返回 死在火星上 首页

上一页 目录 下一页

对火星轨道变化问题的最后解释[1/2页]

天才一秒记住本站地址:[58小说]https://m.58books.com最快更新!无广告!

    作者君在作品相关中已经解释过这个问题,并在此列出相关参考文献中的一篇开源论文。
    以下是文章内容:
    LongtermintegrationsandstabilityofplanetaryorbitsinourSolarsystem
    Abstract
    Wepresenttheresultsofverylongtermnumericalintegrationsofplanetaryorbitalmotionsover109yrtimespansincludingallnineplanets.Aquickinspectionofournumericaldatashowsthattheplanetarymotion,atleastinoursimpledynamicalmodel,seemstobequitestableevenoverthisverylongtimespan.Acloserlookatthelowestfrequencyoscillationsusingalowpassfiltershowsusthepotentiallydiffusivecharacterofterrestrialplanetarymotion,especiallythatofMercury.ThebehaviouroftheeccentricityofMercuryinourintegrationsisqualitativelysimilartotheresultsfromJacquesLaskar'ssecularperturbationtheory(e.g.emax~0.35over~±4Gyr).However,therearenoapparentsecularincreasesofeccentricityorinclinationinanyorbitalelementsoftheplanets,whichmayberevealedbystilllongertermnumericalintegrations.Wehavealsoperformedacoupleoftrialintegrationsincludingmotionsoftheouterfiveplanetsoverthedurationof±5×1010yr.TheresultindicatesthatthethreemajorresonancesintheNeptune?Plutosystemhavebeenmaintainedoverthe1011yrtimespan.
    1Introduction
    1.1Definitionoftheproblem
    ThequestionofthestabilityofourSolarsystemhasbeendebatedoverseveralhundredyears,sincetheeraofNewton.Theproblemhasattractedmanyfamousmathematiciansovertheyearsandhasplayedacentralroleinthedevelopmentofnonlineardynamicsandchaostheory.However,wedonotyethaveadefiniteanswertothequestionofwhetherourSolarsystemisstableornot.Thisispartlyaresultofthefactthatthedefinitionoftheterm‘stabilityisvaguewhenitisusedinrelationtotheproblemofplanetarymotionintheSolarsystem.Actuallyitisnoteasytogiveaclear,rigorousandphysicallymeaningfuldefinitionofthestabilityofourSolarsystem.
    Amongmanydefinitionsofstability,hereweadopttheHilldefinition(Gladman1993):actuallythisisnotadefinitionofstability,butofinstability.Wedefineasystemasbecomingunstablewhenacloseencounteroccurssomewhereinthesystem,startingfromacertaininitialconfiguration(Chambers,Wetherillamp;amp;Boss1996;Itoamp;amp;Tanikawa1999).AsystemisdefinedasexperiencingacloseencounterwhentwobodiesapproachoneanotherwithinanareaofthelargerHillradius.Otherwisethesystemisdefinedasbeingstable.HenceforwardwestatethatourplanetarysystemisdynamicallystableifnocloseencounterhappensduringtheageofourSolarsystem,about±5Gyr.Incidentally,thisdefinitionmaybereplacedbyoneinwhichanoccurrenceofanyorbitalcrossingbetweeneitherofapairofplanetstakesplace.Thisisbecauseweknowfromexperiencethatanorbitalcrossingisverylikelytoleadtoacloseencounterinplanetaryandprotoplanetarysystems(Yoshinaga,Kokuboamp;amp;Makino1999).OfcoursethisstatementcannotbesimplyappliedtosystemswithstableorbitalresonancessuchastheNeptune?Plutosystem.
    1.2Previousstudiesandaimsofthisresearch
    Inadditiontothevaguenessoftheconceptofstability,theplanetsinourSolarsystemshowacharactertypicalofdynamicalchaos(Sussmanamp;amp;Wisdom1988,1992).Thecauseofthischaoticbehaviourisnowpartlyunderstoodasbeingaresultofresonanceoverlapping(Murrayamp;amp;Holman1999;Lecar,Franklinamp;amp;Holman2001).However,itwouldrequireintegratingoveranensembleofplanetarysystemsincludingallnineplanetsforaperiodcoveringseveral10Gyrtothoroughlyunderstandthelongtermevolutionofplanetaryorbits,sincechaoticdynamicalsystemsarecharacterizedbytheirstrongdependenceoninitialconditions.
