Single Runs¶
Use a single run to inspect one configured simulation, explore the graphical interface, or test a model change. For repeated runs and YAML configuration, use Multiple Runs.
Before you start¶
Complete Environment Setup and Running Your First Simulation. Run the commands below from the repository root, where input/ and the executable JAR files are located.
The bundled training population uses 2019. It is suitable for learning and testing, not for substantive research. Research runs require authorised input data and a compatible start year.
Run without the graphical interface¶
Build the executables if needed:
For a first run with the bundled UK training data, prepare the inputs:
The --rewrite-policy-schedule option rebuilds the policy schedule from the available donor files. Do not use it to overwrite a research scenario's custom schedule unintentionally.
The run-only path reads the saved country/year metadata. Its -s argument does not replace that saved year, so complete setup for the intended start year first.
Then run the simulation:
-Setup prepares inputs and exits; -Run runs without setup. They cannot be used together. With neither option, the entry point performs setup and then runs.
Run interactively¶
Follow the first GUI run instructions to select the year and complete setup. Adjust the exposed model and collector settings before building the simulation, then start it. Population size and other initialisation settings require a rebuild to take effect.
Choose the right configuration route¶
The single-run command accepts country, start year, setup/run mode, GUI mode and policy-schedule rewriting. It does not accept the multi-run -config, -p or -e options. For a reproducible run with explicit population size, end year and output settings, use Multiple Runs with maxNumberOfRuns: 1.
Inspect the options supported by your executable:
Check the result¶
Read the console for completion or exceptions, then inspect the newly created run directory under output/. The files present depend on the collector settings; see Statistical display. Keep the code revision, input and parameter versions, seed, settings and policy schedule with the results.
If the run reports a missing population table, check that setup and simulation use the same country and year. If donor files or a policy year are missing, check Tax-Benefit Donors (UK) before rebuilding.
These commands follow SimPathsStart. For an older release, use that release's --help output.