2d cirular domain modeling aiming to mimic the intersection of LAPD

[liangwang0734]

Using the Cornerific Tapered2 Mapping from this paper, originally brought to me by @manauref

• This is 2d only to see how the grid performs.

• The time step size seems to scale as 1/N^2 instead of 1/N, where N=Nx=Ny is the grid number.

  • I don't quite understand why.
  • This might not matter for 3d extension since dz is guaranteed to scale as 1/N.

• The results have slightly strange matters near bad cells (on the diagonal corners) but overall are OK.

• Friction is allowed but not enabled by default. See the occurrences of the has_collision option and the associated collision frequency table.

• The mapping looks like the following (with color lines to help indicating mapped lines):

• The Ex and Bz from a 128x128 run look like the following:
  

[manauref]

This is pretty cool Liang. You say there are strange things near the corner cells. Can you show an example? those two plots above look smooth near the corners

…

On Mon, 11 Sept 2023 at 12:15, Liang Wang ***@***.***> wrote: Using the Cornerific Tapered2 Mapping from this paper <https://arxiv.org/pdf/1709.07875.pdf>. - This is 2d only as a starting point. - The time step size seems to scale as 1/N^2 instead of 1/N. - Why? - Might not matter for 3d extension since dz is guaranteed to scale as 1/N. - Have slightly strange matters near bad cells but overall OK. The mapping looks like the following (with color lines to help indicating mapped lines): [image: indexCornerific Tapered2 Mapping] <https://user-images.githubusercontent.com/1000547/267123923-8626e829-e1d7-4285-9f95-4eb5c588ac83.png> The Ex and Bz from a 128x128 run looks like the following [image: 03-frame30-ExBz] <https://user-images.githubusercontent.com/1000547/267123985-8a7e4aa3-51b0-4186-8e3f-0a18c81d2ee5.png> ------------------------------ You can view, comment on, or merge this pull request online at: #198 Commit Summary - eb5ba63 <eb5ba63> 2d cirular domain modeling aiming to mimic the intersection of LAPD File Changes (1 file <https://github.com/ammarhakim/gkylzero/pull/198/files>) - *A* regression/rt_5m_lapd_2d_circ.c <https://github.com/ammarhakim/gkylzero/pull/198/files#diff-4d63595ba1731ce80f912edb63fcc729f1e673eaadafaafb1f14cc7ae321883c> (560) Patch Links: - https://github.com/ammarhakim/gkylzero/pull/198.patch - https://github.com/ammarhakim/gkylzero/pull/198.diff — Reply to this email directly, view it on GitHub <#198>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAQ6DKGCVL57ED72MGVHU6TXZ5POTANCNFSM6AAAAAA4TZLT3E> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[ammarhakim]

This looks very nice. I wonder if what is going is that the cells near the "corners" are getting so squished that their local area scales as 1/N^2, leading to a more stringent time-step limit that one would expect. I guess this is equivalent to saying that we now moved the small sized cell from the center to the edges but kept the time-step restriction still the same.