Title of article
Hard multidimensionalmultiplechoiceknapsackproblems,anempiricalstudy
Author/Authors
Bing Han، نويسنده , , JimmyLeblet، نويسنده , , GwendalSimon، نويسنده ,
Issue Information
ماهنامه با شماره پیاپی سال 2010
Pages
10
From page
172
To page
181
Abstract
Recent advancesinalgorithmsforthemultidimensionalmultiplechoiceknapsackproblemshaveenabled
us tosolveratherlargeprobleminstances.However,thesealgorithmsareevaluatedwithverylimited
benchmark instances.Inthisstudy,weproposenewmethodstosystematicallygeneratecomprehensive
benchmark instances.Someinstanceswithspecialcorrelationpropertiesbetweenparametersarefound
to beseveralordersofmagnitudeharderthanthosecurrentlyusedforbenchmarkingthealgorithms.
Experiments onanexistingexactalgorithmandtwogenericsolversshowthatinstanceswhoseweights
are uncorrelatedwiththeprofitsareeasiercomparedwithweaklyorstronglycorrelatedcases.Instances
with classescontainingsimilarsetofprofitsforitemsandwithweightsstronglycorrelatedtotheprofits
are thehardestamongallinstancegroupsinvestigated.Thesehardinstancesdeservefurtherstudyand
understanding theirpropertiesmayshedlighttobetteralgorithms.
Keywords
Multiple choice , Multidimensional , Knapsack problem , Algorithm , Performance evaluation
Journal title
Computers and Operations Research
Serial Year
2010
Journal title
Computers and Operations Research
Record number
927633
Link To Document