Springer Sete Bee Fee Bored sage onto ube Stl Be Sot Coe Tobe eats in Applied Mathematics 3 Eee FJoho dened ‘TE. Marston M, Gols Weer deur GoonTexts in Applied Mathematics roc: tmodasonw Applied Maer ‘az Insnacn 9 App Xonitese yeaa Syme and ne Inna Osis wb Bee ‘eohandive eel Equons: A Dyarcl Sysens Approx ‘rn Deer! Eunos. Sag Nebr Cone! Tees: Deere Fae Dimes ‘Pore: Divert Egat sed Dyan Sten 2a Pipa s Cone a eral Eons operon Psi Maes nese athe Life ences Sam cn Te he Renard Rogers: & Pest Gadute Course Pata Dferental Equcans anc Growth ian Perorens Snenate Framewats "Repent, rene Sct Tae Maberatat Theory of ie lent Mets onde vr, Conuren Soarfe Computing Maren atu niet Meshes ye eon Dien Egon © Oyama Sem Apress gee Diners Sens 19. aplanGlerUntensnting Noninst Dyn 29. Howas.nmesctene Fomaton Mee. 272 Thomas Numer Pua iets Egon Fn Diewnce 23. Tenor Paria fetal eet: Boo Te. BA Mere tain oe Toy of Som Jack KJHale Hiiseyin Kocak Dynamics and Bifurcations Win 314 teas é SpringerJak K Hale Hori Kosa ‘Shot of aera Defame of Matbemscane ‘eave ot Techoobsy ‘Compa Sener ose orl able, FEO leat asechede eg HES ames tore AE. Marden 1 sitonsen Cait and Dynamical Syems, 10484 Bran a Ape? Satemats EattomisstiuteasTehwoegs Brown Unent Pasadena, Ca 31128 Proves, RLOI2 es ce Deparment of Mathematics Departs of Appi Manes, Utes of Houeon CeRersac sacar Hanon. 777008 Jey Seuemeime ed 294 ts ‘Soo Hedstere, FRO ‘Conran tet st by Hal Baan, Sate Sea Ca ayo ag Cai Ca ne 92 8 £19 Sore er New Yo ‘ran eae Th non eee reid in in white ‘mma pe ae, 1 ew oN, ‘neice wih a) forma formation stages eae emtcticaanasa, Som Petree sw mr mets So owen bee Sr ‘meet pen dcp ame ade at de, nh Bein, as rm neo agree ofl aio te ake ev ages Selogon ye fas tt Mei yma egy ee 85 Prong oy sean he mre fe Frere iec RE Sy Sn fre 9427654 (Comms it ping.) 5 oam-7e- Sra Vrs Sr Yor. Bara Hott ‘pu borne SpugeVerat Se Nate SreYork SPERGreeting “Thank vou for opening xr book. lasde you wl id ides sad exampes about the gpometey af dvaanies and iu . eons of ordinary difeemal and aiference equations. ) tis a usual Bock i both content and se let ‘a explain how jt evel fam our courses in tae Dive toa of Applied Mathematics at Brows Univesity dung ‘thro year period, and cae ato bene "The subject of ifeenial and difernce aquatons, alle dynaria! ‘eters, a old aod mush-honoredehsper in seen, one which ge ‘ated In ppled lds such a cota echancs, salipear oscilatons, oid dynamics Over ibe nti ara el of the efforts of sin fists and matheraicisns alike an ataoctve and far-reaching theory bas fsnecged, In recent yar, due primar 9 the prolferation of computer. ‘dynamical syste ho once more tures to aot i epson with perhape = more mature outlook Cure, the lve of excitement aad city, aor only onthe mathematical fom but in als allied Bes Stlezaing. unique Teeth aim of oir book to prove © modest fous ‘ation for taking partie certain tooerial and practical facets ofthese ‘acing development “The subjet of dyamical ystems avast one aot csly accesible to undergraduate and being graduate medene inmates or sense {od engiosring. Man of the avlae books and expository narratives ‘ther rqure extensive mathematialpepaaton. ora aoc deed to ‘be psod as textbooks, Tris with the dene fll this vid tbat e have ‘writen the presect book 1 both our convision and oer experience that many of the fun damental ens of dynaice and bifurcations can be explained ia a sim ple sting, one thats mathemati iigntfl yet devo of extensiveva recy fovmaism, Accordingly, we Dave opted in the preset book to proceed ty loecdnenronal dyaanss! systems. We wil momenta give Piet soramaay of some ofthe rena topes of oi hook ope which wrest ontins seme cecil torte Ifsou area Dogncing cadet of daa {Snvhowever se amured tat pres mathematical definitions ofall thee tema, a wile ample enzations ofthe deramieal phenomena in speci ‘Suetins wll unfold a You cur te pages. ‘Equations in dioensios one, cone and pe half and two consitute the majorgy of the text Lae neato hundred pages are deed tosalar equations where espte thes spit? and apparel nang of the eontemprss tae of ur subjet ate alread vee. We ‘Memoasteut. i parler chat th Yaienotons of tate and urea ions of sector fa ar onal raplained for sala autonomies euatons— UImmension onebecause tts How are deternined om the equisum points, We ao explore how mumerical solutions of such equations led oacalae woaps, and show coe of the “atonal,” lke pofand and ating, that fac aree mies aumeriel approumario is poot—pectd ‘Boating bifurcation. chavs te. We then arm tothe dyanms and bic Turatons of pero slions of nonautonomous equntians with period ouiclents —dension oe adobe half—whete sear maps apes nat ‘rally as Poincare snaps. In oer dscusion of the stability of periodic so Tntions of such equosins, we demonstrate hw one natal’ encounters ‘tmnencary but eset eas fom the transformation teary of difecen- ial equations--normal frm theory. These des, presented inthe contest tf salar equations, and nue inporanty. the plosphicaL outlook of the fubject cat thw iene conse. recur Ereqoeatyinlter chapters, wih & {ow techies embelishmens 2 ‘ho next proceed to iavesigats the dynamics of planar autonomous ‘equations-dmenson tso~vhere. in addition 9 equilibria, new dynam ‘al bchavior such as pee and bomoetinie obs appears. In stuling the atts of an agua poi, ee couch upon cerain sue topo lomial aspects of linear ster ae well atthe standard theory of Lapuror fnetione The bifreation theory of equilibrium poins of paar equations ides ae to a umber of new ident When. for example, the hfaretion ENb other spin, one i hd marraly to introduce conver manifolds fh the method of Lispunoe Schmidt to tne vrodosin 0 a sear au {ohomous equotion ‘The ober important bfuration Eom an eqlibiar point i toe peiaicorbit~ Poincare Andronor-Hop! infureation—and Re analysis can be dices to tat of a gonantonomous period eaustion. ‘Thove are of course. eter properties of planar dierntl uations thst fe more elbal in ehracter and ence cannot be Invsigted in ers Gf scalar equatios. Among these interesting opis, weave coven £0 Include te Peincart-Seadimon theory of planar limit sets, geomet aot ‘ifurestion of comarvative abd grades stems, abd a discuson of tue Gretng tral stabilty—withy ofcourse, an emphasis on the Meas rather tha on tensive techni deals ‘We subsequell clude an abbreviate: seusion of eta aspees| of eke ther of plana maps, Asi the case ale equations, we explore (one ofthe difeitie smaciated wih gucenieel soltans of diferent onto in light ofthe dynamics and ifrcarions of sich maps, To ind te not ons the reese but al the beware complet of this rope. ‘ne ice compner silations of some oe famous maps "The nal pat ofthe Book costa .ral substi examples in Aimensione ta mad one hal tee ae ut This sec on is more ds sie tia he previ nes: Kis ne Hs a pene desizned Ye provide {ermooth sus i cersin ares of cute “Cee free seatons, Shnangeaccectons chase completely stegsae Hamitotian seems. et For ore deta lst ofthe topicsSee i We ak, Yo are, couse nvted to brome thrgh the Tab of Coneses. 48 900 ps hoe the ces bower, bea ind tht a daatles, stems. Jost parte of mathematics. wile geeral ote eta: tcp a en fal place, one mat ulimatels foe the tsk o naling the dsmaice fr speete equation, expels in applications. Moreover, even forthe ‘ost abstractly inelned, rapping oth =oeiae examples sali prow {ove an vephaceable source of general thecal obsrerns. With As Plilsopi in ithe text ad the eercas ale are incervoven with merous spice diferentnl and diference equations of teoreial and proctica ness. Unfrrmtas unravels ve dynamics of speci equa Toons often turns out to be anally ssewountabe. The ompue ‘alts presen verti. lawever, begining to prove its ult i this purse, Oa this new font ovr favorite sonpucer program i. of cure. BHASER. tn Animater/Simaltor for Drama Systems. accompany ing one of our eater books Difereital ard Diflerence Equations throngh Computer Experiment. Oat student have found PHASER robe a veal ‘pollu co -te the svoanics ia dsmamsl stems a odo some of ‘heir amiguaneate we. too. ued fo proste ean of te Tstratons fr tour book” 2 ‘Denaclcal systems ia vast ad vbrar are. We hope that Dye & Byjuretvns wll rose your Steet cation theory suey hat you wil be ind to explore this eating subject further using, of Course. our ater favorite Book Method Bifration Theory We wocid like to record sn slong out grata co chose wh have contributed unselfishly o the ralation of cur book. In parte. he Sothusints parpation of our seadets—aUvely group consisting of ‘Sengradnare and graduate scents of pure and applied mathematics, and ‘scence tnd engineeingbelpedconscesbt fm fing our ides and ‘sng reste bounds for our own entboiaan. Critical ceasing of the text and sigh auggetons by Nathan Chafee. Brian Coomes, PhilpDavis, Robert Geils, Henry Hamman. bert Harunt-Alaree, Arnold Hort Sain Korak, Nanev Lather. Alan Laser, Konstantin Mcheliow. Kenneeh Palmer, Jack Pipi. Plicilo Tsoras. Natalia Sternberg, Micha Wotton, Gastato Zampier ad Ler Zi save ben italia ehapig ‘the manuscript. We are equally gatehl =o Hall Butta for his clase ar work, and Pred Bisshopp. Kim Foster-Casner, Sum Fuleomer. Stat Geman, Donald MeClue. Andrew MossSerz. and Jim Yorke for thle a sistance on toutes de made rend, 2 Hale ane Hixeyn Koga put £291, Drei ils Contents PART I: Dimension One Chapter 1. Sealar Autonomous Equations LL Bxitence and Uniqueness ‘ 12. Goomeery of Flows 3 19. Stablty of Eqans 6 1A, Equations on 8 Cc 2 Chapter 2. Blementary Bifurcations 21, Dependence on Paraeters~ Exams 5 22 The npc Function Theorem a 23, Local Penusbatione Near Ela 2 24. An Exampio ona Cle pe 25. Computing Bifurcation Diagrams 36 28, Equivalence of Flows a (Chapter 8. Sealar Mops SLL Euler's Algorithm and Mops 8 32. Geometry of Seale Maps R 23, Bifurcation of Mowotoe Maps a Te, Peod-doubling Bilsreaion| 7 35, An Example: The Logstic Mapssi Contne PART II: Dimension One and One Half (Chepter 4. Sealar Nonautonomeus Equations LE. General Properties of Sluions 412, Geometry of Pring Bqsneions 443, Periodic Equations on a Cylinder 444. Example of Pode uatons 45, Stability of Pedic Solutions Chapter 5. Bifureation of Periodic Equations Su. Bifurcation: of Poincaré Maps 52. Stability of Noubyperbolie Perozic Solutions 53. Perurbatins of Vector Fields Ghopter 6. On Tor end Circles ‘G4, Dierential Eqsations on Tors 462. Rotation Number (63, An Example: The Standard Ciclo Map PART III: Dimension Two Chepter 7. Planar Autonomous Systems Ta “Natural” Examples of Planar Systems 12. General Properis and Geometry 13. Product Systems 114, Fiat inegrae and Comervative Sytem 75, Brample of lemecaay Biareaions Chepter 8. Linear Systems 1, Properties of Solitons of Linear Stems 82. Reduction to Canonical Forms 53. Qualtatve Egubalene in Linear Systems 8:4. Bifureatios in Linear Systems 185. Nonhomogencous Linear Systems 36. Linear Syuens with Lperodie Coeicents Chapter 9. Near Equilibria. 