On the structure of continuous simulation languages

In his comments on the article by Birta,2 Francis Cellier raises an interesting general point on the structure of CSSL’s. The particular problem which was addressed by Birta and Cellier was concerned with the embedding of a CSSL model of a system within a parameter optimisation package. In studies of this kind it is generally necessary for the model to be called from the optimisation routine. In effect the optimisation routine invokes one or more runs of the model so as to generate a current value of the performance criterion. As Birta and Cellier point out, most current CSSL’s are not well adapted to this mode of operation.