Distributed Learning Algorithms and Lossless Convex Relaxation for Economic Dispatch with Transmission Losses and Capacity Limits

Figure 3

A schematic diagram for decentralized iterative computations and message passing. The aggregator (Node ) broadcasts the values of and , and each generator (Node ) sends a private message to the aggregator, where implies the th generator decides to produce its lower limit, i.e., , implies the th generator decides to produce its upper limit, i.e., , and implies the th generator wants to continue negotiation by receiving new values of , , and from the aggregator.