Demonstrate inset_axes in scatter_hist example.

Currently, the scatter_hist example needs to force the main axes to be
square by carefully adjusting the figure size -- shared axes and fixed
aspects don't work well together (and manually resizing the figure shows
that the aspect is indeed not fixed).  In fact, the
scatter_hist_locatable_axes example explicitly states that the advantage
of using axes_grid1 is to allow the aspect to be fixed.

I realized that one can also use inset_axes to position the marginals
axes relative to the main axes *and* support a fixed aspect ratio for
the main axes.  Perhaps this can be considered as a slight API abuse,
but I think this is a solution for a real limitation; the question is
whether we want to promote this use?

Author: anntzer

examples/axes_grid1/scatter_hist_locatable_axes.py

@@ -3,16 +3,16 @@
 Scatter Histogram (Locatable Axes)
 ==================================
 
-Show the marginal distributions of a scatter as histograms at the sides of
+Show the marginal distributions of a scatter plot as histograms at the sides of
 the plot.
 
 For a nice alignment of the main axes with the marginals, the axes positions
-are defined by a ``Divider``, produced via `.make_axes_locatable`.
+are defined by a ``Divider``, produced via `.make_axes_locatable`.  Note that
+the ``Divider`` API allows setting axes sizes and pads in inches, which is its
+main feature.
 
-An alternative method to produce a similar figure is shown in the
-:doc:`/gallery/lines_bars_and_markers/scatter_hist` example. The advantage of
-the locatable axes method shown below is that the marginal axes follow the
-fixed aspect ratio of the main axes.
+If one wants to set axes sizes and pads relative to the main Figure, see the
+:doc:`/gallery/lines_bars_and_markers/scatter_hist` example.
 """
 
 import numpy as np

examples/lines_bars_and_markers/scatter_hist.py

@@ -3,18 +3,22 @@
 Scatter plot with histograms
 ============================
 
-Show the marginal distributions of a scatter as histograms at the sides of
+Show the marginal distributions of a scatter plot as histograms at the sides of
 the plot.
 
 For a nice alignment of the main axes with the marginals, two options are shown
-below.
+below:
 
-* the axes positions are defined in terms of rectangles in figure coordinates
-* the axes positions are defined via a gridspec
+.. contents::
+   :local:
+
+While `.Axes.inset_axes` may be a bit more complex, it allows correct handling
+of main axes with a fixed aspect ratio.
 
 An alternative method to produce a similar figure using the ``axes_grid1``
-toolkit is shown in the
-:doc:`/gallery/axes_grid1/scatter_hist_locatable_axes` example.
+toolkit is shown in the :doc:`/gallery/axes_grid1/scatter_hist_locatable_axes`
+example.  Finally, it is also possible to position all axes in absolute
+coordinates using `.Figure.add_axes` (not shown here).
 
 Let us first define a function that takes x and y data as input, as well
 as three axes, the main axes for the scatter, and two marginal axes. It will
@@ -52,60 +56,53 @@ def scatter_hist(x, y, ax, ax_histx, ax_histy):
 
 #############################################################################
 #
-# Axes in figure coordinates
-# --------------------------
-#
-# To define the axes positions, `.Figure.add_axes` is provided with a rectangle
-# ``[left, bottom, width, height]`` in figure coordinates. The marginal axes
-# share one dimension with the main axes.
-
-# definitions for the axes
-left, width = 0.1, 0.65
-bottom, height = 0.1, 0.65
-spacing = 0.005
-
-
-rect_scatter = [left, bottom, width, height]
-rect_histx = [left, bottom + height + spacing, width, 0.2]
-rect_histy = [left + width + spacing, bottom, 0.2, height]
-
-# start with a square Figure
-fig = plt.figure(figsize=(8, 8))
-
-ax = fig.add_axes(rect_scatter)
-ax_histx = fig.add_axes(rect_histx, sharex=ax)
-ax_histy = fig.add_axes(rect_histy, sharey=ax)
-
-# use the previously defined function
-scatter_hist(x, y, ax, ax_histx, ax_histy)
-
-plt.show()
-
-
-#############################################################################
-#
-# Using a gridspec
-# ----------------
+# Defining the axes positions using a gridspec
+# --------------------------------------------
 #
-# We may equally define a gridspec with unequal width- and height-ratios to
-# achieve desired layout. Also see the :doc:`/tutorials/intermediate/gridspec`
-# tutorial.
-
-# start with a square Figure
-fig = plt.figure(figsize=(8, 8))
+# We define a gridspec with unequal width- and height-ratios to achieve desired
+# layout.  Also see the :doc:`/tutorials/intermediate/gridspec` tutorial.
 
-# Add a gridspec with two rows and two columns and a ratio of 2 to 7 between
+# Start with a square Figure.
+fig = plt.figure(figsize=(6, 6))
+# Add a gridspec with two rows and two columns and a ratio of 1 to 4 between
 # the size of the marginal axes and the main axes in both directions.
 # Also adjust the subplot parameters for a square plot.
-gs = fig.add_gridspec(2, 2,  width_ratios=(7, 2), height_ratios=(2, 7),
+gs = fig.add_gridspec(2, 2,  width_ratios=(4, 1), height_ratios=(1, 4),
                       left=0.1, right=0.9, bottom=0.1, top=0.9,
                       wspace=0.05, hspace=0.05)
-
+# Create the Axes.
 ax = fig.add_subplot(gs[1, 0])
 ax_histx = fig.add_subplot(gs[0, 0], sharex=ax)
 ax_histy = fig.add_subplot(gs[1, 1], sharey=ax)
+# Draw the scatter plot and marginals.
+scatter_hist(x, y, ax, ax_histx, ax_histy)
+
 
-# use the previously defined function
+#############################################################################
+#
+# Defining the axes positions using inset_axes
+# --------------------------------------------
+#
+# `~.Axes.inset_axes` can be used to position marginals *outside* the main
+# axes.  The advantage of doing so is that the aspect ratio of the main axes
+# can be fixed, and the marginals will always be drawn relative to the position
+# of the axes.
+
+# Create a Figure, which doesn't have to be square.
+fig = plt.figure(constrained_layout=True)
+# Create the main axes, leaving 25% of the figure space at the top and on the
+# right to position marginals.
+ax = fig.add_gridspec(top=0.75, right=0.75).subplots()
+# The main axes' aspect can be fixed.
+ax.set(aspect=1)
+# Create marginal axes, which have 25% of the size of the main axes.  Note that
+# the inset axes are positioned *outside* (on the right and the top) of the
+# main axes, by specifying axes coordinates greater than 1.  Axes coordinates
+# less than 0 would likewise specify positions on the left and the bottom of
+# the main axes.
+ax_histx = ax.inset_axes([0, 1.05, 1, 0.25], sharex=ax)
+ax_histy = ax.inset_axes([1.05, 0, 0.25, 1], sharey=ax)
+# Draw the scatter plot and marginals.
 scatter_hist(x, y, ax, ax_histx, ax_histy)
 
 plt.show()
@@ -118,8 +115,8 @@ def scatter_hist(x, y, ax, ax_histx, ax_histy):
 #    The use of the following functions, methods, classes and modules is shown
 #    in this example:
 #
-#    - `matplotlib.figure.Figure.add_axes`
 #    - `matplotlib.figure.Figure.add_subplot`
 #    - `matplotlib.figure.Figure.add_gridspec`
+#    - `matplotlib.axes.Axes.inset_axes`
 #    - `matplotlib.axes.Axes.scatter`
 #    - `matplotlib.axes.Axes.hist`