KEEP_ACC="false"
INIT_EM_ITER=1
CPU=linux
GMTKEMTRAIN=/homes/chiaping/src/gmtk/tksrc/gmtkEMtrain
WORKDIR=.
PARAMS_DIR=$WORKDIR/PARAMS
LEARNED_PARAMS_DIR=$WORKDIR/LEARNED_PARAMS
MISC_DIR=$WORKDIR/MISC; if [! -d $MISC_DIR ]; then mkdir $MISC_DIR; fi
OBS_FILE=$WORKDIR/DATA/train_multi.scp
MASTER_FILE=$PARAMS_DIR/masterFileTrain.params
INITIAL_GMP=$PARAMS_DIR/initialParams
EMOUT_FILE=$LEARNED_PARAMS_DIR/learned_params_multi.gmp
STRFILE=$PARAMS_DIR/aurora_train.str
nV=1e20 ; split=1e-15; nS=1e20
MCVR_ARRAY="$nV $nV $nV $nV $nV $nV $nV  $nV   $nV $nV $nV $nV $nV $nV $nV   $nV   $nV $nV $nV $nV $nV $nV $nV  $nV   $nV $nV $nV $nV $nV $nV $nV  $nV   $nV $nV $nV $nV $nV $nV $nV $nV"
MCSR_ARRAY="$nS $nS $nS $nS $nS $nS $nS  $split  $nS $nS $nS $nS $nS $nS $nS   $split  $nS $nS $nS $nS $nS $nS $nS  $split  $nS $nS $nS $nS $nS $nS $nS  $split  $nS $nS $nS $nS $nS $nS $nS $nS"

MEANCLONEFRAC=0.25
VARCLONEFRAC=0.0
DLINKCLONEFRAC=0.0
STREAM1_PARAMS=" -of1 $OBS_FILE -nf1 39 -fmt1 htk -iswp1 true "
MISC_PARAMS="-dlinkCloneSTDfrac $DLINKCLONEFRAC"
LABEL_FILE=$WORKDIR/labels/Multi08TR_sp.mlf
UTT_FILE=$WORKDIR/labels/multi_train.list
GENERATE_CHUNK_DTS="$WORKDIR/scripts/generate_chunk_dts ${LABEL_FILE} ${UTT_FILE}"

NUM_TRN_SENTS=8440
MAX_EM_ITER=40
LOG_LIKE_THRESH=0.001
EMTRAIN_PARALLELISM=40
EMTRAIN_CHUNKS=80
NUM_LOCAL=0

NODES="bird1 OR bird2 OR bird3 OR bird4 OR bird5 OR bird6 OR bird7 OR bird8 OR bird9 OR bird10 OR bird11 OR bird12 OR bird13 OR bird14 OR bird16 OR bird17 OR bird18 OR bird19 OR bird20 OR bird21 OR bird22 OR bird23 OR bird24 OR bird25 OR bird26 OR bird27 OR bird28 OR bird33 OR bird34 OR bird35"

#!/bin/bash
set -x
. $1
BREAK_RANGE=/g/ssli/research/gmtk/scripts/breakRange.pl
PMAE=/g/ssli/research/gmtk/scripts/pmae
CP=/bin/cp
GENERATE=/g/ssli/research/gmtk/scripts/generate_masterfile.pl
if [ ! -s $EMOUT_FILE ];
then
	MAX_TRN_SENT=`expr $NUM_TRN_SENTS - 1`
	SENT_RNG="0:$MAX_TRN_SENT"
	SENT_RNGS=`$BREAK_RANGE -rng $SENT_RNG -n $EMTRAIN_CHUNKS`
	FOR_RANGE=`perl -e "print join(' ', 1..$EMTRAIN_CHUNKS)"`

	for i in $FOR_RANGE;
	do
		CUR_RNG=`echo $SENT_RNGS | awk "{ print \\$$i }"`
		CUR_MASTER_FILE=$MISC_DIR/masterfile.$CUR_RNG.params
		$GENERATE_CHUNK_DTS $CUR_RNG $MISC_DIR
		$GENERATE $CUR_RNG $MASTER_FILE > $CUR_MASTER_FILE
	done

	MAX_EM_ITER_NO_LL=`perl -e "@mcvr = split(/\s+/, '$MCVR_ARRAY'); \\$num = @mcvr; print \\$num;"`
	if [ $MAX_EM_ITER_NO_LL -gt $MAX_EM_ITER ] ;
	then
		MAX_EM_ITER_NO_LL=$MAX_EM_ITER
	fi