91, Asymptotic Stabiy frm Lineaiation 92, Tasablty fom Linewieation 108 1 18 im 9 1 15 nM 0 mm 216 218 2" ar PS 26 266 Contents 93, Lingua Functions m 4.4 Ap Invariance Principe at 15, Brewemation ofa Saddle am 8, Flow Equivalence Near Hyperbolic Exulria sm 97, Saddle Connections ma Chapter 10, fn the Presence of a Zer Eigenvalue Wo. Seay 208 102, Bifwreations au 102. Conter Manifolds ui Chapter 11, In the Presence of Purely Imaginary Eigercelues INT Seabilty 112, Ponca: Andronov-Hopt Bifurcation 113, Computing Bifueaton Cores Chapter 12. Periodic Orbits Tin Poineré-Bendlsson Theorem 12. Stablity of Perle Orbits 123, Local Biuretions of Pre Orci 124, A Homediai Bifurcation Chapter 18. All Planar Things Considered ‘ir. Strvrurlly Stable Veco Fields 382, Disapative Sates 133, Oae parameter Generic Bifrexcins 2A. Biurcatos in the Premace of Sramercy 125. Local Twoparaneter Bifreatins Chapter 14. Conservative ond Gradient Systems TL Second-order Conserative Syste 142, ifcaton ia Conservative ystems LL. Gradient Vector Fields Chapter 15. Planar Mops 13 Linear Maps - 152, News Fixed Pos 15. Sumerial Algothins and Maps {50h Saddle Node and Peiod Doubing 1S, Poloeare-Andeonow Hopf Bikraton 136, Area preeving Maps so a 368 38 Rages BE as & a a8 88 3xiv Contes PART IV: Higher Dimensions Chapter 16, Dimension Two and One Half 16, Fored Van der Pal 08 162, Freed Dung so 16.3. Nears Transvereal Homoclic:s Point 304 164, Forced and Damped Duling te Chapter 17. Dimension Three 1. Ptiod Doubling a2 Beatin to invariant Tors sid 173. Sintor Orbits a8 FA The Lotene Equations 59 Chapter 18, Dimension Four 1: Iotesable Hamitonians 2 182. A Nonitegrable Haritonian a Paneweu, a7 APPENDIX: A Catal of Fundamental Theores 0 Rarenesces 38 txoex a0- Scalar Autonomous Equations4 Chapter Setar Astonarous Exations 1.1. Existence and Uniqueness 1s tis xoductor section ne stabi or acto dill tow and one sSucans “Tics sa mes Samp oe sates base existence and unions thar ‘eT be an pen itera the el ne Rand pete erie independent variable. Also, let rT ba geen rennin fueslon. Ea Chapter 1 we wl consider eagles apter 1. el onder dierent ESC. beled lr uonomms ident equation cas betas aS ‘will study 2 satistying Semin sy Tose 2s), ta) =a a Bguation (1.2) is referred so an an siitlve problers and any of its sions i alle sion trngh uf A tl cane of She atmo hrc oh sia tion a Be (2) ht |S o ass of general tm aauring thatthe iit vate problems specie with tp = 0, and we wll often tacitly do so. To wi ler 2() be {solution of Eq (12) though rat te and define x) = 21 +f). Now, ‘beer that y(t solution of Ba. (12) through 25 at aero gine HE) = E610) = flat +40) = FLO) aod yf) As you may recall ram your prevoue studies, a slution of Ea (1.2) through 24 ato & given in he method of hone a fo geen imply, ing the method of pain of ise tate, as) 1, ence and Ungunes ‘nben the itegral i defined, One obras st bw finding the inverse of Tpefuncron on the efhand side af this eozation, Orcasionalis. we wil TR tke rem to exiiselanions of peca difleenialequatoas for che Surpones of ilvtrations. Homeer ip get ici inposibe to perform ‘te inteseations and one shoal so expec“ obtain exh formulas for ‘Stone Ir i mportaat to real tia som the begianing, In fact, Rrabjecve inthe book s to understand se mach a pssble about the havior solucions of erential equaiens shout she Sow of an ot orsauafor the seltons ‘et us now give several examples of Sorel equations and hei stuns inorder fo elie some of the ities hat ase niin the ‘Station for The theory, that i, the ocean and the uriquenes of ‘ont of Eq (12) Example 11, The frst example: Consier*e dierentia equation fe ay 1 can be soem hy simple dlezeinton chat sf) = e~Ezy is sluion ‘ino ne net 0. ad it defined fo ll R Question: i isthe ay soloion of, (1) satis the tl ylne 210) = 20? @ Exanaple 1.2. Fnfe time: Consider ee ialalve probless 2, nla as) sean o verify by direct subatution. or wing formu (2.3, that he fneton » O° a ‘wa solucon. Notice that aithough the fnction f(ei = 2 ematkably vce” toe aston xt) edo Sned onthe eral (=>. 2/30) for z0> 0 don (oe. eae) foray = 0, aad (1/2p. 2 forty < 0. The importance SF ds cumple is hat the seutlon isnot lease dined op all of Mand {he itera of denon ofthe pluton wares with the initia condition, [arcbormore, the solution ecomns aaboanced as # approaches Lz, the ‘oudar ofthe ine of dein.) Exaraple 1.3. Malte solutions: Consider the ini problem fave n= ay wth 220. ‘A colution is sven by 2(0) = (¢+2VZ5)° & Hhzy = 0. then there aso ‘aevsolurion which sienteay ero fr alt. Therefore, the initial-value blem above docs ot have unique solution Brough 29 at 20.6 Chapter: Setar Autonomous Egon, rn cgm ta fo: vay apt oot petted (os tummies: eerie Tryon aes eye ols fy So ae a sige ansredan erbaeoe soe Sateen ore aeaoniaees epee oon anteater EN om elas liens perce matate See hinigessemeantare Sooner amereeee Ber Tocianaaracc Solin eitiarnied mcg gates cee tear a eum gn 2 nici ara Tape eget toe ce Sg Sa ER py Folie Fes) x0)=7, defn for il? € siting the lial eonition 0.9) = zp. Ao, ifag, fe then im felt. sole tm, oe i Hite. hes li fot. zl = 430, (W) Hm adtvis. f= CR, RY), them ol 49) is anigue 0 Lag {ate} i corns in (C2) topetr with ies fs patil dere fives that 2-25) 8 Faneten, ‘Tho larg posiie ice J, Sa pare (ofthe theresa above is ‘aed the maz iterel of exstenee ofthe slution (5). The os tal interval of exseice oa auton of Example 2 shown a Figo 10. Figure 1.0. Maxi interval of xstence ot sition of = 2 with (Bel ae 200) = 1 8 (=50 0 mins in) nati common vation ita fe C™(U, R), where Ulsan open and bonded ets Senet SP em aceraeet sons ere ary th tain hatin on ee eee esa aT uetnfTeowem ip hae alfa specie ides ward Soma eae ce ste rarer ciate eons uci gimnat cma emewhet complicated because the doenain of ¢ may depend om 20, af S860 fae Te fe ‘structures of flows will be one ae main ooeea ae cenunty (imeilge Seah sae eee Serer se ree oe a i amebiasis Seb Saisie abe apeeara mares Tie an aun waar simaemeeie pa ome eee ee alae nene tm ee ee engrtmiranetm pri tals chante es a See esc eet gunn ges re 0 dbase ge8 Chap: Setar dtonomons Erato: fon R. There ar also other ways of ebtining dypamicl systems ad we ‘ll see one such iporant cela Chapter Comes azao 1. Vering hypotheses Show hatte ec probiem = = L434. (0) = 0 fe une sition by veri the peter of Thee 1 Fd ‘he soluon: Win hn musa nero nsoee of se sean? 12, Maile sottons ond mimeres Reeanine Example 1h # = 4. wth ‘eile 210) =O ete of These. Wich ippche of toe there srt y the wap opens th ues salu Soh this inal ue pole sumer on scapes you have et Sted meal solos fafeeniasgaton i, Sou tay eat ‘ocean otis ta aerating Cher Sor sms erence belo Wie soaton do yu orate” Ty acter diferent sce elon Do yu suee n aang he tea slat? ‘el athe vege ornate exersee inde neil experiments with dite untions To cbminate the arden! pogamining, we sgt tbe computer progres PHASER: Sm Animator Sister for Dy ‘eral Sytem which acompanin on fo ensins sain Kote 388, 15, fly many latins: Show vst the rl equation 2 = 2°” has labs acy olson ating 2(0)= 0 on eer tee. 1 Mo selon: Cane tention fte)= Lissa 2 Wee Stow ths te dire equation ¢ ~ fle) Ine 9 sion satiing (0) = 0 on ay open inter sont 1.2. Geometry of Flows [preparation for the qualitative study ofdifeeuclal aqasons we now ‘weonsier the vealr autonomous difervntial equation vt) aed ts Bow t ) Hom a geometee poi of ew, At each polat on the (-2)-plane where fe) is define, eb eight hand side of Ba, (1) gives a salve of the destasive ded which can be thou of atthe slope of line seymecr passing tough cht poi The colton ofall such ine Segment called the dition fd of the ferent equation (LL), ‘The raph of solution of Bq. (1.2 through zy, tha i she subset of| the (t zane deed by { (yt £9] :1 2 Lug) cal the taertory ‘hrough 25. A trajectory i angen to the line legments of the teen Bold at each point onthe panei pases through ‘The diction fede an several representative trajectrias of Examples 11 and 12 are shown in Figures La and 12s, respective 18 Goomerry of lows Figure (a) Dict eld log nts teeter. a) we forbes of = 2 Figure 02, (6) Dston Bld long with eeal ration and () we (orld of" Sa = fame ese ip Seems ee at nega ner Eves ; See meer necrtuos one tee wae man marae areee teens Saas tas aren Definition 1.8. The postive erit*(zy), negative arty (zo), aed arbit10 Copter Sela Antonomens Egutons 1.3. (a) Poste ori. aagative cb eit Figure Fst Cs trough su and 9) 30) af ae defined, respectively, as te lloming subsea the x-axis vled= alter en, Uae an) eerste ia hen hoe Sreeeeceracns Hee eee reson to) changing st inerencer The fo odie santa norton Soe siaeemad imamene iain See Se ew guar Sora eeperiansea ate Ruaivateeees ahcaaaree [Sica eee orci ine sivinetaaan eee ees SERS ee with x9 > Op of Example 1 orbit 7¥(55) = fro.) fd a 12 Geomeiy of Fons 11 Figure 1-4. (a) Post rte. mgative obi ri through and) phe para ur? = “Therese some obi which are especialy simp, but they play 8 cenvral role the gualitaive dy of difeenial equations, as well 5 i= Sppbeations. Definition 1.6. A poin 3 € Ris called ax emia pont (lo ool int, sendy sate slution, et) of = (2) J) = 0- “When 2 as equilibrium point, the constant Fonesion 2) tis solution, and thus the orbit) is? si Ts very easy so draw orbits of Bq, (1) from the gragh of f(2) aft the sign off determines the dees ofthe motion slog aD or Ue TE 7(ax) 0. ten the sation Is deresing inf dt 2) ithe hprosche an sgulibrium pitt or tends co ~=0 88 — Bay. Sila {Thug > 8 ha ton ineatig el thes 1 ‘roa an euilbsium pat o° tends to ~20 8 =», Soe Figures 1-58 Ea Loe Furthermore if lutions of the axialvalve problem for Ba. (1) ‘St ulque thee the wuslons through two diferent ital conditions with Se caaify lt 20) © lt ya). Thus we have the flowing lemme: ‘Lecama 1.7. Suppose that the solution Oy depending 2 sch ta. me ro en sh thesia la) of (12) troop 0a 32a the nop (39) #1 Osc that et 24) 210 eral ae stg = 31< {Weshould pit out tat me have rps todd a relundty se af compre say te cs of ala erent 2 ee clin wi ntl valve ce 0 ?apptoacer # 6 creatine Sree fos hat is sable, Bowes. this aot 0 ‘Ges dinetons and the dfn of anyon aii even wi Te tSowing lemma. whose poo is at an exec is bel va doceag te ay aan elias pot of 11 om the Faces J} oma 1.2, Ar equiva point # of = fu) i stable if thre 6 8 dorama hue alan ofr lees <6 Siar. 3 apempracaly Sa nad way if there is 4 8.