Fractal Dimensions for Poincaré Recurrences MONOGRAPH SERIES ON NONLINEAR SCIENCE AND COMPLEXITY SERIESEDITORS AlbertC.J.Luo SouthernIllinoisUniversity,Edwardsville,USA GeorgeZaslavsky NewYorkUniversity,NewYork,USA ADVISORYBOARD ValentinAfraimovich, SanLuisPotosiUniversity,SanLuisPotosi,Mexico MauriceCourbage, UniversitéParis7,Paris,France Ben-JacobEshel, SchoolofPhysicsandAstronomy,TelAvivUniversity, TelAviv,Israel BernoldFiedler, FreieUniversitätBerlin,Berlin,Germany JamesA.Glazier, IndianaUniversity,Bloomington,USA NailIbragimov, IHN,BlekingeInstituteofTechnology,Karlskrona,Sweden AnatolyNeishtadt, SpaceResearchInstituteRussianAcademyofSciences, Moscow,Russia LeonidShilnikov, ResearchInstituteforAppliedMathematics&Cybernetics, NizhnyNovgorod,Russia MichaelShlesinger, OfficeofNavalResearch,Arlington,USA DietrichStauffer, UniversityofCologne,Köln,Germany JianQiaoSun, UniversityofDelaware,Newark,USA DimitryTreschev, MoscowStateUniversity,Moscow,Russia VladimirV.Uchaikin, UlyanovskStateUniversity,Ulyanovsk,Russia AngeloVulpiani, UniversityLaSapienza,Roma,Italy PeiYu, TheUniversityofWesternOntario,London,OntarioN6A5B7,Canada Fractal Dimensions for Poincaré Recurrences V.AFRAIMOVICH SanLuisPotosiUniversity,Mexico E.UGALDE SanLuisPotosiUniversity,Mexico J.URÍAS SanLuisPotosiUniversity,Mexico AMSTERDAM BOSTON HEIDELBERG LONDON NEWYORK OXFORD • • • • • PARIS SANDIEGO SANFRANCISCO SINGAPORE SYDNEY TOKYO • • • • • Elsevier Radarweg29,POBox211,1000AEAmsterdam,TheNetherlands TheBoulevard,LangfordLane,Kidlington,OxfordOX51GB,UK Firstedition2006 Copyright©2006ElsevierB.V.Allrightsreserved Nopartofthispublicationmaybereproduced,storedinaretrievalsystemortransmittedinanyform orbyanymeanselectronic,mechanical,photocopying,recordingorotherwisewithoutthepriorwritten permissionofthepublisher PermissionsmaybesoughtdirectlyfromElsevier’sScience&TechnologyRightsDepartmentinOxford, UK:phone(+44)(0)1865843830;fax(+44)(0)1865853333;email:[email protected] nativelyyoucansubmityourrequestonlinebyvisitingtheElsevierwebsiteathttp://elsevier.com/locate /permissions,andselectingObtainingpermissiontouseElseviermaterial Notice Noresponsibilityisassumedbythepublisherforanyinjuryand/ordamagetopersonsorpropertyas amatterofproductsliability,negligenceorotherwise,orfromanyuseoroperationofanymethods, products,instructionsorideascontainedinthematerialherein.Becauseofrapidadvancesinthemedical sciences,inparticular,independentverificationofdiagnosesanddrugdosagesshouldbemade LibraryofCongressCataloging-in-PublicationData AcatalogrecordforthisbookisavailablefromtheLibraryofCongress BritishLibraryCataloguinginPublicationData AcataloguerecordforthisbookisavailablefromtheBritishLibrary ISBN-13:978-0-444-52189-7 ISBN-10:0-444-52189-5 SeriesISSN:1574-6917 ForinformationonallElsevierpublications visitourwebsiteatbooks.elsevier.com PrintedandboundinTheNetherlands 06 07 08 09 10 10 9 8 7 6 5 4 3 2 1 Preface Researchbyseveralprominentresearchgroups(includingPennStateUniversity, CNRSatLuminyandEcolePolithecnique,UniversitySimónBolivar,ourgroup inSanLuisPotosí,etc.)hasshownthatthedimensiontheoryofdynamicalsys- tems is a powerful tool to analyze the (multi)fractal behavior which appears in realsystemsandtheirmathematicalmodels.Thisbookisdevotedtoanimportant branch of this theory: the study of the fine (fractal) structure of Poincaré