Possible Limitation with Dirichlet BCs

[cmhamel]

Hi, first off great framework!

I started brainstorming on using Trixi to map some old grad student code for processing modeling of digital light processing (DLP, it's a light based 3D printing method) after seeing the recent contributions of hyperbolic diffusion and got hung up on the current implementation of Dirichlet BCs in Trixi. The Beer-Lambert law is an equation that shows up in the system of equations I'm interested in which is hyperbolic but with a time derivative assumed to be zero due to the differing time scales of the problem. The Beer-Lambert law (in 1D for simplicity) only has a one BC on the end points of the domain, physically it's the point at which light is penetrating the media. The other end point is free of a BC constraint. It seems that Trixi's Dirichlet BCs currently assumes "symmetric" Dirichlet BCs in the sense that if you have one BC on -x, you need one on +x. Am I correct in thinking this?

I was wondering how deep this convention goes in the framework? Is it hard-coded all the way down? This equation I described is likely a unique case so probably not worth refactoring the framework unless others have expressed similar concerns. I'm mainly curious of the right places to look if I wanted to modify my own local version and open up a discussion if others have ran into this limitation as well.

Craig

[jlchan]

Hi Craig! While I'm not too familiar with the details of your system, it sounds like you might be looking for boundary_condition_do_nothing at the right endpoint. Does that help?

[cmhamel]

Thanks so much. That definitely does help!

Is this documented and I somehow missed it?

[jlchan]

Hmm...I'm not sure that it is. I can add one soon. Thanks for bringing that to our attention!

[jlchan]

Ah, nevermind - my mistake, boundary_condition_do_nothing is still in the dev branch https://github.com/trixi-framework/Trixi.jl/blob/dev/src/basic_types.jl.

We probably won't be merging dev for at least a few weeks, so if you'd like to use boundary_condition_do_nothing, you can just add the routine locally and use it.

[cmhamel]

Thanks @jlchan this was very helpful.

You can close the issue if you'd like, unless y'all would like to use it to track documentation status.

[jlchan]

Sounds good. I think we can close this once the changes in dev are merged, this so good to keep around for reference.

[jlchan]

When #1149 is merged, this should be closed