"CPU not defined" error

[Lichriszz]

Hi,

When I run the shallow water model, I got an error message saying:

"LoadError: UndefVarError: CPU not defined
in expression starting at /Users/.../Desktop/Research/shallow water test.jl:8
(::Core.var"#Type##kw")(::NamedTuple{(:grid, :gravitational_acceleration),Tuple{RegularCartesianGrid{Float64,Periodic,Periodic,Bounded,OffsetArrays.OffsetArray{Float64,1,StepRangeLen{Float64,Base.TwicePrecision{Float64},Base.TwicePrecision{Float64}}}},Nothing}}, ::Type{ShallowWaterModel}) at shallow_water_model.jl:75
top-level scope at shallow water test.jl:8
include_string(::Function, ::Module, ::String, ::String) at loading.jl:1088"

Here are my codes:

using Pkg
pkg"add Oceananigans, JLD2, Plots"
using Oceananigans, Oceananigans.Advection, Oceananigans.Architectures

grid = RegularCartesianGrid(size=(128, 128, 1), extent=(2π, 2π, 2π))

model = ShallowWaterModel(grid = grid,
gravitational_acceleration = nothing,
)

Can anyone help? Some examples of the ShallowWaterModel function would be helpful.
Thanks!

[navidcy]

Hi @Lichriszz, could you say what version of Oceananigans.jl you used? E.g. the output of

julia> using Pkg; Pkg.status()

and

julia> versioninfo()

[navidcy]

First of all it doesn't seem that you imported ShallowWaterModel, e.g., using Oceananigans.ShallowWaterModel`. But also, I'm not sure if that's part of the v0.47.1 release.

Can you give a bit more background on what you are trying to do? E.g., are you trying to reproduce an example from the docs?

[Lichriszz]

When I try to do "using Oceananigans.ShallowWaterModel", or using Oceananigans.ShallowWaterModels (with an s), it says shallowwatermodel(s) not defined.
I am not working on a specific problem right now. I am just trying to understand the codes. In the future, we may use the shallow water model to simulate equatorial waves.

[navidcy]

May I suggest you try from one of the examples in the Docs?

The ShallowWater model is bot quite “cooked” yet. But it’s getting there!

Other may have more insight on this. @francispoulin, @glwagner

[Lichriszz]

Actually we are interested in the shallow water model specifically. But I will go over those examples first. Thanks!

[navidcy]

I see you. But unfortunately it's not quite ready. Until you see the ShallowWaterModel appearing in the Documentation then I'd advise not to use it or only "use it at your own risk". :)

[ali-ramadhan]

Hi @Lichriszz thanks for reporting this error. As @navidcy said, I think the shallow water model is not yet exported but @francispoulin can better comment on its status.

I've fixed the bug in PR #1363 so it should make it's way into the next tagged release.

If you want to use the shallow water model, for now to get around the bug I think you need to explicitly specify the architecture. This version of your initial example worked for me:

using Oceananigans, Oceananigans.Advection, Oceananigans.Architectures
using Oceananigans.Models: ShallowWaterModel

grid = RegularCartesianGrid(size=(128, 128, 1), extent=(2π, 2π, 2π))

model = ShallowWaterModel(architecture = CPU(), grid = grid, gravitational_acceleration = 10.0)

I think you need to specify a number for gravitational_acceleration otherwise you get a different error.

[Lichriszz]

Thank you!

[francispoulin]

Thanks @Lichriszz for pointing this out and sorry there was a problem to begin with.

I have played with an example for geostrophic adjustment, both one- and two-dimensional, and now working on the instability of a barotropic Bickley jet. Coming soon I hope!

What kind of example would you be looking for?

[Lichriszz]

Thanks for the reply. I am looking for an example that simulates the evolution of a system, like the incompressible model example here: https://clima.github.io/OceananigansDocumentation/stable/generated/two_dimensional_turbulence/.

[francispoulin]

Very good idea! I'll try and put together an example this afternoon and let you know what I come up with.

[Lichriszz]

Thank you!

[francispoulin]

@Lichriszz , I created an example and you can find the code at this link. It's more of a proof of concept at this stage as the results are not anything special. Are there particular parameters you want to run?

I also created a PR in case you wanted to make some comments on the code and what make it easier to use, depending on what you want to do with it.

[Lichriszz]

Thank you so much. I don't have particular parameters that I want to run.
This example is very helpful.