Ensuring Identical Stochastic Behavior in Separate Models

Issue

In cases where two different modeling approaches are being compared it may be desirable to ensure both models use the same stochastic inputs.

Resolution

If two independent models have identical stochastic inputs it is possible to synchronize stochastic behavior by synchronizing the random number seeds and number of realizations in both models. In addition to the random number seed specified in the Monte Carlo settings every element has its own random seed. In order to make the stochastic behavior identical in the two models you need to ensure that the seed for corresponding randomly-behaving (e.g. Stochastic, Event Generator, Delay) elements in both models is also identical.

It's not possible to put both models into separate containers in the same super-model, because GoldSim does not allow elements in the same model to have duplicate random numbers seeds. So the random number seeds of corresponding random elements in both models must be checked. This can be done by clicking a randomly-behaving element in the graphics pane and then pressing CTRL+ALT+SHIFT+F12. If the random number seeds don’t match you can either manually edit one element to make them identical, or you can delete one element and copy and paste in the equivalent element from the other model.

The first time that you run a model where you have modified the element seeds check the run log. If you accidentally gave two elements identical seeds you will see a warning message, and one will have been modified.