Question

With, say, 3 rows of subplots in matplotlib, xlabels of one row can overlap the title of the next. One has to fiddle with pl.subplots_adjust(hspace), which is annoying.

Is there a recipe for hspace that prevents overlaps and works for any nrow?

""" matplotlib xlabels overlap titles ? """
import sys
import numpy as np
import pylab as pl

nrow = 3
hspace = .4  # of plot height, titles and xlabels both fall within this ??
exec "\n".join( sys.argv[1:] )  # nrow= ...

y = np.arange(10)
pl.subplots_adjust( hspace=hspace )

for jrow in range( 1, nrow+1 ):
    pl.subplot( nrow, 1, jrow )
    pl.plot( y**jrow )
    pl.title( 5 * ("title %d " % jrow) )
    pl.xlabel( 5 * ("xlabel %d " % jrow) )

pl.show()

My versions:

• matplotlib 0.99.1.1,
• Python 2.6.4,
• Mac OSX 10.4.11,
• backend: Qt4Agg (TkAgg => Exception in Tkinter callback)

(For many extra points, can anyone outline how matplotlib’s packer / spacer works, along the lines of chapter 17 “the packer” in the Tcl/Tk book?)

Answer 1

I find this quite tricky, but there is some information on it here at the MatPlotLib FAQ. It is rather cumbersome, and requires finding out about what space individual elements (ticklabels) take up…

Update:
The page states that the tight_layout() function is the easiest way to go, which attempts to automatically correct spacing.

Otherwise, it shows ways to acquire the sizes of various elements (eg. labels) so you can then correct the spacings/positions of your axes elements. Here is an example from the above FAQ page, which determines the width of a very wide y-axis label, and adjusts the axis width accordingly:

import matplotlib.pyplot as plt
import matplotlib.transforms as mtransforms
fig = plt.figure()
ax = fig.add_subplot(111)
ax.plot(range(10))
ax.set_yticks((2,5,7))
labels = ax.set_yticklabels(('really, really, really', 'long', 'labels'))

def on_draw(event):
   bboxes = []
   for label in labels:
       bbox = label.get_window_extent()
       # the figure transform goes from relative coords->pixels and we
       # want the inverse of that
       bboxi = bbox.inverse_transformed(fig.transFigure)
       bboxes.append(bboxi)

   # this is the bbox that bounds all the bboxes, again in relative
   # figure coords
   bbox = mtransforms.Bbox.union(bboxes)
   if fig.subplotpars.left < bbox.width:
       # we need to move it over
       fig.subplots_adjust(left=1.1*bbox.width) # pad a little
       fig.canvas.draw()
   return False

fig.canvas.mpl_connect('draw_event', on_draw)

plt.show()