I have a “macro” on my vim configuration file which expands the text “figgnuplot” to the following template. This is useful to create a new gnuplot figure because I tend to forget about my own name, let alone computer syntax… Use it as you want but understand that this is meant to be run using latex and some snippets.

# gnuplot epslatex template # Pau Amaro Seoane, Berlin, 25/Apr/2020 # # This style is meant to be used along with Gnuplot.sh # and GpTex.sh # # To be used as # # $ Gnuplot.sh myscript.gnu # # This produces out.pdf and opens it with a pdf viewer # ******************** load snippets ******************** # # This assumes that you have a symlink ~/.gnuplot pointing to gnuplotrc # In my case it is .gnuplot@ -> /home/pau/fitx_confg/gnuplot/gnuplotrc # In gnuplotrc you add the line: # set loadpath "/home/user/path/snippets:/home/user/path/palettes" # Do not use $HOME because it won't expand. load 'header.cfg' load 'lines.cfg' load 'grid.cfg' load 'lightborder.cfg' load 'arrows.cfg' load 'constants.cfg' load 'convfactors.cfg' # ******************** definitions and styles ******************** # # Scales, ranges and tics #set logscale xy # set xtics rotate by 45 right # if you want them rotated by 45 (suprise!) set mxtics 2 # 10 for log style set mytics 2 # 10 for log style set xrange [*:*] set yrange [*:*] #set y2range [*:*] # unset if second y-axis, and multiply by the # conversion factor both, ymin and ymax, unless you want # different curves in the plot. E.g. # set yrange [1, 10] in meters # set y2range [1/1e3, 10/1e3] in km #set xtics add ('0' 0, '1' 1) # "add" is important in log #set ytics add ('0' 0, '1' 1) # otherwise mx- and mytics disappear # set my2tics 2 # unset this if second y-axis #set y2tics ystart,ystep # the first parameter is the starting value at the bottom of the graph, # and the second is the interval between tics on the axis (optional) #set ytics nomirror # unset this if second y-axis # Power styles # Sometimes this might be bugy # if used with set logscale xy # Try to comment it out if strange # problems appear #set format x "$%.0s \\times 10^{%S}$" # Not setting the $$ might lead to #set format y "$%.0s \\times 10^{%S}$" # problems if using y2 # Another possibility: "$10^{%L}$" # ******************** key ******************** # set key left bottom #unset key #set key at X,Y # X on the right side of the legend ------ Y # Y at the top of the legend | X #set key font ",20" # sets it to 20 # ******************** labels ******************** # set xlabel "Xlabel (pc)" set ylabel "Ylabel ($t^{-1}$)" # offset -1.5,2,0 # moves ylabel -1.5 characters # away the x axis, 2 towards y, # and 0 towards z # (relative to its original position) # set y2label "Y2label ($T^{-1}$)" # offset 0.1,0 # if second y-axis is used # ******************** main plot ******************** # # set samples 1000 # get smoother curves # NB: If using log scale, gnuplot might cut the # curve because of a lack of resolution; make # samples much larger #set multiplot # uncomment only if you have embedded, smaller plots plot f(x) title "" with lines ls 1, \ g(x) title "" with lines ls 2, \ h(x) title "" with lines ls 3, \ i(x) title "" with lines ls 4 # Want a power-line? Do this: #plot f(x) title "" with lines ls 1, \ # [1:2] 3e4 * x**(-1.75) with lines ls 1 dashtype 3 linewidth 2 title "bo" # [rang in x axis] multiplying factor * x**(yourpower) with lines ls 1 dashtype 3 linewidth 2 title "straight line" # If second y-axis present, then #plot f(x) title "" with lines ls 1 axis x1y1, \ # g(x) title "" with lines ls 2 axis x1y1, \ # h(x) title "" with lines ls 3 axis x1y1, \ # i(x) title "" with lines ls 4 axis x1y1, \ # j(x) title "" with lines ls 1 axis x1y2, \ # k(x) title "" with lines ls 2 axis x1y2, \ # l(x) title "" with lines ls 3 axis x1y2, \ # m(x) title "" with lines ls 4 axis x1y2 # Reading data files # set datafile commentschars { "#" | "%" } # set datafile separator { "," | whitespace } # ******************** embedded, smaller plots (zooms) ******************** # # You need to uncomment the "set multiplot" line above # It's better to remove the grid from the larger plot and # add only grids to the smaller plots, because they otherwise # overlap # Grid #set style line 102 lc rgb '#808080' lt 0 lw 1 #set grid back ls 102 # Origin #set origin X,Y # between [0, 1] #set size X-length, Y-length # between [0, 1] #set xrange [xmin:xmax] #set yrange [ymin:ymax] # Labels #unset xlabel #unset ylabel #unset label #set xtics ('1' 1, '2' 2) #set