Gnuplot Manual: Data Plot

Outline:

Data plot

Basics

In all the previous sections, we used the built-in functions of gnuplot, but you can also plot extrernal data files. For example, you have the following forrmatted data saved as test01.d:

# X Y
1.0 2.0
2.0 4.0
3.0 6.0

The symbol, #, can comment out; in a word, gnuplot ignores the whole referred line. Now, let's plot this by gnuplot:

gnuplot> plot [0:3.5] [0:6.5] 'test01.d'

Note that you MUST use quotations or double quotations for the file name in the command line. You also have to locate the folder that the file exists for gnuplot. gnuplot hirophysics

In the default setting, it plots only the points of given data. If the data has only two columns, gnuplot recognizes the first and second columns as x and y values, respectively.
You can specify the sizes, colors and shapes of the plotted points. The line connecting between points can also be displayed. You just type "with line", or "with linespoint" after the plot command. They can be abbreviated as "w l" and "w lp." All the options are listed as follows:

gnuplot> plot 'example.d' w p # Plot with points

gnuplot> plot 'example.d' w l # Plot with lines

gnuplot> plot 'example.d' w i # Plot with impulses

gnuplot> plot 'example.d' w d # Plot with dots (tyny little points)

gnuplot> plot 'example.d' w lp # Plot with lines and points

Vector

There is also a convenient option that is "plot with vector." In order to plot a set of data as a vector, you need 4 values. The first two are x and y coordinates of the start point. The next two are x and y components; in other words, they are lengths of x and y. Take a look at the following data file (vector.d):

# Start Length
# (x, y) (x, y)
0.0 0.0 1.0 1.0

Then, command as follows:

gnuplot> plot 'vector.d' w vector

The coordinate of the start point (tail) is (0, 0) from the first and second values. Then, each component is 1 as each length according to the third and the fourth values.

Let's look at another example.

1.0 1.0 1.0 -1.0

The start point is (1, 1); the x component takes length of 1 in the positive direction; then the y component takes length of 1 in the negative direction. Make sure this with the output figure: gnuplot hirophysics

Bar graph

There are several options to plot the data with a bar graph. The "box" option is a simple bar graph, but the "step" options can produce variations of steps to be plotted.

gnuplot> plot 'example.d' w boxes # Plot with boxes (cannot be abbreviated)

gnuplot> plot 'example.d' w steps # Plot with steps (cannot be abbreviated)

gnuplot> plot 'example.d' w fsteps # Plot with steps (cannot be abbreviated)

gnuplot> plot 'example.d' w histeps # Plot with steps (cannot be abbreviated)

Let's plot erf.d with "box" option.

gnuplot> plot 'erf.d' w boxes

You can also change the width of boxes as follows:

gnuplot> set boxwidth 0.1
gnuplot> plot 'erf.d' w boxes

Then, plot step.d with three different options of "steps" in one frame as follows:

gnuplot> plot [0:9][1:10] 'step.d' w steps, 'step.d' w fsteps, 'step.d' w histeps

Note that you may have to adjust the ranges and other settings to have exactly the same as the following. gnuplot hirophysics

The data above are integers from 1 through 10. As you can see, "steps" goes horizontally first, "fsteps" goes vertically first, and "histeps" takes the half between them.

Line and point styles

You may need a thicker line or a larger size of point for some reason. Here are the options:

gnuplot> plot 'example.d' lw 2 # Specify the line width

gnuplot> plot 'example.d' lt 3 # Specify the line color (must be done before linewidth)

gnuplot> plot 'example.d' pt 3 # Specify the point type

gnuplot> plot 'example.d' ps 4 # Specify the point size

gnuplot> plot 'example.d' dt 1 # Specify the dash type

Let's plot some example with erf.d.

gnuplot> plot 'erf.d' lt 4 lw 3 pt 3 ps 3 w lp

In order to make sure the above options and the other information, check with "test" command, and you will see the following diagram:

gnuplot> test

Multiple data columns

Suppose you have the following data (bessel.d):

0.000000 0.000000 0.000000 0.000000
0.166667 0.083044 0.003464 0.000096
0.333333 0.164363 0.013761 0.000766
0.500000 0.242268 0.030604 0.002564
0.666667 0.315155 0.053526 0.006003
0.833333 0.381529 0.081890 0.011542
1.000000 0.440051 0.114903 0.019563
1.166667 0.489557 0.151643 0.030362
1.333333 0.529095 0.191076 0.044134
1.500000 0.557937 0.232088 0.060964
1.666667 0.575599 0.273510 0.080825
1.833333 0.581853 0.314153 0.103571
2.000000 0.576725 0.352834 0.128943
2.166667 0.560498 0.388412 0.156570
2.333333 0.533701 0.419812 0.185977
2.500000 0.497094 0.446059 0.216600
2.666667 0.451651 0.466299 0.247798
2.833333 0.398533 0.479825 0.278867
3.000000 0.339059 0.486091 0.309063
...

The default of the plot command picks out the first and second as x and y values.

gnuplot> plot 'bessel.d' w l

If you need to plot the first and third as x and y, specify with "using" command.

gnuplot> plot 'bessel.d' u 1:3 w l

The option, "u", represents "using", and "1:3" specifies the first and third columns in the data file. The "using" command MUST be used before "with lines" command. gnuplot hirophysics

If all three graphs have to be plotted in one frame, enter as follows:

gnuplot> plot 'bessel.d' u 1:2 w l, 'bessel.d' u 1:3 w l, 'bessel.d' u 1:4 w l

You can modify data columns in command lines. In the "using " command, specify the number of column. Say, pick out the second column, and operate square root on this.

gnuplot> plot 'bessel.d' u 1:(sqrt($2)) w l

Here is the plot with the first column and square root of the second column. gnuplot hirophysics

Multiplot

You can display multiple coordinates in single output. Just follow this instruction:

gnuplot> set multiplot layout 2,1
multiplot> plot tan(x)
multiplot> plot cos(x)

You will obtain the following output: gnuplot hirophysics

The option of "layout" gives how plots are arranged in the output. If you do not use it, all the plots will be superposed in the same coordinate.

