Parallel Independence of Amalgamated Graph Transformations Applied to Model Transformation

The theory of algebraic graph transformation has proven to be a suitable underlying formal framework to reason about the behavior of model transformations. In order to model an arbitrary number of actions at different places in the same model, the concept of amalgamated graph transformation has been proposed. Rule applications of certain regularity are described by a rule scheme which contains multirules modeling elementary actions and a common kernel rule for their synchronization (amalgamation). The amalgamation theorem by Böhm et al. ensures that for two multi-rules, the application of the amalgamated rule yields the same result as two iterative rule applications, respecting their common kernel rule application. In this paper, we propose an extension of the amalgamation theorem to an arbitrary finite number of synchronous rule applications. The theorem is used to show parallel independence of amalgamated graph transformations by analyzing the underlying multi-rules. As example, we specify an excerpt of a model transformation from Business Process Models (BPM) to the Business Process Execution Language (BPEL).