算法设计与分析算法设计与分析 (6).pptxVIP

Chapter4

GreedyAlgorithms

SlidesbyKevinWayne.

Copyright©2005Pearson-AddisonWesley.Allrightsreserved.

1

4.5MinimumSpanningTree

Minimumspanningtree.GivenaconnectedgraphG=(V,E)withreal-valuededgeweightsce,anMSTisasubsetoftheedgesTEsuch

thatTisaspanningtreewhosesumofedgeweightsisminimized.

G=(V,E)T,eTce=50

MinimumSpanningTree

11

14

9

6

3

Applications

MSTisfundamentalproblemwithdiverseapplications.

nNetworkdesign.

–telephone,electrical,hydraulic,TVcable,computer,road

nApproximationalgorithmsforNP-hardproblems.

–travelingsalespersonproblem,Steinertree

nClusteranalysis.

4

GreedyAlgorithms

Kruskalsalgorithm.StartwithT=.Consideredgesinascendingorderofcost.InsertedgeeinTunlessdoingsowouldcreateacycle.

Primsalgorithm.StartwithsomerootnodesandgreedilygrowatreeTfromsoutward.Ateachstep,addthecheapestedgeetoTthathas

exactlyoneendpointinT.

Remark.AllthesealgorithmsproduceanMST.

5

CyclesandCuts

Cycle.Setofedgestheforma-b,b-c,c-d,…,y-z,z-a.

Cutset.AcutisasubsetofnodesS.ThecorrespondingcutsetDisthesubsetofedgeswithexactlyoneendpointinS.

CutS={4,5,8}

CutsetD=5-6,5-7,3-4,3-5,7-8

CycleC=1-2,2-3,3-4,4-5,5-6,6-1

4

6

Cycle-CutIntersection

Claim.Acycleandacutsetintersectinanevennumberofedges.

Pf.(bypicture)

7

GreedyAlgorithms

Simplifyingassumption.Alledgecostscearedistinct.

Cutproperty.LetSbeanysubsetofnodes,andletebethemincostedgewithexactlyoneendpointinS.ThentheMSTcontainse.

Cycleproperty.LetCbeanycycle,andletfbethemaxcostedgebelongingtoC.ThentheMSTdoesnotcontainf.

f

fisnotintheMST

eisintheMST

C

S

8

GreedyAlgorithms

Sim