In addition to the simulation control parameters there are a number of variables with special meaning. These variables do not have to appear in a model, but when they do Vensim will reference them for certain actions.
Normally simulations are performed as quickly as possible. You can, however, cause the speed with which a simulation occurs to be slowed by adding the variable SIMULATION PAUSE to the model. This gives the number of seconds that need to elapse between each successive SAVEPER in the model. For example if SIMULATION PAUSE was .5 and SAVEPER was .5 Months then for each month of simulated time 1 second of time would pass. If the computations themselves take longer than SIMULATIONPAUSE then no pause will occur, otherwise computation will be paused for the difference between the specified time and the time the computations took. SIMULATION PAUSE is ignored in SyntheSim and during optimization and sensitivity simulations.
NOISE SEED is used to change the behavior of the random number generator. See NOISE SEED.
NOISE RNG is used to select the pseudorandom number generator used. See NOISE RNG.
Acceptable Integration Errors
If you use Runge-Kutta integration with automatically adjusted step size, you can specify the acceptable integration error. The acceptable error is used to determine whether decreasing the step size is necessary to achieve the desired integration accuracy. You can specify the absolute and relative tolerances.
See Simulating Models for more discussion of the integration techniques.
Use ABSOLUTE TOLERANCE to specify that amount of error in all levels that is acceptable to confirm convergence of the variable step size Runge Kutta integration techniques. If this is not included it is the same as giving it a value of 0.001. If the error in a Level exceeds ABSOLUTE TOLERANCE it is checked to see if it also exceeds RELATIVE TOLERANCE.
Use RELATIVE TOLERANCE to specify that amount of relative error in all levels that is acceptable to confirm convergence of the variable step size Runge Kutta integration techniques. If this is not included it is the same as giving it a value of 0.001. When performing the variable step size integration the step size is decreased until for every Level, the error in its computation is less than ABSOLUTE TOLERANCE or RELATIVE TOLERANCE * Base Level Value.
RC START TIME
RC START TIME is used to determine when the Reality Check functions take effect. The Reality Check functions are described in more detail in Chapter 14 of the User Guide. Normally RC START TIME would be a constant
|RC START TIME = 10|
|~ The time to start Reality Check testing.|
If a model does not include RC START TIME then it is assumed to be INITIAL TIME + TIME STEP.
Use MAX TIMES for models used for gaming where one or more of SAVEPER, INTIAL TIME or FINAL TIME is a variable. Normally Vensim computes the number of times data will be saved as (FINAL TIME-INITIAL TIME)/SAVEPER + 1. If, however, any of these are variable this computation may need to be adjusted. MAX TIMES should be the maximum number of saved values that will occur.
Use GAME INTERVAL to specify how far a model should advance during gaming. This value can be changed later using the Gaming Control Dialog or GAME commands. If it is not included the value of SAVEPER is used.
Normally, time series data inputs and data equations are computed at the start of a simulation, and do not update during iterated simulations (optimization and sensitivity analysis). Defining RECOMPUTE DATA = 1 will force recomputation, at the cost of some extra computation.
This option is provided primarily so that one may iterate over data using VECTOR ELM MAP with a constant input.
The recomputation does not update TIME SHIFT data equations.
Spreadsheets and other input data sources will not be re-read.