Plotting
Making Plots
The AnalysisToolbox also has some wrappers for Matplotlib, which allow you to make Matplotlib
plots easily using python. All of them can be found in
import latqcdtools.base.plotting
We tried to write this module such that you can pass Matplotlib arguments to our custom wrappers
like plot_file, so hopefully you can comfortably weave in your own stylistic choices.
Be sure to also check out the example in examples/main_plotting.py.
The basics
You probably want to get started quickly, so here’s a minimal example. Please be mindful of the order of the commands.
latexify()
plot_file('susceptibility.d',1,2,3,label='$56^3\\times8~~m_s/m_l=80$',color=colors[1],marker=markers_1[1])
set_params(xlabel="T [MeV]",
ylabel="$N_s^3\\,\\chi^{\\rm{bare}}$",
title ="$N_\\tau=8$")
plt.savefig("susceptibility.pdf")
plt.show()
The above python code will create a plot using the data in susceptibility.d, save it in
susceptibility.pdf, and also show the plot to you. Let us go through each line. You probably
noticed that the labels and titles use \(\LaTeX\) syntax. In order for this to work, a call to
latexify()
is needed at the beginning. In addition to using nice \(\LaTeX\) font this command also automatically puts the axes in scientific notation if that would look better. I recommend that you always use this command in your plots. The command
set_params()
lets you specify things needed for every plot such as titles and axis labels. The set_params
works best after your plot_file calls and before your plt.savefig and
plt.show calls. For a full list
of parameters have a look at this module. The command
plot_file(filename, xcol, ycol, xecol, yecol=None)
is what adds the data to the plot. The arguments xcol, ycol, xecol, and yecol are the column
numbers of the data file for the x-data, y-data, x error data, and y error data, respectively. After
this there follows some optional arguments. The marker is the symbol used to mark the data on the
plot, and the color is the color of those markers. There are 8 colors in the colors list;
other possible colors to use are given in the module. There are 11 markers in the markers list,
and since this module is based on Matplotlib, you find other possibilities for markers
here. The last two commands are self-explanatory.
In case you wish to change the style of your plot, this can be accomplished with the
style='dots'
option in your plot_file arguments. The allowed styles include dots, lines, fill, and
band. Just as a friendly tip: If you use a style other than dots, which is the default,
you may need to use marker=None or marker=',' to suppress the data points. This will be
necessary if the data points are close enough together and larger than the line itself.
Again since we want to be able to plot quickly within python scripts, you might find yourself in a situation where you want to plot list elements rather than a file. As a minimal example, one can use
plot_dots(xdata,ydata,yedata=edata)
instead of plot_file to plot the list ydata with its error bars edata against xdata.
Other plot style options besides plot_dots include plot_lines, plot_fill, and plot_band.
If you are using a single python script to plot multiple files, python will remember the labels
from the last plot for some reason. Thankfully there is a command which addresses this problem:
After you have written your code to generate some plot plot1, you can call clearPlot()
before your code to generate plot2. This will reset all the plot defaults and legend handles
to get you ready for your next figure.
Some options you might commonly require
Changing axis ranges
The ranges of plots are chosen automatically, but this automatic choice may not always suit your purposes. To do some custom choice, use
set_params(xmin=0,xmax=1,ymin=0,ymax=1)
Changing the location of the legend
The legend position is normally chosen automatically, but you may have other ideas. The
legendpos parameter in set_params can be used to adjust the position of the legend.
For example
legendpos = 5
will put the legend on the right, within the boundary of the plot. A full list of positions that lie within the boundary of the plot can be found in the plotting module. You may need to put the legend outside of the plot. In that case use
legendpos = (1,0)
to put the legend on the right, outside the plot, at the bottom. You can increase the second coordinate to put the legend higher up.
Vertical and horizontal lines; bands
Sometimes you may wish to add a vertical or horizontal line to indicate the position of some important quantity. These can be created with, for example
plt.axvline(x=0, color=colors[0])
plt.axhline(y=0, color=colors[0])
You also have the option of making them dashed using linestyle="dashed". You may also want
to create an error band or something like this. Matplotlib also has a nice command for this,
namely
plt.axvspan(xmin, xmax, alpha=0.5, color=colors[0])
plt.axhspan(ymin, ymax, alpha=0.5, color=colors[0])
Here the min and max arguments are used to specify the width of the band, and 0 < alpha < 1
is the opacity. If you have a file with data and error bars, and you would like to plot these as
error bands, you can use
style='fill'
as an argument to your plot_file command.
Printing only an inset to pdf
When you use plt.show() to look at your figure, a small region in the graph may catch your
interest, and you might zoom in on that. It is possible to print this inset to pdf.
Use
plt.show()
plt.ion()
plt.savefig("example.pdf")
Note that the relative order of plt.show() and plt.savefig() has changed.
Adding PRELIMINARY
Do you have a preliminary result, which requires you to write PRELIMINARY in? You can accomplish this with
preliminary(xcoordinate, ycoordinate)
The xcoordinate and ycoordinate are measured according to the axes themselves; for example if
my x-axis is temperature and I want to put the text at 150 MeV, I just set xcoordinate=150.
Rescaling an axis
We need to do this pretty often, for example when changing from lattice units to physical units. To
rescale the x-axis by a factor 2, pass to your plot_ command xscale=2. yscale rescales the y-axis.
Having nice colors
There is a set of eight base colors that the AnalysisToolbox automatically cycles through. When plotting more
eight things, for example a range of temperatures, it may be useful to generate a graded scale of colors.
For this you can create your own list of ncolors colors myColors with
myColors = getColorGradient(ncolors)
Having one script generate multiple plots
By default, the plotting module and its associated methods like plot_dots uses the axis object plt.
Hence if you are having one script generate multiple plots, it may happen that labels or data from a previous
plot carry over to the next one. You can clear all plot labels, options, and data by calling
clearPlot()