Ability to scale axis by a fixed order of magnitude

To scale an axis by a fixed order of magnitude, set the *scilimits* argument of

``Axes.ticklabel_format`` to the same (non-zero) lower and upper limits. Say to scale

the y axis by a million (1e6), use ``ax.ticklabel_format(style='sci', scilimits=(6, 6), axis='y')``.

The behavior of ``scilimits=(0, 0)`` is unchanged. With this setting, matplotlib will adjust

the order of magnitude depending on the axis values, rather than keeping it fixed. Previously, setting

``scilimits=(m, m)`` was equivalent to setting ``scilimits=(0, 0)``.

Add ``AnchoredDirectionArrows`` feature to mpl_toolkits

A new mpl_toolkits class

:class:`~mpl_toolkits.axes_grid1.anchored_artists.AnchoredDirectionArrows`

draws a pair of orthogonal arrows to inidcate directions on a 2D plot. A

minimal working example takes in the transformation object for the coordinate

system (typically ax.transAxes), and arrow labels. There are several optional

parameters that can be used to alter layout. For example, the arrow pairs can

be rotated and the color can be changed. By default the labels and arrows have

the same color, but the class may also pass arguments for costumizing arrow

and text layout, these are passed to :class:`matplotlib.text.TextPath` and

`matplotlib.patches.FancyArrowPatch`. Location, length and width for both

arrow tail and head can be adjusted, the the direction arrows and labels can

have a frame. Padding and separation parameters can be adjusted.

Add ``minorticks_on()/off()`` methods for colorbar

A new method :meth:`.Colobar.minorticks_on` is

introduced to correctly display minor ticks on the colorbar. This method

doesn't allow the minor ticks to extend into the regions beyond vmin and vmax

when the extend `kwarg` (used while creating the colorbar) is set to 'both',

'max' or 'min'.

A complementary method :meth:`.Colobar.minorticks_off`

is introduced to remove the minor ticks on the colorbar.

Colorbar ticks can now be automatic

The number of ticks on colorbars was appropriate for a large colorbar, but

looked bad if the colorbar was made smaller (i.e. via the ``shrink`` kwarg).

This has been changed so that the number of ticks is now responsive to how

large the colorbar is.

Cyclic colormaps

Two new colormaps named 'twilight' and 'twilight_shifted' were added.

These colormaps start and end on the same color, and have two

symmetric halves with equal lightness, but diverging color. Since they

wrap around, they are a good choice for cyclic data such as phase

angles, compass directions, or time of day. Like viridis, twilight is

perceptually uniform and colorblind friendly.

Stop adding a suffix to suggest unique file name

Previously, when saving a figure to a file using the GUI's

save dialog box, if the default filename (based on the

figure window title) already existed on disk, Matplotlib

would append a suffix (e.g. `Figure_1-1.png`), preventing

the dialog from prompting to overwrite the file. This

behaviour has been removed. Now if the file name exists on

disk, the user is prompted whether or not to overwrite it.

This eliminates guesswork, and allows intentional

overwriting, especially when the figure name has been

manually set using `fig.canvas.set_window_title()`.

Legend now has a title_fontsize kwarg (and rcParam)

The title for a `.Figure.legend` and `.Axes.legend` can now have its

fontsize set via the ``title_fontsize`` kwarg. There is also a new

:rc:`legend.title_fontsize`. Both default to ``None``, which means

the legend title will have the same fontsize as the axes default fontsize

(*not* the legend fontsize, set by the ``fontsize`` kwarg or

:rc:`legend.fontsize`).

Implemented support for axes.prop_cycle property markevery in rcParams

The Matplotlib ``rcParams`` settings object now supports configuration

of the attribute `axes.prop_cycle` with cyclers using the `markevery`

Line2D object property. An example of this feature is provided at

`~matplotlib/examples/lines_bars_and_markers/markevery_prop_cycle.py`

Multipage PDF support for pgf backend

The pgf backend now also supports multipage PDF files.

.. code-block:: python

Pie charts are now circular by default

We acknowledge that the majority of people do not like egg-shaped pies.

Therefore, an axes to which a pie chart is plotted will be set to have

equal aspect ratio by default. This ensures that the pie appears circular

independent on the axes size or units. To revert to the previous behaviour

you may set the axes' aspect to automatic, ax.set_aspect("auto") or

plt.axis("auto").

Add ``ax.get_gridspec`` to `.SubplotBase`

New method `.SubplotBase.get_gridspec` is added so that users can

easily get the gridspec that went into making an axes:

.. code::

Axes title will no longer overlap xaxis

Previously an axes title had to be moved manually if an xaxis overlapped

(usually when the xaxis was put on the top of the axes). Now, the title

will be automatically moved above the xaxis and its decorators (including

the xlabel) if they are at the top.

If desired, the title can still be placed manually. There is a slight kludge;

the algorithm checks if the y-position of the title is 1.0 (the default),

and moves if it is. If the user places the title in the default location

(i.e. ``ax.title.set_position(0.5, 1.0)``), the title will still be moved

above the xaxis. If the user wants to avoid this, they can

specify a number that is close (i.e. ``ax.title.set_position(0.5, 1.01)``)

and the title will not be moved via this algorithm.