Chapter 6 Double Pendulum Analysis (Part 1)

18
6-1 Chapter 6 Double Pendulum Analysis (Part 1) ABSTRACT for Chapters 6,7,8 and 9: Behavior of the double pendulum was explored by making several hundred runs with a simulation model. Only a small part of the parameter space was explored but it showed how complex behavior can be in a very simple system. Patterns of behavior at different levels of total system energy are reported. At low to moderate energy levels bobs perform a rhythmic dance-like sequence of moves that repeats periodically, however with different size steps thus creating patterns whose shape gradually changes. This is called quasi-periodic. At high energy levels the system experiences sensitive dependence on initial conditions and is therefore chaotic. Waveforms within each period are complex. The ability to predict future behavior with these patterns is discussed. Its hypothesized the double pendulum is governed by a 4-dimensional strange attractor similar in concept to the Lorenz butterfly. It’s found that different phase space patterns occur at the same energy level depending on the positions of the bobs when released. Tightly constrained instances of perfectly periodic behavior were found within otherwise quasi-periodic ranges. A sharp, energy defined, qualitative change in behavior was found between behavior where the bob occasionally goes over the top or spins and behavior where it always reverses and falls back. This occurred only above the energy level where the system first became chaotic and therefore does not demark the threshold into chaos. The way energy moves from one form and one part of the system to another is treated. The random occurrence of spikes in the magnitude of system variables after relatively calm periods is highlighted as perhaps the most important practical implication of chaos. Suggestions for further research are offered. Outline 6.1 Research objectives 6.2 Description: 6.3 Types of Behavior 6.4 Double pendulum simulation model 6.1 Research objectives The purpose of this Chapter is to explore and categorize the behavior of the double pendulum over a wide range of conditions and especially to see how its behavior changes as a function of the total energy within the system. Attempts were also made to see what practical difference it makes whether a system is chaotic or not. A key objective was to discover the root cause of chaos in this system, but it was unsuccessful.

Transcript of Chapter 6 Double Pendulum Analysis (Part 1)

Page 1: Chapter 6 Double Pendulum Analysis (Part 1)

6-1

Chapter6DoublePendulumAnalysis(Part1)ABSTRACTforChapters6,7,8and9:Behaviorofthedoublependulumwasexploredbymakingseveralhundredrunswithasimulationmodel.Onlyasmallpartoftheparameterspacewasexploredbutitshowedhowcomplexbehaviorcanbeinaverysimplesystem.Patternsofbehavioratdifferentlevelsoftotalsystemenergyarereported.Atlowtomoderateenergylevelsbobsperformarhythmicdance-likesequenceofmovesthatrepeatsperiodically,howeverwithdifferentsizestepsthuscreatingpatternswhoseshapegraduallychanges.Thisiscalledquasi-periodic.Athighenergylevelsthesystemexperiencessensitivedependenceoninitialconditionsandisthereforechaotic.Waveformswithineachperiodarecomplex.Theabilitytopredictfuturebehaviorwiththesepatternsisdiscussed.Itshypothesizedthedoublependulumisgovernedbya4-dimensionalstrangeattractorsimilarinconcepttotheLorenzbutterfly.It’sfoundthatdifferentphasespacepatternsoccuratthesameenergyleveldependingonthepositionsofthebobswhenreleased.Tightlyconstrainedinstancesofperfectlyperiodicbehaviorwerefoundwithinotherwisequasi-periodicranges.Asharp,energydefined,qualitativechangeinbehaviorwasfoundbetweenbehaviorwheretheboboccasionallygoesoverthetoporspinsandbehaviorwhereitalwaysreversesandfallsback.Thisoccurredonlyabovetheenergylevelwherethesystemfirstbecamechaoticandthereforedoesnotdemarkthethresholdintochaos.Thewayenergymovesfromoneformandonepartofthesystemtoanotheristreated.Therandomoccurrenceofspikesinthemagnitudeofsystemvariablesafterrelativelycalmperiodsishighlightedasperhapsthemostimportantpracticalimplicationofchaos.Suggestionsforfurtherresearchareoffered.

