Random Distributions

Parent Previous Next

Simulation of process models relies on random distributions also called stochastic distributions.

Why? Not all potential influences on a system can be detected and modeled deterministically. E.g. the execution time of an operation exported by a human being is always being subject to fluctuations. Typical other examples are order sizes, processing times or success probabilities. Nevertheless, the realistic variability of such characteristic factors can be indicated with help of pseudo-random-numbers, which are generated by stochastic distributions.

However, the use of stochastic requires a sufficiently large number of simulation experiments. Thus, only the aggregation of results of several experiments and/or very long running experiments will provide reliable information about a system.

Note: The generated 'random-numbers' are so-called pseudo-random-numbers, because micro-processors work deterministically. No micro-processor and no software will ever be able to produce 'real' random-number like a dice does. That's why random-number-generators need seeds for initializing, and as a consequence we are luckily able to reproduce any set of random-numbers. A dice never could do that.