Matplotlib renders wrong ticks on the y axis

Question

I am trying to render specified ticks on the y axis using matplotlib Python package. The library for some reason renders more ticks than I give it to render and use scientific notation for them. For me it seems like a bug in the library, I was able to reproduce the issue on more devices. Is there a workaround?

The source code to reproduce the issue:

import matplotlib.pyplot as plt
import matplotlib.ticker as mticker
plt.figure(figsize=(6,4))
plt.xlim(0, 1000)
plt.xlabel('x')
plt.yscale('log')
plt.ylim(20, 200)
plt.yticks([20, 28, 39, 54, 75, 110, 150, 200])
plt.gca().get_yaxis().set_major_formatter(mticker.ScalarFormatter(useOffset=False))
plt.gca().get_yaxis().set_tick_params(which='minor', width=0)
plt.ylabel('y')
plt.savefig('test.png')
plt.close()

The rendered graph:

EDIT Calling plt.gca().get_yaxis().clear() after the plt.yscale('log') seems to solve the issue.

Answer 1

Commenting plt.yscale('log') out, seems to remove scientific notation from the y axis while keeping the desired tick values.

If you still need it, putting plt.yscale('log') after plt.ylim() and plt.yticks() seems to work:

import matplotlib.pyplot as plt
import matplotlib.ticker as mticker

plt.figure(figsize=(6,4))
plt.xlim(0, 1000)
plt.xlabel('x')
plt.ylim(20, 200)
plt.yticks([20, 28, 39, 54, 75, 110, 150, 200])
plt.yscale('log')
plt.gca().get_yaxis().set_major_formatter(mticker.ScalarFormatter(useOffset=False))
plt.gca().get_yaxis().set_tick_params(which='minor', width=0)
plt.ylabel('y')
plt.savefig('test.png')
plt.close()

To replace 100 for 10^2, I commented the first plt.gca() line: