ebook img

Positive Dynamical Systems in Discrete Time: Theory, Models, and Applications PDF

366 Pages·2015·1.967 MB·English
by  KrauseUlrich
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Positive Dynamical Systems in Discrete Time: Theory, Models, and Applications

UlrichKrause PositiveDynamicalSystemsinDiscreteTime De Gruyter Studies in Mathematics | Editedby CarstenCarstensen,Berlin,Germany NicolaFusco,Napoli,Italy FritzGesztesy,Columbia,Missouri,USA NielsJacob,Swansea,UnitedKingdom Karl-HermannNeeb,Erlangen,Germany Volume 62 Ulrich Krause Positive Dynamical Systems in Discrete Time | Theory, Models, and Applications MathematicsSubjectClassification2010 15B48,37B55,39B12,47B65,47H07,47N10,60J10 Author Prof.Dr.Dr.UlrichKrause UniversitätBremen Fachbereich03–Mathematik/Informatik Bibliothekstr.1 28359Bremen Germany [email protected] ISBN978-3-11-036975-5 e-ISBN(PDF)978-3-11-036569-6 e-ISBN(EPUB)978-3-11-039134-3 Set-ISBN978-3-11-036571-9 ISSN0179-0986 LibraryofCongressCataloging-in-PublicationData ACIPcatalogrecordforthisbookhasbeenappliedforattheLibraryofCongress. BibliographicinformationpublishedbytheDeutscheNationalbibliothek TheDeutscheNationalbibliothekliststhispublicationintheDeutscheNationalbibliografie; detailedbibliographicdataareavailableontheInternetathttp://dnb.dnb.de. ©2015WalterdeGruyterGmbH,Berlin/Munich/Boston Typesetting:PTP-Berlin,ProtagoTEX-ProductionGmbH Printingandbinding:CPIbooksGmbH,Leck ♾Printedonacid-freepaper PrintedinGermany www.degruyter.com | ToCarolaandDaniel Preface Positivedynamicalsystemscomeintoplaywhenrelevantvariablesofasystemtakeon valueswhicharenonnegativeinanaturalway.Thisisthecase,forexample,infields asbiology,demographyandeconomics,wherethelevelsofpopulationsorpricesof goodsarepositive.Positivitycomesinalsoiftheformationofaveragesbyweighted meansisrelevantsinceweights,forexampleprobabilities,arenotnegative.Thisis thecaseinquitediversefieldsrangingfromelectricalengineeringoverphysicsand computersciencetosociology.Therebyaveragingtakesplacewithrespecttosignals in a sensor network or in a swarm (of birds or robots) or with respect to velocities ofparticlesortheopinionsofpeople.Inthefieldsmentionedthedynamicsisoften modeledbydifferenceequationswhichmeansthattimeistreatedasdiscrete.Thus, inrealityonemeetsahugevarietyofpositivedynamicalsystemsindiscretetime. Inmanycasesthesesystemscanbecapturedbyalinearmappinggivenbyanon- negativematrix.Thedynamics(indiscretetime)thenisgivenbythepowersofthe matrixor,equivalently,bytheiteratesofthelinearmappingwhichmapsthepositive orthantintoitself.ApowerfultoolthenisthePerron–FrobeniusTheoryofnonnega- tivematrices(includingtheasymptoticbehaviorofpowersofthosematrices)which hasbeensuccessfulsinceitsinceptionbyO.PerronandG.Frobeniusoverabouthun- dredyearsago.Concerningtheoryaswellasapplicationstherearetwoinsufficient aspectsofPerron–FrobeniusTheorywhichlaterondrovethistheoryintonewdirec- tions.Thefirstaspectisthatthistheoryisnotjustaboutnonnegativematricesbut applies happily also to certain matrices with negative entries. This means that the theory should be understood as dealing with linear selfmappings of convex cones infinitedimensionsnotjustofthestandardcone,thepositiveorthant.Thesecond aspectis thatevensimple positive dynamicalsystems are not linear.Thus,whatis neededisanextensionofclassicalPerron–FrobeniusTheorytononlinearselfmap- pingsofconvexconesinfinitedimensions.Moreover,withrespecttotheoryaswell asapplications,suchanextensionisneededalsoininfinitedimensions.Sinceclassi- calPerron–FrobeniusTheoryhasalreadysomanyapplicationsonecanimaginethe greatvarietyofapplicationssuchanextensiontononlinearselfmappingsininfinite dimensionswillhave. It is the aim of the present book to provide a systematic, rigorous and self– contained treatment of positive dynamical systems based on the analysis of the it- erations of nonlinear selfmappings of a convex cone in some real vector space. To pursuethistask,helpcomesfromabeautifulapproachdevelopedforthelinearcase byG.BirkhoffconsideringJentzsch’sTheoremininfinitedimensionsand,indepen- dently, by H. Samelson considering Perron–Frobenius Theory in finite dimensions. Thecrucialpointofthisapproachisthetranslationofastrongpositivitypropertyof Preface | vii thelinearmappingintoacontractivitypropertywithrespecttosomemetricinternal totheconvex cone.Thismetrichasbeenused longbefore byD.Hilbertwithinthe completelydifferentareaofthefoundationsofgeometryandiscalledHilbert’spro- jectivemetric(aquasi–metric,actually).Theextensionofthisapproach,alsocalled theBirkhoffprogram,tononlinearselfmappingsofconvexconesisacornerstoneof thepresentbook.Asitturnsouttheinvestigationofthenonlinearityismadeeasier byhavingitbasedonaconvexconeanditsanalysis.Sincetheconvexconereflects thepositivityofthesystemonemightsaythatpositivityhelpstotamenonlinearity. Manybeautifulresultsareavailablewhichareimpossiblewithoutpositivity. Thefollowingparagraphssketchbrieflythecontentofeachoftheeightchapters ofthebook. Chapter1motivatesthestudyofpositivedynamicalsystems(indiscretetime)by means of examples from biology and economics. As for biology a nonlinear exten- sionof theclassicalLeslie modelusedinpopulationdynamicsanddemographyis presented by taking population pressure into account. Considering economics, for thelikewiseclassicalLeontiefmodelofcommodityproductionanonlinearextension istreatedwhichcapturesthechoiceoftechniques.Therearemuchmoreexamples of nonlinear positive dynamical systems. The example of opinion dynamics under boundedconfidencehasrecentlyattractedmuchattentionandwillbeinvestigated inthelastchapterofthebook. Chapter2on“ConcavePerron–FrobeniusTheory”presentsanextensionofclassi- calPerron–FrobeniusTheoryfromlineartoconcavemappings(includinglinearones). IntheproofsHilbert’sprojectivemetricmakesitsfirstappearance.Thepointtherebyis thatforthismetricconcavemappingsarecontractionsandtheinteriorofthestandard coneiscomplete.BythisPerron–FrobeniustheoremscanbeprovedusingBanach’s contraction mapping principle. Though only a particular form of nonlinearity con- cavitycoversthenonlinearitiesinthemodelsofLeslieandLeontief.Whereasinlater chaptersmoregeneralnonlinearitieswillbetackledon,thischapterconcentratesjust on concave mappings since for these a variety of results is possible comparable to thoseofclassicalPerron–FrobeniusTheory.Itshouldbenoted,however,thateven concaveselfmappingsofthestandardconeinfinitedimensionsexhibitalreadyspec- tralpropertiesinsharpcontrasttothelinearcaseinthattheremaybeinfinitelymany eigenvalues. Whereasinthefirsttwochapterspositivityisrestrictedtothestandardconein finitedimensions,theoryaswellasapplicationsinlaterchaptersrequiremoregeneral convexconesininfinitedimensions. Chapter3on“Internalmetricsonconvexcones”treatsgeneralconvexconesin topologicalvectorspaceswithafocusoninternalmetrics.Thelatterare(quasi–)met- ricssolelydeterminedbythecone’sconvexstructure.Hilbert’sprojectivemetricand theThompsonmetricorpartmetricarethemostrelevantinternalmetricsbutthere aremuchmore.Besidescertaingeometricalpropertiesofinternalmetricsthechapter concentratesoncriteriaforaconvexconetobecompleteforaninternalmetric.For viii | Preface laterusethetopologyofthevectorspaceisrelatedtotheoneinducedbyaninter- nalmetricandcriteriaforinternalcompletenessareobtainedintermsofthevector space topology. A particularcase is the result obtained first by G. Birkhoff that the positiveconeofaBanachlattice–aswellasitsinterior–arecompleteforHilbert’s projectivemetric.ExtendingthemethodappliedinChapter2forfinitedimensions,by Chapter3selfmappingsofaconvexconecanbelookedatasselfmappingsofacom- pletemetric space. Sincelater on contractivitywith respect to internalmetrics will playarole,Chapter4on“Contractivedynamicsonmetricspaces”investigatesvar- ioustypesofcontractivityingeneralmetricspaces.Conditionsarespecifiedwhich