Some novel locality results for the blob code spanning tree representation
Proceedings of the 9th annual conference on Genetic and evolutionary computation
SEAL '08 Proceedings of the 7th International Conference on Simulated Evolution and Learning
Evolutionary design of oriented-tree networks using Cayley-type encodings
Information Sciences: an International Journal
On the bias and performance of the edge-set encoding
IEEE Transactions on Evolutionary Computation
CEC'09 Proceedings of the Eleventh conference on Congress on Evolutionary Computation
The property analysis of evolutionary algorithms applied to spanning tree problems
Applied Intelligence
Evolutionary optimization of service times in interactive voice response systems
IEEE Transactions on Evolutionary Computation
Evolutionary optimization of electric power distribution using the dandelion code
Journal of Electrical and Computer Engineering - Special issue on Applications of Heuristics and Metaheuristics in Power Systems
Hi-index | 0.00 |
There are many ways to represent spanning trees in genetic algorithms (GAs). Among them are Cayley codes, which represent each tree on n vertices as a string of (n-2) integers from the set [1,n]. In 2003, Thompson showed that the Dandelion Code, a Cayley code with high locality, offers consistently better performance in a GA than all other known Cayley codes, including the Prüfer Code and the Blob Code. In this paper, we study the Dandelion Code and its properties. We give linear-time implementations of the decoding and encoding algorithms, and prove that the representation has bounded locality and asymptotically optimal locality, unlike all other known Cayley codes. We then modify the Dandelion Code to create bijective spanning tree representations for graph topologies other than the complete graph. Two variations are described: the bipartite Dandelion Code (for encoding the spanning trees of a complete bipartite graph) and the Rainbow Code (for encoding the spanning trees of a complete layered graph). Both variations inherit the Dandelion Code's desirable properties, and have the potential to outperform existing GA representations for computationally hard transportation problems (including the Fixed Charge Transportation Problem) and multistage transportation problems, particularly on large instances.