From what I know one needs as many halo points on either side of the domain as the order or number pf boundary conditions.
@dhruvbhagtani2105 you haven’t done that but argued that your formulation is still ok. Can you explain what you did here or someplace? I remembered you explained to me over zoom one day and it sounded reasonable. But then I remember some results with funny behaviour at boundaries and I'm suspicious that something funny might be happening with the boundary conditions.
It’s just hard for me to believe that what the rest people are doing is not needed but would be convinced immediately when I see it demonstrated working fine. :)
From what I know one needs as many halo points on either side of the domain as the order or number pf boundary conditions.
@dhruvbhagtani2105 you haven’t done that but argued that your formulation is still ok. Can you explain what you did here or someplace? I remembered you explained to me over zoom one day and it sounded reasonable. But then I remember some results with funny behaviour at boundaries and I'm suspicious that something funny might be happening with the boundary conditions.
It’s just hard for me to believe that what the rest people are doing is not needed but would be convinced immediately when I see it demonstrated working fine. :)