> 0 such tae ~ 2) f3) <0 0 < a ee ctaltertus pone of = (ew nstablf tere isa6 > D eS ee Sites re br enterO <2—2 O such that 2f(2) <0 ee cans 'Se sen not pofble wo improve Lena 1.12 by changing te fix statement 1 exy “fa ony i€" 018 Chapter 1: Seta Auonomos Equations eet a Figure 18, Phase por oz = 2a! to ernst abi ine" The gn stable tat es = ses enti pin ww penis area To) in t, 9 (see Theorem Tah Por ‘instance. enc ramet oe ee 1 otal ta a portant concepts, & certain amount of technical language is unavoic be ema Sr rine tone mara aes Ceara Given any interval (0, rj, and any = oo Brot en a we nme st a oe ee "te i ie ed ‘Rate that the munber &( 's Example 12 must epprowch zeo usr — 20, On the other hand i stable, then & depends only om 2 and not on r: thats, the sabi firm po # sno te suns tat Pt) ‘contimnous in zo wt £ uniformiy with respect to t > 0. reas Bi ‘evident from Definitions 1.10 and 1.11 that tie stability of an submis 7 OS Som Tre eg we s o e ‘naka ropes ore ene ca ‘is asymptotically stable if c <0 and unstable if © > 0, we. civied to the following statement: see onnrely ‘Theorem 1.14, Suppose that fis a C* fonction ad is en equim Point of = f(z), tht i, fit} = 0. Suppose ale that J) #0, Then {he wai pla ‘ ‘Peep pin i atmpiocaly sale (8) <0, ad unstable 19, Sabi of Eee 19 ‘Poof. We Best shit the z-axis by iatioducing the new variable y = =~ 2. tle the aqulbiom # of = f(z} comeepnds 10 the equbbeium of ‘he dierent equation w= f(2-ry) at y= 0 1f we expand he function Fier os Taylor expesion near 0. we obain y= few ea nhc can he considered as «pevarbatin ote net diferent equation ToS ina in fat. the funcion gfy) sates 910) = @ and o's = 0. Bee ah o,f ange > Os hae ea > O such chat yh < if Fe mg the forma iy) = fe Ca). follows tha ay! he i only soe ort and its direcon oom lett right, Howse, if cee Sctuleas of how sinell am amouat cs vaied. the namber of orbits danger here are ovo ulibeia ar any ¢ <0, and none for c> 0 Fora salar dierential eqeation 2 = (the eqtibium point sad the sign of the fonetin fe between the Aula determine the aumbet TEARS dd the iretion of the ow onthe ris, Werefr tote ber oreo ad the dnesio ofthe Row on te rbics a the orbit taste rite igerarinlequsion or the quaiative structure of the fos “The sews of changes ta te qualitative structare of the How of» dif Ferenc cqution ax parameters ate vai scaled Marton theory. AL er cteteervlue a ciferetal easton aid to have sce obit see be qualitative structure of the Bor doesnot change for ul Fie smalvritions of the parameter. A sale fe whieh nts ce aoe have svble ache structure elle breton ea ae nae cha i uid to be at a Bfursion pent Tt eden om TEESE SE Show thar Ba. (21) hn stable bie strctre fra values ofFigure 2.8. Biication diagram of sie node ication. Nok that {he patente is signed 10 the oral atthe abl uae Arawn in solids aad the setble quia dad See Wo ol) Tell these comeation n Marston grams ‘and that Bq (22) hap sable obit strucore for any € #0, bu is at bifurcation pons fore = 0. The partilar bifureaton behavior of Eq (2.2) Aeseibed bow is called seddenode tiferation. The choice of terminal. ‘ogy sadale-node wil become apparent wien we dscse tatditensonal ‘ystems in Part I of our book. ‘There another very tsful graphical method for doping some of ‘he important dyzamical features i equations += Fle. 2) depending co ' parameter e.This method cont of drawing curvson the (6 2)-Pae, where ti curves depict he equilibrium pats tr each vale othe param ster, More specs, a point (co) les on one of these curves if and only if Fey, 24) = 0. Ak. to eeprewe: the stability types ofthese eu Uri. we label stable equlria with soli cars and unstable equi smth dotted curves, The resulting pictare is called bfretion shana For istance, the bifueation diagram ofthe sede nde bifurcation in Ex: saiple 22. 0 23, the parabola o= ~z* labled afin Figure 23, Example 2.8. Tronscria! ifeeaion: Consider te difereatil equation contig areal parameter beerte, es which i another perurbation of Ep. (5). Valle the previous example, {his perturbation i ota transition ofthe uxpererbed vector fal. Nov certbolss. iti il easy to determine the phase porralt of Eg (28) fom ‘he graph of the function F(e 2) = r+ 2? aesbowa ia igure 24. Novice ‘hat the origi isan equim point forall ales ofthe paracoeter © For c < 0, the origin is asymptotically stable and thee is another equ ‘brium pot # ='~c which is unstable. The paroeter valve © = is 21, Degendnce om Parometes~ rales ase portals of =z +2 fer several als of 2030. Chater & Bementry Biferetons Figure 2.8. Bierce dingram of mecca frat, Iafurcation value at which thermo aqulbria coalesce at the orgs. which 's 8 uonhyperblic ustable equilibrium pomt. For ©> 0, che ona be: ‘comes unstable Ly ansfering ls stailty te another equim point, {Ec or this reason, the bifurcation th: Bq (23) uedevgoes sell ‘ronsriial bifurcation see Figure 25.9 ‘Example 2.4. ytewens: Consider the cabie diferental equation cone taining 3 veal paramere ecteot a) Varying «corespnds to a verte shit of the zt inthe plot of the api of Fle.) = =~ 2 se Figure 26 forthe flows of Eq (24), Fore=0, Bq (24s Eq (6) and it has stable cei trou. The ow ‘continues to have stable orbit strutre fr small alte ofthe parameter hat i for mer <6 < cj. where cy = gly is the lead eam valve ‘and ~cy isthe local minimum vale of FQ, 2). Fat e= a, or ¢ =, ‘the equation ata bifreation olat. Fer the parameter vals ¢-< ey nde oy, the equation again ha stable orbit rose, The bfeation ingrar of. (24) i shown in Figure 27 Because of is request occurrence la spplicaes, tis wortwle to tsplore the dynamics of Eq. (24) ina bit tore cea. Let us suppose ‘that the diferent equation is a model of some physical system and the parameter eis a changeable charactertieof the model. fe eat the fystem with very large native value of, afer log tie, regards of the initial contition zy, the sesem wil be very ner stable oquibrsem ‘stave on the let eg of she cube. Now, lets eauiauouly ncraee he ‘alue ofthe parameter c Since the syste was neat the sable state wes we began foary ci wll stay neat ths cable lave for small variations shi Dependense Paemetrs~Esamer 3 mete 2 fr sewal lies of Figure 26, Phase porter of _ Fiera 24 Paarony fh eee i oie oo Figure 2.7, Bircain dagram of = c= # sn flow ee fact, we nnn se pete teenie we het tt So 1 re arteries wa ll ce TSE" CEE Te tad ns in Fe 28st the see so a rnd os es ee erie abe 28 aa ee ow neater deisg oe parame32 Chapter 2 loony Biretons Figure 28, Hysteresis ip ‘very large postive value, the yom il ll the egulibela onthe right Jeg ofthe subie wnt ee, at wich point = jump to he lat Ig ‘The sold ines in Figure 29 indicate the equi the systers wil oll (Se deren frm very lage postive ve toa very lage negative ‘alu, The important oberon abou his experupent is tht the satan, Crpercoces a jsp at two diferent vale of the Parameter: moreover. {ie parameter male at which tho jump tats place deterined by the Girection in which the physeal parameter Is Vased! This phenomener ip veered to Aguero and the part in Figure 28 tha vesembles a palelograr ica he hysterests lp. © “The perturbation ofthe cubic diferensal equation given in Ea. 24s. Ina nay the simples: one since isequralor to translation ofthe as Th the ext two examples we mill tad the ets of bee perturbations of Ea, (1.4) Example 2.6, Piteort furetion: Conse the diferentat equation Bede-2, (3) where dis ral parametr. Te lfc of varying ds equivalant ro changing The slope of th eubie atthe origin while kepng the sans the same Reasoning as before, ts easy to ee chat Eq (25) fas thee equilibria and able omic struct forall €> O. At = 0, the equa come ropstier ‘the origin and the system is at bfureation pln. For all d <0, the ‘untionaghin bar stable eri strutur, with one aeympocialy stable ‘qullbrlum pein; ee Figure 234 Chapter 2 Elena Brsexons ‘bts Dependence on Paroates Brampiet 38 Figure 2.1, Sopreial pitcinkbfaratoa ie = de ~ 2 sam re nh foe bs ap hn ow Steet bbe ieee oe See es uae Sie Foes Sete sates A wp in cece — ean sicaneaiiet nists, Scie aegis recoecieas etter ran = eaten anoat, sue marbonrineaan sas Beample 2:8. Fld cup Lt ow rw cone eto des : os nw soni the wo alien turbations above and consider the cubic diferestial equation oe poerde-P 2 Fleda), es) eenig on wo va partes nd To vet So ee an ret be Shop ig au tw gn 1S {Ee 2100, Ee epced in Figure 28, whew nave alo Gav 2 cierto oo tht serene of Ea. 20), You me overs ‘ice hace beatin diagrams m appeepeteplanat te isu ont bfurston diagram sven Fuse 214 mar orn 1 ei th Obevaion ha As 0 eee grction igre are elent whe the Pee seers (nc, ol and (A e- Komeves fo en 0 2 in loin he emer ro eae Soe core done st pase hough eo aden MAT int secon awe Eq (29) nth tec ata ete se not coomneca” 0igure 2.15. Sls ofthe tifcoton diagno teva =| superna! pi see- 28 (6) bys , eb fre ad () wpe 124, Deendenet an Paras ~ Examen 39 El Figure 216. Bifcatca dngrame of ¢ = A+(md= tie Bem > melanin <1 ‘it these examples at our dipesal, me ow turn to generalities, avec ee ee 21. dey sb rouge of ampein Exar 1S wt tbe sae ot esi 122, Provide te detail othe compaton of he Bifurcation gras ix Ex opie Fi40 Ghaper ® Elementary Brtins 23, Content haveing: Sopot it » potion ows st Sepa hdc at ea eT pe st =f ata pops ened he feed ets hee lh otic tsa posi: kan rea han owt te ‘of the population and his the rate of harvesting. Nocice that pope ion ‘gon stararoetea eae nen opin sal ese when he pope et sb no cine ewig eK ts fae Han to secre he or of harvest nts peptone he npn des ana beogatn. wet ‘Ser ne min tee 22 Fors ae Debate he patina he oe A) =. i gps af hea ‘ation sho be tot To Thang hand << t/t) ce ee ee ‘eit te nals the pan ut thar ea sat Shot deel ett aro seme ‘he abe ha if he atl ste bone te theshold ale. th 7 pepeltion approaches an equilibrium point. ™ a te hares neh css h Can the opi ‘exterminated regardless of it initial size oars 24 Popa event Spa ht ct een Sop a» prin gon sotig “me mt wae pom anaes based potions not pps bate! he cad hae postive conta, Stow ihe al ety Zo sch popuaton ele snc eran Site fetter aie ine Movant poplin nthe cane = and Fh 35, Hydrplne: The exis motion of hydra soln determine by vale deel saa ofthe fr e) We, ee the sect the la misma. ising mechan, a the ean. ieonabe sae Sip hate hc sappy conan ean he Stet sho nev sal and tay ee crt eds Dae th se er ee femora of he ootan en te asopase 128, The Impl Potion Theor 42 2.2. The Implicit Function