On relation classes and solution relations [Elektronische Ressource] / André Große

Andre Grosse

Sujets

• Mathematics

Informations

• Publié par: friedrich-schiller-universitat_jena
• Publié le: 01 janvier 2004
• Nombre de lectures: 31
• Langue: English

Extrait

derOn16.RelationonClassesf?randhiller-UnivSolutionGro?eRelationsinDissertationundzurriedricErlangungJenadeserakorenademiscebruarhenakult?tGradesMathematikdoInformatikctorFrerumh-Scnaturaliumersit?t(Dr.vrer.Diplom-Mathematiknat.)Andr?vgeborgelegtamdemFRat1975derJenaFerteidigung:Gutac2004hdester:?en1.letztenProf.4.Dr.agGerdhenWJuliecPr?funghsungRigorosums:2.JuniProf.TDr.derIngotlicAlth?ferVT1.ag2004derarTpoentsmyp
.chieyknoeen.wledgemenbtsciationFirstmofassistanceallpresenIallwinouldortlik,eIftocreatingthankHempmpapyIadvisorofProfessorlastGerdhaWencouragemenecuchsungtoforyhisloinspir-bingelectureouldonIComplexitgenerousyofTheory]duringFmgratefulyytimeWasmakinganearsundergraduateasstudenetorandandforamhisobligedexcellenparentysupallervisionand,omvgirlfrienderhadtheforpasthayeenears.thesisWithvhisimpinimitablethankvforervermissionetheandjoinen[thChapterusiasmhneider.heurthermore,inamtrotoducedmmefriendsintheAtourzel-AssotheforeldtheofyresearcashterestingandtheywithoutvhisbsuppForttheirneithertmsuppyIDiplom-thesismnorhthetocurrenytts,thesismwfamilyouldtohamvfriends,ecbtoeenypvingossible.Sabine.SpitecialnotthankseentotheallImvybformerreceiving,andthiscurrenwthacolleaguesefromeenJena,ossible.espfurthereciallyHarald,elHaraldhisHemppel,toJ?rgtRothe,resultsStefanourSctherwGH03arz,inand3ThomasSc.Condten.tsIn1.In.tro.duction.1as2ResultsPreliminaries3.55.2.1.W.ords.and.Languages..........Classes.3.6.42...3.8.................................F5.2.2.Tanduring.Macthehines.and.Reductions..........Solution.........i.............3.3......7.2.3.Imp.ortan.t3.4Complexit.y.Classes............erators.and.........Inclusion.Consequences.....Bey.erator..........9.2.3.1enThe.P.olynomial.Hierarc.h.y....50.53.duction................................9.2.3.2.The.Bo.olean.Hierarc17hRelationsyOracles...........................20.General..............11.2.4.Miscellaneous............21.Op.on.unction.Relation...............30.The.Structure.Structural...........3.7.ond.Op.Metho..13.3.F.unction.and.Relation.Classes.15.3.1.In.tro48ductionOp.Problems...............................4.Relations.4.1.tro........................15.3.2.Basic.Denitions..53..Easy8
Easy9
..Sym.........wsol...and.......Bibliograph.......Some.....Op.........Sub.Op.ssol......55.4.2.1..61Languagesecial.Classes.......63.Problems...............List.ols.Index.75.erators.and........................554.44.2.2SpEasywsol4.2ssol.....................4.5.en...............................68.of.b.71.ject.7359y4.3iiTheEasy Easy9 8
.The.P.olynomial.Hierarc.hFy4.1...Classes...3.2.Classes...of.....The.........Structure.and.I.....Classes.List......10.2.260TheandBo.olean.Hierarc.h.y....iii.43.Inclusion.of.unction.Relation.I.........44.The.Figures2.1and........................124.23.1wsolInclusionssolStructure.of.F.unction.and.Relation.Classes.I..........69..x
x
x p(jxj)
p
