Invoke mask_immersed_field! when converting args for Makie

[glwagner]

@navidcy I think you can use this in the internal tide example?

[navidcy]

lgtm!

only minor thing is that we need to call mask_immersed_field!(f, NaN) in three different places... bit trickier to modify if needed.

we could be calling it within flattened_cpu_interior but the method name doesn't suggest that... if you think of a way to improve do it!

[navidcy]

See https://clima.github.io/OceananigansDocumentation/previews/PR3725/literated/internal_tide/ !

[glwagner]

we could be calling it within flattened_cpu_interior but the method name doesn't suggest that... if you think of a way to improve do it!

Good idea. Find a new name for the function! What it does:

1. Drops singleton dimensions
2. Converts from CPU to GPU
3. masks immersed regions with NaN

Perhaps simple convert_field, because in the context of the extension "convert" means to change an argument from the input into something plottable. Or perhaps make_plottable_field(f)

[navidcy]

I'm tempted to merge this. But I don't understand why the code snippet below is not working:

Look:

using Oceananigans, CairoMakie

underlying_grid = RectilinearGrid(size = (250, 125),
                                  x = (-1000e3, 1000e3),
                                  z = (-2e3, 0),
                                  halo = (4, 4),
                                  topology = (Periodic, Flat, Bounded))

bottom(x) = - underlying_grid.Lz + 500 * exp(-x^2 / (2*(20e3)^2))

grid = ImmersedBoundaryGrid(underlying_grid, GridFittedBottom(bottom))

lines(grid.immersed_boundary.bottom_height)

gives:

which is all good! But:

x, y, z = nodes(grid.immersed_boundary.bottom_height)

lines(x, grid.immersed_boundary.bottom_height)

Perhaps this is a bit informative

lines(x/1e3, grid.immersed_boundary.bottom_height)

[navidcy]

Hm...

julia> using Oceananigans, CairoMakie

julia> underlying_grid = RectilinearGrid(size = (250, 125),
                                         x = (-1000e3, 1000e3),
                                         z = (-2e3, 0),
                                         halo = (4, 4),
                                         topology = (Periodic, Flat, Bounded));

julia> bottom(x) = - underlying_grid.Lz + 500 * exp(-x^2 / (2*(20e3)^2));

julia> grid = ImmersedBoundaryGrid(underlying_grid, GridFittedBottom(bottom));

julia> c = Field{Center, Nothing, Nothing}(grid)
250×1×1 Field{Center, Nothing, Nothing} reduced over dims = (2, 3) on ImmersedBoundaryGrid on CPU
├── grid: 250×1×125 ImmersedBoundaryGrid{Float64, Periodic, Flat, Bounded} on CPU with 4×0×4 halo
├── boundary conditions: FieldBoundaryConditions
│   └── west: Periodic, east: Periodic, south: Nothing, north: Nothing, bottom: Nothing, top: Nothing, immersed: ZeroFlux
└── data: 258×1×1 OffsetArray(::Array{Float64, 3}, -3:254, 1:1, 1:1) with eltype Float64 with indices -3:254×1:1×1:1
    └── max=0.0, min=0.0, mean=0.0

julia> set!(c, rand())
250×1×1 Field{Center, Nothing, Nothing} reduced over dims = (2, 3) on ImmersedBoundaryGrid on CPU
├── grid: 250×1×125 ImmersedBoundaryGrid{Float64, Periodic, Flat, Bounded} on CPU with 4×0×4 halo
├── boundary conditions: FieldBoundaryConditions
│   └── west: Periodic, east: Periodic, south: Nothing, north: Nothing, bottom: Nothing, top: Nothing, immersed: ZeroFlux
└── data: 258×1×1 OffsetArray(::Array{Float64, 3}, -3:254, 1:1, 1:1) with eltype Float64 with indices -3:254×1:1×1:1
    └── max=0.952381, min=0.952381, mean=0.952381

julia> lines(c)

julia> c
250×1×1 Field{Center, Nothing, Nothing} reduced over dims = (2, 3) on ImmersedBoundaryGrid on CPU
├── grid: 250×1×125 ImmersedBoundaryGrid{Float64, Periodic, Flat, Bounded} on CPU with 4×0×4 halo
├── boundary conditions: FieldBoundaryConditions
│   └── west: Periodic, east: Periodic, south: Nothing, north: Nothing, bottom: Nothing, top: Nothing, immersed: ZeroFlux
└── data: 258×1×1 OffsetArray(::Array{Float64, 3}, -3:254, 1:1, 1:1) with eltype Float64 with indices -3:254×1:1×1:1
    └── max=NaN, min=NaN, mean=NaN

After plotting a field then it has NaNs. Is this OK?
Of course, we expect that because we masked it in immersed regions. Is the issue here the show method? Should it be ignoring immersed regions?

[glwagner]

Is the issue here the show method? Should it be ignoring immersed regions?

Yes, that's a bug. Hopefully @simone-silvestri can help who designed the conditional reductions

[glwagner]

I'm tempted to merge this. But I don't understand why the code snippet below is not working:

Look:

using Oceananigans, CairoMakie

underlying_grid = RectilinearGrid(size = (250, 125),
                                  x = (-1000e3, 1000e3),
                                  z = (-2e3, 0),
                                  halo = (4, 4),
                                  topology = (Periodic, Flat, Bounded))

bottom(x) = - underlying_grid.Lz + 500 * exp(-x^2 / (2*(20e3)^2))

grid = ImmersedBoundaryGrid(underlying_grid, GridFittedBottom(bottom))

lines(grid.immersed_boundary.bottom_height)

gives:

which is all good! But:

x, y, z = nodes(grid.immersed_boundary.bottom_height)

lines(x, grid.immersed_boundary.bottom_height)

Perhaps this is a bit informative

lines(x/1e3, grid.immersed_boundary.bottom_height)

Can you check whether this behavior is actually changed by this PR? It seems like maybe this would have happened before this PR anyways.

[navidcy]

I'm tempted to merge this. But I don't understand why the code snippet below is not working:
Look:

using Oceananigans, CairoMakie

underlying_grid = RectilinearGrid(size = (250, 125),
                                  x = (-1000e3, 1000e3),
                                  z = (-2e3, 0),
                                  halo = (4, 4),
                                  topology = (Periodic, Flat, Bounded))

bottom(x) = - underlying_grid.Lz + 500 * exp(-x^2 / (2*(20e3)^2))

grid = ImmersedBoundaryGrid(underlying_grid, GridFittedBottom(bottom))

lines(grid.immersed_boundary.bottom_height)

gives:

which is all good! But:

x, y, z = nodes(grid.immersed_boundary.bottom_height)

lines(x, grid.immersed_boundary.bottom_height)

Perhaps this is a bit informative

lines(x/1e3, grid.immersed_boundary.bottom_height)

Can you check whether this behavior is actually changed by this PR? It seems like maybe this would have happened before this PR anyways.

Right!. Indeed it's not a result of this PR.
So we'll merge and deal with this bug in a different PR.