best practice for "inactive" subdomains?

[jmeier]

Check these boxes if you have followed the posting rules.

• Q&A General is the most appropriate section for my question
• I have consulted the posting Guidelines on the Discussions front page
• I have searched the Discussions forum and my question has not been asked before
• I have searched the MOOSE website and the documentation does not answer my question
• I have formatted my post following the posting guidelines (screenshots as a last resort, triple back quotes around pasted text)

Question

Dear Moose Community,

For various tasks in the field of engineering and mining, the excavation (and possibly refilling) of defined volumes must be modelled. In Moose this is usually to be done by "parking" elements in "inactive subdomains" via one of the SubdomainModifiers (e.g CoupledVarThresholdElementSubdomainModifier or TimedSubdomainModifier). Initially inactive subdomains may be created via add_subdomain_ids and/or add_subdomain_names in MooseMesh.

With this discussion entry I'd like to gain insight on the best practice on how to define the inactive subdomain.

As recommented here and shown there for the "inactive" subdomains a Diffusion or MatDiffusion kernels may be used. As stated in the first link, kernel_coverage_check = false must also be set. In the setup used in this example, also material_coverage_check = false must be set.

I assume that the usual requirements for the inactive domain are as follows (open for discussion):

• Nothing should happen to the elements in the inactive domains (no numerical instabilities, etc.)
• Elements in the inactive domain should not add as less weight as possible to the memory and/or CPU consuption.
• Elements from an inactive subdomain can be reactivated at the user's request.

Now my questions start, all dealing on what are the best practice to define these inactive domains:

1. Is it really neccessary to define kernels in these inactive domains? I assume these kernels add some computational weight?
2. If kernels are to be added, what is the "best" kernel to add? Diffusion? MatDiffusion? From the description, NullKernel seems to be aimed at tests and debugging and not for "production".
3. Is there a way to avoid setting kernel_coverage_check = false and material_coverage_check = false while keeping the inactive subdomains lightweight? I'd love Moose to perform these checks in my "active" subdomains,
4. If the answer to point 3 above would be "no", what would be an acceptable solution as PR? A Parameter skip_kernel_coverage_check_on = 'list of subdomains' in [Problem] (same for skip_material_coverage_check_on)?

Kind regards,
Jörg

[GiudGiud]

Nothing should happen to the elements in the inactive domains (no numerical instabilities, etc.)

There should be either no equations there because the variables are block-restricted away from them
OR
Only trivial equations (dc/dt = 0) if for some reason the variables need to still be there

Elements in the inactive domain should not add as less weight as possible to the memory and/or CPU consuption.

Yes that seems fair. To do that, remove the variables and auxvariables from those subdomains

Elements from an inactive subdomain can be reactivated at the user's request.

This seems fine if you move them back to the active subdomains

Is it really neccessary to define kernels in these inactive domains? I assume these kernels add some computational weight?

No it's not. Just turn off the kernel_coverage_check

If kernels are to be added, what is the "best" kernel to add? Diffusion? MatDiffusion? From the description, NullKernel seems to be aimed at tests and debugging and not for "production".

NullKernel is fine. A TimeDerivative is fine too

Is there a way to avoid setting kernel_coverage_check = false and material_coverage_check = false while keeping the inactive subdomains lightweight? I'd love Moose to perform these checks in my "active" subdomains,

Not really. For now these checks are on/off. But you could change the checks to ignore the inactive subdomains. We would need new code to identify the inactive subdomains

If the answer to point 3 above would be "no", what would be an acceptable solution as PR? A Parameter skip_kernel_coverage_check_on = 'list of subdomains' in [Problem] (same for skip_material_coverage_check_on)?

This seems fine. An alternative could be this:

skip_kernel_coverage_check = 'false / true / active'

[jmeier]

On the things discussed on the kernel and material coverage checks:
I created issue #28117 and PR #28118 proposing an implementation of the ideas here.
Thanks for your support!

[GiudGiud]

@jmeier would you happen to have a good figure to show the use of these subdomain modifiers?
I am presenting on new systems in MOOSE (here mesh modifiers) at ANS Winter 2024 and could use it. I would credit you

[jmeier]

@GiudGiud: let me check...

[jmeier]

Dear @GiudGiud, please check your inbox. Please do not hesitate to contact me if you need something changed.

[GiudGiud]

This is really cool work, thank you!
If you get the chance this would be worth presenting here
https://inl.gov/mooseworkshop2025/