Outline

6.1Researchobjectives6.2Description:6.3TypesofBehavior6.4Doublependulumsimulationmodel

6.1ResearchobjectivesThepurposeofthisChapteristoexploreandcategorizethebehaviorofthedoublependulumoverawiderangeofconditionsandespeciallytoseehowitsbehaviorchangesasafunctionofthetotalenergywithinthesystem.Attemptswerealsomadetoseewhatpracticaldifferenceitmakeswhetherasystemischaoticornot.Akeyobjectivewastodiscovertherootcauseofchaosinthissystem,butitwasunsuccessful.

Page 2: Chapter 6 Double Pendulum Analysis (Part 1)

6-2

TheapproachwastouseaJavasimulationmodel.Theresearchwasdoneintwophasesspacedmonthsapart.ThefirstphaseisdocumentedinChapters7and8andthesecondinChapter9.Thesecondphasewasdonemainlytofindandexploretheboundarybetweenquasi-periodicandchaoticbehavior,whichunfortunatelywasnotdonewellinphaseonedotoaperceivedlackofanyeasywaytodetectoneofthemaincriteriaforchaos,namelysensitivedependenceoninitialconditionsorSDIC.Althoughhundredsofsimulationrunsweremade,manymorewouldbeneededtofullymapthebehaviorofthedoublependulum.HopefullyotherswilltakethisonandalsouseadifferentmodeltoensurethesefindingsarenotbiasedbysomeunknownandunsuspecteddefectinthemodelIused.Aftersearchingthetechnicalliterature–notexhaustivelybutfairlywell-theauthorbelievestheanalysisinthisChapteristhemostthroughexplorationofthedoublependulumdonetodate.Itcontainsnewresearchandwillhopefullysuggestideasforadditionalresearch.Notonlyisthefulloperatingmapofthedoublependulumincompletetherootcauseofchaosinthissystemisnotyetunderstood.Howthedoublependulum–atoysystem-behavesisofnopracticalvaluebyitselfunlessitcanshedlightonhowimportantreal-worldsystemsliketheclimate,ecosystem,economy,andgovernmentbehave.Thusasmallattemptwasmadetoseehowthesefindingsmayapplytotheselargersystems.6.2Description:Thedoublependulumisperhapsthesimplestphysicaldevicethatexhibitschaoticbehavior.Itsoftenmentionedbutapparentlyhasreceivedrelativelylittlesystematicstudy.It’salsosomethingyoucaneasilybuildandexperimentwith.Thefirstphotobelowshowstheauthorspendulum,constructedwithaluminumarmsabout8incheslong.Iusedgoodqualityballbearingspress-fitintocarefullyreamedholestominimizefrictionandwobble.Theextraholeswereexperimentstoseewhichproducedthemostinterestingbehavior,butactuallyneitherthelengthofthearmsortheplacementofholesiscritical.

Page 3: Chapter 6 Double Pendulum Analysis (Part 1)

6-3

Theimagebelowshowsasimilardesignandalsothatthekeyvariables,angle1and2,aremeasuredfromthevertical.

Actualdoublependulumsareusuallyconstructedfrommetalarmsbutforcomputingsimplicitysimulationmodelsoftenusepointmassescalledbobs.Thefollowingimageshowsthepathsthebobswouldtakeafterbeingreleasedfromafairlyhighposition.Theactofliftingthemtothatpositioninputsenergyintothesystem,andiftherewerenoairorbearingfrictionthatenergywouldberetainedwithinthesystemandthesamegeneralbehaviorwouldcontinueindefinitely.Withfrictionitdecreaseinintensityuntilthereisnoenergyleftandthearmshangstraightdown.

Page 4: Chapter 6 Double Pendulum Analysis (Part 1)

6-4

Thefollowingtimeexposureshowshowthetraceswouldlookafteralongerperiodoftime.Itwasmadeusinglightsattachedtoanactualdoublependulum.http://www.iontrap.wabash.edu/adlab/papers/F2011_foster_groninger_tang_chaos.pdf

Page 5: Chapter 6 Double Pendulum Analysis (Part 1)

