# set term png small color
set term png small
set out 'binned.png'
# set data style linespoints
# set data style points
#################################################
# Used to generate the binned word frequency counts
set term postscript eps enhanced color 28
set out 'binned-word-counts.eps'
set xrange [1:400]
set yrange [1:100000]
set logscale x
set logscale y
set format y "10^{%L}"
set title "Frequency of word observations"
set xlabel "Number of observations"
set ylabel "Number of words"
# plot "binned-word-counts.dat" using 2:3 with point pointsize 2 title "obs", \
plot "binned-word-counts.dat" using 2:3 with lines linewidth 5 title "", \
"binned-word-counts.dat" using 2:(53076 * $2**(-1.5)) with lines linewidth 5 title "N^{-1.5}"
#################################################
# Used to generate the binned word-logli counts
#
set term postscript eps enhanced color 28
set out 'binned-word-logli.eps'
set key left
set xrange [10:]
set yrange [1:100000]
set logscale y
set format y "10^{%L}"
set title "Distribution of word Log-liklihoods"
set xlabel "Log liklihood"
set ylabel "Number of words"
plot "binned-word-logli.dat" using 2:3 with lines linewidth 5 title "", \
"binned-word-logli.dat" using 2:(2**(1.5*($2-13.7))) with lines linewidth 5 title "2^{1.5(logli-13.7)}", \
"binned-word-logli.dat" using 2:(2**(0.8*($2-10))) with lines linewidth 5 title "2^{0.8(logli-10)}"
#################################################
# Used to generate the binned word-logli counts
#
set term postscript eps enhanced color 28
set out 'binned-word-logli-60.eps'
set key left
set xrange [10:]
set yrange [1:100000]
set logscale y
set format y "10^{%L}"
set title "Distribution of word Log-liklihoods"
set xlabel "Log liklihood"
set ylabel "Number of words"
plot "binned-word-logli-60.dat" using 2:3 with lines linewidth 5 title "", \
"binned-word-logli-60.dat" using 2:(2**(1.5*($2-13.7))) with lines linewidth 5 title "2^{1.5(logli-13.7)}", \
"binned-word-logli-60.dat" using 2:(2**(0.5*($2+2.5))) with lines linewidth 5 title "2^{0.5(logli+2.5)}", \
"binned-word-logli-60.dat" using 2:(2**(0.9*($2-5.5))) with lines linewidth 5 title "2^{0.9(logli-5.5)}"
set key right
#################################################
# Used to generate the binned average observations
#
set term postscript eps enhanced color 28
set out 'binned-avg.eps'
set yrange [1:3000]
set xrange [1:10]
set xtics (1,2,3,4,5,6,7,8,9,10)
set logscale x
set logscale y
set format y "10^{%L}"
set title "Distribution of average observations"
set xlabel "Average observations per disjunct"
set ylabel "Number of words"
plot "binned-avg.dat" using 2:3 with lines linewidth 5 title "", \
"binned-avg.dat" using 2:(1500/($2**5)) with lines linewidth 5 title "<obs>^{-5}"
#################################################
# Used to generate the binned connector-set MI graphs
#
set term postscript eps enhanced color 28
set out 'binned-cset-mi.eps
set key left
set yrange [1:1e6]
set logscale y
set format y "10^{%L}"
set title "Distribution of connector set MI"
set xlabel "Mutual information of c-set"
set ylabel "Count"
plot "binned-cset-mi.dat" using 2:3 with lines linewidth 5 title "", \
"binned-cset-mi.dat" using 2:(160000 * exp(-0.5*(0.23*($2-11))**2)) with lines linewidth 5 title "exp-(0.23(mi-11))^2/2"
# "binned-cset-mi.dat" using 2:(160000 * 2**(-0.038*($2-11)**2)) with lines linewidth 5 title "2^{-0.038(mi-11)^2}"
# "binned-cset-mi.dat" using 2:(160000 * 2**(-(0.2*($2-11))**2)) with lines linewidth 2 title "1.6e5 2^"
#
#################################################
# Used to generate the weighted binned connector-set counts
#
set term postscript eps enhanced color 28
set out 'weighted-cset-mi.eps'
# set key left
set yrange [1.0e-7: 1]
set logscale y
set format y "10^{%L}"
set title "Weighted distribution of connector set MI"
