Skip to content

Running multiple scenarios

Use a normal Python loop when the same model setup should be solved for several parameter values, for example a range of carbon budgets or discount rates:

from mimosa import MIMOSA, load_params

for budget in ["500 GtCO2", "700 GtCO2", "1000 GtCO2"]:

    params = load_params()

    params["emissions"]["carbonbudget"] = budget

    model3 = MIMOSA(params)
    model3.solve()

    model3.save(f"run_example3_{budget}")  # (1)!
  1. Every run needs a unique output name. Otherwise a later iteration overwrites the CSV and parameter file from an earlier run.

This example performs three sequential optimisations and creates one CSV and one .params.json file for each carbon budget. The files can be loaded together in the MIMOSA Dashboard to compare the scenarios.

Points to keep in mind

  • Call load_params() inside the loop so that every scenario starts from the same defaults.
  • Include the varied value in the output name, and use a filename-safe label for more complicated parameter values.
  • The loop stops if model construction or optimisation raises an error. For a large unattended experiment, add logging and decide explicitly whether a failed scenario should stop the remaining runs.
  • This pattern runs scenarios one after another. It does not run them in parallel.