radius modification in contains_point function when linewidth is specified

[Feyn-Man]

Bug report

Bug summary
When specifyng linewidth in a patch object and then using the contains_point function the radius of this is set to linewidth and returns inconsistent results. This is somewhat documented here but I think this is an unwanted behaviour since linewidth is specified in points and not in data coordinates, while the contains_point function gonna search for those data points enclosed between the border of the path and half the linewidth.

Code for reproduction

import matplotlib.patches as mpatches
import matplotlib.pyplot as plt

circle_ec = mpatches.Circle((0,0), radius=1, ec='r', lw=1)
circle = mpatches.Circle((0,0), radius=1)
point = (0,1.5)
print(circle_ec.contains_point(point))        
print(circle.contains_point(point))

Actual outcome

True
False

Expected outcome

The expected output is False in both cases since the point (0, 1.5) is neither in the patch (Circle in this case) neither on the line of the edge. I propose to multiply _radius = self.get_linewidth() in the _process_radius function in the source code by a corrective factor.

Matplotlib version

• Operating system: Windows 10
• Matplotlib version: 3.3.4
• Matplotlib backend: TkAgg
• Python version:
• Jupyter version (if applicable): 3.9.0
• Other libraries:

Installation from pip

[github-actions]

This issue has been marked "inactive" because it has been 365 days since the last comment. If this issue is still present in recent Matplotlib releases, or the feature request is still wanted, please leave a comment and this label will be removed. If there are no updates in another 30 days, this issue will be automatically closed, but you are free to re-open or create a new issue if needed. We value issue reports, and this procedure is meant to help us resurface and prioritize issues that have not been addressed yet, not make them disappear. Thanks for your help!

[tacaswell]

https://matplotlib.org/stable/api/_as_gen/matplotlib.patches.Patch.html#matplotlib.patches.Patch.contains_point The input to contains_point is understood to be in the coordinate system of patch.get_transform() which is (for the vast majority of rendered Patches) screen space.

I opened a PR to document the logic to pick the default radius, but I think there is still a latent bug with screen space being in pixels, but the linewidth being in points so we are (in all cases with reasonable dpi) not extending the range far enough.

I am not sure we know enough about the source / target of the transforms to be able to reliably fix that problem.