Experiment Types

In general, Modelon Impact can run two types of experiments:

• Single-run
• Multi-run

Single-run

If this is selected, Impact performs one simulation (1 "case") and you get exactly one result for each variable. For more information please check chapter: Simulate a model.

Multi-run

Here, Impact performs a batch simulation in which Impact independently changes parameters in specific categories set by the user. You get results for each successfully executed case.

In the Experiment layer, via the Diagram View, you can set up multi-run experiments, following these steps to perform Parameter sweeps.

The Experiment Setup View offers a more intuitive way to set up Parameter Sweeps and also Statistical Analyses.

For each of these multi-run types, you have several options to modify a parameter:

• Define a range or a set of value choices to perform Parameter sweeps.
• Distribute parameter values statistically using several distribution operators for a Statistical analysis.

Workflow

1. In Experiment layer, open Experiment View
2. Select
   • "Single run" for single-run experiments (selected by default), or
   • "Parameter sweep", a multi-run scenario to setup full factorial DoEs, or
   • "Statistical", another multi-run scenario to set up statistical Monte-Carlo analyses.
3. If "Statistical" was selected, you can furthermore specify the "Expansion" algorithm and number of samples
4. Continue with setting up modifications for parameters

The next step is then to define modifications for:

• Single run,
• Parameter sweep, or
• Statistical analysis.

Finally, you can set Outputs (if needed) and optionally Filters.

Related articles

Single run | Parameter sweep | Statistical analysis | Filters | Simulate a model | Perform Multi-run Experiments | Experiments in Modelon Impact