata \
 -Dplusone.crossValidation.FOLD=10 \
 -Dplusone.crossValidation.run=false \
 -Dplusone.testWordPercents=0.3 \
 -Dplusone.trainPercents=0.95\
 -Dplusone.kValues=1 \
 -Dplusone.closestKValues=5,10,25,50,100,250,500 \
 -Dplusone.closestKSVDishValues=5,10,25,50,100,250,500 \
 -Dplusone.svdDimensions=5,15,25 \
 -Dplusone.dtrw.walkLength=1 \
 -Dplusone.dtrw.stoch=true \
 -Dplusone.dtrw.nSampleWalks=100 \
 -Dplusone.dtrw.normalizeDocsInWord=true \
 -Dplusone.dtrw.normalizeWordsInDoc=true \
 -Dplusone.svdishknn.nLevels=3
 -Dplusone.svdishknn.docEnzs=20,60,120
 -Dplusone.svdishknn.termEnzs=40,120,240
 -Dplusone.svdishknn.dtNs=10,20,50
 -Dplusone.svdishknn.tdNs=10,20,50
 -Dplusone.svdishknn.numLVecs=800,400,200
 -Dplusone.svdishknn.walkLength=3
 -Dplusone.localCO.docEnzs=1500
 -Dplusone.localCO.termEnzs=15000
 -Dplusone.localCO.dtNs=500
 -Dplusone.localCO.tdNs=1200
 -Dplusone.enableTest.localCO=false \
 -Dplusone.enableTest.lda=false \
 -Dplusone.enableTest.knnc=false \
 -Dplusone.enableTest.knn=false \
 -Dplusone.enableTest.knnrw=false \
 -Dplusone.enableTest.dtrw=false \
 -Dplusone.enableTest.baseline=true \
 -Dplusone.enableTest.lsi=false \
 -Dplusone.enableTest.svdknn=false \
 -Dplusone.enableTest.svdishknn=true \
 -Dplusone.outPath=$out_dir \
 -Dplusone.dumpMeta=false \
 -Dplusone.randomSeed=9"

# options to sample cpu
# -agentlib:hprof=cpu=samples
# -agentlib:hprof=heap=sites

# options for memory
# -Xmx4096m

#java -agentlib:hprof=cpu=samples -Xmx4096m -cp Plusone.jar:lib/jars/ejml-0.17.jar $args plusone.Main
java -Xmx4096m -cp Plusone.jar:lib/jars/ejml-0.17.jar:lib/jars/gson-1.7.1.jar:lib/colt-1.2.0.jar:lib/commons-lang-2.6.jar $args plusone.Main