Bernard P. Zeigler, Keynote at SummerSim ’17, RTSync Corp, USA
The notions of state and discrete event were allied abstractions onceived to characterize models underlying the operations research techniques that emerged in the mid-1950s. At that time the limitations of queueing analysis methods for supporting industrial design became evident and led to new abstractions, first spurring machine language attempts to handle specific complex systems, then eventually to general purpose simulation modeling tools. Innovator K.D. Tocher’s core idea was that of a system of individual components progressing through ‘states’ that change only at discrete ‘events’ as time unfolds.
Later, the Discrete Event Systems Specification (DEVS) formalism took the discrete event abstraction one step further using the set theory of logicians and mathematicians. This talk starts with the proposition that both abstraction and its concrete realization are essential for progress – abstractions enable uncluttered thought while concrete realizations are closer to reality but necessarily messy. Our theme riffs on the progression: new abstractions lead to new model representations which in turn drive requirements for new computational environments to exercise and explore these models. When the new abstractions can be related to existing ones, the process of integrating new models and extending legacy simulation systems can be smoother encountering less resistance to change. To make these points we recount some history of DEVS and its progress from a research abstraction to industrial strength realizations supported by high-performance simulation technology.
We indicate how progress continues with deeper theory and new abstractions that can enable modeling complex systems of systems from neuron networks to aviation and cyber-systems.