Changes between Version 4 and Version 5 of Documentation/Howto/ImplementANewVRPEvaluator


Ignore:
Timestamp:
01/28/14 15:09:02 (6 years ago)
Author:
pfleck
Comment:

--

Legend:

Unmodified
Added
Removed
Modified
  • Documentation/Howto/ImplementANewVRPEvaluator

    v4 v5  
    286286}}}
    287287
     288= Set the new Evaluator in HeuristicLab =
     289
     290After we finished implementing the insertion infos and the evaluator, compile the plugin and start HeuristicLab.
     291
     2921. Open a VRP sample (e.g. Tabu Search – VRP)
     293   a. Choose a CVPR instance (e.g. Augerat A-n36-k5)
     2942. Show hidden parameters in problem
     295   a. Select the Evaluator parameter and set the value to the new PRPEvaluator
     2963. In the ProblemInstance set following parameter
     297   a. Limit the number of Vehicles (e.g. 5) (depends on problem; compare the number of vehicles for best known quality)
     298   b. Set DistanceFactor to 0 (distance is already considered in energy consumption)
     2994. Set parameter of the algorithm
     300   a. Increase number of Iterations (e.g. 20000)
     3015. Start the algorithm
     302
     303Now the algorithm tries to find routs where the stops with high demands are delivered first to avoid carrying around heavy loads. (Do demonstrate the effect cities with greater demands are displayed bigger.)
     304
     305||= with PRP =||
     306|| [[Image(6-PRP_result.png, 600px)]] ||
     307
     308
     309Compared to the standard CVRP evaluator, the solution for PRP can differentiate significantly.
     310
     311||= with CVRP =||
     312|| [[Image(7-CVRP_result.png, 600px)]] ||