Bug report

Bug summary

The color bar has multiple labels with the value 0 if matplotlib.colors.SymLogNorm is used leading to a "bold-looking" label.

Code for reproduction

from matplotlib import pyplot as plt
import matplotlib as mpl

fig, axes = plt.subplots(1, 2, True, True)

for i, ax in enumerate(axes): 
    im = ax.imshow([[0]], norm=mpl.colors.SymLogNorm(1e-5, vmin=-1, vmax=1))
    cb = plt.colorbar(im, ax=ax)

# Clean up the labels
zero_labelled = False
for label in cb.ax.yaxis.get_ticklabels():
    if label.get_text() == r'$\mathdefault{0}$':
        if zero_labelled:
            label.set_visible(False)
        zero_labelled = True

Actual outcome on the left, expected outcome on the right

Matplotlib version

• Operating system: Linux 1061836d2c20 4.9.49-moby Fix autofmt_xdate() when using in conjunction with twinx() #1 SMP Fri Dec 8 13:40:02 UTC 2017 x86_64 GNU/Linux
• Matplotlib version: 2.1.1
• Matplotlib backend (print(matplotlib.get_backend())): module://ipykernel.pylab.backend_inline
• Python version: 3.6.3
• Jupyter version (if applicable): 4.4.0
• Other libraries: numpy (1.13.3)

All libraries were installed using pip in a docker image derived from python:3.