[Ebooks PDF] download Proofs that really count the art of combinatorial proof 1st edition arthur t.

Page 1


https://ebookgate.com/product/proofs-that-really-

More products digital (pdf, epub, mobi) instant download maybe you interests ...

An Introduction to Proof Theory Normalization Cut Elimination and Consistency Proofs 1st Edition Paolo Mancosu

https://ebookgate.com/product/an-introduction-to-proof-theorynormalization-cut-elimination-and-consistency-proofs-1st-editionpaolo-mancosu/

The Multifidus Back Pain Solution Simple Exercises That Target the Muscles That Count 1st Edition Jim Johnson

https://ebookgate.com/product/the-multifidus-back-pain-solutionsimple-exercises-that-target-the-muscles-that-count-1st-editionjim-johnson/

Really essential medical immunology 2nd Edition Arthur Rabson

https://ebookgate.com/product/really-essential-medicalimmunology-2nd-edition-arthur-rabson/

Art of India 1. Auflage Edition Vincent Arthur Smith

https://ebookgate.com/product/art-of-india-1-auflage-editionvincent-arthur-smith/

The Nuts and Bolts of Proofs Third Edition An Introduction to Mathematical Proofs Antonella Cupillari

https://ebookgate.com/product/the-nuts-and-bolts-of-proofs-thirdedition-an-introduction-to-mathematical-proofs-antonellacupillari/

Be Still It Is The Wind That Sings 3rd Edition Arthur Osborne

https://ebookgate.com/product/be-still-it-is-the-wind-thatsings-3rd-edition-arthur-osborne/

A Brief History of Modern Psychology 3rd Edition Ludy T. Benjamin

https://ebookgate.com/product/a-brief-history-of-modernpsychology-3rd-edition-ludy-t-benjamin/

The Art of Professional Communication Strategies that Advance Careers 1st Edition Daniel. Plung

https://ebookgate.com/product/the-art-of-professionalcommunication-strategies-that-advance-careers-1st-edition-danielplung/

The Thoreau You Don t Know What the Prophet of Environmentalism Really Meant First Edition Robert Sullivan

https://ebookgate.com/product/the-thoreau-you-don-t-know-whatthe-prophet-of-environmentalism-really-meant-first-editionrobert-sullivan/

Count Proofs the Really

The Art of Combinatorial Proof

Arthur T. Benjamin

Jennifer J. Quinn

c 2003by TheMathematicalAssociationofAmerica(Incorporated)

LibraryofCongressCatalogCardNumber2003108524

PrinteditionISBN:978-0-88385-333-7

ElectroniceditionISBN:978-1-61444-208-0

PrintedintheUnitedStatesofAmerica

CurrentPrinting(lastdigit): 10987654

NUMBERTWENTY-SEVEN

ArthurT.Benjamin

Harvey MuddCollege and

UniversityofWashingtonTacoma

DOLCIANIMATHEMATICALEXPOSITIONS

CommitteeonPublications

GERALDALEXANDERSON, Chair

DolcianiMathematicalExpositionsEditorialBoard

DANIELJ.VELLEMAN, Editor

EDWARDJ.BARBEAU

DONNAL.BEERS

ROBERTBURCKEL

ROBERTDEVANEY

JERROLDGROSSMAN

LESTERH.LANGE

SANFORDSEGAL

WILLIAMS.ZWICKER

TheDOLCIANIMATHEMATICALEXPOSITIONSseriesoftheMathematicalAssociationof AmericawasestablishedthroughagenerousgifttotheAssociationfromMaryP.Dolciani,Professor ofMathematicsatHunterCollegeoftheCityUniversityofNewYork.Inmakingthegift,Professor Dolciani,herselfanexceptionallytalentedandsuccessfulexpositorofmathematics,hadthepurpose offurtheringtheidealofexcellenceinmathematicalexposition.

TheAssociation,foritspart,wasdelightedtoacceptthegraciousgestureinitiatingtherevolving fundforthisseriesfromonewhohasservedtheAssociationwithdistinction,bothasamemberof theCommitteeonPublicationsandasamemberoftheBoardofGovernors.Itwaswithgenuine pleasurethattheBoardchosetonametheseriesinherhonor.

Thebooksintheseriesareselectedfortheirlucidexpositorystyleandstimulatingmathematical content.Typically,theycontainanamplesupplyofexercises,manywithaccompanyingsolutions. Theyareintendedtobesufficientlyelementaryfortheundergraduateandeventhemathematically inclinedhigh-schoolstudenttounderstandandenjoy,butalsotobeinterestingandsometimes challengingtothemoreadvancedmathematician.

1. MathematicalGems, RossHonsberger

2. MathematicalGemsII, RossHonsberger

3. MathematicalMorsels, RossHonsberger

4. MathematicalPlums, RossHonsberger(ed.)

5. GreatMomentsinMathematics (Before1650),HowardEves

6. MaximaandMinimawithoutCalculus, IvanNiven

7. GreatMomentsinMathematics (After1650),HowardEves

8. MapColoring,Polyhedra,andtheFour-ColorProblem, DavidBarnette

9. MathematicalGemsIII, RossHonsberger

10. MoreMathematicalMorsels, RossHonsberger

11. OldandNewUnsolvedProblemsinPlaneGeometryandNumberTheory, VictorKleeand StanWagon

12. ProblemsforMathematicians,YoungandOld, PaulR.Halmos

13. ExcursionsinCalculus:AnInterplayoftheContinuousandtheDiscrete, RobertM.Young

14. TheWohascumCountyProblemBook, GeorgeT.Gilbert,MarkKrusemeyer,andLorenC. Larson

15. LionHuntingandOtherMathematicalPursuits:ACollectionofMathematics,Verse,and StoriesbyRalphP.Boas,Jr., editedbyGeraldL.AlexandersonandDaleH.Mugler

16. LinearAlgebraProblemBook, PaulR.Halmos

17. FromErd}ostoKiev:ProblemsofOlympiadCaliber, RossHonsberger

18. WhichWayDidtheBicycleGo?...andOtherIntriguingMathematicalMysteries, Joseph D.E.Konhauser,DanVelleman,andStanWagon

19. InPolya'sFootsteps:MiscellaneousProblemsandEssays, RossHonsberger

20. DiophantusandDiophantineEquations, I.G.Bashmakova(UpdatedbyJosephSilverman andtranslatedbyAbeShenitzer)

21. LogicasAlgebra, PaulHalmosandStevenGivant

22. Euler:TheMasterofUsAll, WilliamDunham

23. TheBeginningsandEvolutionofAlgebra, I.G.BashmakovaandG.S.Smirnova(Translated byAbeShenitzer)

24. MathematicalChestnutsfromAroundtheWorld, RossHonsberger

25. CountingonFrameworks:MathematicstoAidtheDesignofRigidStructures, JackE.Graver

26. MathematicalDiamonds, RossHonsberger

27. ProofsthatReallyCount:TheArtofCombinatorialProof, ArthurT.BenjaminandJennifer J.Quinn

28. MathematicalDelights, RossHonsberger

29. Conics, KeithKendig

30. Hesiod'sAnvil:fallingandspinningthroughheavenandearth, AndrewJ.Simoson

31. AGardenofIntegrals, FrankE.Burk

32. AGuidetoComplexVariables (MAAGuides#1),StevenG.Krantz

33. SinkorFloat?ThoughtProblemsinMathandPhysics,KeithKendig

34. BiscuitsofNumberTheory,ArthurT.BenjaminandEzraBrown

35. UncommonMathematicalExcursions:PolynomiaandRelatedRealms,DanKalman

36. WhenLessisMore:VisualizingBasicInequalities,ClaudiAlsinaandRogerB.Nelsen

37. AGuidetoAdvancedRealAnalysis (MAAGuides#2),GeraldB.Folland

38. AGuidetoRealVariables (MAAGuides#3),StevenG.Krantz

39. Voltaire'sRiddle:Micromegasandthemeasureofallthings,AndrewJ.Simoson

40. AGuidetoTopology,(MAAGuides#4),StevenG.Krantz

41. AGuidetoElementaryNumberTheory,(MAAGuides#5),UnderwoodDudley

42. CharmingProofs:AJourneyintoElegantMathematics,ClaudiAlsinaandRogerB.Nelsen

43. MathematicsandSports,editedbyJosephA.Gallian

44. AGuidetoAdvancedLinearAlgebra,(MAAGuides#6),StevenH.Weintraub

45. IconsofMathematics:AnExplorationofTwentyKeyImages,ClaudiAlsinaandRogerB. Nelsen

46. AGuidetoPlaneAlgebraicCurves,(MAAGuides#7),KeithKendig

MAAServiceCenter

P.O.Box91112

Washington,DC20090-1112

1-800-331-1MAAFAX:1-301-206-9789

Dedicatedtoourfamilies|thepeoplewhocountthemostinourlives.

ToDeena,Laurel,andAriel. A.T.B. TothebadboysofLoleta. J.J.Q.

Foreword

Everyproofinthisbookisultimatelyreducedtoacountingproblem|typicallyenumeratedintwodifferentways.Countingleadstobeautiful,oftenelementary,andvery concreteproofs.Whilenotnecessarilythesimplestapproach,itoffersanothermethodto gainunderstandingofmathematicaltruths.Toacombinatorialist,thiskindofproofisthe only rightone.Weoffer ProofsThatReallyCount asthecountingequivalentofthevisual approachtakenbyRogerNelsenin ProofsWithoutWordsI&II [37,38].

Whycount?

Ashumanbeingswelearntocountfromaveryearlyage.Atypical2yearoldwillproudly countto10forthecoosandapplauseofadoringparents.Thoughmanyadultsreadily claimineptitudeinmathematics,nooneeverownsuptoaninabilitytocount.Counting isoneofourfirsttools,anditistimetoappreciateitsfullmathematicalpower.The physicistErnstMachevenwentsofarastosay,\Thereisnoprobleminallmathematics thatcannotbesolvedbydirectcounting"[36].

Combinatorialproofscanbeparticularlypowerful.Tothisday,I(A.T.B.)remember myfirstexposuretocombinatorialproofwhenIwasafreshmanincollege.Myprofessor provedtheBinomialTheorem

bywriting

andasking\Inhowmanywayscanwecreatean xkyn k term?"Suddenclarityensued. Thetheoremmadeperfectsense.Yes,IhadseenproofsoftheBinomialTheorembefore, buttheyhadseemedawkwardandIwonderedhowanyoneinhisorherrightmindwould createsucharesult.Butnowitseemedverynatural.ItbecamearesultIwouldnever forget.

Whattocount?

Wehaveselectedourfavoriteidentitiesusingnumbersthatarisefrequentlyinmathematics (binomialcoefficients,Fibonaccinumbers,Stirlingnumbers,etc.)andhavechosenelegant countingproofs.Inatypicalidentity,weposeacountingquestion,andthenansweritin

x PROOFSTHATREALLYCOUNT

twodifferentways.Oneansweristheleftsideoftheidentity;theotheransweristhe rightside.Sincebothanswerssolvethesamecountingquestion,theymustbeequal.Thus theidentitycanbeviewedasacountingproblemtobetackledfromtwodifferentangles.

Weusetheidentity

toillustrateaproofstructurefoundthroughoutthisbook.Thereisnoneedtousethe formula n! k!(n k)! for n k Instead,weinterpret n k asthenumberof k-elementsubsets ofan n-elementset,ormorecolorfully,asthenumberofwaystoselectacommitteeof k studentsfromaclassof n students.

Question: Fromaclassof n students,howmanywayscanwecreateacommittee?

Answer1: Thenumberofcommitteeswith 0 studentsis n 0 .Thenumberofcommitteeswith 1 studentis n 1 .Ingeneral,thenumberofcommitteeswithexactly k studentsis n k .Hencethetotalnumberofcommitteesis n k=0 n k .

Answer2: Tocreateacommitteeofarbitrarysize,wedecide,studentbystudent whetherornottheywillbeonthecommittee.Sinceeachofthe n studentsiseither \on"or\off"thecommittee,thereare 2 possibilitiesforeachstudentandthus 2n waystocreateacommittee.

Sinceourlogicisimpeccableinbothanswers,theymustbeequal,andtheidentity follows.

Anotherusefulprooftechniqueistointerprettheleftsideofanidentityasthesizeof aset,therightsideoftheidentityasthesizeofadifferentset,andthenfindaone-to-one correspondencebetweenthetwosets.Weillustratethisproofstructurewiththeidentity

Bothsumsarefinitesince n i =0 whenever i>n.Hereitiseasytosee what bothsides count.Thechallengeistofindthecorrespondencebetweenthem.

Set1: Thecommitteeswithanevennumberofmembersformedfromaclassof n students.Thissethassize k≥0 n 2k .

Set2: Thecommitteeswithanoddnumberofmembersformedfromaclassof n students.Thissethassize k≥0 n 2k+1

Correspondence: SupposeoneofthestudentsintheclassisnamedWaldo.Any committeewithanevennumberofmemberscanbeturnedintoacommitteewithan oddnumberofmembersbyasking\Where'sWaldo?"IfWaldoisonthecommittee, thenremovehim.IfWaldoisnotonthecommittee,thenaddhim.Eitherway,the parityofthecommitteehaschangedfromeventoodd.

Sincetheprocessof\removingoraddingWaldo"iscompletelyreversible,we haveaone-to-onecorrespondencebetweenthesesets.Thusbothsetsmusthavethe samesize,andtheidentityfollows.

Oftenweshallproveanidentitymorethanoneway,ifwethinkasecondproofcan bringnewinsighttotheproblem.Forinstance,thelastidentitycanbehandledbycounting thenumberofevensubsetsdirectly.SeeIdentity129andthesubsequentdiscussion. Whatcanyouexpectwhenreadingthisbook?Chapter1introducesacombinatorial interpretationofFibonaccinumbersassquareanddominotilings,whichservesasthe foundationforChapters2{4.WebeginherebecauseFibonaccinumbersareintrinsically interestingandtheirinterpretationascombinatorialobjectswillcomeasadelightfulsurprisetomanyreaders.Aswithallthechapters,thisonebeginswithelementaryidentities andsimpleargumentsthathelpthereadertogainafamiliaritywiththeconceptsbefore proceedingtomorecomplexmaterial.ExpandingontheFibonaccitilingswillenableus toexploreidentitiesinvolvinggeneralizedFibonaccinumbersincludingLucasnumbers (Chapter2),arbitrarylinearrecurrences(Chapter3),andcontinuedfractions(Chapter4.) Chapter5approachesthetraditionalcombinatorialsubjectofbinomialcoefficients. Countingsetswithandwithoutrepetitionleadstoidentitiesinvolvingbinomialcoefficients.Chapter6looksatbinomialidentitieswithalternatingsigns.Byfindingcorrespondencesbetweensetswithevennumbersofelementsandsetswithoddnumbersof elements,weavoidusingthefamiliarmethodofovercountingandundercountingprovided bythePrincipleofInclusion-Exclusion.

Harmonicnumbers,likecontinuedfractions,arenotintegral|soacombinatorialexplanationrequiresinvestigatingthenumeratoranddenominatorofaparticularrepresentation.HarmonicnumbersareconnectedtoStirlingnumbersofthefirstkind.Chapter7 investigatesandexploitsthisconnectioninadditiontoidentitiesinvolvingStirlingnumbersofthesecondkind.

Chapter8considersmoreclassicalresultsfromarithmetic,numbertheory,andalgebraincludingthesumofconsecutiveintegers,thesumofconsecutivesquares,sumof consecutivecubes,Fermat'sLittleTheorem,Wilson'sTheorem,andapartialconverseto Lagrange'sTheorem.

InChapter9,wetackleevenmorecomplexFibonacciandbinomialidentities.These identitiesrequireingeniousarguments,theintroductionofcoloredtiles,orprobabilistic models.Theyareperhapsthemostchallenginginthebook,butwellworthyourtime. Occasionally,wedigressfromidentitiestoprovefunapplications.LookforadivisibilityproofonFibonaccinumbersinChapter1,amagictrickinChapter2,ashortcutto calculatetheparityofbinomialcoefficientsinChapter5andgeneralizationstocongruencesmoduloarbitraryprimesinChapter8.

Eachchapter,exceptthelast,includesasetofexercisesfortheenthusiasticreader totryhisorherowncountingskills.Mostchapterscontainalistofidentitiesforwhich combinatorialproofsarestillbeingsought.Hintsandreferencesfortheexercisesanda completelistingofalltheidentitiescanbefoundintheappendicesattheendofthebook.

Ourhopeisthateachchaptercanstandindependently,sothatyoucanreadina nonlinearfashionifdesired.

Whoshouldcount?

Theshortanswertothisquestionis\Everybodycounts!"Wehopethisbookcanbe enjoyedbyreaderswithoutspecialtraininginmathematics.Mostoftheproofsinthis bookcanbeappreciatedbystudentsatthehighschoollevel.Ontheotherhand,teachers mayfindthisbooktobeavaluableresourceforclassesthatemphasizeproofwriting andcreativeproblemsolvingtechniques.Wedonotconsiderthisbooktobeacomplete

surveyofcombinatorialproofs.Rather,itisabeginning.Afterreadingit,youwillnever viewquantitieslikeFibonaccinumbersandcontinuedfractionsthesamewayagain.Our hopeisthatanidentitylike

forFibonaccinumbersshouldgiveyouthefeelingthatsomethingisbeingcountedand thedesiretocountit.Finally,wehopethisbookwillserveasaninspirationformathematicianswhowishtodiscovercombinatorialexplanationsforoldidentitiesordiscover newones.Weinviteyou,ourreaders,toshareyourfavoritecombinatorialproofswithus for(possible)futureeditions.

Afterall,wehopeallofoureffortsinwritingthisbookwillcountforsomething.

Whocounts?

Wearepleasedtoacknowledgethemanypeoplewhomadethisbookpossible|either directlyorindirectly.

Thosewhocamebeforeusareresponsiblefortheriseinpopularityofcombinatorial proof.Bookswhoseimportancecannotbeoverlookedare ConstructiveCombinatorics byDennisStantonandDennisWhite, EnumerativeCombinatoricsVolumes1&2 by RichardStanley, CombinatorialEnumeration byIanGouldenandDavidJackson,and ConcreteMathematics byRonGraham,DonKnuth&OrenPatashnik.Inadditiontothese mathematicians,otherswhoseworkscontinuetoinspireusincludeGeorgeE.Andrews, DavidBressoud,RichardBrualdi,LeonardCarlitz,IraGessel,AdrianoGarsia,Ralph Grimaldi,RichardGuy,StephenMilne,JimPropp,MartaSved,HerbertWilf,andDoron Zeilberger.

Oneofthebenefitsofseekingcombinatorialproofsisbeingabletoinvolveundergraduateresearchers.ManythankstoRobinBaur,TimCarnes,DanCicio,KarlMahlburg, GregPreston,andespeciallyChrisHanusa,DavidGaebler,RobertGaebler,andJeremy Rouse,whoweresupportedthroughundergraduateresearchgrantsprovidedbytheHarvey MuddCollegeBeckmanResearchFund,theHowardHughesMedicalInstitute,andthe ReedInstituteforDecisionSciencedirectedbyJanetMyhre.Colleaguesprovidingideas, identities,input,orinvaluableinformationincludePeterG.Anderson,BobBeals,Jay Cordes,DuaneDeTemple,PersiDiaconis,IraGessel,TomHalverson,MelvinHochster, DanKalman,GregLevin,T.S.Michael,MikeOrrison,RobPratt,JimPropp,James Tanton,DougWest,BillZwicker,andespeciallyFrancisSu.Itcouldn'thavehappened withouttheencouragementofDonAlbersandtheworkofDanVellemanandtheDolciani boardoftheMathematicalAssociationofAmerica.Finally,weareevergratefulforthe loveandsupportofourfamilies.

1FibonacciIdentities

1.1CombinatorialInterpretationofFibonacciNumbers

2GibonacciandLucasIdentities

2.1CombinatorialInterpretationofLucasNumbers...............17

2.2LucasIdentities..

2.3CombinatorialInterpretationofGibonacciNumbers

3.1CombinatorialInterpretationsofLinearRecurrences............36

3.2IdentitiesforSecond-OrderRecurrences

3.3IdentitiesforThird-OrderRecurrences

3.4Identitiesfor k thOrderRecurrences.....................43

3.5GetReal!ArbitraryWeightsandInitialConditions

5BinomialIdentities63

5.1CombinatorialInterpretationsofBinomialCoefficients...........63

5.2ElementaryIdentities .............................64

5.3MoreBinomialCoefficientIdentities ....................68

5.4Multichoosing... ..............................70

5.5OddNumbersinPascal’sTriangle......................75

5.6Notes.....................................77

5.7Exercises...................................78

6AlternatingSignBinomialIdentities81

6.1ParityArgumentsandInclusion-Exclusion..................81

6.2AlternatingBinomialCoefficientIdentities... ..............84

6.3Notes.....................................89

6.4Exercises...................................89

7HarmonicandStirlingNumberIdentities91

7.1HarmonicNumbersandPermutations....................91

7.2StirlingNumbersoftheFirstKind......................93

7.3CombinatorialInterpretationofHarmonicNumbers.............97

7.4RecountingHarmonicIdentities. ......................98

7.5StirlingNumbersoftheSecondKind....................103

7.6Notes.....................................106

7.7Exercises...................................106

8NumberTheory109

8.1ArithmeticIdentities .............................109

8.2AlgebraandNumberTheory.........................114

8.3GCDsRevisited................................118

8.4Lucas’Theorem................................120

8.5Notes.....................................123

8.6Exercises...................................123

9AdvancedFibonacci&LucasIdentities125

9.1MoreFibonacciandLucasIdentities .....................125

9.2ColorfulIdentities. ..............................130

9.3Some“Random”IdentitiesandtheGoldenRatio ..............136

9.4FibonacciandLucasPolynomials ......................141

9.5NegativeNumbers..............................143

9.6OpenProblemsandVajdaData.......................143

SomeHintsandSolutionsforChapterExercises147 AppendixofCombinatorialTheorems171 AppendixofIdentities173 Bibliography187 Index 191 AbouttheAuthors194

1

FibonacciIdentities

Definition The Fibonaccinumbers aredefinedby F0 =0, F1 =1, andfor n ≥ 2, Fn = Fn 1 + Fn 2.

ThefirstfewnumbersinthesequenceofFibonaccinumbersare 0, 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144,...

1.1CombinatorialInterpretationofFibonacciNumbers

Howmanysequencesof1sand2ssumto n?Let'scalltheanswertothiscounting question fn.Forexample, f4 =5 since4canbecreatedinthefollowing5ways:

, 1+1+2, 1+2+1, 2+1+1, 2+2

Table1.1illustratesthevaluesof fn forsmall n.Thepatternisunmistakable; fn begins liketheFibonaccinumbers.Infact, fn willcontinuetogrowlikeFibonaccinumbers,that isfor n> 2, fn satisfies fn = fn 1 + fn 2 .Toseethiscombinatorially,weconsider thefirstnumberinoursequence.Ifthefirstnumberis1,therestofthesequencesums to n 1,sothereare fn 1 waystocompletethesequence.Ifthefirstnumberis2,there are fn 2 waystocompletethesequence.Hence, fn = fn 1 + fn 2

Forourpurposes,wepreferamorevisualrepresentationof fn.Bythinkingofthe 1sasrepresenting squares andthe2sasrepresenting dominoes, fn countsthenumber ofwaysto tile aboardoflength n withsquaresanddominoes.Forsimplicity,wecalla length n boardan n-board. Thus f4 =5 enumeratesthetilings:

Figure1.1. Allfivesquare-dominotilingsofthe4-board

Welet f0 =1 counttheemptytilingofthe 0-boardanddefine f 1 =0 Thisleads toacombinatorialinterpretationoftheFibonaccinumbers.

CombinatorialTheorem1 Let fn countthewaystotilealength n boardwithsquares anddominoes.Then fn isaFibonaccinumber.Specifically,for n ≥−1,

fn = Fn+1

1.2Identities

ElementaryIdentities

Mathematicsisthescienceofpatterns.Asweshallsee,theFibonaccinumbersexhibit manybeautifulandsurprisingrelationships.AlthoughFibonacciidentitiescanbeproved byamyriadofmethods,wefindthecombinatorialapproachultimatelysatisfying. Forcombinatorialconvenience,weshallexpressmostofouridentitiesintermsof fn insteadof Fn.AlthoughothercombinatorialinterpretationsofFibonaccinumbersexist (seeexercises1{9),weshallprimarilyusethetilingdefinitiongivenhere. Intheproofofourfirstidentity,aswithmostproofsinthisbook,oneoftheanswersto thecountingquestionbreakstheproblemintodisjointcasesdependingonsomeproperty. Werefertothisas conditioning onthatproperty.

For

,f

Question: Howmanytilingsofan (n +2)-boarduseatleastonedomino?

Answer1: Thereare fn+2 tilingsofan (n +2)-board.Excludingthe\allsquare" tilinggives fn+2 1 tilingswithatleastonedomino.

Answer2: Conditiononthelocationofthelastdomino.Thereare fk tilingswhere thelastdominocoverscells k +1 and k +2.Thisisbecausecells1through k can betiledin fk ways,cells k +1 and k +2 mustbecoveredbyadomino,andcells k +3 through n +2 mustbecoveredbysquares.Hencethetotalnumberoftilings withatleastonedominois f0 + f1 + f2 + ··· + fn (orequivalently n k=0 fk ).See Figure1.2.

Identity2 For n ≥ 0,f0 + f2 + f4 + ··· + f2n = f2n+1

Question: Howmanytilingsofa (2n +1)-boardexist?

Answer1: Bydefinition,thereare f2n+1 suchtilings.

Table1.1.

Answer2: Conditiononthelocationofthelastsquare.Sincetheboardhasodd length,theremustbeatleastonesquareandthelastsquareoccupiesanodd-numbered cell.Thereare f2k tilingswherethelastsquareoccupiescell 2k +1,asillustrated inFigure1.3.Hencethetotalnumberoftilingsis n k=0 f2k

ManyFibonacciidentitiesdependonthenotionofbreakabilityatagivencell.Wesay thatatilingofan n-boardis breakable atcell k,ifthetilingcanbedecomposedintotwo tilings,onecoveringcells 1 through k andtheothercoveringcells k +1 through n.Onthe otherhand,wecallatiling unbreakable atcell k ifadominooccupiescells k and k +1

Toseethat

-boardwithsquares anddominoesandconditiononthelocationofthelastdomino.

(2

+1)-boardwithsquares anddominoesandconditiononthelocationofthelastsquare.

Figure1.2.
Figure1.3.

Figure1.4. A10-tilingthatisbreakableatcells 1, 2, 3, 5, 7, 8, 10 andunbreakableatcells 4, 6, 9.

Forexample,thetilingofthe 10-boardinFigure1.4isbreakableatcells 1, 2, 3, 5, 7, 8, 10, andunbreakableatcells 4, 6, 9.Noticethatatilingofan n-board(henceforthabbreviated an n-tiling)isalwaysbreakableatcell n.Weapplytheseideastothenextidentity.

Identity3 For m,n ≥ 0,fm+n = fmfn + fm 1 fn 1

Question: Howmanytilingsofan (m + n)-boardexist?

Answer1: Thereare fm+n (m + n)-tilings.

Answer2: Conditiononbreakabilityatcell m

An (m + n)-tilingthatisbreakableatcell m,iscreatedfroman m-tilingfollowed byan n-tiling.Thereare fmfn ofthese.

An (m + n)-tilingthatisunbreakableatcell m mustcontainadominocovering cells m and m +1.Sothetilingiscreatedfroman (m 1)-tilingfollowedbya dominofollowedbyan (n 1)-tiling.Thereare fm 1 fn 1 ofthese.

Sinceatilingiseitherbreakableorunbreakableatcell m,thereare fmfn + fm 1 fn 1 tilingsaltogether.SeeFigure1.5.

mnm +tilings breakable at: mnm +tilings unbreakable at:

Figure1.5. Toprove fm+n = f

count (m + n)-tilingsbasedonwhetherornot theyarebreakableorunbreakableat m

ThenexttwoidentitiesrelateFibonaccinumberstobinomialcoefficients.Weshall saymoreaboutcombinatorialproofswithbinomialcoefficientsinChapter5.Fornow, recallthefollowingcombinatorialdefinitionforbinomialcoefficients.

Definition The binomialcoefficient n k isthenumberofwaystoselect k elementsfrom an n-elementset.

Noticethat n k =0 whenever k>n, sothesumintheidentitybelowisfinite.

Identity4 For n ≥ 0,

Question: Howmanytilingsofan n-boardexist?

Answer1: Thereare fn n-tilings.

Answer2: Conditiononthenumberofdominoes.Howmany n-tilingsuseexactly i dominoes?Fortheanswertobenonzero,wemusthave 0 ≤ i ≤ n/2.Suchtilings necessarilyuse n 2i squaresandthereforeuseatotalof n i tiles.Forexample, Figure1.6isa 10-tilingthatusesexactlythreedominoesandfoursquares.The dominoesoccurasthefourth,fifth,andseventhtiles.Thenumberofwaystoselect i ofthese n i tilestobedominoesis n i i .Hencethereare i≥0 n i i n-tilings.

Figure1.6. Thereare `7 3´ 10-tilingsthatuseexactlythreedominoes.Sucha 10-tilingusesexactly seventilesandisdefinedbywhichthreeoftheseventilesaredominoes.Herethefourth,fifth,and seventhtilesaredominoes.

Identity5 For n ≥

Question: Howmanytilingsofa (2n +1)-boardexist?

Answer1: Thereare f2n+1 (2n +1)-tilings.

Answer2: Conditiononthenumberofdominoesoneachsideofthe median square. Anytilingofa (2n +1)-boardmustcontainanoddnumberofsquares.Thusone square,whichwecallthemediansquare,containsanequalnumberofsquaresto theleftandrightofit.Forexample,the 13-tilinginFigure1.7hasfivesquares.The mediansquare,thethirdsquare,islocatedincell 9. Howmanytilingscontainexactly i dominoestotheleftofthemediansquare andexactly j dominoestotherightofthemediansquare?Suchatilinghas (i + j) dominoesandtherefore (2n +1) 2(i + j) squares.Hencethemediansquarehas n i j squaresoneachsideofit.Sincetheleftsidehas (n i j)+ i = n j tiles,ofwhich i aredominoes,thereare n j i waystotiletotheleftofthemedian square.Similarly,thereare n i j waystotiletotherightofthemediansquare. Hencethereare n i j n j i tilingsaltogether.

As i and j vary,weobtainthetotalnumberof (2n +1)-tilingsas

i j n j i

≥0 j≥0

Figure1.7. The 13-tilingabovehasthreedominoesleftofthemediansquareandonedominoto therightofthemediansquare.Thenumberofsuchtilingsis `5 3

Identity6 For n ≥ 0,f2n 1 = n k=1 n k fk 1

Question: Howmany (2n 1)-tilingsexist?

Answer1: f2n 1

Answer2: Conditiononthenumberofsquaresthatappearamongthefirst n tiles. Observethata (2n 1)-tilingmustincludeatleast n tiles,ofwhichatleastoneis asquare.Ifthefirst n tilesconsistof k squaresand n k dominoes,thenthesetiles canbearranged n k waysandcovercells 1 through 2n k.Theremainingboard haslength k 1 andcanbetiled fk 1 ways.SeeFigure 1 8

Figure1.8. Thereare `n k´fk 1 tilingsofa (2n 1)-boardwherethefirst n tilescontain k squares and n k dominoes.

Forthenextidentity,weusethecombinatorialtechniqueoffindingacorrespondence betweentwosetsofobjects.Inparticular,weusea 1-to-3 correspondencebetweenthe setof n-tilingsandthesetof (n 2)-tilingsand (n +2)-tilings.

Identity7 For n ≥ 1, 3fn = fn+2 + fn 2

Set1: Tilingsofan n-board.Bydefinition,thissethassize fn

Set2: Tilingsofan (n +2)-boardoran (n 2)-board.Thissethassize fn+2 + fn 2

Correspondence: Toprovetheidentity,weestablisha 1-to-3correspondence betweenSet1andSet2.Thatis,foreveryobjectinSet1,wecancreatethreeobjects inSet2insuchawaythateveryobjectinSet2iscreatedexactlyonce.HenceSet 2isthreetimesaslargeasSet1.

Specifically,foreach n-tilinginSet1,wecreatethefollowingthreetilingsthat havelength n +2 orlength n 2.Thefirsttilingisan (n +2)-tilingcreatedby appendingadominotothe n-tiling.Thesecondtilingisan (n +2)-tilingcreated byappendingtwosquarestothe n-tiling.Sofar,sogood.Butwhataboutthethird tiling?Thiswilldependonthelasttileofthe n-tiling.Ifthe n-tilingendswitha square,weinsertadominobeforethatlastsquaretocreatean (n +2)-tiling.Ifthe n-tilingendswithadomino,thenweremovethatdominotocreatean (n 2)-tiling. SeeFigure1.9.

Toverifythatthisisa 1-to-3 correspondence,oneshouldcheckthateverytiling oflength n +2 orlength n 2 iscreatedexactlyoncefromsome n-tiling.Fora given (n +2)-tiling,wecanfindthe n-tilingthatcreatesitbyexaminingitsending andremoving i)thelastdomino(ifitendswithadomino)or ii)thelasttwosquares(ifitendswithtwosquares)or iii)thelastdomino(ifitendswithasquareprecededbyadomino). Foragiven (n 2)-tiling,wesimplyappendadominoforthe n-tilingthatcreates it.

SinceSet2isthreetimesthesizeofSet1,itfollowsthat fn+2 + fn 2 =3fn.

Aone-to-threecorrespondence.

PairsofTilings

Inthissubsection,weintroducethetechniqueof tailswapping,whichwillprovetobe veryusefulinseveralsettings.

Considerthetwo 10-tilingsoffsetasinFigure1.10.Thefirstonetilescells 1 through 10;thesecondonetilescells 2 through 11.Wesaythatthereisa fault atcell i,for 2 ≤ i ≤ 10,ifbothtilingsarebreakableatcell i.Wesaythereisafaultatcell 1 ifthe firsttilingisbreakableatcell 1.Putanotherway,thepairoftilingshasafaultatcell i,for 1 ≤ i ≤ 10,ifneithertilinghasadominocoveringcells i and i +1.Thepairof tilingsinFigure1.10hasfaultsatcells1,2,5,and7.Wedefinethe tails ofatilingpair tobethetilesthatoccurafterthelastfault.ObservethatifweswapthetailsofFigure 1.10weobtainthe 11-tilingandthe 9-tilinginFigure1.11,andithasthesamefaults.

Tailswappingisthebasisfortheidentitybelow,sometimesreferredtoasSimson's FormulaorCassini'sIdentity.Atfirstglance,itmayappearunsuitableforcombinatorial proofduetothepresenceofthe ( 1)n term.Nonetheless,wewillseethatthistermis merelythe\errorterm"ofan\almost"one-to-onecorrespondence.

Figure1.10. Two 10-tilingswiththeirfaults(indicatedwithgraylines)andtails.

Figure1.11. Aftertailswapping,wehavean 11-tilinganda 9-tilingwithexactlythesamefaults.

Figure1.9.

Identity8 For n ≥ 0,f 2 n = fn+1 fn 1 +( 1)n

Set1: Tilingsoftwo n-boards(a top boardanda bottom board.)Bydefinition,this sethassize f 2 n

Set2: Tilingsofan (n +1)-boardandan (n 1)-board.Thissethassize fn+1 fn 1.

Correspondence: First,suppose n isodd.Thenthetopandbottomboardmusteach haveatleastonesquare.Noticethatasquareincell i ofeitherboardensuresthat afaultmustoccuratcell i orcell i 1.Swappingthetailsofthetwo n-tilings producesan (n +1)-tilingandan (n 1)-tilingwiththesamefaults.Thisproduces a 1-to-1 correspondencebetweenallpairsof n tilingsandalltilingpairsofsizes n +1 and n 1 thathavefaults.Isitpossibleforatilingpairofsizes n +1 and n 1 tobe\fault-free"?Yes,preciselywhenalldominoesarein\staggeredformation"as inFigure1.12.Thus,when n isodd, f 2 n = fn+1 fn 1 1

Similarly,when n iseven,tailswappingcreatesa 1-to-1 correspondencebetween faultytilingpairs.Theonlyfault-freetilingpairisthealldominotilingofFigure 1.13.Hencewhen n iseven, f2 n = fn+1 fn 1 +1.Consideringtheoddandeven casetogetherproducesouridentity.

Figure1.12. When n isodd,thereisonlyonefault-freetilingpair.

Figure1.13. When n iseven,thereisonlyonefault-freetilingpair.

Identity9 For n ≥ 0, n k=0 f 2 k = fnfn+1

Question: Howmanytilingsofan n-boardand (n +1)-boardexist?

Answer1: Thereare fnfn+1 suchtilings.

Answer2: Placethe (n +1)-boarddirectlyabovethe n boardasinFigure1.14, andconditiononthelocationofthelastfault.Sincebothboardsbeginatcell 1,we

Figure1.14. Thereare fnfn+1 waystotilethesetwoboards.

Figure1.15. Thereare f2 k tilingswithlastfaultatcell k

shallconsideranytilingpairtohaveafaultat\cell0".Howmanytilingpairshave theirlastfaultatcell k,where 0 ≤ k ≤ n?Thereare f 2 k waystotilebothboards throughcell k.Toavoidfuturefaults,thereisexactlyonewaytofinishthetiling, asinFigure1.15.(Specifically,alltilesaftercell k willbedominoesexceptfora singlesquareplacedoncell k +1 intherowwhosetaillengthisodd.)Summing overallpossiblevaluesof k,givesus n k=0 f 2 k tilings.

AdvancedFibonacciIdentities

Inthissubsectionwepresentidentitiesthatinouropinionrequireextraingenuity.Forthe firstidentity,weutilizeamethodofencodingtilingsasbinarysequences.

Specifically,forany m-tiling,createthelength m binarysequencebyconvertingeach squareintoa\1"andconvertingeachdominointoa\01".Equivalently,the ithterm ofthebinarysequenceis 1 ifandonlyifthetilingisbreakableatcell i.Theresulting binarysequencewillhavenoconsecutive 0sandwillalwaysendwith 1.Forexample, the 9-tilinginFigure1.16hasbinaryrepresentation 011101011

Figure1.16. The 9-tilingabovehasbinaryrepresentation 011101011

Conversely,alength n binarysequencewithnoconsecutive 0sthatendswith 1 representsaunique n-tiling.Ifsuchasequenceendswith 0, thenitrepresentsan (n 1)tiling(sincethelast0isignored).

Wemaynowinterpretthefollowingidentity.

Identity10 For

Question: Howmanybinarysequencesoflength n exist?

Answer1: Thereare 2n length n binarysequences.

Answer2: Foreachbinarysequence,weidentifyatiling.Ifasequencehasno consecutivezeros,weidentifyitwithauniquetilingoflength n or n 1 depending onwhetheritendedwith 1 or 0,respectively.Otherwise,thesequencecontainsa 00 whosefirstoccurrenceappearsincells k +1 and k +2 forsome k, 0 ≤ k ≤ n 2.For suchasequenceweassociatethe k-tilingdefinedbythefirst k termsofthebinary sequence(notethatif k> 0,thenthe kthdigitmustbe 1.)Forexample,thelength 11 binarysequence 01101001001 isidentifiedwiththe 5-tiling\domino-squaredomino",aswouldanybinarysequenceoftheform 0110100abcd where a,b,c,d

Figure1.17. The 5-tilingshownisgeneratedby16differentbinarysequencesoflength11,all beginningwith0110100.

caneachbe 0 or 1.SeeFigure 1 17.Ingeneral,for 0 ≤ k ≤ n 2, each k-tiling willbelisted 2n 2 k times.Inparticular,theemptytilingwillbelisted 2 n 2 times.

Thenextidentityisbasedonthefactthatforany t ≥ 0 atilingcanbebrokeninto segmentssothatallbutthelastsegmenthavelength t or t +1.

Identity11 For m,p,t ≥ 0,fm+(t+1)p = p i=0 p i f i t f p i t 1 fm+i

Question: Howmany (m +(t +1)p)-tilingsexist?

Answer1: fm+(t+1)p . Answer2: Foranytilingoflength m +(t +1)p,webreakitinto p +1 segmentsof length j1,j2, ..., jp+1 .For 1 ≤ i ≤ p, ji = t unlessthatwouldresultinbreaking adominoinhalf|inwhichcasewelet ji = t +1. Segment p +1 consistsofthe remainingtiles.Countthenumberoftilingsforwhich i ofthefirst p segmentshave length t andtheother p i segmentshavelength t +1. These p segmentshavetotal length it +(p i)(t +1)=(t +1)p i.Hence jp+1 = m + i.Sincesegmentsof length t canbecovered ft waysandsegmentsoflength t+1 mustendwithadomino andcanbecovered ft 1 ways,thereareexactly p i f i t f p i t 1 fm+i suchtilings.See Figure1.18. j3 = 5 j2 = 4 j1 = 5 remaining tiles

Figure1.18. When t =4 and p =3,thetilingaboveisbrokenintosegmentsoflength j1 =5, j2 =4, j3 =5,and j4 =6

Thenextidentityreadsbetterwhenstatedintermsofthetraditionaldefinitionof Fibonaccinumbers(where F0 =0 and F1 =1 andthus fn 1 = Fn forall n ≥ 0).

Theorem1 For m ≥ 1,n ≥ 0, if m|n,then Fm|Fn

Ourcombinatorialapproachallowsustoprovemore.

Theorem2 For m ≥ 1,n ≥ 0, if m divides n,then fm 1 divides fn 1.Infact,if n = qm,then fn 1 = fm 1 q j=1 f j 1 m 2 fn jm

Question: When n = qm,howmany (n 1)-tilingsexist?

Answer1: fn 1

Answer2: Conditiononthesmallest j forwhichthetilingisbreakableatcell jm 1.Sucha j existsandhasvalueatmost q sincethetilingisbreakableatcell n 1= qm 1.Given j,thereare j 1 dominoesendingatcells m, 2m,..., (j 1)m Thecellsprecedingthesedominoescanbetiledin f j 1 m 2 ways.Cells (j 1)m + 1, (j 1)m +2,..., (jm 1) canbetiled fm 1 ways.Therestoftheboardcan thenbetiled fn jm ways.SeeFigure1.19.

Figure1.19. Thereare fj 1 m 2fm 1fn jm waystotilean (n 1)-boardwhen j isthesmallest integerforwhichthetilingisbreakableat jm 1

1.3AFunApplication

Althoughtheapplicationinthissectionisnotprovedentirelybycombinatorialmeans,it utilizessomeoftheidentitiesfromthischapter.Sincewehavedonemostoftheworkto proveitalready,itwouldbeashametoomitit.

Forintegers a and b,thegreatestcommondivisor,denotedby gcd(a,b),isthelargest positivenumberdividingboth a and b.Itiseasytoseethatforanyinteger x,

gcd(a,b)=gcd(b,a bx), (1.1)

sinceanynumberthatdividesboth a and b mustalsodivide b and a bx,andviceversa. Twospecialcasesarefrequentlyinvoked:

gcd(a,b)=gcd(b,a b) (1.2) and Theorem3(EuclideanAlgorithm) If n = qm + r,then gcd(n,m)=gcd(m,r)

IntheEuclideanalgorithmwetypicallychoose q = n m ,sothat 0 ≤ r<m.For example,whenweapplytheEuclideanalgorithmtofind gcd(255, 68),weget

gcd(255, 68)=gcd(68, 51)=gcd(51, 17)=gcd(17, 0)=17.

ItimmediatelyfollowsthatconsecutiveFibonaccinumbersarerelativelyprime,that is

Lemma4 For n ≥ 1, gcd(Fn,Fn 1)=1.

Proof. Thisistheworld'sfastestproofbyinduction.When n =1, gcd(F1,F0)= gcd(1, 0)=1.Assumingthelemmaholdsforthenumber n,thenusing(1.2),weget

gcd(Fn+1 ,Fn)=gcd(Fn,Fn+1 Fn)=gcd(Fn,Fn 1)=1.

NextweexploitIdentity3toobtain

Lemma5 For m,n ≥ 0, Fm+n = F

Proof. Fm+n = fm+(n 1) = fmfn 1 +

.

Finally,werecallthatTheorem1statesif m divides n,then Fm divides Fn. Weare nowreadytostateandproveoneofthemostbeautifulpropertiesofFibonaccinumbers.

Theorem6 For m ≥ 1,n ≥ 0, gcd(Fn,Fm)= Fgcd(n,m)

Proof. Suppose n = qm + r,where 0 ≤ r<m.ByLemma5, Fn = Fqm+r = Fqm+1 Fr + FqmFr 1. Thus

gcd(Fn,Fm)=gcd(Fm,Fqm+1 Fr + FqmFr 1 ) butby(1.1),wecansubtractmultiplesof Fm fromthesecondtermandnotchangethe greatestcommondivisor.SincebyTheorem1, Fqm isamultipleof Fm,itfollowsthat gcd(Fn,Fm)=gcd(Fm,Fqm+1 Fr)=gcd(Fm,Fr), (1.3)

wherethelastequalityfollowssince Fm (adivisorof Fqm)isrelativelyprimeto Fqm+1 byLemma4.

Butwhatdowehavehere?Equation(1.3)isthesameastheEuclideanAlgorithm, butwith F sontop.Thus,forexample,

gcd(F255,F68)=gcd(F68,F51)=gcd(F51,F17)=gcd(F17,F0)= F17, since F0 =0.Thetheoremimmediatelyfollows.