set xlabel "Mutual information of c-set"
set ylabel "Count"
plot "weighted-cset-mi.dat" using 2:3 with lines linewidth 5 title "", \
"weighted-cset-mi.dat" using 2:(0.0015 * 2**(1.3*$2)) with lines linewidth 5 title "2^{1.3mi}", \
"weighted-cset-mi.dat" using 2:(0.006 * exp(-(0.15*($2-12))**2)) with lines linewidth 5 title "exp -(0.15 (MI-12))^2 /2"
#################################################
# Used to generate the binned word marginal MI graph
#
set term postscript eps enhanced color 28
set out 'binned-word-mi.eps'
set key left
set xrange [4:23]
set yrange [1:10000]
set logscale y
set format y "10^{%L}"
set title "Distribution of word fractional marginal MI"
set xlabel "Fractional marginal MI of word"
set ylabel "Count"
plot "binned-word-mi.dat" using 2:3 with lines linewidth 5 title "", \
"binned-word-mi.dat" using 2:(1900*exp(-0.5*(0.24*($2-20))**2)) with lines linewidth 5 title "exp -(0.24 (FMMI-20))^2 /2"
#################################################
# Used to generate the binned disjunct MI counts
#
set term postscript eps enhanced color 28
set out 'binned-dj-mi.eps'
set xrange [2:23]
set logscale y
set format y "10^{%L}"
set title "Distr. of disjunct fractional marginal MI"
set xlabel "Fractional marginal MI of disjunct"
set ylabel "Count"
plot "binned-dj-mi.dat" using 2:3 with lines linewidth 5 title "", \
"binned-dj-mi.dat" using 2:(50000*exp(-0.5*(0.27*($2-13.7))**2)) with lines linewidth 5 title "exp -(0.27 (FMMI-13.7))^2 /2"
#################################################
# Used to generate the binned word entropy counts
#
set term postscript eps enhanced color 28
set out 'binned-word-ent.eps'
set key left
set xrange [16:24]
set yrange [1:10000]
set logscale y
set format y "10^{%L}"
set title "Distr. of word fractional marginal entropy"
set xlabel "Fractional marginal entropy of word"
set ylabel "Count"
plot "binned-word-ent.dat" using 2:3 with lines linewidth 5 title "", \
"binned-word-ent.dat" using 2:(20*exp(0.5*(0.7*($2-19.5))**2)) with lines linewidth 5 title "exp (0.7 (FE-19.5))^2 /2"
#################################################
# Used to generate the binned squared-len/norm ratio
#
set term postscript eps enhanced color 28
set out 'binned-sqlen-norm.eps'
set yrange [1:500]
set xrange [2:1000]
set logscale y
set logscale x
set format x "10^{%L}"
set format y "10^{%L}"
set title "Mean-square/size ratio distribution"
set xlabel "Mean-square to size ratio"
set ylabel "Number of words"
plot "binned-sqlen-norm.dat" using (2**$2):3 with lines linewidth 5 title "", \
"binned-sqlen-norm.dat" using (2**$2):(900/(2**$2)) with lines linewidth 5 title "ratio^-1.0"
#################################################
# Used to generate the binned average observations
#
set term postscript eps enhanced color 28
set out 'binned-sims.eps'
set logscale y
set format y "10^{%L}"
set title "Cosine similarity distribution"
set xlabel "Cosine Similarity"
set ylabel "Number of pairs of words"
plot "binned-good-sims.dat" using 2:3 with lines linewidth 5 title "", \
"binned-good-sims.dat" using 2:(3.0e5*exp(-8*$2)) with lines linewidth 5 \
title "exp(-8 cos)"
#################################################
# Used to generate the binned average observations
#
set term postscript eps enhanced color 28
set out 'binned-sims-all.eps'
set xrange [0.1:]
set yrange [1:]
set logscale y
set format y "10^{%L}"
set title "Cosine similarity distribution"
set xlabel "Cosine Similarity"
set ylabel "Number of pairs of words"
plot "binned-sims-all.dat" using 2:3 with lines linewidth 5 title "", \
"binned-sims-all.dat" using 2:(4e4*exp(-12*($2-0.1))) with lines linewidth 5 \
title "exp(-12 cos)"
# plot "binned-good-sims.dat" using (1-(2*acos($2)/3.14159)):3 with lines linewidth 2 title "acos"
#################################################