matplotlib - How to set a fixed/static size of circle marker on a scatter plot?

Question

I want to plot the location of some disks generated randomly on a scatter plot, and see whether the disks are 'connected' to each other. For this, I need to set the radius of each disk fixed/linked to the axis scale.
The 's' parameter in the plt.scatter function uses points, so the size is not fixed relative to the axis. If I dynamically zoom into the plot, the scatter marker size remains constant on the plot and does not scale up with the axis.
How do I set the radius so they have a definite value (relative to the axis)?

Answer 1

Instead of using plt.scatter, I suggest using patches.Circle to draw the plot (similar to this answer). These patches remain fixed in size so that you can dynamically zoom in to check for 'connections':

import matplotlib.pyplot as plt
from matplotlib.patches import Circle # for simplified usage, import this patch

# set up some x,y coordinates and radii
x = [1.0, 2.0, 4.0]
y = [1.0, 2.0, 2.0]
r = [1/(2.0**0.5), 1/(2.0**0.5), 0.25]

fig = plt.figure()

# initialize axis, important: set the aspect ratio to equal
ax = fig.add_subplot(111, aspect='equal')

# define axis limits for all patches to show
ax.axis([min(x)-1., max(x)+1., min(y)-1., max(y)+1.])

# loop through all triplets of x-,y-coordinates and radius and
# plot a circle for each:
for x, y, r in zip(x, y, r):
    ax.add_artist(Circle(xy=(x, y), 
                  radius=r))

plt.show()

The plot this generates looks like this:

Using the zoom-option from the plot window, one can obtain such a plot:

This zoomed in version has kept the original circle size so the 'connection' can be seen.

---

If you want to change the circles to be transparent, patches.Circle takes an alpha as argument. Just make sure you insert it with the call to Circle not add_artist:

ax.add_artist(Circle(xy=(x, y), 
              radius=r,
              alpha=0.5))