	EM_ITER=$INIT_EM_ITER
	LL_OUT_FILE=$MISC_DIR/loglikestorage.EM${EM_ITER}.data
	CUR_OUT_FILE=$MISC_DIR/learned_params"$EM_ITER".gmp
	if [ $EM_ITER -le 1 ];
	then
		CUR_IN_FILE=$INITIAL_GMP
	else
		LAST_EM_ITER=`expr $EM_ITER - 1`
		CUR_IN_FILE=$MISC_DIR/learned_params"$LAST_EM_ITER".gmp
	fi

	while [ $EM_ITER -le $MAX_EM_ITER_NO_LL ];
	do
		if [ $KEEP_ACC = "true" ]
		then
			ABSTRACT_ACC_FILE=$MISC_DIR/acc_file_${EM_ITER}_@D.data
		else
			ABSTRACT_ACC_FILE=$MISC_DIR/acc_file_@D.data
		fi

		LL_OUT_FILE=$MISC_DIR/loglikestorage.EM${EM_ITER}.data
		CUR_OUT_FILE=$MISC_DIR/learned_params${EM_ITER}.gmp
		(
		for i in $FOR_RANGE;
		do
			if [ $KEEP_ACC = "true" ]
			then
				ACCFILE=$MISC_DIR/acc_file_${EM_ITER}_$i.data
			else
				ACCFILE=$MISC_DIR/acc_file_$i.data
			fi

			CUR_RNG=`echo $SENT_RNGS | awk "{ print \\$$i }"`
			CUR_MASTER_FILE=$MISC_DIR/masterfile.$CUR_RNG.params

			echo "$GMTKEMTRAIN $STREAM1_PARAMS $STREAM2_PARAMS $STREAM3_PARAMS -inputMasterFile $CUR_MASTER_FILE -inputTrainableParameters $CUR_IN_FILE -binInputTrainableParameters false -strFile $STRFILE -maxEmIters 1 -random false -trrng $CUR_RNG -storeAccFile $ACCFILE -accFileIsBinary true $MISC_PARAMS "
		done
		) | $PMAE "$NODES" | tee $MISC_DIR/emacc.EM${EM_ITER}.makefile | pmake -L $NUM_LOCAL -J $EMTRAIN_PARALLELISM -f - > $MISC_DIR/pmake_emtrain.${EM_ITER}.out 2>&1
		CUR_MCVR=`echo $MCVR_ARRAY | awk "{ print \\$$EM_ITER }"`
		CUR_MCSR=`echo $MCSR_ARRAY | awk "{ print \\$$EM_ITER }"`

		RNG=`echo $SENT_RNGS | awk "{ print \\$1 }"`
		FIRST_MASTER_FILE=$MISC_DIR/masterfile.$RNG.params

		$GMTKEMTRAIN $STREAM1_PARAMS $STREAM2_PARAMS $STREAM3_PARAMS -inputMasterFile $FIRST_MASTER_FILE -inputTrainableParameters $CUR_IN_FILE -binInputTrainableParameters false -outputTrainableParameters $CUR_OUT_FILE -binOutputTrainableParameters false -strFile $STRFILE -maxEmIters 1 -mcvr $CUR_MCVR -mcsr $CUR_MCSR -loadAccFile $ABSTRACT_ACC_FILE -loadAccRange "1:$EMTRAIN_CHUNKS" -trrng nil -accFileIsBinary true -llStoreFile $LL_OUT_FILE -random false -meanCloneSTDfrac $MEANCLONEFRAC -covarCloneSTDfrac $VARCLONEFRAC $MISC_PARAMS >> $MISC_DIR/pmake_emtrain.${EM_ITER}.out 2>&1
		
		CUR_IN_FILE=$CUR_OUT_FILE
		EM_ITER=`expr $EM_ITER + 1`
	done

	LAST_EM_ITER=`expr $EM_ITER - 1`
	LAST_LL_OUT_FILE=$MISC_DIR/loglikestorage.EM${LAST_EM_ITER}.data
	 
	LAST_LL=`cat $LAST_LL_OUT_FILE`
	MCVR=`echo $MCVR_ARRAY | awk "{ print \\$$MAX_EM_ITER_NO_LL }"`
	MCSR=`echo $MCSR_ARRAY | awk "{ print \\$$MAX_EM_ITER_NO_LL }"`

