A 0.5358-Approximation for Bandpass-2
Journal of Combinatorial Optimization
The Bandpass-2 problem is a variant of the maximum traveling salesman problem arising from optical communication networks using wavelength-division multiplexing technology, in which the edge weights are dynamic rather than fixed. The previously best approximation algorithm for this NP-hard problem has a worst-case performance ratio of 227/426. Here we present a novel scheme to partition the edge set of a 4-matching into a number of subsets, such that the union of each of them and a given matching is an acyclic 2-matching. Such a partition result takes advantage of a known structural property of the optimal solution, leading to a 70−2√/128≈0.5358-approximation algorithm for the Bandpass-2 problem.
Huang, Liqin, Weitian Tong, Randy Goebel, Tian Liu, Guohui Lin.
"A 0.5358-Approximation for Bandpass-2."
Journal of Combinatorial Optimization, 30 (3): 612-626.