guaranteepointwiseconvergenceoftheiteratesofaselfmappingtoafixedpoint.An importantprinciplestatesthatthisglobalpropertyappliesalreadyifitholdslocally incaseofpower–lipschitzianmappings(includingnonexpansivemappings).Forlater applicationstononautonomouspositivesystemsthecompositionofinfinitelymany selfmappingsanditsasymptoticbehaviourisanalyzed. Both,Chapter3andChapter4supplyinageneralsettingtoolsneededinsubse- quentchapters.Besidethis,bothchapterspresentknownandnewresultswhichare interestinginitself. Chapter5on“Ascendingdynamicsinconvexconesofinfinitedimension”presents afar–reachingextensionofChapter2toconvexconesininfinitedimensionsandcor- respondingselfmappingsincludingconcaveones.Anascendingoperatoris,roughly speaking,aselfmappingofaconvexcone,thevaluesofwhich,onasubsetofthecone, increasewithrespecttothecone’sorderingonvectorsaswellaswithrespecttothe commonorderonpositivescalars.Itisanimportantfeatureofascendingoperators tobepositivewithoutbeingnecessarilymonotone. InthelinearcasetheuniformlypositivelinearoperatorsintroducedbyG.Birkhoff areexamples.Nonlinearexamplesaretheu –concaveoperatorsstudiedbyM.Kras- 0 noselskiiandhiscollaborators.Whereasthesemappingsneedtobemonotone,this is,however,notthecaseforascendingoperatorsingeneral.UsingHilbert’sprojective metricforascendingoperatorsrelativestabilityisproven,meaningtheiteratesofthe normalizedoperatorsdoconvergetoaneigenvector.Usingthepartmetricforweakly ascendingoperators,absolutestabilityisshownthatistheiteratesconvergetoafixed point.Applicationsconcernnonlineardifferenceequationsandanonlinearversion ofJentzsch’sTheoremonintegraloperators,includinganapproximationalgorithmto computetheuniquesolution. Chapter6on“Limitsettrichotomy”investigatesafundamentalphenomenonof positivedynamicalsystemswhichmeansthateitherallorbitstendtoinfinityorall orbitstendtozeroorallorbitstendtoafixedpointintheinteriorofthecone.Vari- ousconditionsforthisphenomenontohappenarespecified.Limitsettrichotomycan beusedinmanyways,itguarantees,forexample,theexistenceofagloballystable fixedpointintheinteriorifthereexistsanorbitpositivelyboundedfrombelowand above.Forthecaseofdifferentiableselfmappingsinfinitedimensionseasytocheck conditionsforlimitsettrichotomyaregiven.Anapplicationistononlineardifference Preface | ix equationsincludingageneralizednonlinearFibonacciequation.Anotherapplication considerscooperativesystemsofdifferentialequationswithabiochemicalcontrolcir- cuitasaparticularexample. Chapter7on“Nonautonomouspositivesystems”dealswiththeasymptoticbehav- iorofcompositionsofinfinitelymanyselfmappingsofaconvexcone.Variouskinds ofbehaviouraspathstability,asymptoticproportionality,weakandstrongergodicity areanalyzed.Theresultonconcaveweakergodicityisanextensionofthefamous(lin- ear)Coale–LopezTheoremindemography.Anothernonlinearextensionconcernsthe classicalstrongergodicitytheoremfornonnegativematrices.Furthermore,abeautiful theoremofH.Poincaréonnonautonomouslineardifferenceequationsisextendedto includenonlineardifferenceequations.Also,thenonlinearversionsofthemodelsof LeslieandLeontiefintroducedinthefirstchapterareinvestigatedforsurvivalreates dependentontimeandfortime–dependenttechnicalchange,respectively.Finally, forpopulationsbeingunderenforcementfromtheenvironmentconditionsonpopu- lationpressurearegivenwhichstillyieldpathstability. Thelastandlongestchapter,Chapter8,isonthe“Dynamicsofinteraction:Opin- ions,meanmaps,multi–agentcoordination,andswarms”.Itistheaimofthischapter todevelopasystematicandrigorousanalysisforthedynamicsofseveralfascinating kindsofinteraction.Suchinteractionshavebeenaddressedrecentlyinawidespread and fastly growingliteratureby researchers from quitedifferentfields whichrange fromelectricalengineeringoverphysicsandcomputersciencetosociologyandeco- nomics.Theleadingquestiontherebyasksunderwhatconditionsagroupofagents, beingitrobotersorhumansorotherkindsofanimals,isabletocoordinatethemselve