	CUR_LL_RATIO=100000000
	CONVERGED=0

	while [ $EM_ITER -le $MAX_EM_ITER ] && [ $CONVERGED -eq 0 ] ;
	do
		if [ $KEEP_ACC = "true" ]
		then
			ABSTRACT_ACC_FILE=$MISC_DIR/acc_file_${EM_ITER}_@D.data
		else
			ABSTRACT_ACC_FILE=$MISC_DIR/acc_file_@D.data
		fi

		LL_OUT_FILE=$MISC_DIR/loglikestorage.EM${EM_ITER}.data
		CUR_OUT_FILE=$MISC_DIR/learned_params${EM_ITER}.gmp

		(
		for i in $FOR_RANGE;
		do
			if [ $KEEP_ACC = "true" ]
			then
				ACCFILE=$MISC_DIR/acc_file_${EM_ITER}_$i.data
			else
				ACCFILE=$MISC_DIR/acc_file_$i.data
			fi

			CUR_RNG=`echo $SENT_RNGS | awk "{ print \\$$i }"`
			CUR_MASTER_FILE=$MISC_DIR/masterfile.$CUR_RNG.params

			echo "$GMTKEMTRAIN $STREAM1_PARAMS $STREAM2_PARAMS $STREAM3_PARAMS -inputMasterFile $CUR_MASTER_FILE -inputTrainableParameters $CUR_IN_FILE -binInputTrainableParameters false -strFile $STRFILE -maxEmIters 1 -random false -trrng $CUR_RNG -storeAccFile $ACCFILE -accFileIsBinary true $MISC_PARAMS"
		done
		) | $PMAE "$NODES" | tee $MISC_DIR/emacc.EM${EM_ITER}.makefile | pmake -L $NUM_LOCAL -J $EMTRAIN_PARALLELISM -f - > $MISC_DIR/pmake_emtrain.${EM_ITER}.out 2>&1

		RNG=`echo $SENT_RNGS | awk "{ print \\$1 }"`
		FIRST_MASTER_FILE=$MISC_DIR/masterfile.$RNG.params

		$GMTKEMTRAIN $STREAM1_PARAMS $STREAM2_PARAMS $STREAM3_PARAMS -inputMasterFile $FIRST_MASTER_FILE -inputTrainableParameters $CUR_IN_FILE -binInputTrainableParameters false -outputTrainableParameters $CUR_OUT_FILE -binOutputTrainableParameters false -strFile $STRFILE -maxEmIters 1 -mcvr $MCVR -mcsr $MCSR -loadAccFile $ABSTRACT_ACC_FILE -loadAccRange "1:$EMTRAIN_CHUNKS" -trrng nil -accFileIsBinary true -llStoreFile $LL_OUT_FILE -random false -meanCloneSTDfrac $MEANCLONEFRAC -covarCloneSTDfrac $VARCLONEFRAC $MISC_PARAMS >> $MISC_DIR/pmake_emtrain.${EM_ITER}.out 2>&1

		CUR_LL=`cat $LL_OUT_FILE`
		LAST_LL_MAGNITUDE=`perl -e "if ($LAST_LL >= 0) { print $LAST_LL; } else { print -1*$LAST_LL; }"`
		CUR_LL_RATIO=`perl -e "print 100*($CUR_LL - $LAST_LL)/$LAST_LL_MAGNITUDE;"`

		CONVERGED=`perl -e "if ($CUR_LL_RATIO < 0) { print "0";} else { \\$abs = $CUR_LL_RATIO; if (\\$abs > $LOG_LIKE_THRESH) { print "0"; } else { print "1"; } }"`

		echo curr ll is $CUR_LL, last ll is $LAST_LL, ratio is $CUR_LL_RATIO, thresh is $LOG_LIKE_THRESH, converged is $CONVERGED
		
		LAST_LL=$CUR_LL;
		CUR_IN_FILE=$CUR_OUT_FILE
		EM_ITER=`expr $EM_ITER + 1`
	done
	echo -----------------------------------------------------
	echo `date`: Stopped after `expr $EM_ITER - 1` iterations.
	echo Copying $CUR_OUT_FILE to $EMOUT_FILE
	echo -----------------------------------------------------

	$CP $CUR_OUT_FILE $EMOUT_FILE

	echo `date`: Done training!	Now go have a drink.

else 
	echo `date`: NOT EM Training, EM output .gmp file $EMOUT_FILE exists
fi