Plot numpy.histogram2d()

Question

I am wondering if it is maybe possible to plot the output of numpy.histogram2d() using matplotlib.pyplot.hist2d? In the one-dimensional case, this can be done by using:

counts, bins = np.histogram(something, bins=no_bins, range=(range_min, range_max))

plt.hist(bins[:-1], bins, weights=counts)

Is there a similar solution for the two-dimensional case? I do not want to plot the 2d histo with the methods suggested on https://docs.scipy.org/doc/numpy/reference/generated/numpy.histogram2d.html The idea behind this is that I would like to apply some corrections to the inital histogram (ie bin-by-bin background subtraction using data from another 2d histogram) and then plot the corrected histogram.

Many thanks in advance

Answer 1

Plotting 2d histograms is typically done with imshow() . If you're used to ROOT or some other plotting libraries, be especially careful what argument you give for origin and extent .

Answer 2

The following solution works as expected:

counts_bkg, bins_x_bkg, bins_y_bkg = np.histogram2d(x_bkg, y_bkg, bins=(x_bins, ybins))
counts, bins_x, bins_y = np.histogram2d(x, y, bins=(x_bins, y_bins))

diff = counts - counts_bkg
diffT = diff.T

fig, ax = plt.subplots(1)
pc = ax.pcolorfast(bins_x, bins_y, diffT)
plt.show()

Answer 3

In the docs , you can find three examples of how to plot the output from np.histogram2d() using the matplotlib functions imshow , pcolormesh and NonUniformImage .

Answer 4

I googled my way here. I am going to write here what worked for me, just in case, some else finds themselves in similar situtation.

I was looking at the source code of pyplot, where I found reference to axes class. The hist2d funtion is actually defined in 'matplotlib/axes/_axes.py'. There I found hist2d calls np.histogram2s and then uses the xedges, yedges and bins in plt.pcolormesh as follow:

pc = self.pcolormesh(xedges, yedges, h.T, **kwargs)

Remember there is no option to use keywords 'Range', 'density' and 'bins' in pcolormesh but these are taken into account by the np.histogram2d function.

TL;DR: Using pcolormesh is the simplest way for plotting 2D histogram from the output of np.histogram2d.