• Aucun résultat trouvé

Application of the Nested Rollout Policy Adaptation Algorithm to the Traveling Salesman Problem with Time Windows

N/A
N/A
Protected

Academic year: 2021

Partager "Application of the Nested Rollout Policy Adaptation Algorithm to the Traveling Salesman Problem with Time Windows"

Copied!
14
0
0

Texte intégral

Loading

Figure

Fig. 1. This figure explains three steps of a level 1 search. At each step of the playout of level 1 shown here with a bold line, an NMC of level 1 performs a playout (shown with wavy lines) for each available decision and selects the best one.
Fig. 2. Score as a function of T . Average on 30 runs. Plateaus are reached for the first three levels, and increasing the level of the algorithm improves the results.
Fig. 3. Comparison between the NRPA algorithm and the NRPA EK algorithm of the problem rc204.3
Table 1. Results on all problems from the set from Potvin and Bengio [11]. First Column is the problem, second column the number of nodes, third column the best score found in [9], fourth column the best score found by the NMC algorithm with heuristics fro

Références

Documents relatifs

A branch-and-cut algorithm for the generalized traveling salesman problem with time windows.. Yuan Yuan, Diego Cattaruzza, Maxime Ogier,

Other variants of the maximum traveling salesman problem that have been considered are among others: the maximum symmetric traveling salesman problem (MAX TSP), in which the

This hierarchical objective function, called Solomon’s objective in the sequel, is unusual in vehicle routing: for instance, most articles on the VRP minimize the total distance

A NEW HEURISTIC FOR THE TRAVELING SALESMAN PROBLEM 251 Though simulated annealing [4] gives good results, it seems that the best heuristic for the symmetrie TSP is the one of Lin

The nonexistence of a tour with exactly two endpoints at distance 2 not connected by the tour can be shown by a similar case analysis as in the proof of Lemma 3.6 (c). The

We first prove that the minimum and maximum traveling salesman prob- lems, their metric versions as well as some versions defined on parameterized triangle inequalities

[12] Tristan Cazenave and Fabien Teytaud, ‘Application of the nested rollout policy adaptation algorithm to the traveling salesman problem with time windows’, in Learning

We suppose that we know that the solutions of the planar isoperimetric problem are circles. We consider a very old variant of this problem called Dido’s problem [18]. It is related