ytics ('0' 0, '2' 2) # Format of box and tics #set tics scale 0.5 front #set border linewidth 3 # Plot #plot f(x) title "" with lines ls 1, \ # g(x) title "" with lines ls 2, \ # h(x) title "" with lines ls 3, \ # i(x) title "" with lines ls 4 # After the embedded, smaller plots, we need to unset multiplot #unset multiplot # ******************** back to x11 ******************** # set terminal x11 # ******************** dump analytic function to a table ******************** # # set table "out.dat" # plot '+' u (sprintf("%g,%g,%g,%g,%g",x , f(x) , g(x) , h(x) , i(x))) w table # unset table # ******************** use python programmes ******************** # # You can use python programmes from "outside" which allow you to do more things. # One example is the Bessel functions of any order. Gnuplot is limited to first order. # # To defined a new function, simply create a programme.py and call it like this # # bessel(n,x) = real(system(sprintf("python bessel1stkind.py %g %g", n, x))) # # where # # $ cat bessel1stkind.py # import sys # import numpy as np # from scipy.special import * # # n=float(sys.argv[1]) # x=float(sys.argv[2]) # # print (jn(n,x)) # # Then simply use it in gnuplot # # gnuplot> print bessel(1,1) # 0.440050585744934 # gnuplot> print bessel(-2,1) # 0.114903484931901 # ******************** style: hints ******************** # # Some help on style: # http://en.wikipedia.org/wiki/Wikipedia:How_to_create_charts_for_Wikipedia_articles#gnuplot # http://en.wikipedia.org/wiki/File:Ps_symbols_color.png # http://en.wikipedia.org/wiki/File:Ps_symbols_color_solid.png # http://upload.wikimedia.org/wikipedia/commons/2/2a/Ps_symbols_bw.png # See also the local files: # qiv -t ~/treball/documents/manuals/gnuplot/Ps_symbols_color.png # qiv -t ~/treball/documents/manuals/gnuplot/Ps_symbols_color_solid.png # qiv -t ~/treball/documents/manuals/gnuplot/Ps_symbols_bw.png # ******************** format tics ******************** # # Some style about format: # http://lowrank.net/gnuplot/tics-e.html # gnuplot> set format x "%10.3f" # # The syntax for the digit is "%" + (total length).(precision). The floating # number 6.2 represents that the total length is six and there are two digits # following the decimal point, so that tic-labels are shown as "5.00". It is # possible to omit the length or precision, like 6 or .2. The default values are # used for the omitted number. # # The display format is expressed by one letter -- 'f', 'e', 'E', 'g', 'x', 'X', # 'o', 't', 'l', 's', 'T', 'L', 'S', 'c', and 'P'. The default is "%g". When the # tics-labels can be expressed by appropriate length and precision, those are # written by "%.0f" format, otherwise "%e" format is used. The next table shows # the difference among the format 'f', 'e', 'x', and 'o'. The formats 'e' and 'E' # are the same except that the written text is 'e' or 'E'. The format "%O" exists # in the gnuplot manual, but it does not work (bug ?) # # Format Explanation Example (underscore means blank) # f decimal %6.3f __6.00 # e,E exponential %11.4e _5.0000e+01 # x,X hexadecimal %x fffffffb # o,O octal %o 37777766 # # The formats, 't', 'l', 'T', and 'L' are related to log-scale plot. # ******************** samples ******************** # # Gnuplot doesn't really draw curves for functions - it actually computes the # functions at multiple points and connects them with straight lines, similarly # to what would happen if you were plotting a data file # # Sometimes when plotting analytical functions which span over many orders of # magnitude, gnuplot does not have enough resolution, even if you set it to the # maximum, which is about 8e6. For _one_ curve it should be enough but sometimes # we are plotting various curves, so that the sampling is divided by the number # of curves and hence, each curve has a lower sampling. This should be avoided # because some extreme values might be missing. # # The fix for this is the following # # 1) output to a file with the data # # set table "g.dat" # # 2) Plot each of the curves. If you have the functions g(x), h(x) and i(x), # # set samples 30*number of maxima (as a rule of thumb) # set table "g.dat" # plot g(x) # unset table # # then # # set table "h.dat" # plot h(x) # unset table # # then # # set table "i.dat" # plot i(x) # unset table # # After that, in the gnu script you use # # plot "g.dat" u 1:2 with lines ... ,\ # "h.dat" u 1:2 with lines ... ,\ # "i.dat" u 1:2 with lines ... ,\ # # And this way the functions g(x), h(x) and i(x) will have a higher resolution.