not so FAQ

• Legend
• Tics
• Label
• 2-D Plot
  • 
    fixed-size plot

  • 
    both sides of Y-axes

  • 
    erase axes

  • 
    fixed aspect ratio figure

  • 
    zero axis

  • 
    small bars of error bars

  • 
    larger letters

  • 
    smooth curves.

  • 
    points on border

  • 
    several figures

  • 
    grid at minor tics

  • 
    two axes

  • 
    grid at an arbitrary position

  • 
    figure in figure

  • 
    bar-chart

  • 
    Align two figures

  • 
    normal probability axis

  • 
    print data values
• 3-D Plot
• Polar Coordinate
• Parametric Functions
• Data File
• After Plotting
• Miscellaneous Stuff

About 2-Dimensional Plot (No.7)

Display a graph with normal probability axis.

In a field of powder or aerosol engineering, they often plot the
size distribution of particles. It often shows a “log normal
distribution”, which becomes a straight line on the “normal
probability”(fraction) vs log(size) plot, and so called “normal
probability axis” is used. With two built-in functions – the normal
distribution norm() and the inverse normal distribution invnorm() –
of Gnuplot one can draw the normal probability axis. The following
script shows how to make “normal probability axis” for y. Other
modified scales can be made in the same way.

gnuplot> set ytics ("0.1" invnorm(0.001),"1" invnorm(0.01),"5" invnorm(0.05),
>          "10" invnorm(0.1),"20" invnorm(0.2),"30" invnorm(0.3),
>          "40" invnorm(0.4),"50" invnorm(0.5),"60" invnorm(0.6),
>          "70" invnorm(0.7),"80" invnorm(0.8),"90" invnorm(0.9),
>          "95" invnorm(0.95),"99" invnorm(0.99),"99.9" invnorm(0.999))

gnuplot> set yrange [invnorm(0.0001):invnorm(0.9999)]
gnuplot> set ylabel "Cumulative mass fraction (%)"
gnuplot> set xlabel "Diameter (mm)"
gnuplot> set logscale x
gnuplot> set grid
gnuplot> plot "sample.dat" using 1:(invnorm($2)) notitle w lp

In this example the X-axis is in the log-scale since the
log-normal distribution is assumed for the data.

How can I print values at each datapoint ?

Sometimes it is helpful if values of the points are shown numerically
near those points. However gnuplot cannot treat the values in a data
file as a text, we have to read those numbers with some program,
and show them by set label . Suppose we have the following
data, those data are shown by points, and we print those values
near the points.

0.40  80
3.00  70
3.00  60
6.00  50
9.00  40
12.00  30
13.00  20
5.00  20
8.00  20

Since we cannot make it with gnuplot alone, here we generate
the label commands with Perl. Gnuplot needs the next lines.

gnuplot> set label "(0.4,80)" at 0.4,80
gnuplot> set label "(3.0,70)" at 3.0,70

Those lines can be generated as:

% perl -ane 'print "set label \"($F[0],$F[1])\" at $F[0],$F[1]\n"' file.dat > label.plt

Now we read those gnuplot commands at plotting, the numerical values
are shown at each point.

gnuplot> load "label.plt"
gnuplot> plot 'file.dat' u 1:2 smooth csp with lines ,
>             'file.dat' u 1:2 w points