recur- rences–instantsoftimewhenthesystemalmostrepeatsitsinitialstate.Because ofthisrestrictionwewereabletowriteanentirelyself-containedtextincluding manyinsightsandexamples,aswellasprovidingcompletedetailsofproofs.The onlyprerequisitesareabasicknowledgeofanalysisandtopology.Thusthisbook canserveasagraduatetextorself-studyguideforcoursesinappliedmathematics ornonlineardynamics(inthenaturalsciences). To motivate our study of Poincare recurrences, imagine that the phase space (or invariant subset) is partitioned into different colors according to their “tem- perature”.Theorbitforinitialconditionschoseninhotareasreturnsnearbyfaster thanforinitialconditionschosenfromcoldareas.Moreprecisely,thereturntime foranεballcenteredonahotinitialconditionismuchlessthanforacoldinitial condition.Itistrueevenforuniformlyhyperbolicsystems,providedthattheyare “sufficientlynonlinear”,i.e.,theirlinearizationdependsonthepoint. Onecanthenfixalargecollectionofεballsandcalculatean“average”return time, and study this average as ε goes to 0. For many hyperbolic systems the average behaves as γ logε and for non-chaotic systems as ε γ, where γ is a − − keydimension-likecharacteristicobtainedfromthefractaldimensionmachinery: thedimensionforPoincarérecurrences.Itdependsonthesetofinitialpointswe dealwith,i.e.,itisafunctionofaset.Ifwechooseasetsituatedaroundhotspots, evidentlywewillobtainadimensionthatisdifferentfromthataroundcoldspots. Thishasprofoundpracticalapplications,foritprovidesauseful“measure”of howchaoticisadynamicalsystemwithintheclassofchaoticdynamicalsystems while for nonchaotic dynamical systems it is a new and useful measure of the complexity of the orbit structure. Furthermore, given an invariant measure, it is naturaltointroduceadimensionofmeasurewhichequals,asusual,thedimension ofthesmallestsetofthefullmeasure.So,ourdimensioncandistinguishdiffer- ent measures according to the behavior of Poincaré recurrences. If the measure is ergodic then the behavior of Poincaré recurrences is asymptotically the same for any ball centered at a typical point. Thus, one can obtain the dimension of v vi Preface measureforPoincarérecurrences,whichisaglobalquantity,byknowingalocal dimension. Webelievethatthedimensioncouldbequiteusefulformanyappliedproblems. Letusemphasizethatthedimensioniscomputableandcanbefoundnumerically for specific systems. Let us mention now a problem related to synchronization phenomena. If two (or more) coupled subsystems are synchronized then their behaviorintimehastobesimilarandsincethedimensiondefinitelyreflectssuch abehavior,thenthedimensionforPoincarérecurrencesinasynchronizedregime hastobethesameforallindividualsubsystems.Thus,itcanserveasanindicator ofsynchronization. Thesecondproblemwewanttomentionistheproblemoffractalandmultifrac- tal features of Poincaré recurrences. In situations where a system is nonergodic andcontainsboth,chaoticinvariantsubsetsandsubsetswithzerotopologicalen- tropy (such as in standard map) a normalized distribution of return times to a regionbehavesasfollows P(τ) τ γ, τ − ∼ →∞ (whereP(τ)dτ isinfacttheprobabilitytoreturntotheregionduringtheinter- val of time (τ,(τ dτ))). We explained in Chapter 15 that this exponent γ is + directlyrelatedtothedimensionforPoincarérecurrences.So,ourquantityhasan importantphysicalmeaning. The book includes figures already published in our papers. Kind permissions