Building a better heuristic for the traveling salesman problem: combining edge assembly crossover and partition crossover

A genetic algorithm using Edge Assemble Crossover (EAX) is one of the best heuristic solvers for large instances of the Traveling Salesman Problem. We propose using Partition Crossover to recombine solutions produced by EAX. Partition Crossover is a powerful deterministic recombination that is highly exploitive. When Partition Crossover decomposes two parents into q recombining components, partition crossover returns the best of 2q reachable offspring. If two parents are locally optimal, all of the offspring are also locally optimal in a hyperplane subspace that contains the two parents. One disadvantage of Partition Crossover, however, is that it cannot generate new edges. By contrast, the EAX operator is highly explorative; it not only inherits edges from parents, it also introduces new edges into the population of a genetic algorithm. Using both EAX and Partition Crossover together produces better performance, with improved exploitation and exploration.

[1]  Yuichi Nagata,et al.  Fast EAX Algorithm Considering Population Diversity for Traveling Salesman Problems , 2006, EvoCOP.

[2]  A. Mobius,et al.  COMBINATORIAL OPTIMIZATION BY ITERATIVE PARTIAL TRANSCRIPTION , 1999, cond-mat/9902034.

[3]  Keld Helsgaun,et al.  An effective implementation of the Lin-Kernighan traveling salesman heuristic , 2000, Eur. J. Oper. Res..

[4]  Doug Hains,et al.  Tunneling between optima: partition crossover for the traveling salesman problem , 2009, GECCO.

[5]  Doug Hains,et al.  Improving Lin-Kernighan-Helsgaun with Crossover on Clustered Instances of the TSP , 2012, PPSN.

[6]  Thomas Stützle,et al.  The Impact of Automated Algorithm Configuration on the Scaling Behaviour of State-of-the-Art Inexact TSP Solvers , 2016, LION.

[7]  L. Darrell Whitley,et al.  A New Generalized Partition Crossover for the Traveling Salesman Problem: Tunneling between Local Optima , 2019, Evolutionary Computation.

[8]  L. Darrell Whitley,et al.  Partition Crossover for Pseudo-Boolean Optimization , 2015, FOGA.

[9]  Shigenobu Kobayashi,et al.  Edge Assembly Crossover: A High-Power Genetic Algorithm for the Travelling Salesman Problem , 1997, ICGA.

[10]  Yuichi Nagata,et al.  The EAX Algorithm Considering Diversity Loss , 2004, PPSN.

[11]  L. Darrell Whitley,et al.  The Traveling Salesrep Problem, Edge Assembly Crossover, and 2-opt , 1998, PPSN.

[12]  L. Darrell Whitley,et al.  Generalized asymmetric partition crossover (GAPX) for the asymmetric TSP , 2014, GECCO.

[13]  L. Darrell Whitley,et al.  A New Evaluation Function for Clustering: The NK Internal Validation Criterion , 2016, GECCO.

[14]  Yuichi Nagata.  New EAX Crossover for Large TSP Instances , 2006, PPSN.

[15]  J. Lenstra,et al.  In Pursuit of the Traveling Salesman: Mathematics at the Limits of Computation , 2016 .

[16]  Shigenobu Kobayashi,et al.  A Powerful Genetic Algorithm Using Edge Assembly Crossover for the Traveling Salesman Problem , 2013, INFORMS J. Comput..