How Multiscale Composite Models Execute#
An application is a model configured with ModelSpec: it says where the model runs and where its inputs come from. For example, ModelSpec(model; on=Many(scale=:Leaf)) runs that model once on each leaf. Each leaf keeps its own values in Status. You define the objects, their labels, and their parent relationships when building the simulation.
Selectors choose objects by these labels and relationships. Self() means the object whose model is currently running. SelfPlant() limits a search to that object's plant, starting from the top of its structure. SceneScope() allows the search to include the whole simulation. Wrap a selection in One, OptionalOne, or Many to require exactly one match, allow zero or one, or accept a collection.
Common choices are:
| Relationship | Pattern |
|---|---|
| application targets every leaf | ModelSpec(model; on=Many(scale=:Leaf)) |
| input from this same object | omit inputs when exactly one model supplies the matching output |
| input from one ancestor | One(Ancestor(scale=:Plant)) |
| input from this plant's leaves | Many(scale=:Leaf, within=SelfPlant()) |
| input from shared soil | One(scale=:Soil, within=SceneScope()) |
| optional named organ | OptionalOne(name=:fruit, within=SelfPlant()) |
For a model running on a leaf, Self() means that leaf, not its plant or its species. Use object IDs and labels to select specific objects or groups. Scope(name) refers to a named search area that you must define in the simulation before using it.
Saved results identify the application, object, and variable. Each leaf's history therefore stays separate, including when you use the same process more than once. Removing an organ stops its future calculations but keeps the results already saved for it.
Before a long run, use Diagnostics.explain_objects to check the plant structure, Diagnostics.explain_applications to check where models run, and Diagnostics.explain_bindings to check where they read their inputs.