6-5

Thesystem(i.e.thependulum)wasoperatingchaoticallyduringtheperiodthisexposurewastaken.Thegreenlightisatthemainpivotpoint.Theredlightwasattachedatthepivotpointbetweenthetwoarms,andthebluelightwasattachedtotheendoftheouterarm.Notethatthemovementsofthesearmsarephysicallyconstrainedtomoveonlywithinthisspacenomatterhowfastorcomplicatedtheirmovements.Thelargearmwasphysicallytiedtothemainpivotpointsotheonlythingitcoulddowasswingbackandforth,orrotate.Theouterarmwastiedtotheendoftheinnerarmsoitsouterendcouldonlyswingbackandforthrelativetotheredbobandtravelwithinthedonutshapedareaevidentfromtheenvelopeofbluetraces.Thepositionandspeedofthearmsmustatalltimesreflect,indeedcontainormanifest,alltheenergywithinthesystem,whichremainsconstantthroughouttherunifthesystemisfrictionless.Evenifthepartsinthissystembehavechaoticallytheycan’tjustflyawayanyoldway.Thefactthatthemovementswereconfinedisimportantbecausethemovingpartsinmanychaoticsystemsareconfinedeitherbyphysicalbarrierslikewallsor

Page 6: Chapter 6 Double Pendulum Analysis (Part 1)

6-6

attachments.Othersystemslikethesolarsystemareconfinedbytheamountofenergyinthesystemwhichlimitshowfarapartthepartscantravel.Thetotalenergy–potentialpluskinetic-inthissystemmustbeexpressedintheheightandspeedofthetwoarms.Inotherwordsifonemeasuredtheheightofthearmsabovethelowestpointtheycouldreach(withinthesystem,notrelativetothefloororearthscenter)onecouldcomputetheirgravitationalpotentialenergyatanypointintime.Hangingstraightdownitszero.Knowingtheirspeedwouldallowonetocomputetheirkineticenergy.ThesumofGPEandKEwouldremainconstantinafrictionlesssystem.Howeveritoscillatesbetweenitsdifferentforms,namelyPEandKE,anditoscillatesbackandforthbetweenthetwodifferentbobsorarms.Occasionallyalloralmostallthetotalenergyinthesystemcanmomentarilyconcentrateinjustoneboborarm.Thisisakeypoint.Thebestwaytoseechaosinactionistoshowarealoneinoperation.Trythissite:http://video.mit.edu/watch/double-pendulum-6392/Thissimulationvideoshowsoneoftheabovetracesbeingdrawn:http://scienceworld.wolfram.com/physics/DoublePendulum.htmlHereisanothergooddblpendvideo:http://isites.harvard.edu/icb/icb.do?keyword=k16940&pageid=icb.page80864&pageContentId=icb.pagecontent278143&state=maximize&view=view.do&viewParam_name=indepth.html#a_icb_pagecontent278143Whattowatchfor:Theeasiestaspectsofchaoticbehaviortoobservebywatchingthependuluminactionarereversalsinthesmallpendulumsdirectionofrotation,orit’stransitionsbetweenswingingandrotating.Otherparameterssuchasangles,elevation,andratesofrotationarealsochangingbuttheyhappensofasttheyarenotreadilyobserved.Thekeypointisthatallthesechangesinbehaviorhappeninaseeminglyrandommannerwhenthesystemoperateschaotically.SometimesthesmallpendulumrotatesXtimesbeforechangesdirectionorstopsrotating,othertimesitsYtimes.Whenwatchingthedoublependulumonefeelsthateacharmistryingtodoitsownthing,namelyswinglikeasimplependulumfollowingitsownnaturalrhythm.Unfortunatelytheotherarmisn’tquiteinsyncandcontinuestodisturbthatrhythm.Sometimes,likewell-timedpushesonachild’sswing,armAboostswhatArmBistryingtodonaturally.Othertimesthepushesorpullsareoutofsyncandtheforcesfighteachother.Often–anditsbestobservedwhentheupperarmisswingingbackandforth-itlooksliketheinnerarmisslingingorwhippingtheouterarmaround.Thuswetendtoseeashortseriesoforderlybackandforthswingswheretheouterarmswingsabitmoreeachtimeuntilitfinallygainsenoughenergytomakeacompleterevolution.Thereisafeelingthatenergyissurgingbackandforth(itis)insomewavelikemanner.Theideathattheintensityofonepartsoscillationis

Page 7: Chapter 6 Double Pendulum Analysis (Part 1)

6-7

pumpedupbytugsorpullsfromanotherpartuntilthereisadrasticchangeinbehaviorseemskeytounderstandingchaos.Whenrunningatfairlylowenergynotehowtheinnerarmisoftenhangingnearlystraightdownandismotionless–andthuscontainslittleenergy-whentheouterarmswingsoverthetop.Thisindicatestheinnerarmhastransferreditsenergytotheouterarm.Behaviorofauthorsownpendulum:IusedmyowndoublependulumtoexperimentwithhigherenergylevelsthantheJavasimulationsreportedthroughoutthisbookcouldaccommodate.

