This article was accepted into the corpus but its outbound wikilinks were never NER-processed — typical at the deepest BFS hop or when the run's entity cap was reached. No expansion funnel to show.
| Lin–Kernighan | |
|---|---|
| Name | Lin–Kernighan algorithm |
| Authors | Charles Lin; Brian W. Kernighan |
| Introduced | 1973 |
| Domain | Combinatorial optimization |
| Problem | Traveling Salesman Problem |
| Classification | Heuristic; Local search; Variable k-opt |
Lin–Kernighan
The Lin–Kernighan algorithm is a heuristic for the Traveling Salesman Problem devised by Charles Lin and Brian W. Kernighan in 1973. It refines solutions using variable-depth local moves inspired by the k-opt family and is influential in the development of heuristics such as Chained Lin–Kernighan, Kernighan–Lin algorithm, and metaheuristics used in competitions like the DIMACS Implementation Challenges. The method has been analyzed and extended by researchers affiliated with institutions such as Bell Labs, Princeton University, Carnegie Mellon University, and Massachusetts Institute of Technology.
The algorithm generalizes the 2-opt and 3-opt ideas of edge exchange to a variable k-opt procedure, selectively performing sequences of edge deletions and insertions to reduce tour length. It explores moves guided by nearest-neighbor structures popularized in work at Stanford University and University of Waterloo, using candidate lists akin to those in Lin and Kernighan (1973), Kirkpatrick, and later enhancements by David S. Johnson and William J. Cook. The search employs gain criteria reminiscent of procedures in Simulated Annealing research led by Scott Kirkpatrick and Gene Weiner, and it often uses data structures introduced in studies at Bell Labs and IBM Research.
Practical implementations rely on adjacency and candidate lists, edge-weight matrices, and fast move-evaluation routines developed in projects at AT&T Research and NEC Corporation. Efficient variants use priority queues and nearest-neighbor heuristics inspired by work from Jon Bentley and Michael Held, and performance tuning draws on algorithm engineering from Donald Knuth and Robert Sedgewick. Implementations often integrate with solvers like those from Concorde TSP Solver teams and leverage programming practices standardized by communities around GNU Project and ACM SIGPLAN.
Numerous variants extend the original scheme: Chained Lin–Kernighan, Lin–Kernighan–Helsgaun (LKH) by Keld Helsgaun, and stochastic adaptations influenced by Genetic Algorithms research from John Holland and memetic frameworks tied to work by Panos M. Pardalos. Hybridizations combine Lin–Kernighan moves with tabu search elements explored by Fred Glover and path-relinking concepts investigated by Franco M. Serafini. Parallel and distributed versions were proposed by researchers at University of California, Berkeley, ETH Zurich, and University of Tokyo.
Analytical and empirical studies by David S. Johnson, László Lovász, and Éva Tardos have compared Lin–Kernighan to exact methods like the Branch and Bound framework employed by Cook, Rohe, and Reinelt and approximation schemes discussed by Vazirani. Benchmarking on instances from the TSPLIB library and on challenge datasets used by DIMACS shows Lin–Kernighan and its descendants frequently produce near-optimal tours for large instances arising in studies at Los Alamos National Laboratory and industrial research at Siemens. Complexity analyses reference worst-case scenarios examined by theorists at Princeton University and probabilistic analyses from Stanford University.
The algorithm and its variants have been applied in logistics and routing problems encountered by FedEx, DHL, and UPS studies, in circuit design challenges addressed at Intel and Qualcomm, and in bioinformatics sequencing problems researched at National Institutes of Health and European Molecular Biology Laboratory. It also underpins route optimization in projects by NASA for mission planning and has influenced scheduling systems in financial firms such as Goldman Sachs and Morgan Stanley that require combinatorial optimization.
Lin and Kernighan published the original paper following research at Bell Telephone Laboratories and amidst contemporaneous advances by H. Christofides and Jack Edmonds on combinatorial optimization. The algorithm’s evolution was shaped by later contributions from Keld Helsgaun, David Johnson, and industrial teams at AT&T and IBM Research, and it became a standard reference in textbooks by Cormen, Leiserson, Rivest, and Stein and in surveys authored by G. Reinelt and M. Held. Its legacy endures in algorithm engineering curricula at Massachusetts Institute of Technology and in competitions organized by ACM and SIAM.
Category:Heuristics