Abstract for: Hierarchical Behavioral Decomposition of Complex Problems
Building, testing, and understanding large models can be a challenge. Modules break a model into several submodels that can be developed and tested independently, facilitating hierarchical decomposition, a process of dividing a large problem into smaller parts starting from the highest conceptual level and working down to the concrete level. At each level, a problem can be divided either i) structurally into logical structural groups or ii) behaviorally into groups with identifiable behavior. Behavioral decomposition extends the standard modeling process to several hierarchical levels, each of which has reference modes, dynamic hypotheses, and models. Benefits of this approach extend to all models, not just large ones, and include: (i) the model and its behavior are simpler to understand as the model is presented starting from the high-level view followed by details from the lower levels, (ii) the model can be developed and tested in parallel using separate model files for each piece, (iii) the model can be also be developed from the bottom up making it possible to start with a smaller scope (narrower model boundary) and extend the model as needed, and (iv) standard behavioral components can be built and reused across many modeling projects.