For CSV files

The data in CSV files or any data separated by commas are not plotted properly in gnuplot. For example, (csv.d)

12,6
10,1
2,11
6,8
15,7
5,6
1,4

In order to ignore commas for the proper plot, you should use the following command:

gnuplot> set datafile separator ","

Then, you have gnuplot hirophysics

For Fortran high precision data

If the data calculated from a Fortran program are double or quadrruple precision, use the following command:

gnuplot> set datafile fortran

This enables a special check for reading data sets.

Error bars

A typical plot with error bar requires 3-column of data. (error1.d)

# x y error
1.3 4.5 0.7
1.8 5.5 1.0
2.4 5.9 0.5
3.1 6.7 0.8
3.9 8.5 0.6

As shown above, the third column expresses the range of error. Let's plot this with the error bar of y axis.

gnuplot> plot 'error1.d' w yerrorbars

You can also use it as the error range of x.

gnuplot> plot 'error1.d' w xerrorbars

Then, the following is plotted with the bar graph:

gnuplot> plot 'error1.d' w boxerrorbars

Suppose you need to determine the error range by start and end values. Let's create 4-column data as follows: (error2.d)

# x y error
1.3 4.5 4.0 4.6
1.8 5.5 5.0 5.9
2.4 5.9 5.5 6.2
3.1 6.7 6.2 7.3
3.9 8.5 8.1 9.0

Then, command as follows:

gnuplot> plot 'error2.d' w yerrorbars

Likewise, you can set the parameters for x-error bar.

You can have both x- and y-error bars with a 4-column set of data. (error3.d)

# x y error
1.3 4.5 0.3 0.6
1.8 5.5 0.4 0.9
2.4 5.9 0.5 0.2
3.1 6.7 0.3 0.3
3.9 8.5 0.5 0.4

gnuplot> plot 'error3.d' w xerrorbars

The following data set specifies both x and y error bars of start and end points. (error4.d)

# x y error
1.3 4.5 1.2 1.6 4.0 4.7
1.8 5.5 1.4 2.0 5.2 5.7
2.4 5.9 2.1 2.6 5.8 6.2
3.1 6.7 2.8 3.3 6.4 6.9
3.9 8.5 3.6 4.2 8.3 8.8

gnuplot> plot 'error4.d' w xyerrorbars

Candle chart

If you have a set of data for stock exchange, you can plot it with candle charts. The data must be 5-column, which are date, open, high, low, and close, respectively. Here is an example, (stock.d):

01-15-2008 15155.73 15156.66 12572.68 13592.47
02-15-2008 13517.74 14105.47 12923.42 13603.02
03-15-2008 13412.87 13413.63 11691.00 12525.54
04-15-2008 12539.80 14003.28 12521.84 13849.99
05-15-2008 13802.59 14392.53 13540.68 14338.54
06-15-2008 14342.96 14601.27 13453.35 13481.38
07-15-2008 13514.86 13603.31 12671.34 13376.81
08-15-2008 13276.57 13468.81 12631.94 13072.87
09-15-2008 12936.81 12940.55 11160.83 11259.86
10-15-2008 11396.61 11456.64 6994.90 8576.98
11-15-2008 8702.77 9521.24 7406.18 8512.27

Then, enter as follows in the command lines:

gnuplot> set xdata time #optional to express dates
gnuplot> set timefmt "%m-%d-%Y" #optional to express dates
gnuplot> plot 'stock.d' u 1:2:3:4:5 w candlesticks

Note that you have to use "using 1:2:3:4:5" to avoid any error.

3D data plot

For one of the 3D-plot data, you can have the following, (3d-1.d):

0
0
0
3
0

1
1
4
1
1

2
7
2
1
1

3
3
3
3
5

0
1
0
1
0

This will be a 3D surface plot if you use "splot" command. The data above are values for z axis. In other words, this is equivalent with the following data (3d-1eq2.d):

#x y z
  0 0 0
  1 0 0
  2 0 0
  3 0 3
  4 0 0

  0 1 1
  1 1 1
  2 1 4
  3 1 1
  4 1 1

  0 2 2
  1 2 7
  2 2 2
  3 2 1
  4 2 1

  0 3 3
  1 3 3
  2 3 3
  3 3 3
  4 3 5

  0 4 0
  1 4 1
  2 4 0
  3 4 1
  4 4 0

The both files can generate the same plot as follows:

gnuplot> set multiplot layout 1,2
multiplot> splot '3d-1.d' w l
multiplot> splot '3d-1eq2.d' w l

The number of each data block has to be the same; otherwise, gnuplot will not plot a surface, but multiple lines.

How about if you put a double space for each block? (3d-2.d)

gnuplot> splot '3d-2.d' w l

Each block of the data becomes independent to be plotted as multiple lines; namely, it does not create the lattice.

Suppose you have a set of data in a matrix form (matrix.d).

1 2 1 4 9 0
0 1 0 5 1 2
0 4 95 81 3 1
2 9 60 73 8 1
2 6 7 9 5 9
1 5 3 2 4 4

The rows and columns correspond to x and y coordinates. Use the matrix option to plot this.

gnuplot> splot 'matrix.d' matrix w l

Gnuplot manual: Frequently used methods