Inspect A Simulation#
When a result is unexpected, first check what ran, where it ran, and which values it read. The Diagnostics functions return information you can display as a table. Use them to check how the simulation is configured. Check the scientific equations and parameter values separately.
| Question | Diagnostic |
|---|---|
| Which objects and labels exist? | Diagnostics.explain_objects(model) |
| Which models run on them? | Diagnostics.explain_applications(model) |
| Where does each input come from? | Diagnostics.explain_bindings(model) |
| Which initial values are supplied or missing? | Diagnostics.explain_initialization(model) |
| When does each application run? | Diagnostics.explain_schedule(model) |
| Which weather or spatial source is sampled? | Diagnostics.explain_environment_bindings(model) |
| Which models are called by a controller? | Diagnostics.explain_calls(model) |
| Why were earlier values of an output kept? | Diagnostics.explain_output_retention(simulation) |
Inspect a small working example#
Here leaf surface depends on supplied carbon biomass. Check its input source before running a larger structure:
julia
using PlantSimEngine, DataFrames
using PlantSimEngine.Examples
model = CompositeModel(ToyLeafSurfaceModel(0.02);
status=(carbon_biomass=50.0,), id=:leaf, scale=:Leaf)
DataFrame([
(variable=row.variable, role=row.role, disposition=row.disposition)
for row in Diagnostics.explain_initialization(model)
])2×3 DataFrame
| Row | variable | role | disposition |
|---|---|---|---|
| Symbol | Symbol | Symbol | |
| 1 | carbon_biomass | input | supplied |
| 2 | surface | output | generated |
julia
simulation = run!(model; outputs=:all)
DataFrame(Diagnostics.explain_output_retention(simulation))1×6 DataFrame
| Row | application_id | variable | reasons | retention_steps | current_target_count | current_output_object_count |
|---|---|---|---|---|---|---|
| Symbol | Symbol | Tuple… | Nothing | Int64 | Int64 | |
| 1 | leaf_surface | surface | (:all_outputs,) | 1 | 1 |
If model construction fails, use the error's object and application names to correct the configuration first. A diagnostic cannot inspect a model that was never constructed.
After growth or movement#
After adding or removing organs, PlantSimEngine updates which objects each model runs on and where its inputs come from. This happens after the application that changed the structure finishes. New organs may run models scheduled later in the same time step. Models that already ran are not repeated. If you change the structure between steps, these connections are updated before the next step.
If you move an organ or change its geometry, PlantSimEngine needs to find its new location in the spatial environment. Use functions such as move_object! and update_geometry! so it knows to update this connection. Removing an organ stops its future calculations and keeps results that were already saved. See Modify plant structure for an example that checks the affected objects and their carbon balance.
On this page