Start Class Implementation¶
simpaths.experiment.SimPathsStart is the single-run entry point. For commands and first-use setup, see Single Runs; this page describes the implementation.
Startup sequence¶
main()parses the command-line arguments.- In GUI mode,
runGUIdialog()presents the setup choices. In headless mode,runGUIlessSetup()performs setup when requested. - A setup-only invocation exits. Otherwise, startup reads the saved database country/year metadata before creating the simulation engine.
- The experiment builder registers the model, collector and, where enabled, observer.
- The engine builds the experiment. The GUI controls interactive execution; headless execution starts the simulation and waits for completion.
In the checked revision, run startup explicitly selects the UK and reads its saved start year from DatabaseCountryYear.xlsx. Supplying -s to a run-only invocation does not rebuild or replace that saved metadata. Use setup to prepare the intended start year.
Responsibilities¶
| Component | Responsibility |
|---|---|
parseCommandLineArgs() |
Country/year arguments, mutually exclusive setup-only and run-only modes, GUI and policy-schedule options |
runGUIdialog() / runGUIlessSetup() |
Coordinate preparation of the inputs |
buildExperiment() |
Create and register SimPaths managers |
Parameters.databaseSetup() and input parsers |
Build or load the required population and donor inputs |
SimPathsModel.buildObjects() |
Initialise the simulation population and model state |
SimPathsModel.buildSchedule() |
Define the simulation event order |
Database preparation is not all implemented inside the start class. Follow the called parser and parameter methods before changing input handling.
Extending startup¶
Keep setup and simulation responsibilities separate. A new option should have a defined default, validation and help text, and should behave consistently in GUI and headless use where both are supported.
Test setup-only and run-only paths, missing inputs, saved country/year metadata and the normal training-data workflow. Preserve custom policy schedules unless rewriting them was explicitly requested. Do not reset entity ID counters during setup or between runs.
For repeated execution and YAML configuration, see MultiRun Implementation.