Theorem te phie section. we atte a ndameacal rel mathematica! analyst e206 tne lnptet Puscion Theor, which urns out tobe 22 indi wees wal in freadon thor. The simple version presented below PetGlored forthe stay of furacons of equibra of scala diferent ‘Chon, A ore general frm ofthis ipporsaat beorem is ven inthe pod Te = [asses 2e) bea wetcr 2 RY, For now. we take the “Boe” Ajo Ato be Dye af 4a cebich we may oterpres ag the eogriof 2 We wll say more about norms econ in Captor 7 ‘Theorem 2.8. Suppose thet F: BY RR. O foneion sasring pals Pd aisa oF, F70,0)=0 and SEia.oi+0. ‘hen there ae constants 8> 0 and n> 0. and » C* faction ves AC =R st that (0) 0 snd FIA. 8)=0 fr al ef Mecover. if tert 8 Yo, 29) € RE sich that da <6 andl <7 “and sins tae equation Fld, x)= 0, then to = UO). ‘The implicit Funcoon Theorem can be used co study equine ftoring conte. Let = FOX fi be a cferetal equation depending tak poramecer A= (hx, Ae. 12 = Osa bypeboi equim ‘owe of te deren! equation ¢ = PQ. 2) at = 0, the the coitions rahe Implicit Function Theorem are sified. This guarantees that the (Ghasion F(z} Onny besoves aly for = (8) a fanetion of the hremesers (Ars Duh Soe Figure 217. Furthermore F(A, w(2))/25 # Doe hauciasy ral. Ths, rhe qualitative structare of the flow does tat change gear ? = 0, Conseqsetly, ther ar 20 bifurcations in the Tighboriood of 0 fr sufctent small aes of oe paremeters. Lo he pee meton, we wil dgcse this stution in detail as well as the {mportane roe thet the Lmpiie Function Theo plays in bifurcation theoryigure 2.17, The lpi Function Tere, To app the Implce Fumeton Ty nw eee ote in pee ta, som suave arabia cha th he ign 3 Cae re eee dae Gly + A042) and FO.D}=0 wrens Fah = er fe hci F920 me ei) Recent Po Tho 27. Apply th Imp incton The hw that afte aquatea sly td 228 Show that the condos the pe Sern condos fhe Tp Fc Ten sil (a) Arsng ONL 2am eaia/6+ 2)e eran sane. eee Near Equilibria Thee Seer ES iahiionas Peeters £5 rile ew nthe ean meted dhe seal Sinincsocm 123, Lal Parrtaton Near Enis qu renga sal, erations. Ths our resis ae valid only we cies soll neighboriood around the egulbrem point. Fat 2 we bal assure that = [125 has an equiboam poi ac Of PACS de ca eronslte £0 new coordinate eaten, ain tbe proof OF Theorem 1 Case 1: Hyperbolic equilibria. Suppose that f i a C¥ function ‘nth flo) = tod 7(0) #0. We saw in Theoem 114 cat he stabi areal the equ pot Oot he diferent exumion# = (2) 8 rerined by te iene approtmation ofthe vector Held near 0 tbat “eater order errurattoas inthe Tal expansion ofthe vector fed rons fect te qualtacve stricture of the Sow near zero. However the 2a Stun mains what happens ime rake perturbations Chat infence ‘ReGoticanr and ehe nea tra? We wil show below cbt he 3 aaton In Euample 21 ale prevails inthe general a, "Tbe precue. consider the perturbed difereatal uation era a webee 2 REC RR; (2) FYAC a isa C! fone siting Foon 0) ws Eeuernsoen a) Let ust seveatgat the existence of equa of che perturbed a Co (212) fF 0) 90, then the orga wl 2 longer be an equi Dutt However, fan Bq (215) and theft tt (0) = Oy we ave raoao mt Fon Foren Fence. the Implicit Function Theorem imps that cbere are constants Po Dandy strand 6 funtion c18) defied for |< & with 80) eck that Fasyao 7 Soreover every (2) with [Al] < & and [a] Desoh that for [ll <8 te is of Eq, (218) 6 °Chater & Blementry Biretions the same as that of (0), Therefore. the stable tpe ofthe equim ‘(A ofthe peared equation (2.12) isthe ste sete stability type of ‘he seulibrium O ofthe unperturbed equation 2 = J. ‘We can summarize the eiseustion above by ang thatthe fom near 4 bpsboleeqlina pot ls oma pean ofthe Caso Tl: Equilbria with quadratic degeneracy. Suppose tat J Js aC? hinetion with /(0) 0, (0) = but POh yD This the ext ‘order of complication that occurs when we eainot tale a deion sb he vail of an squirm poi based a the awarzation, ‘et us easier the perturbed diferensal equation B=FlA ah 215) where FRR: |A,2) 6+ F(A 2) aC? Snetion sting rosiete, Lams Boonroen am ‘These conditions gether with (0) = 0 ipl tha he Tylor expan- son of F about the eit has the flowing fre sina) Hie +e ceo, ‘with a(0) = 0, b{0) = 0, (0) = f"(0) # 0, and, for any e.> 0, there are says Maus Maem mae AS an instance of Cave Il, Jet us recall the dierental equation + = se Sateen Hecente M e = Fa2) aeroe ‘We saw in Example 22 cht the Aow of this eqution changes et \ = 0 fom two equilibria when 17"(0) < Oto bo aquibrim sen Af" > 0 sce Figure 22. We wll sow blow thatthe bifraton behavior of this simple example occur for the general cae [Bas (215) ant (216) as well. ‘To verly the assertion abot soni neces to demonstrate that the function F(X, 2) near 2 =O fo any small bass apa which ke » parabola. This can be accomplished by showing te sittence of 9 igue ‘exteme point ofthe funtion (2) near = 0 forall The etzetne Doin of F correspond to che saltions = ofthe equation Foes em 2) tcl Perron Near Epis 8 Figure 218, Lea! ications of aoabepebae sium poe wih (eat depnerc Lue Hi) 2) OFA, 29/92 Then Eq 12.16) ples hat oH, , 7 1.) = P10) 0. w0.0)=0 and 0.0) =" ‘he np Fonction Tere mpi th ths ae contac 8» Fe ata Chines (A) deed fr [cP eh Ba Os odin. tan =o. re aE, Fo. wan =0 aot, more ey sttion (2) Ea (21) with IM < Hand Ure ae ulna pes boat te er Reema coer nthe cio hoe tn the atte setae eos pre nin 5) rd fo 8 To the parame ean, he ten (cone ing the eee coo of Fz) Th, ve hogh here a be ‘Soon fe (uct par a) te ein SER Gace omni (2.5) dens on wage sone SOTA Sone cnc nx iron probe, ws hae etal er Bll fea cinensomone Hon48. Oneper 2 Bementry Bitrctions Example 2.9. Two prometers but cdimanson one: As an exaaple of x ‘codimension one bifurcation of econ Sld depending onto parame‘ers, ‘onsiae the deel equation seheueee where A (A, 2) ae two smal parameters. The function af fr this ‘example corresponds tothe minam val ofthe funtion FO 2s = n= dyes and igen by 0% Aa) = As ‘Ths the iureacons cova a we ross the curse Ay = AY/4in the Ay Ma) Plane: there ae two equlftlam points if) < 34/4 aad none if Dy > Neo ‘Example 2.10. In sppications. « single parameter sav affect veal ‘terms in he Taio expansion of « perturbation, For instance. conde the ‘following perturbation of te eta Bld (2) = given by F2 8 tee tat here Aisa scalar parameter, a @ i a given constant. As before, the bifurcation behavior of the perturbed equation depends on the minim ‘valu of the fusetion F(\.2) = 2+ 25Ae + 2, which i gual toa) — 21 ~ 2). The sign of a) is determined Uy the near tem in the Derturbationila|> 1, and by the constant rm fal <1. The bifuretion tales place at A= 0 inboth eas, the low dicen depending on ‘whether al < Lor] > 1. You ar invited to draw the breation dingra and representative pictures af the fow. 9 {Case 111: Equilibria with cubie degeneracy. Suppor that f isa Cancion with {(0)=0, 7(0) = 0, f"(0} = 0, bat f"(0) #0. Tes {he lve! of complication tht ocr the proper of the ow meat the auilbrium point a aro cant be devermibed ra the ner or seconde ‘order terms ofthe Taylor expansion off). We have sn previcusy (Section 2.1) ia the specie example f(s) = 2! that at les wo parameters were needed fo captore allo the possible ‘ype of behavior of funder parurbsins. [a parca we bave sted the bifurcations in the perturbed equation 2 = 2+ dr 2 soe Exam ple 20 apd Figures 2.12 and 2.16, We wil show below that thie example fs vepreseneative of what can happes in Ue gerel case ear the eli ‘nm point 2 =0 0 f safying the conditions above mien ff subjected to ‘nice perturbations that are small together with their derivaives rough order tree 129, Laat Peetaons Newt Eula 47 Move speily, consider ube pecorbed cietential equation anf) (218) ve FRE ARR: [bya] FO, 2} i680? fiction song F roses, ose £10) #0. 219) aE, ae Mane. Fi oe mica _ commana sag ae {othe to promeer cose the eee sae Saeed erage amplectetrecperameterperibanse ae "ie Set prosit the frcetion ana ot PAA (auc te), perarotion P(A 2) with the fling Taylor expansion about ceneeanyZ ean roo.e. a) ests ven). oA a Lh Pie ee Bee ‘values of the parameters, vet gi te = ca lacoste ena ES oe top diac oa se ean) - eo pavsyan ere eae sant 60.0 ° 2 Gy, 2) (2.22) Sa demeansateMnn ¢ ios dio We a 20a (200) 9 tan of cin ei SSS perumacaly im seme of 3, 70 bt i 20 Bey he ho of Cosi minions conenion se as ote epic Funcion Teor sng Ea rhe rao ftton of By and then ube the reat Geo Ee (221) deri yi ere of48 aplenty iets Tn peptone he np Fanta Tiere, or 2 BE ates dntm a tanies aye + SE sya tons daa snide + Zan ‘The raring fhe als mea oe ugsive tess w tool the tage Funtnn Toe so dy 2) ening Ch) se nase ys the pendent ae Ohm 00,0) 8 ted 310, 00/0 = 1 20. Tay the Ing Pacon Thee ‘there is a C? function w2(1, 2) defined for A, and 2 small and * HO, 2. 20,2) Moroes every small (2239) ang Hy 2.) = 018 gen Ba cath. 3) Hom se pc rm fH a dt ih ‘hae ito computing of cone, tae 30,0) on o Boon Se the Toler expan of he ft a, 2) wth pet 0h winder mae, aap oan ons Wie now sbatitte Ay = WoC ive ne ‘Waa, 2) lato Bq (221) and solve fr As in Casta statin 4 0s =e ethvehs and observe that J[0, 0) = 0 wth 24(0,0)/84; — 1 #0. Ths, by the “mpl Function Tren, her aC" ae 9 dead rx tessa) 0 Maren, ey sol (3) ting Jo) = 0 en by SiG. Bom te pc fomn of isnt dif to aun ‘some computing, that “ ae Yrl0} = 0) = Wl} =O, HHO) = 240). ‘ha the Te gation of eft (a em le) = UO) + em) 1.3 Loa Parubacons Near Equine 4D ecan nw susnaciz theres ofthe computations above bY sang ae Ne ee sluions (2) ad Be of Ea (221) and (2:22) near the Sen ae gira BF fa) = ile) = WO? —~ = dale) = eattale) 2) = "+ “rhe equations ar the parametric representation of csi the iA, Aa FI Chin nar the odin approszmately cvs withthe cure = Oa, 20) 1 70) <0 chen the equation above sce: that of te cas Bo Serer he ait suze of he Sos of the prin EC ar for anal als of 3 ae the same asthe ones gen in SR IR For practice, you sould drow the Hows fr the ce when Pr> 0. setae = ‘havon:2.20) san example “good” tro-ptrncter petarbaslan ia fu with cubic degeneracy In he sense int the bifseasions| 2S ny he oan ad he se fe ‘Solon ofthe recor Sed. ‘The term o3)/2 did not enter no Fer meee ciation vo the csp in the (Ay Dn}lane: sw Eas. (2:25) nd tesa because the unetin (8 dierent nA and 0) = - {Pie the Taslor expansion fr e(3) must be aves br oO) sade where cand are constants: bene. the tera (2% ee forse oe = eh —eptae + Observe at. when zs tal the rm 3s ssa than a0 the see erage! is smaller thes a Therfore esto be expected chat the Term cis! bs Lele aoence on th bento of Ba (2.20) ic mow show thar an atitary two parameter perrbation of the cutie dgeneray under eetnin eaconabe coptions. can be seduced to ee speekl tow parameter perturbation (22D) Let a = (ua) BE 8 ee eee tnd Fomider the persurbetion F(x. 2) withthe folowing Tele xpaasion about the org Guz. 27) aw? Fus) 0, thee sree andy» osc thatthe faction G sto (G(s, 2) <2 For al <6 and fz ay? A snp compat is tb Sra $= V8) Th, he bration vas of a. 22} the (ae fore Stevan Ste heme {he pares dome ener a the mgt inne ws ‘6; the determinant in Eq, (2.28) is zero—*bad.” ¢ ieee ‘As we mosis eaie, 3 complet ais ofthe cule dene is beyond the tend sope of or book ‘There een the rte long section ith an expe of thre prance psturbtoa Example 2:18. Thre parameters: Consher the thee paramet ation of f=) = ~2"/6 given by ae anne ae a + Hana +n + Badly = ‘Theor, i 8 dane evo new parameters owk Ti ae si cape Petty Shuai harem erica te ofthe original parameters (ni, 3) comping to twteteune ™ bifreation points Bin HS) =9 (0s ans bl)? 