LecturesinAppliedMathematicsVolume00,19xxAnOptimalAlgorithmfortheLocalSolutionofIntegralEquationsKarinFrankAbstract.ThelocalsolutionproblemofmultivariateFredholmintegralequa-tionsisstudied.RecentresearchprovedthatforseveralfunctionclassesthecomplexityofthisproblemiscloselyrelatedtotheGelfandnumbersofsomecharacterizingoperators.ThegeneralizationofthisapproachtothesituationofarbitraryBanachspacesisthesubjectofthepresentpaper.Furthermore,aniterativealgorithmisdescribedwhich{undersomeadditionalconditions{realizestheoptimalerrorrate.Thewaythesegeneraltheoremsworkisdemon-stratedbyapplyingthemtointegralequationsinaSobolevspaceofperiodicfunctionswithdominatingmixedderivativeofvariousorder.Thereby,theex-actrateofinformationcomplexityisderived,andanimplementableoptimalalgorithmisformulated.1.IntroductionFredholmintegralequationsofthesecondkindappearinmanyphysicalapplica-tions,e.g.inboundaryvalueproblems.Localsolutionofintegralequationsmeansthatinsteadofthesolutionfunctiononthewholedomainonlythevalueofasinglelinearfunctionalappliedtoitistobecomputed.Thismaybee.g.thevalueinasinglepoint,somesinglebasiscoe cientofthesolutionfunctionoraweightedmean.Usually,innumericalapplicationswedonothavefullinformationaboutboththekernelfunctionandtherighthandside,butratherpartialinformationsuchaspointvaluesorasetofbasiscoe cients.Information{basedcomplexitytheoryconsidersthissituation,studyingthequestionhowmanyinformationandoperationsareatleastrequiredto ndanapproximationtothesolutionwithanerrorofatmost.Thisquantity,whichiscalled{complexity,characterizestheintrinsicdi cultyofanumericalproblem.LocalsolutionofFredholmintegralequationswas rstconsideredbyHeinrich[7],lateronbyFrankandHeinrich[5],[3].Inthesepapers,foreachspecialfunctionclasssimilartheoremswereshownstatingtheequivalenceofthen{thminimalradiusofinformationtoGelfandnumbersofsomeoperators(forde nitionssee1991MathematicsSubjectClassi cation.Primary65Y20,45L10;Secondary47B06,45B05.c 0000AmericanMathematicalSociety0075-8485/00$1.00+$.25perpage12KARINFRANKsections2and3).Furthermore,forseveralfunctionclassesalgorithmswerefoundrealizingtheoptimalerror,whichbaseoncommonprinciples.Similaralgorithmsprovedtobeoptimalalsoforthefullsolutionproblemincertainfunctionclasses(see[2],[4]).Itseemedtobequitenaturaltotrytogeneralizethisapproachtoawiderclassofequations.Thisistheaimofthepresentpaper.Thepaperleadsoutasfollows:Insection2,theproblemisformulatedinthesettingofinformation{basedcomplexity.Therealsothemostimportantde nitionsarerecalled.Insection3,undersomeverygeneralassumptionstheequivalenceofthen{thminimalradiusofinformationandtheGelfandnumbersofthreeoperatorscharacterizingtheproblemisshown.Aniterativealgorithmisproposedinsection4whoseoptimalityinthesenseofinformationcomplexityisshowninsection5undercertainadditionalconditions.ThesegeneralresultsareappliedtoFredholmintegralequationsofthesecondkindinSobolevspacesofperiodicfunctionswithdominatingmixedderivativeHr1;:::;rd([0;1]d)insection6,wheretheexactorderofthen{thminimalradiusofinformationisderived.Moreover,thegeneralschemeofaniterativealgorithmformulatedinsection4isconcretizedandanimplementableoptimalalgorithmforthisfunctionclassisgiven,whichisbasedonahyperboliccrossapproximationofthekernelfunction.2.FormulationoftheproblemLetV,EandKbeBanachspaces.Asusual,forthespaceofallboundedlinearoperatorsfromVintoEweshallwriteL(V;E),andL(E)=L(E;E).E denotesthedualspacetoE,BEtheunitballinE.WeassumethatViscontinuouslyembeddedintoE,i.e.thereexistsalinearandcontinuousembeddingoperatorJV:V!E.Moreover,thereissomelinearcontinuousoperatorTassigningtoeachelementk2KanoperatorTk2L(E).WeconsidersuchsubsetsV0 V,K0 K,thattheoperator(I Tk) 1:E!Eisboundedforanyk2K0.HereandfurtherIdenotestheidentityoperator.SettingX0=K0 V0westudytheclassofoperatorequationsu Tku=f;(k;f)2X0:(1)Theproblemhastobeformulatedwithintheframeworkofinformation{basedcom-plexitytheory(IBC).Hereonlythemostimportantde nitionsareoutlined,formoredetailsthereaderisreferredto[11].Insteadofsearchingthesolutionuofequation(1)onthewholedomain,weareinterestedonlyinthevalueofonesinglefunctional 2E appliedtothesolutionu2E.Thisproblemsettingiscalledlocalsolutionoftheoperatorequation(1).Inspecialcases,whenV;Earefunctionspaces,thefunctional canbee.g.a t0{functionalwhichgivesthevalueofuinasinglepointt0,oraweightedmean,orasinglecoe cientintherepresentationofuinsomebasisinE.TheoperatorS :X0!IRde nedasS (k;f)=h(I Tk) 1f; i(2)iscalledthelocalsolutionoperatorofequation(1).Usually,wehavenofullinformationaboutthedatak2K0,f2V0,butonlypartialinformation.Weassume,thatwearegivenlinearinformationabout(k;f)2X0,ANOPTIMALALGORITHMFORTHELOCALSOLUTIONOFINTEGRALEQUATIONS3sowede netheinformationoperator N=(N;M), N:X0!IRn1+n2asNk=(hk; ii)i=1;::;n1; i2K ;Mf=(hf; ii)i=1;::;n2; i2V :(3)Anyoperator’:IRn1+n2!IRiscalledanalgorithmassigninganapproximatesolution’( N(k;f))toanyinformationvector N(k;f).Theerrorofanalgorithm’:IRn1+n2!IRisde nedastheworst{caseerrore(S ; N;’)=sup(k;f)2X0jS (k;f) ’( N(k;f))j:AnessentialquantityinIBCistheso{calledradi