Distinguished Carl Adam Petri Lecture – T.A. Henzinger

Promises and Challenges of Reactive Modeling: A Personal Perspective

Reactive models offer a fundamental paradigm for predicting the behavior of highly concurrent event-based systems, which includes all systems with significant software components. While much historical emphasis has been put on the analysis and comparison of different models for concurrency, several additional capabilities of reactive models have come into focus more recently: the heterogeneous combination of computational and analytical models, of worst-case and best-effort techniques; interface languages for the decomposition of a system into multiple viewpoints, in addition to temporal, spatial, and hierarchical structuring mechanisms; reasoning about strategic choice, in addition to non-deterministic and probabilistic choice; computing quantitative fitness measures vis-à-vis boolean requirements, in addition to measures of time and resource consumption; design for robustness properties, in addition to correctness and performance properties; theories of approximation in addition to theories of abstraction; methods for system synthesis in addition to model analysis. We review some results and outline some challenges on these topics.