Functional abstractions for UML activity diagrams

Abstraction and encapsulation are core principles of modern sensible engineering. Whereas especially encapsulation is coined by the OOP community the principle could be generalized to other kinds of information flow. Not only implementation information has to be encapsulated, but also possible side effecting parts have to be separated from side effect free parts of the program. In this paper we strive to separate non side effecting parts of activity diagrams from the rest of the model. This is specifically important in modeling domains like cryptography or signal processing which use complex, purely mathematical (i.e. that is non side-effecting) transformations. Additionally, we show how to take abstraction a step further by introducing the concept of arrows which leads naturally to the extraction of side effect free parts and results in both, a decent design and cleaner code.