Batch Simulation

After modeling the system, we can solve it using . The Batch Simulation Module is shown in Figure .

Figure: The Batch Simulation Module

The input parameters are :

• Simulation File Name: this file will store the result of all rewards specified in the model.
• Runs: total number of simulation runs.
• : check it to distribute the simulation runs through a network of workstations.
• : the default value is 95%.
• Stop Condition: the simulation of the model will be interrupted in the following cases :
  • Time - Total time of each run, when only this parameter is specified.
  • Number of Transitions - If only this parameter is specified, each run stops when the specified number of transitions is reached.
  • Only Events - When only this parameter is specified, each runs stops when the specified event triggers the given number of times.
  • (Time and Transition) or (Time and Event) or (Time and State) - When the user specifies one of these pairs, the simulation will be interrupted when one of the conditions is satisfied.
  • State and Time - This stop condition uses a state variable of an object. The user should specify a lower and a upper bound for the state variable and the simulation stops when one of the bounds is reached. NOTE: if the ``state'' option is used, the user must also specify the ``time'' stop condition. This is a safety feature, to force the simulation to halt in case neither the lower nor the upper state variable bounds are reached.
• Reward Option - Press button Options to open the

  Figure: Reward Options Window

  
  • Reward Behavior. This option is used whenever there are two or more conditions/values associated with the same reward in the model. If two or more different values are associated with the same reward at a state, only one reward value can be associated to that state. The user can define the following values to resolve the multiple assignment:
    • Mean - The reward value to be assigned to the state is the arithmetic mean of all values associated with true conditions.
    • Maximum - The reward value is the maximum of all possible values.
    • Minimum - The reward value is the minimum of all possible values.
    • Error - In this case, an error message is given when more than one condition is true for a specific state. This option is useful to catch model specification errors if the user has no intention to allow multiple reward values to be assigned to a state.
  • Reward Trace Files. Click Trace to generate traces from your rewards. The first column of the file is the simulation time and the second is the reward value. Reward values are only written into the trace file when the reward changes its value. NOTE: Use it with caution because you may generate huge files, refer to the FAQ for knowing how to compress these files.
  • Reward Levels Specification. Set max values if you want to know how much time the reward was above that level.