|
Real-time simulation: better suited than discrete-event simulations for short term ecological models
Discrete-event simulation seems to be the current preferred mode of simulation. In this kind of simulation, time is artificial; events are rearranged in order of increasing timestamps before the simulation is finally run. The program has total control over time. This has many advantages: the simulation is deterministic, it can be mathematically analyzed, algorithms like time-wrap may be applied (with time-wrap, in case of dead-ends time may be unrolled and events rearranged). But this does not make a lifelike simulation. Real life is not deterministic. It is much richer in open possibilities than the one-way course of the ideal mathematical world [see Wegner, 1997]. In a real-time simulation, time is an elapsed time, with no possibility to run backward. To see why this is more realistic, consider a resource (for example a location in space, or a trophic resource) which can only be accessed by 1 individual at a time. If two or more individuals are competing for this resource, whatever the time granularity, you may always find an interval short enough so that, at a given moment, only 1 of them may get the resource. Depending on the state of the individual, this may have huge consequences. If the individuals in competition have practically the same chance of appropriating the resource, it will be accessed at random; the one which will get it is unpredictable. In real life, concurrency is the rule. A concurrent, real-time simulation thus more faithfully models what happens in the real world.
(Last modified 2003-03-13)
|