Corner reflection

[mattia-lecci]

Problem:
A rare event was found: a corner reflection.
This can only happen for a number of reflection >1, since 2 adjacent triangles are needed.
The problem arises when the MoI identifies a point on the boundaries between the 2 triangles to be the reflection point P.

This creates 2 problems:

1. Considering, for example, 2 reflections, the identified path will have the following structure: TX -> P -> P -> RX. This creates a sort of "ghost" segment in the middle, which might create bugs somewhere else.
2. Given the symmetry of this path, reversing the 2 triangles will yield an identical ray. This means that 2 rays will be produced, acting a single one, and thus their power will always add coherently, thus being equivalent to the same ray with double the power (+3 dB).

Solution:
No solution for the moment.
This case should be rare enough to be negligible.
Other programs built on top of this should be aware of this issue and deal with it somehow.

Example case:
Rectangular empty room of size (14, 7, 3).
2 nodes in (1, 3, 1.6) and (1, 9, 1.6), respectively.
Corner reflection on floor and bottom wall, from the visualizer file:
(9, 3, 1.6), (5, 0, 0), (5, 0, 0), (1, 3, 1.6)

Notes:
This problem was found in the treetraversal branch, but the theory should hold in general, and the problem should thus be found in the master branch as well.

[neerajvarshney]

Hi Mattia Thank you for pointing out this.

Could you please share this example folder to resolve this bug?

Best

[mattia-lecci]

Hi Mattia Thank you for pointing out this.

Could you please share this example folder to resolve this bug?

Best

Sure, I'll send it to you via mail!

[neerajvarshney]

This issue has been resolved!

[mattia-lecci]

This issue has been resolved!

Update: the issue has been resolved in the master branch, but not in feature/treetraversal.

I will keep this issue open as a remainder.