Model script

Parameter Handover

• from param.in: This file exists during Dakota-runs and it contains the parameter values and the ID (number) of the evaluation.
• via template and preprocessor

Objective function

For model calibration runs, the difference between reference data (typically from experiments) and simulated data must be quantified.

Reference data and simulated data may have different ranges and different values of the independent variable. A difference can only be defined in the union of their ranges. This union is between max(min(x_ref), min(x_sim)) and min(max(x_ref),m max(x_sim)). Within this range, both y_ref and y_sim must be sampled at n equidistant points, where n is dependent on data smoothness.

As a distance measure or objective function the root mean square of the differences between y_sim and y_ref at the sampling points can be used.

For each evaluation the model script must produce a plot containing the following:

• reference data plot
• simulation data plot
• parameter values
• value of objective function