[Python] Adjusting the color bar

Vocabulary Book. Reprinting a lot because it is troublesome to search every time. There is an original URL.

[matplotlib.pyplot.colorbar] (https://matplotlib.org/3.1.0/api/_as_gen/matplotlib.pyplot.colorbar.html) [Qiita @ skotaro: A story dedicated to those who have found out by dissecting the colorbar of matplotlib, or who do not want to have trouble adjusting the colorbar anymore] (https://qiita.com/skotaro/items/01d66a8c9902a766a2c0)

Adjust position: Draw a color bar on the specified Axes

Specified by fig.add_axes

Specified with the cax = option.

fig = plt.figure(figsize=(8,5))
ax  = fig.add_axes((0.05, 0.15, 0.9, 0.8))
im  = ax.imshow(np.random.random((10, 20)), cmap='jet')
cax = fig.add_axes((0.05, 0.05, 0.9, 0.08))
plt.colorbar(im, orientation='horizontal', cax=cax)

Specified by make_axes_locatable

The method of drawing the color bar is the same as above, only the method of making Axes is changed.

from mpl_toolkits.axes_grid1 import make_axes_locatable
divider = make_axes_locatable(ax)

#Make space to the right
cax = divider.new_horizontal(size="2%", pad=0.05)
fig.add_axes(cax)
# OR
cax = divider.append_axes('right', '5%', pad='3%')

plt.colorbar(im, cax=cax)

Qiita @ termoshtt: Align colorbar to figure with matplotlib

However, if you specify projection for ax, it does not work.

Qiita @ nishimuraatsushi: Align the size of colorbar with matplotlib

Draw color bar alone

plt.figure(figsize=(8, 0.5))
norm = mcolors.DivergingNorm(vcenter=0.0, vmin=-5, vmax=5)
im   = ax.imshow(np.array([[0,1]]), cmap='bwr', norm=norm)
plt.gca().set_visible(False)
cax = plt.axes([0.1, 0.2, 0.8, 0.6])
cb  = plt.colorbar(im, orientation='horizontal', cax=cax)

[stackoverflow: Standalone colorbar (matplotlib)] (https://stackoverflow.com/questions/16595138/standalone-colorbar-matplotlib)

The above code is drawn once with ʻimshow → then define ʻAxes for the color bar, and so on.

--Save this color bar alone → Edit with pptx etc. --Create a function that passes an axes to draw a color bar and an image (ʻim` in the above code) as arguments.

In that case, this is enough. The latter example:

def draw_coloarbar(ax, im):
    return plt.colorbar(im, orientation='horizontal', cax=ax)

fig  = plt.figure(figsize=(8, 4))
ax   = fig.add_axes((0.05, 0.05, 0.9, 0.9))
im   = ax.imshow(np.array(np.random.random((10, 20))), cmap='jet')
cax = fig.add_axes([0.1, 0.45, 0.8, 0.1])
draw_coloarbar(cax, im)

However, in the function that does not pass the image as an argument = it is necessary to do a temporary ʻimshow`

When creating a function that draws only the colorbar on the specified ʻAxes (after doing a tentative ʻimshow), the method of specifying the position is different:

def draw_coloarbar2(ax):
    norm = mcolors.DivergingNorm(vcenter=0.0, vmin=-5, vmax=5)
    im = ax.imshow(np.array([[0,1]]), cmap='bwr', norm=norm)
    ax.set_visible(False)
    return plt.colorbar(im, orientation='horizontal', #cax=ax) # cax=If so, the color bar will not be displayed
                        ax=ax, fraction=1.0)

fig  = plt.figure(figsize=(8, 4))
ax   = fig.add_axes((0.05, 0.05, 0.9, 0.9))
im   = ax.imshow(np.array(np.random.random((10, 20))), cmap='jet')
cax = fig.add_axes([0.1, 0.45, 0.8, 0.1])
draw_coloarbar2(cax)

In this case, it is necessary to specify the position with ʻax = instead of cax = . See the link at the top of the page for the differences in the meanings of these arguments. fraction =` is an argument of how much of the specified Axes should be used for the color bar.

Shape adjustment

When changing the thickness, ʻaspect =`

plt.colorbar(aspect=20) #thin
plt.colorbar(aspect=50) #thick

[stackoverflow: How to decrease colorbar WIDTH in matplotlib?] (https://stackoverflow.com/questions/33443334/how-to-decrease-colorbar-width-in-matplotlib)

To add triangles to both ends of the color bar ʻextend =`

plt.colorbar(extend='both') #Triangle on both ends

Adjusting the scale

Use cbar.set_ticks () for the position of the scale and cbar.set_ticklabels () for the character string of the scale.

[Akira Takeshima's Web Page: Color Bar] (http://hydro.iis.u-tokyo.ac.jp/~akira/page/python/contents/plot/general/colorbar.html) [matplotlib: pylab_examples example code: colorbar_tick_labelling_demo.py] (https://matplotlib.org/examples/pylab_examples/colorbar_tick_labelling_demo.html)

Get the scale

Note that the calling method differs between ticks and ticklabels.

ticks = cbar.get_ticks()
ticklabels = [ticklabel.get_text() for ticklabel in cbar.ax.get_xticklabels()]

Even when changing only ticklabels, I get angry with set_ticks () must have been called. The same thing is specified as cbar.set_ticks (ticks).

For example, if you want to add a unit to the colorbar:

#the above-5 to 5 color bars as cbar
ticks = cbar.get_ticks() # [-4. -2.  0.  2.  4.]
ticklabels = [ticklabel.get_text() for ticklabel in cbar.ax.get_xticklabels()]
# ['−4', '−2', '0', '2', '4']
ticklabels[-1] += ' [unit]'
cbar.set_ticks(ticks)
cbar.set_ticklabels(ticklabels)

Change font size

cbar.ax.tick_params(labelsize=10)

[stackoverflow: Change fontsize of colorbars in matplotlib] (https://stackoverflow.com/questions/40184696/change-fontsize-of-colorbars-in-matplotlib)