***insertavideoofitoperatingatlowenergylikemyiPhotoclip#15Iinputthesehighenergiesbyholdingthetwoarmsinlineandswingingthemrapidlyaroundsotheylookedlikeawhirlingairplanepropeller.Thistookfarmoreforceonmypartthansimplyholdingthemverticalbeforerelease.ThisindicatingIwasdefinitelyimputingmoreenergyandthatthesystemhadtocontainormanifestthatenergyassoonasIreleasedit.ItwasmostlyintheformofKEduetorapidrotation.Inthiscasetheinitialwhirlingmotionwasquiteconsistentandorderlyshowingthisparticularsystemcanbehaveinansomewhatperiodicmannerathighenergylevel.Itcouldnotbedeterminedifthishighenergy,propeller-like,behaviorwaschaoticornotsinceminorwobblesinthearmscouldnotbedetectedifindeedtheyexisted.Howeverastherotationslowedduetofrictionbehaviordegeneratedintoamoreviolentchaoticmotionthananyseeninthesimulations.Asfurtherenergywaslosttheouterpendulumnolongermadeoccasionalrevolutionsandeventuallyendedupswinginggentlybackandforthnearitslowpoint.

Alinktovideosofmypendulumdoingthisisfoundat*******((needtoincludeoflinktomyclip#17whereitspinslikeapropellerandclip#18wheretheouterarmspinswhiletheinneronedoesalso.Thistookevenmoreenergy.))

Ialsotriedaddingevenmoreenergybycausingtheouterarmtorotaterapidlyrelativetotheinnerarmwhileswingingthewholeassemblyintorapidrotationaroundthefixedpivot.Thistookquiteabitofphysicalarmenergyonmypart.Theresultingactionwasfastandviolentwiththewholeassemblyrotating.Howevertheouterarmquicklystartedbehaving“chaotically”bywhirlingintermittently.Itthendegeneratedintothenormalchaoticbehaviorseeninmostvideos.Eventhoughthistrialverylikelyinducedmoreenergythantheprioritsbehaviorwasneverorderly,orsoitseemed.Youhavenowobservedactualrealworldbehavior.IntheanalysissectionofthischapterIwillattempttodrawfurtherinsightsintothebehaviorofthedoublependulum.

Page 8: Chapter 6 Double Pendulum Analysis (Part 1)

6-8

6.3TypesofBehaviorTherearefourvariablesthatchangeasthedoublependulummoves:theangleofeachbobrelativetoitsstraightdownposition,whichisproportionaltotheheightofthebobanditspotentialenergy,andthekineticenergyofeachbob,whichisproportionaltoitsspeed.Behaviorishowthesevariableschangeandinteract.Basicallythebobsorarmswanttomovelikesimplependulums,buttheyaren’tusuallysynchronizedsotheypushandpulloneachother.Inotherwordstheydisturbeachother.Behaviorwillbedescribedinthreewaysduringthesesimulations.First,bywatchingthetraceinspaceleftbythebobsintheupperplot.Watchingitonecanalmostvisualizehowtheyalternatelyaccelerateorretardeachotherandslingeachotherabout.Thatdancecanbeseen.Secondweplotawaveformshowinghowoneofthevariables,usuallyangle2,changesovertime.Welookforevidencethatthewaveformduringsometimeintervalorperiodwillrepeattimeaftertimeduringsubsequenttimeintervalsofthesameduration.ThirdwelookatthepatternproducedbyplottingonevariablelikeKE2againstanotherlikeangle1.Doesthatpatternrepeattimeaftertimesooneplotliesatoptheother?Thisplotisatwodimensionalviewofwhatisreallya4-dimensionalphasespaceportrait.Itcanbecomparedwith2-Dviewsofthe3-DLorenzbutterfly.Periodicbehavior:Acommonthemeintheliteratureassertsthatatleastsomedynamicalsystemsbehaveperiodicallywhenfunctioningatlowenergy,butbecomechaoticathighenergy.Whenoperatingperiodically–inthetruesenseofthatword-thebehaviorduringoneperiodoftimeisexactlyrepeatedduringthesecondandallsubsequentequallengthperiodsoftime.Ifwetrackedsomevariablelikethespeedofonearminthependulumitwouldtracesomesortofwaveformandthatwaveformwouldrepeat.Everypeakwouldbethesameheight,everyvalleytheexactsamedepth.Inotherwordsthevalueswouldrepeat.Definitionsofperiodicandaperiodicdonotspecifywhetherabehaviororwaveformmustbeexactlythesametobecalledperiodicorwhetherapproximatelythesameisgoodenough.Thisisafaultintheliteratureandsomeauthorsseemcasualabouttheterm.Thiscreatesaproblemintryingtodescribehowthedoublependulumactuallybehavesandrequiredtheauthortocreatetheterms“patternperiodic”(whichIusedearly-on)and“perfectlyperiodic”whichIusedlater.Thetechnicalliteratureoftenusestheterm“quasi-periodic”tolabelabehaviorthatissomewherebetweenperfectlyperiodicandchaotic.Iuseittoo.Herearesomerelateddefinitions:

Periodic:

Page 9: Chapter 6 Double Pendulum Analysis (Part 1)

6-9

Mathematics.(ofafunction)havingagraphthatrepeatsafterafixedinterval(period)oftheindependentvariable.From:http://dictionary.reference.com/browse/periodic

Aperiodic:Theoppositeofaperiodicsignalisanaperiodicsignal.Anaperiodicfunctionneverrepeats,althoughtechnicallyanaperiodicfunctioncanbeconsideredlikeaperiodicfunctionwithaninfiniteperiod.From:https://en.wikibooks.org/wiki/Signals_and_Systems/Aperiodic_Signals2.Physics.oforpertainingtovibrationsoroscillationswithnoapparentperiod.From:http://www.thefreedictionary.com/aperiodic1:ofirregularoccurrence:notperiodic<aperiodicfloods>From:http://www.merriam-webster.com/dictionary/aperiodic

Imainlyusedplotsoftwovariables,KE2versusa1,tojudgewhetheragivenbehaviorwasperfectlyperiodicorquasi-periodic.WhenplottedtogethertheyformwhatIcalla“partialphasespaceplot”.Itwouldtakefour-dimensionstoplotacompletephasespaceplot.IftheKE2/a1patternwasretracedexactlytimeaftertimesoastoproduceasinglelineitindicatesthatthebehaviorwasperfectlyperiodic.Itdoesn’tmatterhowcomplexthatpatternit,justthatitrepeatexactly.Ifitdidnotrepeatexactly,justapproximately,Ilabeleditquasi-periodic.Belowarethreeslidescomparingperfectlyperiodic,quasi-periodicandchaoticbehavior.Slide95showsperfectlyperiodicbehaviorwhichrequiresthetracetofollowathinlineasitdrawsthepatterntimeaftertime.Notethesymmetryintheplotatright.Notealsothatthebehaviorduringanygivenperiodcanbecomplexcomprisedofdifferentheightandshapedpeaks.Ilikenittoadancewithasequenceofstepsthatrepeatexactlytimeaftertime.

Page 10: Chapter 6 Double Pendulum Analysis (Part 1)

6-10

Slide96showswhatquasi-periodicbehaviorlookslike.Theimagesatleftshowhowthepartialphasespaceplotdevelopsovertime.Thepatternseemstodrift.Thesamegeneraldancemovesrepeatbutthestepsarenotthesamesizeeachtime.Inotherwordsthevaluesareslightlydifferenteachtimealthoughthegeneralpatternstaysmuchthesame.Sometimesthevaluesdriftsoastofilltheentireenvelopeasshown.Becausethepatternsdon’toverlayexactly,northevaluesrepeatexactly,thisisnotwhatoneshouldcallperiodicbehavior.Quasi-periodicistermIuse.Thetwotopimagesatrightshowhowthequasi-periodicwaveform

Slide95Perfectlyperiodicbehavior

Pa2ernrepeatsaboutevery13secondsoverlayingexactlyduringthis150second-longrun.Behavior(“dance”)duringanyoneperiodcanbecomplex.