    Fromthatpointofview,manyofthepreviouslongtermnumericalintegrationsincludedonlytheouterfiveplanets(Sussmanamp;amp;Wisdom1988;Kinoshitaamp;amp;Nakai1996).Thisisbecausetheorbitalperiodsoftheouterplanetsaresomuchlongerthanthoseoftheinnerfourplanetsthatitismucheasiertofollowthesystemforagivenintegrationperiod.Atpresent,thelongestnumericalintegrationspublishedinjournalsarethoseofDuncanamp;amp;Lissauer(1998).Althoughtheirmaintargetwastheeffectofpostmainsequencesolarmasslossonthestabilityofplanetaryorbits,theyperformedmanyintegrationscoveringupto~1011yroftheorbitalmotionsofthefourjovianplanets.TheinitialorbitalelementsandmassesofplanetsarethesameasthoseofourSolarsysteminDuncanamp;amp;Lissauer'spaper,buttheydecreasethemassoftheSungraduallyintheirnumericalexperiments.Thisisbecausetheyconsidertheeffectofpostmainsequencesolarmasslossinthepaper.Consequently,theyfoundthatthecrossingtimescaleofplanetaryorbits,whichcanbeatypicalindicatoroftheinstabilitytimescale,isquitesensitivetotherateofmassdecreaseoftheSun.WhenthemassoftheSunisclosetoitspresentvalue,thejovianplanetsremainstableover1010yr,orperhapslonger.Duncanamp;amp;Lissaueralsoperformedfoursimilarexperimentsontheorbitalmotionofsevenplanets(VenustoNeptune),whichcoveraspanof~109yr.Theirexperimentsonthesevenplanetsarenotyetcomprehensive,butitseemsthattheterrestrialplanetsalsoremainstableduringtheintegrationperiod,maintainingalmostregularoscillations.
    Ontheotherhand,inhisaccuratesemianalyticalsecularperturbationtheory(Laskar1988),Laskarfindsthatlargeandirregularvariationscanappearintheeccentricitiesandinclinationsoftheterrestrialplanets,especiallyofMercuryandMarsonatimescaleofseveral109yr(Laskar1996).TheresultsofLaskar'ssecularperturbationtheoryshouldbeconfirmedandinvestigatedbyfullynumericalintegrations.
    Inthispaperwepresentpreliminaryresultsofsixlongtermnumericalintegrationsonallnineplanetaryorbits,coveringaspanofseveral109yr,andoftwootherintegrationscoveringaspanof±5×1010yr.Thetotalelapsedtimeforallintegrationsismorethan5yr,usingseveraldedicatedPCsandworkstations.OneofthefundamentalconclusionsofourlongtermintegrationsisthatSolarsystemplanetarymotionseemstobestableintermsoftheHillstabilitymentionedabove,atleastoveratimespanof±4Gyr.Actually,inournumericalintegrationsthesystemwasfarmorestablethanwhatisdefinedbytheHillstabilitycriterion:notonlydidnocloseencounterhappenduringtheintegrationperiod,butalsoalltheplanetaryorbitalelementshavebeenconfinedinanarrowregionbothintimeandfrequencydomain,thoughplanetarymotionsarestochastic.Sincethepurposeofthispaperistoexhibitandoverviewtheresultsofourlongtermnumericalintegrations,weshowtypicalexamplefiguresasevidenceoftheverylongtermstabilityofSolarsystemplanetarymotion.Forreaderswhohavemorespecificanddeeperinterestsinournumericalresults,wehavepreparedawebpage(access),whereweshowraworbitalelements,theirlowpassfilteredresults,variationofDelaunayelementsandangularmomentumdeficit,andresultsofoursimpletime?frequencyanalysisonallofourintegrations.
