Add option to input functional groups and size classes parameters #912

Rosejoycrocker · 2025-01-15T22:53:05Z

Adds the option to input the number of functional groups and size classes and associated parameters as an input file when loading the domain. If no file is input, the domain loads the default parameters as normal.

Changes to CoralGrowth so number of functional groups and size classes is not hard coded
If no file is provided, default parameters and number of functional groups and size classes are loaded as normal
coral_spec() requires an input dictionary with the base parameters required to build the spec, but the default can be loaded without inputs using ADRIA.load_default_coral_spec()

@ConnectedSystems, I'm not sure what you think of storing alternative coral parameters in a json. This was used due to being able to load directly into a dict, but open to other formats.

If no file provided, default parameters will be loaded

Add helper function for creating default coral_spec

…` to default parameter set Fix units in calculating colony areas Formatting Fix bugs causing test errors

ConnectedSystems · 2025-01-15T22:57:27Z

YAML is what we use else where and it supports in-file comments, but I'll take a look first

Rosejoycrocker · 2025-01-15T22:58:31Z

YAML is what we use else where and it supports in-file comments, but I'll take a look first

Sure, happy to change to that. I'll sort out the failing formatting test now. If you're happy with the setup I'll also add a test for loading the coral_spec from a file, at the moment the tests only test the default loading.

ConnectedSystems · 2025-01-16T02:58:08Z

Sorry, could you say a few words on what the intended use case being supported is?

Changing the spec won't affect the data or any of the modelling done via ADRIA, so just trying to understand the purpose better

ConnectedSystems

Still waiting on the above but for now I'll say I wish you discussed these changes with someone beforehand.

The principle concerns I have are:

Relies on a separate (and currently undocumented?) JSON file that the user provides when loading a domain.
I suspect this will create mismatches all over ADRIA as some code rely on the "default" coral specification, and others don't
Adds a coral_params field to the Domain type that is not consistently used. It also duplicates the function of the model_spec.
Requires even more changes to the metrics module than what is currently included in this PR.

Specific to just changing the number of functional groups and size classes, a cleaner approach would be to add methods to allow the model_spec and coral_spec to be changed/updated with new coral-relevant factors.

See relevant code, which you would have seen when working on this PR:

ADRIA.jl/src/ecosystem/corals/Corals.jl

Line 465 in 50c3aa0

    
           function _update_coral_factors(spec::DataFrame, coral_params::DataFrame)::DataFrame

ADRIA.jl/src/ecosystem/corals/Corals.jl

Lines 370 to 371 in 50c3aa0

    
           function create_coral_struct(bounds::Tuple{Float64,Float64}=(0.9, 1.1))::Nothing 
        
               _, base_coral_params, p_vals = coral_spec()

ADRIA.jl/src/ecosystem/corals/Corals.jl

Line 412 in 50c3aa0

function to_coral_spec(m::Coral)::DataFrame

ConnectedSystems · 2025-02-02T07:31:49Z

src/Domain.jl

+function load_domain(path::String; coral_spec_path::String="")::Domain
+    return load_domain(path, ""; coral_spec_path=coral_spec_path)


Sorry, I don't see how this works.

The Domain datasets themselves hold data relevant to initial coral cover, susceptibility to cyclones and other info.

ConnectedSystems · 2025-02-02T07:32:32Z

src/ExtInterface/ADRIA/Domain.jl

+                "
+The number of functional groups and size classes specified in the coral parameters file
+does not match the size of the initial coral cover matrix.
+"


Multi-line strings are triple quoted:

""" like so """

ConnectedSystems · 2025-02-02T07:34:01Z

src/ecosystem/corals/CoralGrowth.jl

+    # The following splits into "small, mid and large" indices based on number of size classes and functional groups
+    indices_array = zeros(Int64, (n_groups, n_sizes))
+    for n_g in 0:(n_groups - 1)
+        indices_array[n_g + 1, :] .= collect((n_g * n_sizes + 1):(n_g * n_sizes + n_sizes))
+    end
+
+    mid_ind = collect(indices_array[:, 2:(end - 1)])[:]
+    size_mid = prod(size(indices_array[:, 2:(end - 1)]))
+
+    small::SVector{n_groups} = indices_array[:, 1]
+    mid::SVector{size_mid} = mid_ind
+    large::SVector{n_groups} = indices_array[:, end]
    p = @NamedTuple{
-        small::StaticArrays.SVector{5,Int64},  # indices for small size classes
-        mid::StaticArrays.SVector{25,Int64},   # indices for mid-size corals
-        large::StaticArrays.SVector{5,Int64},  # indices for large corals
-        rec::Matrix{Float64},                  # recruitment values, where `s` relates to available space (not max carrying capacity)
-        sXr::Array{Float64,3},                  # s * X * r
-        X_mb::Array{Float64,3},                 # X * mb
-        r::Matrix{Float64},                    # growth rate
-        mb::Matrix{Float64}                    # background mortality
+        small::StaticArrays.SVector{n_groups,Int64},           # indices for small size classes
+        mid::StaticArrays.SVector{size_mid,Int64},            # indices for mid-size corals
+        large::StaticArrays.SVector{n_groups,Int64},           # indices for large corals
+        rec::Matrix{Float64},                           # recruitment values, where `s` relates to available space (not max carrying capacity)
+        sXr::Array{Float64},                           # s * X * r
+        X_mb::Array{Float64},                          # X * mb
+        r::Matrix{Float64},                             # growth rate
+        mb::Matrix{Float64}                             # background mortality
    }((                        # cache matrix to hold X (current coral cover)
-        # Cached indices
+        # cached indices
        small, mid, large,

