New process or new model?​#

A process names a biological or physical question. A model is one hypothesis or formulation used to answer it. For example, two photosynthesis equations can belong to the same process even if they need different inputs.

Before creating a model, check whether the process already exists. That lets users find your implementation beside the alternatives they may want to compare.

Find the existing family​#

Load the package that provides the models. This example uses the teaching models distributed with PlantSimEngine:

using PlantSimEngine
using PlantSimEngine.Examples

growth_models = Authoring.available_models(AbstractGrowthModel)
growth_models

For a scientific application, load its model package first and inspect its documentation. Authoring.available_processes() lists the process types available in the packages you have loaded. Authoring.available_models(process_type) then lists the model types that implement that process.

Choose a model, give it a parameter value, and inspect its inputs and outputs:

candidate = ToyRUEGrowthModel(0.2)
(
    process=process(candidate),
    inputs=inputs(candidate),
    outputs=outputs(candidate),
)

Loaded model catalog explains discovery and inspection in more detail. These functions only find packages loaded into Julia. Also check the package you plan to use before concluding that a process is missing.

Decide what your new equation changes​#

Give alternative equations separate model types so users can select and test them. To compare parameter values in the same equation, create instances of that model with different parameters. Models in one process may need different inputs, produce different outputs, or run at different frequencies. Use Model compatibility and replacement before substituting one.

Declare a genuinely new process​#

This declaration creates a process for a teaching example of root exudation:

PlantSimEngine.@process "docs_root_exudation" verbose=false

struct DocsLinearExudation{T} <: AbstractDocs_Root_ExudationModel
    fraction::T
end

process(DocsLinearExudation(0.1))

@process creates the abstract type that concrete models inherit from:

When another package already declares the process, import its abstract type and use it after <: in your model definition. This puts your model in the same family as the existing alternatives.

The type above is only the beginning. It still needs input and output declarations, descriptions of the variables' units and meaning, an equation, and tests. Continue with Implement a basic model to complete those steps.