2, Le Perabo ear Epona in te treedimessonal (jy za). Aa one costs tis sae, i Maer of equim pals changes roscoe to ree. ar ces ° soo On quire degenercy For each ofthe vector Hels low, dra the ete tings ith the comeponsig aw pa plc aitc cert) Thm Example 29. buh te we tbe frandormasion #902, enn AD eyes 2. Dene 2 ves for sso aie of a once oe dftrare eween a end 0 > es Onn Neetu 4 Dra 2 es fr vo so ‘No we eiferace beget = Lal > 1 Bus) whan ror ane Beda (RIE) 230 ante oa) orany Bend abr su. On calc dgenesey Obtain the tfreation caves of he lowing oe Se es ciar petainsoos uf be cubic wer Bod f(z) = ~e! am ene ecaamgpine pois Hectares ()bmatng rin ogie—zt 18) anya Oa ale 2 (Qilpcad ime tem teat ide met = {Ln qui igor Tt eerie on ais 00% 8 mopar Prchtatons of ce quae vena Bld f(a) = =. A complete discon htm genetolpetrtsion af he quasi wey ihe For ech of ER portion dev te rete nas along wh the oxesponding eta Bele (oResjseses, () Rd s)=ar424 (Qrcnoet rs le) Fleas) seeders [sPloesicerete | Ades) adecat tat “2, Oftsn th ification cares of he lowing ene ad wo pacer i (lagu end seh some epreenaie pins pores (ieoresnesy Weems Iobom 2222) ex > 1 (fam ce Paso 128, Unloding «peor: Aa ne bun een a ample 28. he pico ie lon ace forthe ecto eld Flys 2) == Heat n=O wher he ERR Poratic clage fom aoe vo the orice vem Am portant ramos punto nan happens to the cation aga tis ot Se earl eect co smal perturbation. Sprang Pecan fat wie down 9 ton geo petrbwsion oe ‘Stl has eer he ei ang ols two atonal parameters Fld dae) = 2 a sy othr perebaton,honever ay pause ay cutan, an be SOSA JS eheparsculrprurbatian above Fer thea, he hee SERRE Scr i tte eed to athe wflig of er5A Chapter 2 Elementary Dirt utuccaion Yes prpas of ths ere 1 deri te hr ei 0 fer Band vale f=" lise State ee gant tm, We = = =~: $a usr hy eed ia npg 84 Ri niet = ee bad 6 5 Spot Nee at tha paca ue pe 28 Dis i yon deta otha ample an: Me wees Chery of Per, cats ae Gola and Seer NESS 2.4, An Example on a Pec omel pn wy he seein sms pertbatin 1 he lo alent t= tins of Sei By eoing he be Pa eee ace wr wl oan rv foe tnt a era Pere ttn pone, nae the ee ramp 2.24, muir th woparaieer peri ofthe pede vcr fia yey nso retains, asp senre¢ and dayil parameters, OF coum, the povesied vector Sell (230) TRG SERA Sia stn fora vans ofthe pret When d= che or ss inh perdi with ay pero. Patberore, Mle =, then Sie nanos of een Fue The se Set ver Inereiag InF row the abies of 2, (238 Doth onthe aR as ce S for ate perater sls “Grearse thn in 388 the lit ete of Eq 234) on? are iter equilbram pons ce dl 2 6) or $Y Wolf (ease fh. tnough rer Gppecne st Flare 210, in te ater cave each sclution also a me abe tailed ava motion on S- Lor us ft cake ths potion precise Definition 2.1% & sioton vt, zl. with yl0, 29) = 20. of sealer (Oiferential eqs “= = 2) sting fla P) = f'2) scaled a pene Sittion of pos sereisaT > 0 such that AUT, 30) ott 20) +P for alle 20. SSE et N29) # lb. 20) + P lor aneO ST, thon Tis cal = “0 ped. Recurins <%= DAMP, forthe difereatal equation (2.8), we have Peete, Were ov stat if < |e then every sation of Eg (234) 24 An Beagle om 2 Cele 88 ee oO ae ee igure 219. Phase porta of petiodion St. Since ot, 2) Isic dein by (20 de tf aedeine ad cess # O forall, there ina unique T such hat (Ty 20) = 2042) = " ae pe de te ee | oste tena of By), te a 9) 2 a Meal) ain an ty anaes (Thor re ent fr alt 20 Ths prow he aertn shoe ene pet’ npn fh nial atte , fu, when < Ih he tie pei ft stn of Ba (24), can i Re jr (2) oe Fo seam’ erg 1 res nih ci eps pen oe pee le ein past rete) dee ut wey bunt tri eared ag fe ae =i58 Chapter 2: Elementary Bifeations a _ Oe sesame Figure 220, Tie peodcy ofsanom of = 6 +dxing a) I solos ae pei 6. 1/8 and > 8 (There are for equi f= 1/8 8 (6) Thre av two squiioes 0-22.28 215. Mo eur: Seppone tht 4 2e-pdi C* fusion wt he propery that /2)"7 0 oral © B. Show tae ery sition af th Aire ‘eanion #'= (2) pres ote eet bie ag he nd Pos of an intel ef lngth 2x Dei frm forthe ert of he ‘lane 2.5. Computing Bifurcation Diagrams Tn exploring the dynamics ofa dfeenil equation # = f(z) Som by step sto locate ts guibrium points. equtalent the sere ofthe function {For most functions. this wually a ormidabe task forthe pen. ow fever, there ar flctive apprasimation procedures that can be performed ‘on the computer rathor eel. We wll ee ome sues prosedure—Newou Method—in Chapter 3. For the moment, we can also use the cbeervaton {in Section 1.2 thatthe equilibrium points of salar diferent equation 2 f(a) are thea and wim ses oats. Thx we ean obtain appa: mate values of the ston of f by folowing few sled eri infrared lor reverse tine, Nuserial eprtion of uth aris can be cacried oat om ‘the computer using one ofthe standard numeral method fo nia ae problems sch ax Ears Method (am Section 3), Th the case where the wetor Bl depends on & partner A a could apply the same procedure to the equation = F(A.) for ech Bed vale (of. By performing mauy aumencal expenmens for suficient umber of ‘aluos Gf", one could then arrive a an approunitesepesetatkn of 26, Compan Biurcaten Digrars ST se bifretion diagzam ofthe diferent equation. For the simplest of ‘Repl bowever tis procet i ory cmbenome and inefcest. I is [PE of ts scton to describ practi procedwe fr determining {fe bres of FU 2) sing diflerential equations. “ev suppoe that eae given aC function, depending 08 scalar paraieer 5 FERAR=R, (Az)- Fa! and ish to determine rh ot of plas (3, x) saising te equation Fe) =0. Cg tration eed ete, amma wt eee DRE Goo (lossy th Flo, zo) = 0 bet us eume tbat i 8 LERSLSSTr (sy onthe alae ee ces of a) Ue ob sate cane can pms tee ty 2 at see wl SOE euone yen) sig NO) = 1 = and we) nde tine atbataed oe FO 0 fs pcan ane a ato feet go sono Meteor suon FU) 2) D apes a + Eo 2)e=0. Pir Beaver nate Hey. (2) mt be acme mip of we BEANS Beal ol ee eon de pramenztba of te cane 1 EO Seca wow en we arne e fel este 235) sotsng the intial conditions NO) = 35. 2(0)= 20 Ti tea cpt my ee cnt Sint en cheer ne sae aes wena oes taster alee Bere OR opr Swe hor crgerin caters aati ig tyra Dang ot(epee 2 Klement Bieations 4 Step 1. Fic A= Ay and use the sal ees lsuation $= F(Qac 2} to Bind the set Ex oF ees OF ie) 1 Step 2. For coch 29 € Bry, compute namercally ‘he selition of the diferencia aquaron (2.5) with inl value 0}"~ 2g ad 2.0) = "in both foe ‘rad and reverse ine (using nega step ie) 1 Step 9. Repeat Step 1 as newest fr othe val fs ofthe paraaoter because thee maybe some ‘components ofthe curves of rs af FU that never ineset the erie lite 2 = {ec us now plement tate stepson Sever examples ad atempt to ‘lan the ambiguity as necessary") eat i rene in Step 3 Example 2.16 Computing hysteresis Coulee the equation Fa)eder-2 (236) ‘The zoo set ofthis function I of cours, the cubic curve A = —z +2 However. forthe sake of practice let us ry to smplement the thre steps Step 1, ket ut fix, for example, Ao “real squnion $= 5 Then, using the sale 2 we find har de ant of te ula Im Step 2, erential equations (2.3) fr she fneton (2.36) become (1-3) 1 ea ith the ital data N0) 5 an (0) = 1.60. Numero “om thine peobeg gives the cubic shows igre 22 Forward inept sig patve ie and ative sep a) of hep sie ort ges the peo ofthe cube above the ney = 5 and he sural iteration (ong np ead tfaine op Se) ft the Fale fi ils the remaing ler par ‘Altoogh thei 0 ne! fr Stop hs singe example et ss what bape fr shotie vale of xe tp =O for eal, Seren B11) Neel cpa of hc £52. ru ar oe pa, a (1) gt Example 2.17. Consider the egtion Fa) trae. (238) 28. Computing ihreenon Dagar 50 igure 23%. Numerals compute iturin dear fr te itr egaton #92 “The pir of difeenil equations (2.85) for the faction (238) are herasae 7 “eet of By (280) uth canine 222 Ul Se ‘shove Sew feo rae ete tree of E28) ae oe ce ene chef foe b= 2 papas G8 B= Fe ea) am eof Ea 299) to Te 1 Ome nie a ther St te lowe Seiki oe er care Ye ind SeEE ae Sol Rr open Sear ser gue 2 cr: Ae fel ex of hi on If xe owes to = 0, the Bo the pine (0 2) gies be oP Erample 2:38, Corr et tte dt be se at thos ea FOa)= de? 20) Te i of fein cquasin (2) fo cs same re fens at aan sexs E24) te a of, (20), bt he ln at Sega tas el ee Tome sci eat ean Dom te En = (0, TH ee eh (0) me mo Figure 2.23,(90. Chapa lenny Bifeations Figure 2.22, Sime manna 2 ime ecmpute raon igram ofthe diferent Figure 2.28.8 (erence fo he eh etion ras thor, of the ae ee ee For tp = 1 we te pn TRE Bi (ot 0.1). Th nite of Ba. (2.41 thon peat I 5 Gelman toes ican oad ree eae Serie eg asstmanctecauen ah Grid ars) mm ee ewe 216. Use the umerical i ‘eas ances hs meena Si Sma 2, Baualnce of Fw OL npstos GUE a pee et ERD een ae teaser aw Oliv rove we. men av e208e— 2 eT cateate aw O)eadat yaa — 1 2.6. Equivalence of Flows get in Section 211 conider sw flows tobe qualitative saint Weg fan ee came orbit strctre that. fthew have equ, umber of eae a deyions ofthe How on the corresponding ona che srt 8 ca scl ier equations hs stot of cusalnce sane Eas boon safcint fr ue purposes, Tate case of more than one ote oe ck we wil coir in intr chagte hs somes ima sae gebation of equivalence proves to be aequce, Te sis scion ‘inde Gootber defiion af flow equrelence wich goers aarualy SEbpe amescons, The old and the now notions of eqns wl of 10 i me te sabe for sar eel equations. 2s we will ow below. rm. za) be the Bow of ¢ = f(a) and eft 20) be she flow of 15 BE)? Soap the rita of ches flow, aarl co determine bang of alas ArR-=R: oils) ich, foreach fat least tes ne flow to the oth that Gol 29)) = (tea!) aa oral as ong os the Paws redefined. Furthermore, since we ae ooking rea tice relation, te cear that h shoud bean tsereble rap ee aa cco p, What addtional properties should we rege of Pe eta we ualitacve features ofthe Hows?) This 5 8 rather Fe eS because if we do not eesti hse. we may aot eee sara feo quatatively diferent fms: atthe other exceme, ee ter mverely then two ws withthe same orbit serra may oe retalen, One Mnatizal” choice would be to requis both hand fe tg be fanevions. Definition 2:19. 