Page 11: Chapter 6 Double Pendulum Analysis (Part 1)

6-11

looksovertime.Thereisoftensymmetryinthewaveformapparentasitdrawstherightandlefthandpartsofthepattern.Awaveformthatshowsagenerallyrepeatingbehaviorformsinlong-termplots.Thebottomrightshowsacommonformofquasi-periodicbehaviorwherethetracesremainwithinwell-definedbands.Itistrickytojudgethelengthofa“period”usingtheke2/a1plotswhenthesystemisquasi-periodic.Afterfindingthatthesystemwasnotperfectlyperiodicevenatverylowenergylevelslikesettinga1to0.1degreeitwasjudgedthatthedoublependulum,atleasttheseconfigurations,isnotconsistentlyperiodicatlowenergylevelsanddonotthereforeexperienceperioddoublingasdoestheiconiclogisticsequation.Thuslittleeffortwasspenttryingtoestimateperiods.Itishoweverperfectlyperiodicatsomeparticularenergylevelsandquasi-periodicaboveandbelowthoselevels.

Page 12: Chapter 6 Double Pendulum Analysis (Part 1)

6-12

Slide97showswhatchaoticplotslooklike.Notethatgivenenoughtimesomeformsofquasi-periodicbehaviorfilltheentireenvelopewithtracesjustlikechaosdoes.

Slide96Quasi-periodicbehavior

Page 13: Chapter 6 Double Pendulum Analysis (Part 1)

6-13

Figure90illustratesthedifferencebetweenthewaveformofsimpleperiodicbehaviorandthecomplexwaveformsapplicabletothedoublependulum.Ifbehaviorisperfectlyperiodicanygivenpeak,liketheonemarked“A”willbeexactlythesameheightduringeveryperiod.

Slide97Chao%cbehavior

Le.-handplotsshowhowpar%alphasespaceplotdevelopedover%metofillen%reenvelop.Right-handplotsshowhowKE2waveformlookedoverdifferent%meintervals.Calmsmarkedingreen,spikesinred.

Runsmadewithdefaultvalues:L1,L2,M1,M2=1.a2=0.drag=0

Page 14: Chapter 6 Double Pendulum Analysis (Part 1)

6-14

Ifasystemhasapatternofbehaviorthatrepeatsfromoneperiodtoanotherthenitfuturebehavioranytimeinthefuturecanbepredicted.Ifitwereperfectlyperiodicitsexactvaluesatanyfuturetimecouldbepredicted,assumingweknewtheshapeofthewaveformbasedonhistoricalobservationandknewwherewewereonthatwaveform.Ifit’snearlyperiodicorquasi-periodicthenwemaybeabletopredictthepatternofbehavioraroundsomefuturetimebutnotexactvalues.

Page 15: Chapter 6 Double Pendulum Analysis (Part 1)

6-15

Perioddoubling:Whetherornotthedoublependulumexperiencesperioddoublingcouldnotbeabsolutelyconfirmedordeniedusingthesesimulations,althoughitseemsveryunlikely.Firstisthequestionofwhatperiod-2behaviorwouldlooklike.Woulditbethesamepatternorwaveform–asitwasinthelogisticsequation-butshiftedbackandforthinvalue?Ifsothetracewouldpresumablytraceonenicepatternonthefirstpass,anoffsetpatternonthesecondpass,andthenreturnonthethirdpasstotraceapatternexactlyontopofthefirstone.Thiswasneverobserved.Alternatelyperhapsthepatterninperiod-2behaviorwouldbeverydifferenteverywaveformduringeveryothercycle.Thiswasnotobservedeither.Finally,asnotedaboveperiod-1behavior,thatisaperfectlyperiodicwaveformorperfectlyrepeatingphasespaceplot,wasnotfoundatlowenergy.Inotherwordstherewasnoperiod-1behavior.Dramaticevents:Whenthelogisticsequationwasoperatingchaoticallythepopulationwouldsometimeszoomtotheupperlimitandthencrashtonearextinction.Icallthisadramaticevent.LikewisethetraceintheiconicstrangeattractorfortheLorenzequationwouldsuddenlyandrandomlymovefromonewingofthebutterflytotheother.AndtheLorenzwaterwheelwillsuddenlyreverseorstartspinning.Thesearedramaticqualitativechangesinbehavior.Thedoublependulumalsoexperiencessuchdramaticeventswhenchaotic.Atrelativelylowenergytheupperbobswingsbackandforthandthelowerbobswingsbackandforth.Nothingdramatic.Thatsamegeneralbehaviorrepeatstimeaftertime.Ontheotherhandathighenergythelowerbobrandomlyswingsrightoverthetopratherthanfallingback.Ilabelthesegoingover-the-top(OTT)eventsasdramaticeventssincetheymarkadistinctchangeinbehavior.ItturnsoutthatthepresenceorabsenceofsuchdramaticOTTeventsisnotacriterionfordecidingwhetherthesystemischaoticornot.Itdoesnotmarkthedemarkorthresholdbetweenquasi-periodicandchaoticbehaviorinthedoublependulum.ThesystemcanbecomechaoticwithlessenergythanittakestoenableOTTeventstooccur.Howeveriftheydooccurthesystemischaotic,atleastintheterritoryI’veexplored.Mostpopularvideosaboutthedoublependulumonlyshowthisgoing-over-the-toporspinningbehaviorbecauseitsquitedramatic,andofcourselookschaoticinanyone’seyes.6.4DoublependulumsimulationmodelIntroductiontosimulationmodel:BehaviorofthedoublependulumwasexploredusingaJavasimulationmodelwrittenbyProfessorDoolingattheUniversityofNorthCarolinaatPembroke.It’sfoundat:http://www2.uncp.edu/home/dooling/applets/double_pen.files/tom/models/doublepen.html.ThesecondmodelIusedafewtimeswaswrittenbyEricNewmanandisfoundat:http://www.myphysicslab.com/dbl_pendulum.html.

