Modular semantics for model-oriented design

Modern systems engineering mandates the integration of heterogeneous models in design and analysis. This has given rise to the notion of model-oriented design where specifications can be defined, translated, and composed. A common aspect that model-centered tools and languages share is the capability of using different models of computation together. As a result, the heterogeneous components of a particular system can be expressed in their most natural representation.; We propose a framework that supports the representation of a variety of computational models. An important part to any representation is the provision of a formal semantics that defines its correctness. We accordingly define a precise and modular semantics that uses the notion of institutions to provide meaning to well-formed syntactic elements. Institutions relate specifications to mathematical models such as algebras and coalgebras. The formal semantics thus defined allows us to derive consistency of designs and to reason about system specifications. These specifications are written in the Rosetta language. Rosetta supports describing the ontology of a formalism or model of computation in specifications known as domains.; A particular representation of a computational model depends on the chosen unifying semantic domain. We identify some unifying semantic domains to be static, state-based and trace-based. Each model of computation uses the vocabulary and semantics associated with these three semantic domains. We express the latter in Rosetta and then define several computational models by using the notion of extension of specifications. Specification relations such as translation and composition are also defined. A translation is a mechanism that relates specifications that use different paradigms. Composition involves constructing a specification from available, already defined specifications. When combined, translation and composition allow the reuse of specifications, the analysis of interaction between different heterogeneous views or components, and the specification of complex systems. We demonstrate this new heterogeneous design methodology with the specification of a system that includes both discrete and continuous dynamics. We also look at integrating different views, some functional and others non-functional, into a complete system.