4! Aineion hs R= R with aC inves is ald 9 (C aieomorphism af R. “To apeeiate some ofthe possible consrqunces of equring 10 Pe 8 (C ditermonpism. let ws wudy a specie example. Consider ce allowing linear votr Bells es852 Chapter 2 Hlametary Biretons “Thee tn cir ls in the sae et sce bea they es Sco upp sberlin pot Fa hs oot Sine, Set ome Aleta) =e 7H) Aye ect histo ith ee oy ad eat te ee mite wats wo =e¥0, Simro emer) #0 thw ih fsutiag tpt iat “=P. Cncuny ncaa Cee ko SeuCtineun aha Comite ocentin hoes ct ssh isto be catty eons wie rte st bk ine Delton 220, Acoma sph : Rw acotinos inte Salt omen a Detniion 221, Teo alr diel toss = (ead = id be oocytes heen Be Se ath rb erie ted euton os ot ther ard prereset i For the purpoes of comparing the qualitative atures of ows of sal eauatons, 6 not a low to require a be merely a bose hr 2.2, Te al ial pate =) ad # =) ted rh sats aoa 220) ‘Fase there eee a = Po, Lt gt nat fo tea tegen ‘emu een eee emg Font i or Sol a eines a Se ‘Sti carga plese aoe Seo a eyes ae te Seong Larter inl oe sree nhc rn ch rm ns tut sy Te yi Pere riersbte yp yemtcr teen te Gin. aia Siow tone Sn ens seu See ate en ee: ie et Stem be f (2) (8) suet pe nr Prey eee a Sie sa ain png ees eo he Reserreomeeerai ‘Beta heh) Seep oy Sears mmm, Torin = sama 26, Brumaence of Flows 63 igure 224, Constring homsomeepisms of avec Sle ses /goa poecengs et sama ent eat saint gos ean ee le groan nt cepemacneerass “ nate paren = sa a em tiie meres heal ciegsemaornamgas ta Samra yu gear SS Game a Soin nt loa Came wo (if ES si ten arb soto nie rasan Pee sooo ——— een vam ent clea = epi amet dnt Drs tae here te ie TRE so et saa Naat ites vegan tm ea eer onde ferment stn Ss ‘Silat. wing the phi Bi diet) = (By64 Chapter & Slmetay estos ni Figure 2.25. Grp of Socom (el equatmcs cle ns cose radi em ore ‘Theorem 2.25. A selardiferential equation = E(3. 2) with «Bite umber of equlibeaa pons be of alii pits src ete faa i hae oa GT cae hs ntey Igy he on ea aan ‘mea diet gations, Nex we undesethe y of accher o ers 217, Identify the fanetions below as homevmc difeomor ae Sher nl dn of phen rations oa GBoEE SS seamen 218 Conse taseomerpbin of Bi ee topo onion ofthe real sqnioes# aed tat 218, Conse he eganton $= 2! = 1.43 depend Show hf thea s'epgenly cat eee cw leiattadiwee ee Show hc ee = 12) a er of gpeibebe theo ibe verar hls tract sabe logaphica! Notes Bical of : er let ne otis tls tiny 6 atom pop nove weap oy A 26 Eminence of Flows 68 gt eney of some of hte eons Arsil BO ea bad Sheer HOE] and Goby, Stewart a8 BEE ons Sours on cats thor: ae aay sine of scm (re Wisecy 195), 075), and Zeeman TA, Cole a Fenn (673) and Poin and Sewn 1078) comin Oa ee For appiceuon tomes se Thopaon ant Hur 2079 2 cemptation of tfweation diagrams of obe arenes Ye tor fe ean umportant explora tol, in dimensions greats hey or Bel Stn ofthe bfeaton crv of one-parametes fie of teers rw ens Deas vey can 30 lnge be obese st se appropriate dereril equations. Some sample sours fF i Sod Gere, 1970, Donde SRS i and Yorn 287, Kubo aero 1085, Peingen and Prfer (979). and Smale (1976! and, 2087 Mae gl na the fis ne to ow an ordinary iferetileaua- on etching fry of ont. terey stablaing the gomeric oF se ty tn spit of te otra coutbutos of Lun se ey tory and Bo [27 on daamical exes» {te Sistion of sopotoieal equivalence didnot sppear wai) Andros aera 10S au we hae are. coneepé ed so snare 8 a Pear on, Weal uy more abot topo] equa -t Tepetanons of planar vector Bes in Chapters 8 and 13300OCO Scalar Maps Wick she cureneprofieaion of computers, umes ‘Miulrics have boot eammon paste, suggeetiBE NeW ‘evfbmeikal gree and new ares of applications: ‘Basic the poner of ramenal apresation acess as Deer eae ook and bi ae of implementation OF ceeretpuce, eres amas tbe dials of deciing the sory of compuceions Even in the cave of salar deren Oe Eee lee be confonted with rather strange maibemaicl Ph cwaton. one SP gar de ro the fact chat ane epproximaion tome erin equation leds 9 a diflrence equation, and eat ile” cae ere, dese tbr innocuous apesrance ca base emazinglt SSOMEINEDdeemince, inthis apts. we ist urate om cites aerate caled maps zine in muse aprorimaions. Bac atone cance in ber contests. we chet undertake che =tady of Fe ATR iorcetins of maps. particular, we sovestnate th Par race of maps. monotone maps which wl later flay @ aaa mur study of diferesialequstons. We ead the aoe career exposition of andrack quadraticChapter 9 Seal Map 3.1. Euler's Algorithm and Maps Senn ir te aT ‘coat et eg Stat vat es mat soos ean, COL ea Se Sens a i pa in, acs a Se esi Trae at tn nc 'Ra, ein peas as eto a ae Seth ema de sf zone lee ie functions found in older calculus books. Now. given 2 = fie) (hom ce pt fo ‘enowing r(t), we determine (tz), and s0 on with the samme, ci a the sreronin nate value of z(fn) by = RSCenan iar cathe cube wes M0 ee ee ct qe =/h The eee ey neama tn obec dgtoee unos ast = Eat hflen on Given llth apposite val Ge er pons foc Se ett lloec eqn wah we Example 3.1. Lopstic and Pepe hr heehee ate aa aa) ses pn pe secs ine, ei Sah Some ences See ectenaeces pheantrarscmaeenne ae a ge nostic = to obtan the allowing diezence equation tes (5? +) 8 ‘nt Salers Aart and Meas 6B _ eee uo an iia ae 1 1 CH 0 by Gen a (ote como, SE Several snp 2 Ea ee Honey anal tm RS eg 38) a dele wb wera) oun of Ea. 33) en a ale to Las te 7 “olution of fe (33) with ca a lot eae in Figure 325, Ts a (39) wa ne behav as aid hE ofthe ‘iret equation 3.2) \ i a ego cce wien) > 1 See A pee nae ee = 18 and b= 0008, Neca ra 3.002 appe+ fe rae tobe asl star Sie tit tne sottion of tbe diffrence oe fei equation. Fo ist ane eaion Fo rg or tabulated in Fate 3.2 es a eng obi of te eens eToATS ‘cunion (28) with = Aue evemblance co the tae pong te inn fe Gen OT (a2). AMheney got monotone aed lees the ae “america expen et co nent free 920 apd 3.2, fn this amie of te ages ein 2) SEPA compl aon. wiih comes WP 8 a only being poplin: cal nag te oe re of cin OS ew ey ee ep ening eal SE a ‘mathe aon diet Sur angie —Sewtos Method Or aM tre of fet ‘Example 8:2. Roots with Newton: 1 eros of a funcelon f foil equationiii o} @ =i ini igure 3.2. Tae fain of diftece eatin ( ee Sole) 2 an ees meee th ay for b= 28 ad ah Spt = em orb 28 ° 4.8, Geometry of Sela Mepr TH ter wing a aproceaon procedure such s Bul’ grt Se in ot obaye “fast” ad, more pores, ich wil be thease whe Usiounates alo gene to vectored fenctons = et erential gations oo te pane 2 Se ord, for ecmpl, Newtons Metod, dines gel lor wing fz} 20 ach once he in spect donbedy ta with cs eto Dut er a het she eth the pit oti sein, ive Teco fer be scl een eqsion fia, Fe mae this ferential equation wig Bal’ igs wth tee appre he resulting deren eqs Nopswe A= 1 the _ Sea eo soni nen HE, a ss the method of Newton. As a example, cosier the funtion f.2) 3 act cesses at 2/2. Fo his fonction, he ference 8 (8.4) becomes yl eo swe nat with a0 itl ale for eamp, 2 = $y shen be slaion of {adores equation comveges (0 srovogo, 1.99388, 146212, 40400, 141421 0 converees to v3. and with iy paielr funeion, an Euler 2090 oe expecied rou, You sould functions and step Ses of dierence equations 2 it sone bate concen ond tn ac, any soation with nal value zo > oO converses 10 V2 ‘aon wth a sotherTrge ep se eS ‘Specimen on tbe computer with aro tn onder to explore the exiting dams fanter, we now te tothe bk of deve — —_—s0a0 es lita dv dierent “ike tear diferent oan f= abe rg sano he 0 uaitave beavion 28 eecees___— SL Baler om ina peat: Using El cP open hw sltons se ee nae of he pate ® Ae Ee free atin Bast ‘vee etnies.12 chapter Sealr Says ete eee a mn ses i nig a a Se SOLES SD alan desoe ee 3.2. Geometry of Scalar Maps Grena face fF. oe to an i ae yet of iterates of 29 under the function J: se settee se. Hao). fiflao). Ssfeo eae cen lp a a ‘sition of a function f with its pF eee tet si ace (gS Siar bey ccc ecirs irst-order difference equation pean Enon = flea 86) For the sake of brevity, we mil set eta J fer to suc a dilerence equation Definition 2.3. A poste orbit fz. ee is the set of points 20, {{20), (20), It is important to realize that ra ut putin bit of (36) 6 points, not an interval. ‘reason oa a a cele “Analogous to equibbeium points of “Serer eqtions. ‘equations have simple distinguished orbits as well, > “iscn Definition 3.4. +2 ‘3.4. The point 2 is called fred point fr f 1 /(2) = we tape ante oe ‘map, asin the case of the equilibrium points of a differenti a ion bane Pinta nate He ‘nema pan ot st determine the 2eros of the function f(z) =z. not {{2) 1 piptgeaette cn hcn/agpar eae tang ee Seam teak tc apiece Me me eae ema ers a ewe 2a, Giaare enamel aa Can arc 122, Cuomary of Sei Mope 78 fen spe ee ine rom hin i ee RNS EE rela I cr "Sores that fa, Be pipe equator te orale of Ea. (38) Be Piaget ora OBES oe food panes of aposauercein ofthe wap o fw he este Gags 0 Lact SAF Sreple 38.Liner mee Const he na nt quaion ast =O a7) ceo ro parner, He 6y woe TON atic of wi ae he vf pis © forma 0022 se i Ste te EP 20.2 = 05. Tape ge 20 ae shown in Figure 29.608 Shon consider the aaseta Ei) Novice that when 20 ‘Pepe obi oronialt ce T peeang oh oe fh B ort cre on When <0 BONE, 8S saps ae oa ie res eee mms 8 teenth pt clare equates © aoe pes eros, ew 0 © eT sessigate the sabity propor, ean ay of nif cesta eaves ake he ‘homing fiom vourmgion aa. A dead pos fo dab Habe for any «> 0. Ditonto fr ere 20 rw 8 ‘be erates ‘fs sry ee iocunlity "C2 or Moe an 2 0. The Bxed point ie oe anton table, reed pun 2 | no cman ae Detain cede, the ant > Oe ee F120) 2 wee oral scshing e0~ 8 <7 tn enlegy with he Trnaveaton aboot sn exis pals of GET en Fee Thre 118, RP He eS ne cial exe Te aed poe oe ee Fe Mh ar pit he fe) ei vide woe oe a pagans ote ews map FS 1 ee fo on mein OP ei, ond ce re uggs he flowing Hoesen theorem about 1 Bed pat. ‘Theorem 38. ec enc! map. A fd pit of wane Trea yal sand its unstable if 21> > ae por omic, we rs anne em 2) slay) Prk For i gee new ele eo #27TA Chapter 3: Sela Mops 8, Joma of Sear Mo se nay fi the new conrinates beams oS ove fee) =f le i Cleats 410) = 0 and studying the mability of ie Axed poise £ of fs Ce 20 a yee A cot giare fi Se oaseam memgisiersl, | Megan SL ice yl = 2 ~ ade if $6 then maul < Tote $M CConeuently. by repented apliaions ofthe cha. 3 mul s|aM(all MM for 20 ipa) < 1. then the ian > Ouch thst My <1, Fehermore, Woujce the ioe e Mp be thon eae eg > O ad fy > Onach that my, > 146. Suppone that 180 and fl Seo. Then, we have gr (oh) 2 ago > + 6 ah ts long a8 "(el < e- Thi inequality sows seat there mas be vale any ierscn chat g(a > em. Since wean He taken arbitral close ee Sao espe thatthe Saal pint zero of > unstable. Example 3.9. Computing V3: Lat seneturn to computing vi with New tobar Eaunpie 32 ‘The actions of the dfresceequstion (3.8) eau nc eo Reatons ofthe map wchave ' ‘We have drowa in Figure $4 the grap of ts map aad che salty b ‘Bagi of foo of Is postive oebts, one wth Zp < 0 and te ote witb a ae SSO Gh at Saace P(2| = Figure 23. Typical tai. 2 a flows fom Theorem 38 tat point are aermptoie ‘peal starstp dagame of lor sap ees = on seer ours. this face which factates the crmpuration of —_—_——— sith only a appresimate initial gvem. ©8 Chapa Sel Maps Figure 3:4. Asrptoily stable aed pinto 2) = 2/21 Wo ioeoduee the following notion for Sse pints of diferencable swaps to which Theorem 338 appicabe: Definition 9.10. 