    InSection2webrieflyexplainourdynamicalmodel,numericalmethodandinitialconditionsusedinourintegrations.Section3isdevotedtoadescriptionofthequickresultsofthenumericalintegrations.VerylongtermstabilityofSolarsystemplanetarymotionisapparentbothinplanetarypositionsandorbitalelements.Aroughestimationofnumericalerrorsisalsogiven.Section4goesontoadiscussionofthelongesttermvariationofplanetaryorbitsusingalowpassfilterandincludesadiscussionofangularmomentumdeficit.InSection5,wepresentasetofnumericalintegrationsfortheouterfiveplanetsthatspans±5×1010yr.InSection6wealsodiscussthelongtermstabilityoftheplanetarymotionanditspossiblecause.
    2Descriptionofthenumericalintegrations
    (本部分涉及比较复杂的积分计算,作者君就不贴上来了,贴上来了起点也不一定能成功显示。)
    2.3Numericalmethod
    WeutilizeasecondorderWisdom?Holmansymplecticmapasourmainintegrationmethod(Wisdomamp;amp;Holman1991;Kinoshita,Yoshidaamp;amp;Nakai1991)withaspecialstartupproceduretoreducethetruncationerrorofanglevariables,‘warmstart(Sahaamp;amp;Tremaine1992,1994).
    Thestepsizeforthenumericalintegrationsis8dthroughoutallintegrationsofthenineplanets(N±1,2,3),whichisabout1\/11oftheorbitalperiodoftheinnermostplanet(Mercury).Asforthedeterminationofstepsize,wepartlyfollowthepreviousnumericalintegrationofallnineplanetsinSussmanamp;amp;Wisdom(1988,7.2d)andSahaamp;amp;Tremaine(1994,225\/32d).Weroundedthedecimalpartofthetheirstepsizesto8tomakethestepsizeamultipleof2inordertoreducetheaccumulationofroundofferrorinthecomputationprocesses.Inrelationtothis,Wisdomamp;amp;Holman(1991)performednumericalintegrationsoftheouterfiveplanetaryorbitsusingthesymplecticmapwithastepsizeof400d,1\/10.83oftheorbitalperiodofJupiter.Theirresultseemstobeaccurateenough,whichpartlyjustifiesourmethodofdeterminingthestepsize.However,sincetheeccentricityofJupiter(~0.05)ismuchsmallerthanthatofMercury(~0.2),weneedsomecarewhenwecomparetheseintegrationssimplyintermsofstepsizes.
    Intheintegrationoftheouterfiveplanets(F±),wefixedthestepsizeat400d.
    WeadoptGauss'fandgfunctionsinthesymplecticmaptogetherwiththethirdorderHalleymethod(Danby1992)asasolverforKeplerequations.ThenumberofmaximumiterationswesetinHalley'smethodis15,buttheyneverreachedthemaximuminanyofourintegrations.
    Theintervalofthedataoutputis200000d(~547yr)forthecalculationsofallnineplanets(N±1,2,3),andabout8000000d(~21903yr)fortheintegrationoftheouterfiveplanets(F±).
    Althoughnooutputfilteringwasdonewhenthenumericalintegrationswereinprocess,weappliedalowpassfiltertotheraworbitaldataafterwehadcompletedallthecalculations.SeeSection4.1formoredetail.
    2.4Errorestimation
    2.4.1Relativeerrorsintotalenergyandangularmomentum
    Accordingtooneofthebasicpropertiesofsymplecticintegrators,whichconservethephysicallyconservativequantitieswell(totalorbitalenergyandangularmomentum),ourlongtermnumericalintegrationsseemtohavebeenperformedwithverysmallerrors.Theaveragedrelativeerrorsoftotalenergy(~10?9)andoftotalangularmomentum(~10?11)haveremainednearlyconstantthroughouttheintegrationperiod(Fig.1).Thespecialstartupprocedure,warmstart,wouldhavereducedtheaveragedrelativeerrorintotalenergybyaboutoneorderofmagnitudeormore.
    RelativenumericalerrorofthetotalangularmomentumδA\/A0andthetotalenergyδE\/E0inournumericalintegrationsN±1,2,3,whereδEandδAaretheabsolutechangeofthetotalenergyandtotalangularmomentum,respectively,andE0andA0aretheirinitialvalues.ThehorizontalunitisGyr.
    Notethatdifferentoperatingsystems,differentmathematicallibraries,anddifferenthardwarearchitecturesresultindifferentnumericalerrors,throughthevariationsinroundofferrorhandlingandnumericalalgorithms.IntheupperpanelofFig.1,wecanrecognizethissituationinthesecularnumericalerrorinthetotalangularmomentum,whichshouldberigorouslypreserveduptomachineεprecision.
