Fix

[AinurZiga]

Trying Sionna-RT for my research, I noticed strange behavior with wedges/edges and, particularly, a method for swapping edge extremities to orient identical edges in the same way. The problem was, as I understood that some coordinates (for example, r0 and r1) might coincide, but due to numerical issues, one might be larger than another (r0 > r1 yields True).

I changed the code, so comparison for nearly equal values starts before checking what number is larger.

My 1st commit contains an example where I have observed the problem. All edges have equal radii in spherical coordinates - due to numerical issues, r0 > r1 may give True.
My 2nd commit contains changes to the method for swapping edge extremities.

The code snippet which I used to reproduce the problem.

scene = load_scene(sionna.rt.scene.sphere)
num_edges = tf.reduce_sum(tf.cast(scene._solver_paths._is_edge, tf.int32))
num_wedges = scene._solver_paths._wedges_origin.shape[0] - num_edges
print(num_edges, num_wedges)

Expected output/output after the fix: 0 120
Output before the fix: 44 98

Signed-off by: Ainur Ziganshin [email protected]

Checklist

• Detailed description
• Added references to issues and discussions
• Added / modified documentation as needed
• Added / modified unit tests as needed
• Passes all tests
• Lint the code
• Performed a self review
• Ensure you Signed-off the commits. Required to accept contributions!
• Co-authored with someone? Add Co-authored-by: user@domain and ensure they signed off their commits too.

[faycalaa]

Hi,
Thank you for this fix @AinurZiga!

Could you please delete the commit that adds the sphere scene and only keep the one implementing the fix?

[AinurZiga]

Hi, Thank you for this fix @AinurZiga!

Could you please delete the commit that adds the sphere scene and only keep the one implementing the fix?

Hi Fayçal,
I removed the unnecessary commit.

[faycalaa]

Thank you!