4 fixed poine 7 of ssid vo be hyperbte it 7(2) #1 From Theorem 38 fa fixed point 2 off hyperbole, then nt be ‘ther asympeotial stable a unstable and the stably type determined ‘nom (2). However, we sould emphasize thatthe Behavior of artis neat 4 Seed point 2 compleay diferent depending on whether (3) > 0 ot (2) <0. fact as evident in the neat map (36) (3) > 0, then in 'neighboracod of porte orbies seman on one sie of Hf J"(2) <0, then pastve orbits calate bout ‘Thestabilty ype of nonhsperblle xed pola cannot be determined fromthe Bt derivative ofthe mip. Tose instances ft, let i erate several specific mape onthe computer, Bxample 8.11. A nomhyperbolic fred point: Consider the qunrate rap fiyaete (ss) ‘The point = 0 is a nouhyperblic ed pole of eis map with $0) = 1 AAs seen in Figure 35, the fed point is ustable bees fis ater ‘hom the lef, bat repoing from the right. IF we consider the ful map, ‘ot jt its war tem, this Behaviors evident. The nonieas term = ‘almays pastive Cansequents. near the origin, the subaequet iterates of zo ae increasing Example 3.12. Another nonhypertli id point: Consider the quratc map fa)=-2-38. cal 438, Gamat of Sela Mast igure 2.3. Noayperbtc feed poo 2= 0 of fla) ~ ~# Aratae ‘Bzample 3.18. More nonhyerioic fed pont: Conse the cubic maps 1310) ste) fejnae2 an) ith 10. = 13 = Om anonhypetblic Sad poi hme maps Teme a Be (810) but coal BQ-810) ee Fgwe 27, Sows ‘onder to te cubic mage 32) se) seja-2tP. 313) pin # = 018 « vonhyperle Sed pon fae ar neva py = ca wt anal E072) ad se tet rk se To ame the eect fh oa arms a Pal come te cmd enn 7-0 7488, Gamevy of Sor Mops 79 7 Chapter Sel ops Figure 2.8, Nonhyperbolc fed plat # = Oa map fe) = 1 Se? ie ‘able: (0) sraph off and () i second Kee Fe) 27 = 18s? ie las 7 You might surpect from these exams that bifurcation behavior of | ‘onhrperboic fed points could be quite diferent depending oa Saether | £12) = Lor #0) = -1. Ths is ladeed the situation Toersore, re wal | ! i @ Investigate these Cwo cases separately in the following two sections. - Bet gag 39, ad wie Gor ts ics oon ae + _—_—— paca VR Coane 1) = 1. Neaypabate ed pos te igs tal ‘Are the ons dfn fr ll i? What sa apopinteetenc ad ae ee) ns rah) usable fle) == and ‘sin tne ri wl ps of ens ey 84 te ong fe wa map eo: ad psd pro od pau het a y i { 1 | I80 Chapter 9 Saar aps ‘ably ofS pits, drow typi sist dag on ah ned pit. Toot ol pate 00 soul on PHASER, fa fei nar lege keh iar Cres 9 fhee (8 J8.2125—aoce~ jem bia nth epi, 29 ot fr to comid she parameter alles 6 5h 138 a 1 55 Show dnt eed lt of che tap raed om Newton's tod fer ‘Sepang {Fs scenecaly stl Acme br Smug Ieral m whet ensured toe 6 Compuce theed its othe map (2. sa) Decne he ale ote prt wih «sen et pl “unstable or asymptotically stable, ne 27, 4 msrp ed oH lating ae lg eof oh oad gam tae f "ah Bey ‘ts nr et Sow te lan (STA ces a a oi ea (21 i inf dere wth Uo 0 6 lls in the domain Of definion of f. the f 8.4 mosctone map. that ix the Poti er (gy) af any lial ction 2-2» montane sequence, Proof Eom the Mean Value Taser, we have sam HlEe) =F 20-0) = Pitney = tat) for some ty. Therefore. yoy ~ 2 forever postive integer, has the sive sgn a that of 229 For he purposso dmamics, xe mil rogue f v0 be a est wih positive derive. an slr fo ica anf ply as» monotone ap EY is monotone en fo! soe iene of festa. We wll we che rotation f™ to denote the mois composition of J! wih sll. Definition 3.15. 1+ monotace hen the negative orbit of zs the se ofpints xo, f"za, Az). and s dented by 9725). The orbits oft i defined to be 24) =*~ 2-19 t0. ‘The geometry af rhs of a monotone map i vey iia to that ofa scala ierensial equation: the Fed points at Ike epibia and we ‘we aro%s to indicate te crecon of ether orbits under frwardierten Consequently to sty ifuratons of fixed pois uf monotone maps me need only to reinterpret results ix Saeton 23. we shall do ow. For simplicity of wottlon et us asume thatthe tp / hae fd point at £ = 0; I ot. we can chango connate to mak it 20. Fir. thermore, suppase thar (0) > 0 so that f i monotone i a sully ‘small neighborhood ofthe crigin. Consider the perurbed soap (3,2) Sepending on F parameters X= 1s, FORPARGR (2) F142) with FU02)= fir) IEF(A. 21 is @ C* function. the Follows that F(A. 2) ato monotone lin 2 foreach smal value of 4. Now. fo ech Bad. the key observation |S thatthe anal of Sead points of F(A.) egialenr tothe analy cof the zeros ofthe Rc g Fisine In Section 2.8 we base analyzed “Be bifurcations of tro of « Faneton. or, equivalently. the bureations <2 equilibria, under various eype of ye Dotheses on the ines. nda. aad cubic terms. We no emlare those results fr biurentiony of ted potas of monotone maps. (Case I: Hyperbolic Fixed Points. Suppo that f is «monotone. (Ct map with f(0)= snd f'(0) = 1. Consider a C™ map F(X, 5) soins Ponts) wot E00 /0 9 418. Biatins of Monotone Mane 83 Figure 3. Anynpeoialy stable ype Bed pit oF FS) = Sie periena ated sa pati Pensa ape re e mares! Exam us cutsider the Hoar ie 3.16. Oneparameer tna map: Let wt cos cea) 2 Se and ie one parazeter perturba eo Fy) = 2-08. : a ancien ee “ot dupe rcenmrenan woobentc mona’ ‘he parameter 8Chapter 9 Sater Mape Case Ul Fixed Points with Quadratic Degeneracy. Suppo that fia monotone, C* map mth f10) =. #0) = 1 But IO) 2 Consider aC? map FA 2) sacsiing er, Fo = rie “Thon there is «fonction os satnving (0) = 0 which corresponds to the extreme value of tbe function P(A. 2} ~ 2 soc that ‘a(N)/"0) <0 = cw Bae points of F. {AIPM =0 == one Rx pint of F QO) > — eo cad pot of F for suicionty smal values of Example 3.17. Oneparameter quadrate mep: Coasider the mep (2 z= ina suicently small aeighboricod of zr vo that it 3 monotone. ‘The local bfureaton of the one-parameter porrurbaton of 7 given by Fivsinherest sxe llucratd in Figure 3.9. fn this case, we have (2) =X and 7") = Therefore. when 10, there aze cw fed pols at \ = 0 theres one fixed point for A> 0, there sno Beed post af FOL 2). Caso IIE: Fixed Points with Cubie Degeneracy. Suppose that { ‘ea monotane, map with f(0} = 0, f'(0) = 1, (0) = 0, but JO) #0. [A compleee bifurcation analis ofan arbieary'C> perturbation FC, 2) of fis dieu. Therefore, we confine our dscuson to the analysis of 1 “piel” example of « two-parameter perturbation Asin Case I of Section 2.3, under mild hypotheses. sty of local ifureatiocs of fixed ots of aay twoparamater map F can be reduced to ths example. The Aeti are denical to the oes given prevouly. Example 3.18. Two-pammeter cubic map: Consider the map f(z) 2 = in sail oighbochood of te orga 30 that fs monotone. For the two-parameter, A= (y, Anh. pertubation off sven by PUg)= A (dasa there cusp i de (y= A-plane such Ut "there ee thee xed pats Of F(X.) fr vas of (2. 2g aide the cusp 1 there sone fied pont of FA. 2) for vals of (i. 5a} ouside the 1+ there are swo fed points of FA.) for wales of (A. 3a) 9 the ean F(A, Aa) (0.0), and one fed poite f(y, Ma) = (0.0) ‘The posible lea hitzeations of Heed plats of F liste above are ls trated in Figure 3.0. 0 : 123, Birt of Monorane Maps 85 gare 3.9, Bien of ed poe of FO rity are D1, Dad 0 ei86 Chapter 8 Solr Maps Pigure 8.10. Bifweaons of F(A 2) = 2s + (0+ oie —2 Br if at moons oS ee smal etn Neype ed at ower a ne Seki hmm cao naa ot sr dle ter lemcpee neces 4089 Appel rr Sr oe Tha 22 SShmp wih We drentl cquacen Secon oe eae 4 en ems cr so ae Pyare ee Riietieumy mee a caer ESS ee cee 54 Peinddobing Bypration 87 4.4. Period-doubling Bifurcation important bifwation that = nonhyper pty to udenge ben fis subjected fo pees ane in previous cxamples, when {2 <0, the map fs mt rape tpn clone to 2 tothe oe sie of If the xed ‘Soo? becomes unstable, an orbit cBBot tos, However. ifthe 2 ey oundeds npn tha theo ierte converts © ee ge and the even Series cxoene 0 f=") Te a en me i) ne wth fin) # tha 2" a pei 7h ae Phin breton (ald porid-ouing tf ferenton 9 recon ding of th impatatbifreacon sows ASP SAT Depicts rmblance. his gram shold not be cof CA That of ech ifuretion of iim pot rear 331 Beste ga sup end ahi Lexus begin our exploration of pevodlublngitzcation by defiing the eenupt of seural periole poi ad some raed emasks Definition 3.30. A point °K called a pete point of minimal preg PUPP) sand che les sch pst inegr, The st of 6H ‘eraten fa pavodi pole is called pene ort petiod ‘ptgh Conant, the notion of stabi of 2” flow tht of 8 et Fe eee aarzation cel in Theorem 38 an be appbed 1 /"t2 ‘eternine the stability type of 288 Chapter 5: Senter ape Definition 3:20. 4 periodic poise = of minimal palo m 8 sid to ea. bes naympttily sale or unstable fxs reapective: sable. a scemprorcall sable or ap unstabieSved poe of “Lats now extabis hat ot Ist cbsevation on the psi of periodedoublg nftaton i face occurs ical perarbations of tonyperboi aed pois with "r= —1. The satenent of the theceem Briow swore general than it mas appear, 25 shall poi out shorty ‘Theorem 3.21. Le f(z) beaC? funecon wih fixed point a the oi satévin the lowing onions: fain ‘i f0) =0. FO) = = Gi POY” #0. Consider one-parameter pecuréarion F(A =f fia) wit FAO Thea threo albedo 3. = (00 whieh ah a Ath PO) 0 hore peri rio nial period? Further ‘te, the period orbit secs sale petty, etal i ‘Seerign an uaa fopeos: aoe abe aed ptt Proof. Parole pois of pciad 2 of he ap F(2)comepond tothe feed poate of PG 2) SFU Ft. 2), sent, oe are of FH) == Rew, bras of conuon th ad pl 20s Sache rian Tre oo Fei locate atthe pee poss of oil pad 2, weed 0 ‘analyze the zeros of the function aa Pua sjne To ccomplsh this me ben. ms aun, b demining the et severe terms of te Taya expan. Lets sete ovation “pine” to dee the derivative th pect to ad compute one derives (P02 = FA FEF 2), (FO a)" FA, FO 20 F°O. 2 FO, A 2PM 2) In pesca, the cig, we have (F*0, 0! Po) (F0.0)"= (Po =0. 