Module Parameters

Parameters are loaded into each module before they are initizalized. The exact steps of a simulation run are shown in Simluation Steps.

Parameters can use any ParaReader module, the JSON input module provided works for this.

The format for JSON parameters is:

{
  "module type": {
    "module name": {
      "parameter name": "parameter value"
    }
  },
  ...
}

Example of a Goals module parameters:

{
  ...
  "goals": {                  // Module Type
    "astar": {                // Module Name
      "endGoalType": "exit",  // Parameter values
      "gridSize": 0.1,
      "closestObstacle": 0.25,
      "goalRange": 0.05
    }
  },
  ...
}

Parameter Sweeps

Parameter sweeps run many simulations with randomized parameters to get a better idea of the possible outcomes.

Parameters

Parameters are provided as they are with a normal run, with some extra rules.

NOTE: parameters are randomized in a non-parameter sweep run if provided in this format

Non-Random Values

Any parameter that should not be randomized is provided as it is normally.

single_string: "example1"

single_float: 1.0

Array's can still be used as they normally would be.

array_param: [1.0, 2.0, 3.0]

Discrete Random Values

Parameters can be randomized between a set of discrete values.

For each simulation run, the actual value for the parameter is randomly chosen from the list given with a uniform distribution.

Examples:

discrete_string_param: [ "option1", "option2", "option3" ]

discrete_float_param: [ 1.0, 1.2, 1.4 ]

NOTE: Array parameters can NOT be randomized

// will NOT work

random_array_param: [ [1,2], [3,4] ]

Continuous Random Values

Numerical parameters can also be randomized within a continuous range.

For each simulation run, the actual value for the parameter is randomly chosen with a uniform distribution from min to max inclusively.

Examples:

continuous_param: {
  "min": 1.0,
  "max": 4.0
}

NOTE: Only float values can be continuous