uswithytheEdm65classicationaofproblem.problemsinwithproblemrespappropriateectthistoossible.theiracomplexitcarriesymatc.theTeroupgivrunningebahardersenseclasstosetsthismacstatemenmactinwtainede],haprimevsetsehine.tooutexplainhine,someconguration.partswithofsolvingit.onlyIfcomplexitweresometimesreferwtobproblemstheweebusuallyformeanandecisionevproblems.InconcernedDoasestrivialandinggivgeneralenKha79obanjectFbtheelongsetooundeda1certainstepssetyoruringnot?theTtheoallosolvclassifyetosucofhgorithm.problemswmeansuppthatforwofelohaoundsvdetermineeisananalgorithmconsiderdecidingrstlywhetherEdmondsaThisgivofencaninputbypbTelongsThattoerythetheresetTorthatnot.inputThisatgivChapteresTheusareaproblems.pandossibilitareyP,tomaximmeasureinho[wprogrammingdicultandsuctestinghtegeraAKS02decisionthereproblemNPis,ofbutcanonlybwithprespuringectproblemstobtheofalgorithmcarriedused.bWtheeTmeasuremachofromwinputmantoynalresourcesThisthewsalgorithmtoneedsproblemsforregarditsthedecision,timedepaendingal-onInthemanner,lengtheofgettheerinputoundscomplexitthe.yTaoOptimalacwhievbearethistoinandathisreasonableimpwAsaexample,ye,theaP,computationaldenedmoydel[is].necessaryis.classInall1936thatTburingdecided[yTdeterministicur36olynomial-time-]oundeddevuringelophine.edmeans,aevunivsetersalPcomputationalexistsmoappropriatedel,uringthehineso-calledforTeryuringductionComputationaloutismosttroy1forpmacmanhine.steps,Wsomeeolynomialdistinguish.tsetswPoconsideredvfeasibleersions,Mananaturaldeterministicnonandproblemsaconnondeterministicinone.includingWhicahumresourceshingareausuallygraphconsidered?Edm65Onelinearp[ossibilit]ytheisoftowhethermeasureinhoisw[m].ucurthermore,hexiststimeclassanalgorithmclassneeds.allFthatorbthisacceptedpurpyosenondeterministicwolynomial-time-beTcounmactAlltheinnum=
=
=
⁄† r2 ¢C () (9B2C)(9p2 )(8x2Σ )
⁄[r(x)=fy2Σ :jyj•p(jxj) ^ hx;yi2Bg]
⁄† f 2 ¢C () f 2 ¢C ^ (8x2Σ )[jjf(x)jj• 1]
InonlysponecrucialofPthese,hardThisproblemstransformisP.inofP,follothenclassesittuitivfollowingwsinthatP.Pthesense:uringNP.olynomial-time-bOnegivsucHW00hhproblemofisesthetoTspreadrathevclasselingthatSalessucPguesserson,Problem:o,ApsalesNPMVpyersonmacwdenitionsan3tsec00toclassesvisitofathencomplexitumTbclassesertoofalsocities.vIswhictherethea],routeandshorterythanaaproblemsgivtheenylength?lotThetainsquestionywhetherthatPfunwingwNParehasaboundedeenhinethethestartingcomputablepnondeterministicoinTt[of].longbandininWtensiv[eandresearctoh.relations.Thisoinresearcsystematichtogaofvweclassesrisecomputationtomacahlotrelationsofonlynewandquestions.notations.ManwsyvothergeneralclassescasesofareproblemsvthanFPWandeNPeratorswwhicerecom-observtoedrelationalloofwingofaaredeepertunderstandingvofthemthisproblems.area.candidatesTherearewsetsereNPmanwyallattemptsnottoinanswareerquestiontheevPcourse.folloNP,NPductionquestion,Chapterbutdeterministicthisolynomial-time-bproblemThasmacbandeenunsolvclassedrelationsbbyatopdaoundedyuring.hineBesideBLS84decisionBLS85problems,ExactrelationswillplaeyenanChapterimp.ortanetwroleWin]computational[complexit]ydene.ofNotTheonlypdottheythisappapproacearisasbasetodenitionolsrelationbutonalsoell-studiedasyobinsteadjectstheofofresearcuringhhines.themselvapproaces.toTheofcomplexitdoynotofleadrelationsnaturalhasinreceiveedItmalloucushproattenetioneryintheorems,theeciallastofdecade.hThiswidelydevoelopmenertliterature.wolloas[essenec00tiallywinuenceddenebopyrelSelmanfuninhtheaearlyplexitninetiesclass(seea[ofSel94or,classSel96functions:]).NP,therelIfhardesttheFirsttheyeyvpropsomeeresults.haooleofexample:ManwhwnforjectionofcarriesavTheretothanofmoreWconwillthatwouldevtheoreticians1complexithttp://www.claymath.org/Millennium_Prize_ProblemsNearlyndinhoaretoNPthissets.thereclassesis:of1relations:$1,000,000FPer,reltheHoclassofoftoallinfunctionsPcom-troputable1.b2yMan.ywdierenproteclassesgeneralofTrelationsgivandanTheasell-knoaprosptheoremecialocaseerclassesclassesrelations.ofefunctionsshowthatereenstudied.SeeTtoooutmenwtionearntmoneywo