4 Pordoing Bifietion 89 sets onthe forme tha the Tyr expansion f FA 1 abot thorn ven wv als Fs) ates seer the functions) and (sai ins =0, 0) = [0 “therfore. the Taylor expaasion we have ess skin it ao, bp a dente ete ince Ho) #0. te aaa he son fi ct Mente 0 esata at ca gent Caw ol Scien 22 wht ah nt ea im pow ch anda deers. Tp et eet tre aon of Ba CTS, HAL ae iby (std wh argent oa Sing fod? Se (ok FID " a Fae) «to determin she se peo he (ee eee ee ton 2-21 # and me a ened frig en. and FH) gapleere PN sae bis oe anon A 1 a a a i tbe then te pr? ct Bale: Example 8.22. Continuation of ample 8.10: Let us nw ecru 1 te Pam tuple 3.12 and conde ts on-pazaites perturbation sven te nae, FA }ea(le ae +a 108 + 9.1 fa ssn > ‘We conlude this setin with several rears regarding Theorem 32) 1 ee eaetoal simply, we assumed £ = 0: tht can almars De chord Us eanaaion. eine ung asmumion on he clr dervatve cAB0" BP sre ree cao om tbe sod desve became (0) = 0 Is alwaye sacised,134, Penton Bieetion 91 Ct tr : s oy A zh i oY . gure ce ete pes Figure 3.12 Continued. The um alimatnn ‘pgs ot fr F tbe sand eae of he Figure 812. Gras of te map F(A. 2) = (0 +A}=~(9-+ Ale par the n= (a or ¥ <9, scp stable lpn thei (0) fo 20, the agin noe bypertosbur stl atti () er 0, tare ‘8 unique anypctely stable proof pati 2 (cata) eOoEe——E—Eeee ee92 Chater 5: Sela ops 4 Theres no nr of genralty n assuming F(3, 0) =O tn fat, wee Gia. 2) = Fld 3), then G0 0) = Dad AGI. 0/05 = =2 Tas, ‘he implicit Funston Theorem ialis that there is unique function {2G}, ith UO) = 0 n'a reghbortiood of (3 2)» (0.0) sue that GlayetA)) =O. that a there e#uique Bed pot of FA 3) 1 we Jee z= gL) 2 then we otal FA, 0) = 0 Hypotes (ie) save tt 0 axtmprotclly stable for 2 < 0 but taal fo > 0 te table ope ange. "The theorem can be woe to establish pend doubling bifurcations of petiole pois of any petiod, got ust xed points. br considering AS approprite erate ofthe map- Ifthe peta large homeves, ‘computing dere could got rcher cumbersome, recs aves 1115, Tao mapa: Fed val of be parameter 2 at wc the map 12,2) = 20 2) ny to unengo w pdedeblng ication Subaten ‘ce your gus eral ing PHASER. Cat ou prove your ober: fica! Pare snr ataal 408 theoreti stig on the ap Pid, z) = 22" Do gouebverve sey stn deren betwen she ‘ai ofthe wo saps? {226 Show ha the map F(A 2) = ~(82harctan undergoes «ped dovblng ‘Wheaton at hor? Can foe determine te ably type ofthe seule peitwo orbit? Dealing perdtre: At pprosmatsy A = 38%, the map FO 2) = Se(ia} nr an sembly sable prod oe of minimal period ‘tree. Vary this athe compar wing PHASER. Inve the paramter [Bacay and acre thatthe ped thee orbit gies up ably C0 pesca of sims period six Can You prove the rat f 30H 3.5. An Example: The Logistic Map In us neon th dees ution zne1 = Brat ~B)8~ tee Ex (9a) of Seton Ire el ar, (12) ty a ths change efouey 24 ft btan tnd aren ~ (028) toon ‘ha te eatin bes ue = 1st enh Puree me st (18) 2, the the diferente equation (33) i equivalent tothe eration ofthe feliowing ove paramter maps fx) dgta3. D1 fm ‘This mop is known us the logistic map and nas bee te subject of rect inten studies” Derpte fx imnocuoss ook, the opis map exhibis “aan Eva: The Lae top 98 erat in dyad sees, ths esomens sonnet te sy iporant shen ry fsb Pope, Se etn ren en nthe bok: our expense tea ai lng its xed points et aye (3.35) ind SY sist si Nena rp. a fn = 1/4 Se arden es ins ttc (1 toner ; oe sists) 8 snore te tri ce ig emma toe eS nt eat a sn ea sO ; mnt (3.15% er anes he ng mD 238 Peep per eT ae eae Seer eee» . ar) ay Tas fs 0 ea ro en E ge because fas no negation fined won a “ man, BS Rar come ener pats re oral (2) west ” ee rani ob ge map 13) swt gore pone ® i 15) ea tad pit Soto Tre 98 08 (820 EEE Screed a 2 NA aT eau tthe aan more abou the aay of EL wt tet fe 2g cen = 08. Ton Fe aly eerie ‘Slo appaseh £a monoto Ta Ogee i til cai ace) = 8 0.1 for enon) So Fare 314 "You sould deterine TE abe bur he cay peal aa 2 TEE Me fo pen 95 nae Fz) = EN4 Chapter 8 Ser age i | A | 2 Pigure 3.13. assmpcoely stable ed pie ofthe lg map or ie porameat eh 18 that ssl balls eeacing). This rugs that A= is bifurcation ‘aloe = BE TST ST ts tis ese, the Rad points seo andy ae both asta De therefore, the erates (zy) eaaane approach 4 ether ofthe od Doin, However, they remain in the interval (0,1), Ii plausible that a8 2s increased tough 2, the fixed point 2 undergoes a period doubling bifurcation giving ris to an asymprotial stable period xb of minimal ero 2; see Figure 3.14 : ‘Ti in fc the cae. ‘The existance and she asymptotic stably of such periodic ore can be estate by aualyzing the zeros of the function (3,2) = Indo, using the fact that 2 =O and 2 = 5, ate rootsof /(3,:)~5,one shows tha the pero ebit of period 2s orate a th rots ofthe equation 33? A(Avl|e+Awi = 0. Ford KEEEIE: Teresa unique asymptotically able pete obit of mink ASE This ert enny to locate othe computer wit Scurry nly erating almost aay isl value long enous 0696s, s(e") = 0480172 Plat) = 0.98, « P(a") = 020888.Figure 315. Sec of prod doling i the fogiie 3 fn the oct ne ‘ey ab peste i pv shan 8 ie st = 3 35 od 38, ropes Fe soe! hdd ae 45 An Beales The Loti op 09 pe igure 18, Apparent case 3 pve nt wih oe Pc. Pore 0 Seen, rate beers 1-100 ae tc toneve, thse in more to the dams of he pte map at AS 2581 dee ly vibe on the compute. asa warng reo of Sraovsi poses ot sl Ea rer, et us ode the pote iatgers inte flowing ny led the Sharkwsat onder psp To vod. 322-502: To- vee BoP BoP Te ppeap@ ged tens oePote2el “To deserve in words. weil the of munbers except 3 then 2 time Be Fee times the od numbers, res, Finals. ie oi am eran ont wath tee: Deve R sre. {hs a tcles alo che postive integer ‘Theorem 8.24. (Sharks) Lt f : RL — Bi be a continous rp eee pas bas erode pi of miniaa!pecod mH m= 8 Suppane that feng. then [ao bas pert pine of minimal ed no Th notewathy consequence ofthis there état ffs «pei pov of nal poiod ten has gerne of ery EI! pov of fica he gi np at 3 = 390, dtl he etd es proc orbit of al mul pesos, However, Decaie Peto dhe ase nok rly deeabl on th ROPE = STRUT: As he parameter is increas he sid mt wade ST ins brcoion aad ges op = aby © OP AYO earned 6 orb: se Figure 336 i rend fret thereFigure 3.17. Conpuerserated comple bifecation dir of he leptemap, 's sequence of pevod doubling bifurcations. Remaztay, the ratio of ‘the distances betwen thom sureasve period-douling bifurcation eet ppronches the sae constant 4.60 Here we end the ints on maps and tur to the tas of puting some of tke results to good use In our dy of tonattonomots salt erential equations inthe next chapter frercies avec 5M, Another form for legate: he mates rata, ie oe op Envi = Aull tm) often tater to the map fav) y=, Fad {Be eaafrmation ety the va of the pray at which the Inter tap undergo snc aode or peradounnsbfweston gare 3.8, Seon fe ‘aver of | 8, Ae Eze: Te Logic Mop 100 aig of period ei te agi i obtaof peri 30a8 12 9 Be Yas, eh and S84. apes102 Chapter: Setar Mape 2119 cable map The one parser coe map ses a(n heh ‘sad in gras inveving 236 oe with mo alles. Thi np undergo 4 saqnnce of pe dublin breton which eceumate othe oe othe pacer 8 = 2.6080. Beperinent onthe machine fede {Bie sera aso aie of he proto at rhich he map uges 4 pod doubling bifreaton abd attrape to fd aware! conta ot Feigrtaim rpe: Be warned ha hs cad be teioas euperimane bce fhe gemele natureo the aque of tifreton var ‘Ths pi stated in the vary af PHASER ander she ane cae Sint the praetor ad Whey 1388 JO)—4 fy=% W125 fM)=1 th nae etme she iatagus: Drow the raph aff. Kotoe tha ‘people ef miisnal pies 5. Flown ton mtn blo sho tat F bat wo perio ple of minal pared 3. Vey ae PL Ae (2,5) fies] = 9) Pe sl = ed so cha f° Nu no fad pols ‘he eras. Hower 3 = (15 and th / bas xed pec [Bl Now show tnt sd pois nique sare sth end pelt off 0. pci pit of misma pad 3. To verify the equ of ‘hs See pin you shold otnerve hat fe oncom Screg [3.4 Tome a thi, you may want to daw he op aff ‘This eames deta Li aod Yorke [197 The eomtrurion bone can be eral to ei the tapes of Sakon odting kr 084 prods Canyon fn map sth ered 7 but 0 a pred 8 pln” The coe Dios ree irs tee Teplopal valence of maps Two maps f ng ae mid tobe lope {ely eaten teplpealyenygae fe ek bump ses he oh ‘Show eat Atak oreo feeble fg tat ho 0A“ = 9° (0) Heri enous example of oplogealcnsurcy. Cone the en ‘nap die onthe unt ateralby tw fren = (Sro-e) #2205, ‘whee ata parameter. Using the Bonsomaphins 2 ros {4 ath tc ap for he paramere = 20 opolgcaly Seiten vetbe agate map (2) = 0) ote prt ae 9.5. An Examples The Lp Map 103 = te at tt ap eee agi ASS ROU Masry tec tate mate scl i aceon mere earn oa ‘Rpoimgeal conjagsey above. Bren if you do aot kaow what sn arnt Seles {ey Tanta maps Sore nthe itary of PHASER de he ae et ° ‘mupericaly for rhe owo inns at the indent Cee ect cane Som otc wcaher 2 ng im the mumperical compa Digs Since Warrier ey Sana ae aa ws ses Aen wt str Cac No.2) alte B92 +28152~18.306") dbl ptt nm ae ete th cs nto A. bee Fee tae nt tayo FRASER unde te mae ser ‘Abe o> Snge TS bee thi map fi pene settee #8 ot ea Teepe aire ee ere at te fae ar pay a Senpeauvarmes wetecsc ae cea oubnten rst hcg sot tog of te mot email tne ot vata soe ‘see also Stefan [1977]. A special case was rediscov ‘aide witha provocative title by Land Yor [1975 ie = "The surprising gpometric nature ofthe sorcestive period- doubt | wns eval rn soar anaeatoer wee1962 and 1987) using the machine in novel way, Universality can be ‘sed 1 make quantitative prediction about bifurcations laa physical sy ‘em whee a expl frmula for a map govering the sacem may ot Ue avlable. Such a preiction hag been demonstrated by Libeabe nad “Maver 1982) in experiments om Rayleigh Benard conection ‘An sementary exposition ofthe algorithms of Buler and Newton ae ‘eneatain for example, Conte ant deBoor (197). Recently, ies fom, jamie systems have begun to open ew avenue in humeral analy, ‘= exemple in Sle (198) and 1087) se alo Sat and Uren [184 fd Tak (1986 [erating a number on a compte ins, alone 100,00 ine aa lead 0 epurious ees because of Sete Batng pitt anthinetiIt ‘as lng been known tha certs wel:-beaved ape exhibit shedowing roperty: near a sumeally computed artis, chee exits «true orbit ‘hich however, mabe the orbit of dierent inital vale than be one mended; soe Anos 1967] and Bowen 1975, This important and com: foriag. property has been established st some parameter valves ofthe Doge map in Hama tal [1987