-        # Cache matrices
-        zeros(n_groups, n_locs),           # rec
-        zeros(n_groups, n_sizes, n_locs),  # sXr
+        # cache matrices
+        zeros(n_groups, n_locs),  # rec
+        zeros(n_groups, n_sizes, n_locs), # sXr
        zeros(n_groups, n_sizes, n_locs),  # X_mb
-        zeros(n_groups, n_sizes),          # r
-        zeros(n_groups, n_sizes)           # mb
+        zeros(n_groups, n_sizes),  # r
+        zeros(n_groups, n_sizes)   # mb


This entire bit needs to be revisited as many of these cache are not used any more (they were for the older ODE method).

ConnectedSystems · 2025-02-02T07:34:33Z

src/ecosystem/corals/Corals.jl

+function bin_widths(bin_edges)
+    return bin_edges[:, 2:end] .- bin_edges[:, 1:(end - 1)]
 end


This will break any code that used the older implementation?

ConnectedSystems · 2025-02-02T07:35:34Z

src/ecosystem/corals/Corals.jl

-function colony_areas()
-    # The coral colony diameter bin edges (cm) are: 0, 2, 5, 10, 20, 40, 80
-    edges = bin_edges(; unit=:cm)
+function colony_areas(edges)


Same comment here. This will break any code that relies on the previous method.

To be clear, anyone who had written ADRIA.colony_areas() will see their code break.

ConnectedSystems · 2025-02-02T07:38:06Z

src/io/inputs.jl

    return ZeroDataCube(;
        timesteps=1:length(timeframe),
        locations=sort(loc_data.reef_siteid),
-        species=ADRIA.coral_spec().taxa_names,
+        species=ADRIA.default_coral_spec().taxa_names,
        scenarios=[1]
    )
 end


I'm unsure of the thinking here as this will only work for the default coral specification and nothing else.

Rosejoycrocker added 5 commits January 14, 2025 09:45

Add option to load file with coral parameter specification

99915a7

If no file provided, default parameters will be loaded

Adjustments to where coral_spec is called when calculating metrics

216acf9

Update/add function docs

a5f0cb1

Add helper function for creating default coral_spec

Formatting

06061d9

Fix references to coral_params in tests and add `planar_area_params…

0d67bd0

…` to default parameter set Fix units in calculating colony areas Formatting Fix bugs causing test errors

Rosejoycrocker requested a review from ConnectedSystems January 15, 2025 22:53

Rosejoycrocker added the enhancement New feature or request label Jan 15, 2025

Rosejoycrocker self-assigned this Jan 15, 2025

Rosejoycrocker changed the title ~~Add option to input functional groups and sizes classes parameters~~ Add option to input functional groups and size classes parameters Jan 15, 2025

Fix formatting

788eb2e

ConnectedSystems reviewed Feb 2, 2025

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add option to input functional groups and size classes parameters #912

Add option to input functional groups and size classes parameters #912

Rosejoycrocker commented Jan 15, 2025

ConnectedSystems commented Jan 15, 2025

Rosejoycrocker commented Jan 15, 2025 •

edited

Loading

ConnectedSystems commented Jan 16, 2025

ConnectedSystems left a comment •

edited

Loading

ConnectedSystems Feb 2, 2025

ConnectedSystems Feb 2, 2025

ConnectedSystems Feb 2, 2025

ConnectedSystems Feb 2, 2025

ConnectedSystems Feb 2, 2025

ConnectedSystems Feb 2, 2025

	function create_coral_struct(bounds::Tuple{Float64,Float64}=(0.9, 1.1))::Nothing
	_, base_coral_params, p_vals = coral_spec()

		function load_domain(path::String; coral_spec_path::String="")::Domain
		return load_domain(path, ""; coral_spec_path=coral_spec_path)

Add option to input functional groups and size classes parameters #912

Are you sure you want to change the base?

Add option to input functional groups and size classes parameters #912

Conversation

Rosejoycrocker commented Jan 15, 2025

ConnectedSystems commented Jan 15, 2025

Rosejoycrocker commented Jan 15, 2025 • edited Loading

ConnectedSystems commented Jan 16, 2025

ConnectedSystems left a comment • edited Loading

Choose a reason for hiding this comment

ConnectedSystems Feb 2, 2025

Choose a reason for hiding this comment

ConnectedSystems Feb 2, 2025

Choose a reason for hiding this comment

ConnectedSystems Feb 2, 2025

Choose a reason for hiding this comment

ConnectedSystems Feb 2, 2025

Choose a reason for hiding this comment

ConnectedSystems Feb 2, 2025

Choose a reason for hiding this comment

ConnectedSystems Feb 2, 2025

Choose a reason for hiding this comment

Rosejoycrocker commented Jan 15, 2025 •

edited

Loading

ConnectedSystems left a comment •

edited

Loading