Higher-Order Non-Causal Modelling and Simulation of Structurally Dynamic Systems

This paper explores a novel approach to the implementation of non-causal modelling and simulation languages supporting highly structurally dynamic systems. One reason the support for structural dynamics is limited in present mainstream non-causal modelling and simulation languages is that they are designed and implemented on the assumption that symbolic processing of models and ultimately compilation of simulation code takes place prior to simulation. We seek to lift that restriction, without sacrificing efficiency, by exploiting just-in-time (JIT) compilation to allow new simulation code, reflecting structural changes, to be generated as the simulation progresses. Our work is carried out in a framework called Functional Hybrid Modelling that supports higher-order modelling, as higher-order modelling lends itself naturally to expressing structural dynamism. However, the central ideas of the paper should be of general interest in the area of structural dynamism. The paper provides an in-depth description of the implementation techniques we have developed as well as a performance evaluation.