A powerful new encoding for tree-based combinatorial optimisation problems

We describe a new encoding scheme and associated operators for tree structures on graphs and test it in the context of an evolutionary algorithm applied to the degree-constrained minimum spanning tree problem (DC-MST). The new encoding is relatively simple and easily copes with degree constraints. We compare with three existing encoding schemes, including edge-set, which represents the current state of the art on the DC-MST. The new encoding demonstrates superior performance on the larger instances of the well-used 'Structured Hard' DC-MST problems, and similar performance on the smaller instances. We conclude that the new encoding is a recommended method for the DC-MST.