I use matplotlib.pyplot to generate plots. When you show the plots in my Spyder or Jupyter Notebook IDEs and save them by clicking, the default format is .png. However, in scientific publications, we use vector images to make sure they work well when embedded in text (i.e. the dots and captions do not blur when zooming in and zooming out).

In Spyder IDE, you can change the default format of image saving to .svg, but it seems that it does not work well when embedded in LaTeX, and some journals do not accept the format (???). The universal way is to save them as .pdf format.

However, when the script for plotting stops running and the figures are already on the screen, it is hard to save them as .pdf format. It could be frustrating if it needed approx 10 hours to run the script. You do not want to re-run it! Therefore, it is important to develop a habit for embedding plot saving code in your script before you run it. Here is the hint. If your script output multiple images, use

import matplotlib.pyplot as plt

# Plot the first figure
def plot_first():
	fig, ax = plt.subplots() # create a figure object
	... i.e. ax.plot()
	plt.savefig('first.pdf', format='pdf')

# Plot the second figure
def plot_desond():
	fig, (ax1,ax2,ax3) = plt.subplots() # create a figure object
	... i.e. ax1.plot()
	plt.savefig('second.pdf', format='pdf')

plot_first()
plot_second()

Note that it is very important to create figure objects for multiple-figure saving!