werereceivedfromthepublishersWorldScientific,DiscreteandContinuousDy- namicalSystemsandTheAmericanPhysicalSocietyforthereproductionofthe following figures of this book: Figures 3.4 and 3.5 from reference [52], Fig- ure15.3fromreference[12]andFigures16.3–16.6fromreference[13]. Contents Preface v Chapter1. Introduction 1 PARTI. FUNDAMENTALS 7 Chapter2. SymbolicSystems 9 2.1. Specifiedsubshifts 9 2.1.1 Ultrametricspace 11 2.2. OrderedtopologicalMarkovchains 12 2.3. Multipermutativesystems 17 2.3.1 Polysymbolicgeneralization 19 2.3.2 Topologicalconjugationofpolysymbolicminimalsystems 20 2.3.3 Nonminimalmultipermutativesystems 23 2.4. Topologicalpressure 28 2.4.1 Dimension-likedefinitionoftopologicalpressure 32 Chapter3. GeometricConstructions 35 3.1. Moranconstructions 35 3.1.1 GeneralizedMoranconstructions 37 3.1.2 InvariantsubsetsofMarkovmaps 40 3.2. TopologicalpressureandHausdorffdimension 43 3.2.1 Hausdorffandboxdimensions 43 3.2.2 Bowen’sequation 45 3.2.3 Morancovers 45 3.3. StrongMoranconstruction 48 3.4. Controlledpackingofcylinders 48 3.5. Stickysets 49 3.5.1 Geometricconstructionsofstickysets 51 vii viii Contents Chapter4. TheSpectrumofDimensionsforPoincaréRecurrences 53 4.1. GeneralizedCarathéodoryconstruction 53 4.1.1 Examples 54 4.2. Thespectrumofdimensionsforrecurrences 57 4.3. Dimensionandcapacities 58 4.4. Theappropriategaugefunctions 59 4.5. Generalpropertiesofthedimensionforrecurrences 63 4.6. Dimensionforminimalsets 65 4.6.1 Thegaugefunctionξ(t) 1/t 66 = 4.6.2 Rotationsofthecircle 66 4.6.3 Denjoyexample 69 4.6.4 Multidimensionalrotation 72 PARTII. ZERO-DIMENSIONALINVARIANTSETS 75 Chapter5. UniformlyHyperbolicRepellers 77 5.1. SpectrumofLyapunovexponents 78 5.2. Thecontrolled-packingcondition 79 5.2.1 ProofofLemma5.1 80 5.2.2 ProofofLemma5.2 82 5.3. Spectraunderthegapcondition 83 Chapter6. Non-UniformlyHyperbolicRepellers 87 6.1. Noorbitsinthecriticalset 88 6.2. Thecriticalsetcontainsanorbit 90 Chapter7. TheSpectrumforaStickySet 95 7.1. ThespectrumforPoincarérecurrences 95 Chapter8. RhythmicalDynamics 99 8.1. Set-up 99 8.2. DimensionsforPoincarérecurrences 100 8.2.1 Thecaseofanautonomousrhythmfunctionφ 100 8.2.2 Thecaseofnon-autonomousrhythmfunctionφ 101 8.3. Thespectrumofdimensions 102 8.3.1 Autonomousφ 102 8.3.2 Non-autonomousφ 103 Contents ix PARTIII. ONE-DIMENSIONALSYSTEMS 107 Chapter9. MarkovMapsoftheInterval 109 9.1. Thespectrumofdimensions 110 Chapter10. SuspendedFlows 117 10.1. Suspendedflowsoverspecifiedsubshifts 117 10.1.1 Poincarérecurrences 118 10.1.2 Suspendedflow 118 10.2. Bowen–Walters’distance 118 10.3. Spectrumofdimensions 119 10.3.1 ThePoincarérecurrence 119 10.3.2 Thespectrum 120 10.3.3 Mainresults 120 10.3.4 ProofofClaim10.1 127 10.3.5 ProofofClaim10.2 129 PARTIV. MEASURETHEORETICALRESULTS 133 Chapter11. InvariantMeasuresandPoincaréRecurrences 135 11.1. Pointwisedimensionandlocalrates 135 11.2. TheSMBtheorem 137 11.3. KolmogorovcomplexityandBrudno’stheorem 137 11.4. Thelocalrateofreturntimes 138 11.4.1 ProofofTheorem11.3basedontheSMBTheorem 138 11.4.2 ProofofTheorem11.3basedonBrudno’sTheorem 140 11.4.3 Rotationsofthecircle 141 11.5. Remarksonlocalrates 143 11.6. Theq-pointwisedimension 145 Chapter12. DimensionsforMeasuresandq-PointwiseDimension 149 12.1. Preliminariesandmotivation 149 12.2. Aformulaformeasures 151 12.3. Theq-pointwisedimension 153 12.4. Stickysets 156 12.5. Remarksontheq-pointwisedimension 161