Implement multi-mixotroph generalization of the NPBD model

[glwagner]

It would seem that the NPBD model might be our first "useful" model? So we should implement that in the source code.

I know @emilyzakem also has aspiration for an "Multi-species NPBD", where I think either P or B can profilerate into NP or NB "species". This model would then have NP + NB + 2 tracers.

Perhaps MultiSpeciesNutrientsPlanktonBacteriaDetritus is a good name for that model?

[emilyzakem]

Can I first show you a similar version of this model that does have multiple types to see what you think? It has nn and nd as well as np and nb. Hopefully that alleviates the "N" name problem too. I'd like to further think about merging p and b, and maybe even z, since there's lots of mixotrophs. The biogeochemical roles of populations would then be defined using matrices of substrate (including light, or other populations, in the case of grazing) inputs and outputs. I only worry this would get opaque quickly. Maybe next week I walk through my current Julia model with you?

…

On Tue, Apr 4, 2023, 5:49 PM Gregory L. Wagner ***@***.***> wrote: It would seem that the NPBD model might be our first "useful" model? So we should implement that in the source code. I know @emilyzakem <https://github.com/emilyzakem> also has aspiration for an "Multi-species NPBD", where I think either P or B can profilerate into NP or NB "species". This model would then have NP + NB + 2 tracers. Perhaps MultiSpeciesNutrientsPlanktonBacteriaDetritus is a good name for that model? — Reply to this email directly, view it on GitHub <#6>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHWPMDVP4JCNOFW22UNMD4LW7S6R7ANCNFSM6AAAAAAWTM6HQM> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[glwagner]

Yes sure, happy to take a look next week! You can also paste or link to code in this issue to further the discussion.

In terms of implementing this within the current Oceananigans interface, the main challenge is software. Right now we require a different function for each tracer. So here we will have, for example, model.tracers.mixotroph_1, model.tracers.mixotroph_2, etc, and we require a different "dynamics function" for each one that dispatches on ::Val{:mixotroph_1}, ::Val{mixotroph_2}, etc.

This challenge is easily overcome however, since we can use metaprogramming to automatically generate such functions. Again more of a julia 102 than julia 101 concept, but nonetheless straightforward in principle.

Once that minor software hurdle is overcome, its trivial to use matrices, tuples, etc to hold any number of parameters that generate different behavior for each mixotroph species within some homogeneous mathematical framework (ie all mixotrophs have their own maximum growth rate, mortality rate, etc).

[glwagner]

I changed the title of this issue to reflect the discussion!