This work is part of our ongoing research activity [31], [32] on formal modeling self-adaptive systems having a decentralized MAPE-K control loop architecture. As in [36], with decentralization, we refer to how control decisions in a self-adaptive software system are coordinated among different components, regardless of how those control components are physically distributed. Specifically, we consider decentralization at the level of the four computations of a MAPE-K loop.