B.W. Repetto. Upper and Lower Bounding Procedures for the Asymmetric Traveling Salesman Problem. PhD thesis, Graduate School of Industrial Administration, Carnegie-Mellon University, 1994.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....Cook, described in [2] Experimental Analysis of Heuristics for the ATSP 3 Finally, the experimental study of heuristics for the ATSP is much less advanced. Until recently there has been no broad based study covering a full range of ATSP heuristics and what ATSP studies there have been, such as [13, 30, 31, 32], have concentrated mostly on the TSPLIB instances and randomly generated instances with no obvious connection to applications. Nor has there been a formal ATSP Challenge like the STSP Challenge that served as the main resource for Chapter 9. Fortunately, a first attempt at a more comprehensive ....

B.W. Repetto. Upper and Lower Bounding Procedures for the Asymmetric Traveling Salesman Problem. PhD thesis, Graduate School of Industrial Administration, Carnegie-Mellon University, 1994.

....500 node problems using the genlarge problem generator, which Bruno Repetto [18] kindly gave to us. Repetto solved these problems with an open ended heuristic approach, where he would produce an initial tour by randomly choosing one of the nearest two neighbors, and then apply his implementation [19] of the Kanellakis Papadimitriou heuristic [12] an adaptation to asymmetric TSP s of the Lin Kernighan heuristic for the symmetric TSP [13] This process would continue for 1200 seconds and the best solution found would be returned. The tours generated were used as initial tours for our ....

B. Repetto, Upper and Lower Bounding Procedures for the Asymmetric Traveling Salesman Problem. Ph.D. Thesis, GSIA, Carnegie Mellon University, April 1994.

....cover the current best approaches. This is unfortunate, as a wide variety of ATSP applications arise in practice. In this paper we attempt to begin a more comprehensive study of the ATSP. The few previous studies of the ATSP that have made an attempt at covering multiple algorithms [Rep94,Zha93,Zha00,GGYZ] have had several drawbacks. First, the classes of test instances studied have not been well motivated in comparison to those studied in the case of the symmetric TSP. For the latter problem the standard testbeds are instances from TSPLIB [Rei91] and randomlygenerated two dimensional point sets, ....

.... below the Held Karp bound (and hence at least that far below the optimal tour length) The Held Karp bound is also a significantly more reproducible and meaningful standard of comparison than the best tour length so far seen, the standard used for example in the (otherwise well done) study of [Rep94]. We computed Held Karp bounds for our instances by performing the NP completeness transformation from the ATSP to the STSP and then applying the publicly available Concorde code of [ABCC98] which has options for computing the Held Karp bound as well as the optimal solution. Where feasible, we ....

[Article contains additional citation context not shown here]

B. W. Repetto. Upper and Lower Bounding Procedures for the Asymmetric Traveling Salesman Problem. PhD thesis, Graduate School of Industrial Administration, Carnegie-Mellon University, 1994.

....cover the current best approaches. This is unfortunate, as a wide variety of asymmetric applications arise in practice. In this paper we attempt to begin a more comprehensive study of ATSP algorithms. The few previous studies of the ATSP that have made an attempt at covering multiple algorithms [Rep94, Zha93, Zha00] have had several drawbacks. First, the classes of test instances studied have not been well motivated in comparison to those studied in the case of the symmetric TSP. For the latter problem the standard testbeds are instances from TSPLIB [Rei91] and randomly generated two dimensional point sets, ....

.... below the Held Karp bound (and hence at least that far below the optimal tour length) The Held Karp bound is also a significantly more reproducible and meaningful standard of comparison than the best tour length so far seen, the standard used for example in the (otherwise well done) study of [Rep94]. We computed Held Karp bounds for our instances by performing the NP completeness transformation from the ATSP to the STSP and then applying the publicly available Concorde code of [ABCC98] which has options for computing the Held Karp bound as well as the optimal solution. Where feasible, we ....

[Article contains additional citation context not shown here]

B. W. Repetto. Upper and Lower Bounding Procedures for the Asymmetric Traveling Salesman Problem. PhD thesis, Graduate School of Industrial Administration, Carnegie-Mellon University, 1994.

No context found.

N. B. Repetto, Upper and Lower Bounding Procedures for the Asymmetric Traveling Salesman Problem, Ph.D. Thesis, GSIA, Carnegie-Mellon University, Pittsburgh, 1994.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC