Figure 73: Kinds of simulation

 

see also:

http://www.informatica-didactica.de/InformaticaDidactica/Thomas2003#Boo99

 

 

In the 1960s researchers used to distingush between analog, digital and hybrid computers used for simulation (e. g. John McLeod, 1968, 5-12)

 

In 1969 Geoffrey Gordon in his book „System Simulation“ (1969, 18-19) distinguishes a)  continuous from discrete simulations (specific 29ff; 123ff) and b) three types of system studies:

·        „Systems analysis aims to understand how an existing system or a proposed system operates.

·        In system design studies, the object is to produce a system that meets some specifications.

·        System postulation is characteristic of the way simulation is employed in social, economic, political, and medical studies, where the behavior of the system is known but the processes that produce the behavior are not.“

 

In 1975 Robert E Shannon (1972, 2) saw three aims of simulation:

1.     describe the behavior of systems;

2.     construct theories or hypotheses that account for the observed behavior;

3.     use these theories to predict future behavior, that ist, the effects that will be produced by changes in the system or in its method of operation.

 

In 1992 Hartmut Bossel distinguishes two possibilities to imitate the behavior of dynamic systems by simulation:

·        by a model which shows the "same" behavior. This leads to the description of historical behavior in the run of time.

·        by a model of the same structure as the system. This requires knowledge of the causal structure and permits the explanation of the behavior as well as the prediction of other behavior.

The describing model requires vast and precise data from behavioral observation. The explanatory model requires a thorough knowledge of the structure and function of the system. Thererefore two different ways of model development are necessary. They result in different formulations of the model.

In spite of these fundamental differences purely explanatory models are hardly to be found. They mostly rely on descriptions of behavior. But they need less data, whilst the „effort to understand“ is much higher.

The "correctness" of such models cannot be proved. Models can be only checked. If a model is wrong we can only say that there is a gap between reality and simulation. Positively said, the model is only "valid" for a certain purpose.

 

 

Bibliography

Modelling & Simulation