Changes between Version 3 and Version 4 of Documentation/Reference/Robust Taboo Search


Ignore:
Timestamp:
07/25/11 11:02:55 (8 years ago)
Author:
abeham
Comment:

updated RTS page

Legend:

Unmodified
Added
Removed
Modified
  • Documentation/Reference/Robust Taboo Search

    v3 v4  
    22= Robust Taboo Search =
    33== Introduction ==
    4 This algorithm was first described in Taillard, E. 1991. Robust Taboo Search for the Quadratic Assignment Problem. //Parallel Computing// 17, pp. 443-455. It is among the most cited algorithms to solve instances of the quadratic assignment problem (QAP).
     4This algorithm was first described in [#Tai91 [Tai91]]. It is among the most cited algorithms to solve instances of the quadratic assignment problem (QAP).
    55
    66There are two main differences to a "standard" tabu search:
     
    1919[[Image(aspiration_default.png)]]
    2020Fig. 1
     21
     22== Parameters ==
     23||= Parameter        =||= Description =||
     24|| !AlternativeAspirationTenure || The time t that a move will be remembered for the alternative aspiration condition. ||
     25|| Analyzer           || The analyzers that are applied after each iteration. ||
     26|| !MaximumIterations || The number of iterations that the algorithm should run. ||
     27|| !MaximumTabuTenure || The maximum tabu tenure. ||
     28|| !MinimumTabuTenure || The minimum tabu tenure (unused if !UseNewTabuTenureAdaptionScheme is set to true) ||
     29|| Seed                || The random seed used to initialize the new pseudo random number generator. ||
     30|| !SetSeedRandomly    || True if the random seed should be set to a random value, otherwise false. ||
     31|| !TerminateOnOptimalSolution || True when the algorithm should stop if it reached a quality equal or smaller to the !BestKnownQuality. ||
     32|| !UseAlternativeAspiration || True if the alternative aspiration condition should be used that takes moves that have not been made for some time above others. ||
     33|| !UseNewTabuTenureAdaptionScheme || In an updated version of his implementation, Eric Taillard introduced a different way to change the tabu tenure. Instead of setting it uniformly between min and max, it will be set between 0 and max according to a right-skewed distribution. Set this option to false if you want to optimize using the earlier 1991 version, and set to true if you want to optimize using the newer version. Please note that the !MinimumTabuTenure parameter has no effect in the new version. ||
     34
     35== References ==
     36[=#Tai91 [Tai91]] Taillard, E. 1991. Robust Taboo Search for the Quadratic Assignment Problem. //Parallel Computing// 17, pp. 443-455.