toreachaconsensus.Mathematically,thelattermeansforadynamicalsystemwith severalcomponentswhethertheseconvergealltothesamestate.Initsmostsimple caseoneconsidersanonnegativematrixwithallrowssumminguptooneandasks forconditionsunderwhichthepowersofthematrixconvergetoamatrixhavingall itsrowsequal.Theanswerinthisspecialcaseisthatthishappenspreciselyifthema- trixhasapowerwhichisscrambling.Thisis(asharpenedversionof)thewell–known BasicLimitTheoremforMarkovChains.Alreadysimplecasesofinteraction,however, arenonlinear(ortime–variant)asforthemodelofopiniondynamicsunderbounded confidence(alsoknownasHegselmann–Krausemodelintheliterature)whichhasat- tractedmanyresearchersinrecentyears.Anonlinearanalogueofa(row–)stochastic matrixisameanmap.Concerningtime–varianceoneconsidersasequenceofstochas- ticmatrices.Bothcasesleadtopositivedynamicalsystemsasconsideredinprevious chapters.Facingtheparticulartypeofconvergencetoconsensustoolsadaptedtothat aredevelopedinChapter8.Incaseoftime–variancethesearetoolstohandleinfinite productsofstochasticmatrices.Whatisneededareconditionsonthestructureand intensityofinteractionwhichmaketheinfiniteproductconvergenttoamatrixwith equalrows.Anoftenusedtool,thetheoremofWolfowitz,isgeneralized.Thechapter concludeswithanapplicationtothedynamicsofswarmsofbirds.Therecentlymuch x | Preface discussedCucker–Smalemodelistreatedandanewmodelofswarmingisdeveloped wherebirdsareformingswarmsundersomeweakerconditionsoninteraction. Eachchapterissubdividedintosectionsthematerialofwhichisillustratedbyex- amplesandcontainsexercisesrangingfromsimpleverificationsoveradditionaltop- icstoopenproblems.Remarkscommentonresultsobtainedandprovidelinkstothe literature.Ineachchapterresults,examples,remarksareconsecutivelynumberedby a.b.cwhereareferstothechapter,btothesectionandctotheparticularitem.Each chapterappendedisabibliographyspecifictoit.Alistofnotationsandanindexcon- cludethebook. The book is directed to researchers from various disciplines and graduate stu- dents,too,whoareinterestedinpositivedynamicalsystems. Thebookisself–containedandorganizedinamannersuchthatitsmaterialcan alsobeusedincoursesandseminars.Chapters1and2requireonlyabasicknowledge inlinearalgebraandanalysisandcouldbeusedforanintroductorycourseinnonlin- earPerron–FrobeniusTheoryincludingapplications.Chapters3,4,and5couldserve asmaterialinacourseorseminarforgraduatestudentsandrequiresomefamiliar- itywithfundamentalconceptsintopologyandfunctionalanalysis.Thesameapplies toChapters6and7whichcouldbeusedasmaterialinanadvancedcourse.Thelast Chapter8canbereadindependentlyofthepreviousonesandcouldserveasanintro- ductionintorecentapplicationsofpositivedynamicalsystems.Thefascinatingtopics aresuitableforgraduatestudentstoworkon,analyticallyaswellasbydoingcom- putersimulations. ThisbookgrewoutofseveralcoursesandseminarsIheldovertheyearsatthe UniversityofBremen.Itwasagreatexperiencetosharewiththestudentstheenthu- siasmforafieldwhichisjustinthebeginning.Iliketothankallthestudentsfortheir contributionsandIwanttomentioninparticularTimNesemannandJanLorenz.The readerwillconsultthereferencesgiveninthebooktotheirworkandthatofotherstu- dentsaswellastotheworkofresearchersIenjoyedtowritejointpaperswith.HereI liketothankChristianBidard,RainerHegselmann,DiederichHinrichsen,TakaoFu- jimoto,TimNesemann,RogerNussbaum,MihályPituk,PeterRanft,DietrichWeller. Furthermore,IwanttothankBirgitFeddersenfromtheDepartmentofMathemat- icsforherexperiencedandnicetranslationofthemanuscriptintoLaTex,including thefigures. FormanycarefulandhelpfulcommentsIhavetothankthethreeanonymousre- viewersofthemanuscript. Finally,IwanttothankthepublisherDeGruyterandinparticularFriederikeDitt- bernerandSilkeHutt,whohavebeenmosthelpfulintheprocessofpublication. ThebookIdedicatetomywifeCarolaandtooursonDanielwhostayedsofriendly tosomeonewholivedwithadeskfordays,monthsandyears. Bremen,November2014 UlrichKrause

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.