`billiard_polygon` with periodic conditions does not work adequately about dynamicalbilliards.jl HOT 5 CLOSED

Hi Diego,
unfortunately you are correct... periodic doesn't work for n=6 for exactly the reason you said. But isn't it straightforward to do the condition:

if n==6 && setting == "periodic"
  normal = normalize([-w[2], w[1]])*sqrt(3)*r
else
  normal = inr*normalize([-w[2], w[1]])
end

? What would the new function do differently?

Also you are right about the plotting: it only works for orthogonal vectors :(
A new function has to be made for triangular/hexagonal periodicity as far as plotting is concerned...

from dynamicalbilliards.jl.

Also, are you using this package? What are you working on? I am very curious and also very happythat somebody else is using this package!

from dynamicalbilliards.jl.

Hm, also, I just took a much closer look at the source code there. I think the only problem is that I use
normal = inr*normalize([-w[2], w[1]]) instead of normal = 2inr*normalize([-w[2], w[1]]) (notice the 2). Since two opposite faces have distance of twice the inner radius, shouldn't this fix all problems? (besides plotting)

from dynamicalbilliards.jl.

Hi there! Oh, you are right.: it seems than normal = 2.0*inr*normalize([-w[2], w[1]]) fix the problem.

I am using the package to calculate the spectrum of Lyapunov exponents in Billiards. Just changing few things in the main code it seems very suitable for this purpose.

I take this chance to thank you for sharing the package :)

Let me know if I should close the issue

from dynamicalbilliards.jl.

yeap we can close this. I just did a commit that changed it to 2inr.

from dynamicalbilliards.jl.