    2.4.2Errorinplanetarylongitudes
    SincethesymplecticmapspreservetotalenergyandtotalangularmomentumofNbodydynamicalsystemsinherentlywell,thedegreeoftheirpreservationmaynotbeagoodmeasureoftheaccuracyofnumericalintegrations,especiallyasameasureofthepositionalerrorofplanets,i.e.theerrorinplanetarylongitudes.Toestimatethenumericalerrorintheplanetarylongitudes,weperformedthefollowingprocedures.Wecomparedtheresultofourmainlongtermintegrationswithsometestintegrations,whichspanmuchshorterperiodsbutwithmuchhigheraccuracythanthemainintegrations.Forthispurpose,weperformedamuchmoreaccurateintegrationwithastepsizeof0.125d(1\/64ofthemainintegrations)spanning3×105yr,startingwiththesameinitialconditionsasintheN?1integration.Weconsiderthatthistestintegrationprovidesuswitha‘pseudotruesolutionofplanetaryorbitalevolution.Next,wecomparethetestintegrationwiththemainintegration,N?1.Fortheperiodof3×105yr,weseeadifferenceinmeananomaliesoftheEarthbetweenthetwointegrationsof~0.52°(inthecaseoftheN?1integration).Thisdifferencecanbeextrapolatedtothevalue~8700°,about25rotationsofEarthafter5Gyr,sincetheerroroflongitudesincreaseslinearlywithtimeinthesymplecticmap.Similarly,thelongitudeerrorofPlutocanbeestimatedas~12°.ThisvalueforPlutoismuchbetterthantheresultinKinoshitaamp;amp;Nakai(1996)wherethedifferenceisestimatedas~60°.
    3Numericalresults?I.Glanceattherawdata
    Inthissectionwebrieflyreviewthelongtermstabilityofplanetaryorbitalmotionthroughsomesnapshotsofrawnumericaldata.Theorbitalmotionofplanetsindicateslongtermstabilityinallofournumericalintegrations:noorbitalcrossingsnorcloseencountersbetweenanypairofplanetstookplace.
    3.1Generaldescriptionofthestabilityofplanetaryorbits
    First,webrieflylookatthegeneralcharacterofthelongtermstabilityofplanetaryorbits.Ourinterestherefocusesparticularlyontheinnerfourterrestrialplanetsforwhichtheorbitaltimescalesaremuchshorterthanthoseoftheouterfiveplanets.AswecanseeclearlyfromtheplanarorbitalconfigurationsshowninFigs2and3,orbitalpositionsoftheterrestrialplanetsdifferlittlebetweentheinitialandfinalpartofeachnumericalintegration,whichspansseveralGyr.Thesolidlinesdenotingthepresentorbitsoftheplanetsliealmostwithintheswarmofdotseveninthefinalpartofintegrations(b)and(d).Thisindicatesthatthroughouttheentireintegrationperiodthealmostregularvariationsofplanetaryorbitalmotionremainnearlythesameastheyareatpresent.
    Verticalviewofthefourinnerplanetaryorbits(fromthezaxisdirection)attheinitialandfinalpartsoftheintegrationsN±1.Theaxesunitsareau.ThexyplaneissettotheinvariantplaneofSolarsystemtotalangularmomentum.(a)TheinitialpartofN+1(t=0to0.0547×109yr).(b)ThefinalpartofN+1(t=4.9339×108to4.9886×109yr).(c)TheinitialpartofN?1(t=0to?0.0547×109yr).(d)ThefinalpartofN?1(t=?3.9180×109to?3.9727×109yr).Ineachpanel,atotalof23684pointsareplottedwithanintervalofabout2190yrover5.47×107yr.Solidlinesineachpaneldenotethepresentorbitsofthefourterrestrialplanets(takenfromDE245).
    ThevariationofeccentricitiesandorbitalinclinationsfortheinnerfourplanetsintheinitialandfinalpartoftheintegrationN+1isshowninFig.4.Asexpected,thecharacterofthevariationofplanetaryorbitalelementsdoesnotdiffersignificantlybetweentheinitialandfinalpartofeachintegration,atleastforVenus,EarthandMars.The

对火星轨道变化问题的最后解释[1/2页]

『加入书签,方便阅读』

上一页 目录 下一页