Page 16: Chapter 6 Double Pendulum Analysis (Part 1)

6-16

Bothsimulationsusebobssuspendedon“strings”asopposedtothemetalarmsusedinmanyreal-worldlaboratoryversions.Bobsorarmsbehaveaboutthesameway.Digitalcomputermodelscannotanddonotpreciselyreplicaterealworldbehavior.Thereisapossibilitythesemodelscontainsomecomputingartifactthatcausesparametervaluestoslowlydriftwhentheyshouldn’t.ForinstancethetotalenergylevelintheNewmanmodelslowlydropsalthoughthesimulationissupposedlyfrictionless.Thismustbekeptinmindwhenviewingplotswherethetracecompletesonepatternandthendriftstodrawasimilarbutoffsetpattern.Itcouldbeacomputingartifact.Neverthelessit’sfeltthatmostifnotalldriftintheDoolingmodelreflectsactualbehaviorandnotcomputingartifactssincetherewasnodriftinseveralruns.Ifthemodelweredriftingtherewouldn’tbesuchexceptions.Bothmodelsareeasilyrunonahomecomputeranditshighlyrecommendedyoudososincethereisnosubstituteforwatchingthebobsmoveinslowspeedandwatchingthewaveformsandpatternsevolveovertime.Notethatalldynamicvariables-suchasangle,position,PE,orKEinteractinthesedynamicsystems.Ifonebehavesperiodically,orchaotically,theotherswillalldolikewise.Thusitsnotthatimportantwhichoneoronesareplottedtodeterminewhetherasystemisperiodicorchaotic.Thekeythingistoobservethewaveformsandpatternsandtheextenttowhichtheyrepeatperiodically.OneminordeficiencywiththeDoolingmodelisthatthelengthsandmassescannotbesetaccuratelywiththeslideranywherebetweenitsfarrightandfarleftpositions.ForinstanceLength2canpresumablybesetaccuratelyto0.5byslidingtheslidertoitsfarleftandaccuratelyto1.0byslidingitfarright.Howeverbetweentheseextremesthenumericalvalueshowndoesnotchangeaccuratelyastheslidermovessinceitsroundedoff.Thelengthofthearmdoeschangeinthemodelasthesliderischangedbutthevaluesetintheboxisroundedoffsodoesnotrepresentanexactvalue.Thusonlytheonlyaccuratewaytousethemodelistohavethesliderfarrightorfarleftsoforinstancem2iseither0.5or1.0.Thereareseveralthingswewanttoexplore:

1) Arethemovementsofthedoublependulumeverperiodicmeaningthatacertainpatternofbehaviorrepeatsoverandoveragain.Ifsowhatdoesbehaviorduringthisperiodlooklike?WesawperiodicbehaviorintheLogisticsequationatlow“r”valuesandwonderifthedoublependulumalsobehavesperiodicallyatlowenergylevels.

