Why does CenterField contain no data for BoxModelGrid?

[glwagner]

I noticed that CenterField is defined as nothing on BoxModelGrid:

OceanBioME.jl/src/BoxModel/boxmodel.jl

Line 190 in 4b7b120

CenterField(::BoxModelGrid, args...; kwargs...) = nothing

But if I understand this correctly, a box model grid is effectively 0-dimensional. That means that any field would actually have 1 point (not 0 points). Otherwise, what is the prognostic state of a BoxModel?

I'm asking because I think the concept of a 0D grid could be possibly useful in Oceananigans (but right now we use a triply-Flat grid for the same purpose when we need it).

[iuryt]

I am curious about the utility of having 0D grid. Can you explain that?

[glwagner]

There are non-trivial dynamics on a 0D grid --- inertial oscillations, chemical and biological reactions, etc

[jagoosw]

This is a good question, I think at the time of writing it was just a workaround to allow models to be constructed (through the default setup of the BGC models light attenuation models get constructed which setup fields I think). I'll have a look at how it's used.

Tbh all of the box model code could do with some rethinking to be better aligned with the rest of the code. No one's used it much so it's not had a lot of testing or thought.

[glwagner]

How is the prognostic state of a BoxModel stored?

[jagoosw]

Just as a Tuple of Floats (vs a tuple of fields in Oceananigans)

[glwagner]

Ah, I realize that you are not using an Oceananigans model in this case. So the question is really why / where CenterField gets called at all.

[navidcy]

@jagoosw this was resolved with #164, right?

[jagoosw]

yeah, sorry should have said!