CLEM - Combining Live and Exploratory Modeling

Behavioral models are widely used in science and engineering for representing different kinds of systems, such as software-intensive, cyber-physical, and economic systems. Such models usually need to be first designed, then implemented, and finally executed before results can be observed to answer particular questions about the systems. Each time the model is changed to reflect evolving requirements, alternative scenarios or potential improvements, the entire process must be restarted. The initial model state must be prepared and the full previous execution history must be replayed to see the effects on the results.

Live modeling challenges the strict design-first/execute-later approach by enabling users to seamlessly transition between design and execution phases. This makes it possible to render the effects of a design change visible directly on an already running execution without requiring a full restart and replay of the execution history. Furthermore, combining live modeling with exploratory modeling enables the direct exploration of a model’s design space as well as its runtime states. This allows design alternatives to be compared in experimental settings and accelerates the exploration of the design space.

Although preliminary studies of both live and exploratory modeling have been conducted, there is still no systematic modeling language engineering approach for designing and executing behavioral models with the support of such facilities. Consequently, these facilities must be redeveloped for each new modeling language. Furthermore, to the best of our knowledge, there is currently no guiding framework for the sound combination of live and exploratory modeling. This project will develop a tool-supported methodology to enable live and exploratory modeling. In particular, scenarios such as the interactive use of live modeling in the context of the exploration and the efficient comparison of several alternative model designs will be studied for the first time. This sound combination of live and exploratory modeling will provide the necessary support for design space exploration, trade-off analysis, and decision making.

The integration of several advanced modeling techniques is required for the efficient replay of model interaction and effective design model synthesis to realize immediate feedback and direct manipulation for behavioral models. We will therefore develop a novel architecture for model execution engines that offers interactive model execution. This architecture will also support the necessary tool interfaces and language protocols for integrating interactive model exploration tools. Moreover, we will establish language design principles to enable these capabilities for any modeling language.