Laplace & Stokes support

[tanderson92]

It would be useful to me if this package would wrap Laplace and Stokes. In addition to FMM2D being described as an improvement to FMMLIB2D it has clear API advantages: it does checking for $x == x_j$ and drops such terms from the sum. This would be useful in a project I am working on, except the julia package doesn't support Laplace or Stokes.

Not essential, but these features would be appreciated (by myself and @maltezfaria)!

[mipals]

Hi there @tanderson92,

I've been wanting to add Laplace and Stokes for a while now, but due to missing documentation I've been putting it aside. In January I was told that the documentation would be added soon, but it does not seem like it has happened yet. Maybe ping Manas in the other issue?. In any case I will have a look tomorrow to see how easy/difficult it would be to simply look at the Fortran code to create the missing wrappers.

[mipals]

Hi again,

Manas was kind enough to quickly add the mathematical notation for the problems to the documentation. The branch adding_laplace_wrapper now adds the wrappers for Laplace. I've only tested the potential, so I do not know if the gradients and Hessians works as intended.

Note also that cfmm2d is broken as the _jll library is outdated. When I get them updated I will merge the branch to the main branch.

How much do you need Stokes? (I think it will require a little more effort to include)

Cheers,
Mikkel

[tanderson92]

Hi Mikkel,
I think (but am not sure yet) that I need Laplace more. (The reason being that IFGF should support 2D Stokes, so if nothing else I can use that.)

I will test your branch when I get a chance, hopefully very soon.

Thank you!
--Thomas

[mipals]

Hi Thomas,

Ok, I will see if I have time to look into Stokes. Its not something I have experience with, so I might wait a little until the actual documentation is up-to-date (currently only the mathematical description is there).

Note that I have already merged the branch into main and made a new release that includes updated binaries. Unfortunately it still seem as cfmm2d still does not work as intended with non-zero charges. So be careful if you use that functionality.

Cheers,
Mikkel