2) Whatisthenatureofbehavior,especiallywithinoneperiod?3) Doesperioddoublingoccur?ItdidintheLogisticsequationandexpertssay

itoccursinothersystemsaswell.4) Howdoesenergyinthissystemmovebetweenitsdifferentforms?5) Whatdoeschaoticbehaviorlooklikeinthedoublependulum?

Page 17: Chapter 6 Double Pendulum Analysis (Part 1)

6-17

6) Doesobservationofpastbehaviorofferanyhelpinpredictingfuturebehavior,bothwhenthesystemisoperatingperiodicallyandwhenitisoperatingchaotically?

Otherthingsofpotentialnotecanbegleanedformthesesimulations.Forinstancethewayvariableschangeisneverabruptorinstantaneousbutratherhappensinafluidlikemanner.Thisisbecausethependulumbobshaveinertiaandittakestimetoacceleratethem.Ittakestimetoconvertpotentialenergyintomovementandkineticenergy.ThemainthreadIwantedtoexplorewashowthetotalamountofenergywithinthesystemeffectsitsbehavior.Thusaseriesofrunsweremadewitheachhavingmoreenergythantheprior.Thiswasachievedbyreleasingthependulumatalowangle,namely5degrees,forthefirstrun.Angle1wasthenincreaseduntilthelastrunsweremadeat170degrees.Thebluebobwasalmostalwaysstartedata2=0,thatishangingstraightdown.Detailsofmodelruns:Allrunsassumedafrictionlesspendulumsothetotalenergylevelinthesystemremainedconstantduringtherun.Theoreticallywecouldtrackanyvariable–likea1,a2,KE1orKE2-overtimetolearnifthesystemisoperatingperiodicallyornot.Howeverwewouldneedtocutandpasteonewaveformatopanothertoseethatclearly.Itseasiertoplotonevariablevs.anotherovertime.Iftheyformapatternthatrepeatsexactlytimeovertimethesystemisperfectlyperiodic.Thepatternmakeduringthesecond,thirdandallsubsequentperiodswillliedirectlyatopthepatternproducedduringthefirstperiodofoperation.GenerallyI’vechosentoplotKE2versusangle1(calledtheta1inthemodel).KE2isproportionaltothespeedoftheouterbob.I’vealsoincludedafewplotsofangle1vs.angle2.Theyconveythesameinformationbutarenotaseasytointerpret.Thependulumarmsorbobsdoacomplexdanceasitwere.Oftenfollowingthesamegeneralpatternbutusuallytakinglargerandlargerorsmallerandsmallerstepseachtime.Asystemwithevenmoremovingpartswouldprobablyhaveanevenmorecomplicateddance.Itsnotnecessarytopaymuchattentiontoallthedifferentparametersbeingmeasuredinthesesimulationsortheirvalues.Wecansimplyfocusoftheshapesofthepatternsorwaveformstogetagoodfeelforhowthedoublependulumbehaves.Notehowtheparameters,whichrepresenttheinitialconditionsorconstants,areset.Mostaredefaultsettingsbutmass2wasgenerallysetto0.5inphase1and1.0inphasetwo.LaterIemphasizethatchangingthevalueofanyparameter,likemass2,canaltertheresultsdramatically.Inshortchangingthesize,shape,massoranyotherphysicalpartofasystemwillchangeitsbehaviorjustlikechangingthe

Page 18: Chapter 6 Double Pendulum Analysis (Part 1)

6-18

lengthofaclockpendulumdoes.Achangeinoneoftheseso-calledconstantsmightshiftitfromperiodictochaoticoperation.Whyisthisworthmention?LetssupposeweweremodelingthecirculationoftheGulfstream.Possiblyachangeinoneoftheconstants,likeaveragewatertemperatureorsalinity,orevenmaybewaterdepthoversomesubsurfaceobstructionwouldchangethegeneralpatternofcirculationfromperiodictochaotic,orfromonepatternofcirculationtoanother.Couldsomethingthatishappingtipitoverthethresholdintochaos?WeknowacceleratingmeltingoftheGreenlandicesheetischangingthesalinityofthenorthAtlantic.Ihavenospecificknowledgeaboutthissituationandthisispurelyahypotheticalandoversimplifiednotion.Wesimplynotethatthebehaviorofasystemisconstrainedbyitsphysicalbarriersordimensionsaswellasbytotalenergy.Itmustoperatewithinthoselimits.Changethoselimitsandthebehaviorwillchangeinresponse.