ForthereaderinterestedinseeingevenmoreadvancedFibonacciidentities,werecommendreadingChapters2and9.OneofthetreatsinstoreisaproofofBinet'sFormula, anexactformulaforthe nthFibonaccinumber.Specifically

Aneagerreaderactuallyhasallthetoolsnecessarytotacklethecombinatorialproofand canjumpstraighttoIdentity 240.

1.4Notes

Fibonaccinumbershavealongandrichhistory.TheyhaveservedasmathematicalinspirationandamusementsinceLeonardoPisano(filiusdeBonacci)firstposedhisoriginal rabbitreproductionquestionatthebeginningofthe13thcentury.Fibonaccinumbershave touchedthelivesofmathematicians,artists,naturalists,musiciansandmore.Forapeek attheirhistory,werecommendRonKnott'simpressivewebsite,FibonacciNumbersand

theGoldenSection[32].ExtensivecollectionsofFibonacciidentitiesareavailablein Vajda's Fibonacci&LucasNumbers,andtheGoldenSection:TheoryandApplications [58]andKoshy's FibonacciandLucasNumberswithApplications [33].

TheFibonacciSocietyisaprofessionalorganizationfocusingonFibonaccinumbers andrelatedmathematics,emphasizingnewresults,researchproposals,challengingproblems,andnewproofsofoldideas.Theypublishaprofessionaljournal, TheFibonacci Quarterly, andorganizeabiennialinternationalconference.

CombinatorialinterpretationsofFibonaccinumbershaveexistedforalongtimeand canbesurveyedinBasinandHoggatt'sarticle[1]intheinauguralissueofthe Fibonacci Quarterly orStanley's EnumerativeCombinatoricsVol.1 Chapter1exercise14[51]. We'vechosenthetilinginterpretationandnotationpresentedinBrighamet.al.[15]and furtherdevelopedin[8].

Finally,abijectiveproofofCassini'sformulasimilartotheonegivenforIdentity8 withouttilingswasgivenbyWermanandZeilberger[60].

1.5Exercises

Proveeachoftheidentitiesbelowbyadirectcombinatorialargument.

Identity16 For

Identity17 For

Identity18 For

Identity19demonstrateshowanyfourconsecutiveFibonaccinumbersgenerateaPythagorean Triple.

Identity19 For

Identity20 For

Identity21 For

Identity22 For

For

Identity24 For

TherearemanycombinatorialinterpretationsforFibonaccinumbers.Showthatthe interpretationsbelowareequivalenttotilingaboardwithsquaresanddominoesby creatingaone-to-onecorrespondence.

1.For n ≥ 0, fn+1 countsbinary n-tupleswithnoconsecutive0s.

2.For n ≥ 0, fn+1 countssubsets S of {1, 2,...,n} suchthat S containsnotwo consecutiveintegers.

3.For n ≥ 2, fn 2 countstilingsofan n-boardwherealltileshavelength2orgreater.

4.For n ≥ 1, fn 1 countstilingsofan n-boardwherealltileshaveoddlength.

5.For n ≥ 1, fn countsthewaystoarrangethenumbers 1 through n sothatforeach 1 ≤ i ≤ n,the ithnumberis i 1 or i or i +1.

6.For n ≥ 0, f2n+1 countslength n sequencesof 0s, 1s,and 2swhere 0 isnever followedimmediatelyby 2.

7.For n ≥ 1, f2n 1 = a1a2 ar ,where r ≥ 1 and a1,...,ar arepositive integersthatsumto n.Forexample, f5 =3+2 1+1 2+1 1 1=8.(Hint: a1a2 ar counts n-tilingswithtilesofanylength,where aj isthelengthofthe jthtile,andonecellcoveredbyeachtileishighlighted.)

8.For n ≥ 1, f2n counts 2numberof ai thatequal 1,summedoverthesameset asbefore.Forexample,when n =3=2+1=1+2=1+1+1, f6 = 20 +21 +21 +23 =13

9.For n ≥ 1,fn+1 countsbinarysequences (b1,b2,...,bn),where b1 ≤ b2 ≥ b3 ≤ b4 ≥ b5

UncountedIdentities

Theidentitieslistedbelowareinneedofcombinatorialproof.

1.For n ≥ 1,f 3 0 + f 3 1 + + f 3 n = f3n+4 +( 1)n6fn 1 +5 10

2.For n ≥ 0,f1 +2f2 + ··· + nfn =(n +1)fn+2 fn+4 +3

3.Thereareidentitiesfor mfn analogoustoIdentities16{18foreveryinteger m

(a)For n ≥ 4, 5fn = fn+3 + fn 1 + fn 4

(b)For n ≥ 4, 6fn = fn+3 + fn+1 + fn 4

(c)For n ≥ 4, 7fn = fn+4 + fn 4

(d)For n ≥ 4,

(e)For n ≥ 4,

(f)For n ≥ 4, 10fn = fn+4 + fn+2 + fn 2 + fn 4.

(g)For n ≥ 4

(h)For n ≥ 6, 12

TheseidentitiesareexamplesofZeckendorf'sTheoremwhichstatesthatevery integercanbeuniquelywrittenasthesumofnonconsecutiveFibonaccinumbers. Thecoefficientsintheaboveformulasarethesameasintheuniqueexpansionof positiveintegersinnonconsecutiveintegerpowersof φ =(1+√5)/2 Forexample 5= φ3 + φ 1 + φ 4 and 6= φ3 + φ1 + φ 4 Isthereaunifyingcombinatorial approachforalloftheseidentities?

4.For n ≥ 4,f 3 n +3f 3 n 3 + f 3 n 4 =3f 2 n 1 +6f 3 n 2.JayCordeshasshownus acombinatorialproofthatrequiresbreakingthetilingtriplesintooveradozen differentcases.Doessomethingsimplerexist?

5.Findacombinatorialinterpretationforthe Fibonomialcoefficient

Another random document with no related content on Scribd:

The Project Gutenberg eBook of Pikku Dorrit I

This ebook is for the use of anyone anywhere in the United States and most other parts of the world at no cost and with almost no restrictions whatsoever. You may copy it, give it away or re-use it under the terms of the Project Gutenberg License included with this ebook or online at www.gutenberg.org. If you are not located in the United States, you will have to check the laws of the country where you are located before using this eBook.

Title: Pikku Dorrit I

Author: Charles Dickens

Translator: Helena Kesäniemi

Release date: July 9, 2024 [eBook #73994]

Language: Finnish

Original publication: Hämeenlinna: Arvi A. Karisto Oy, 1926

Credits: Juhani Kärkkäinen and Tapio Riikonen *** START OF THE PROJECT GUTENBERG EBOOK PIKKU DORRIT I

PIKKU DORRIT I

Kirj.

Charles Dickens

Englanninkielestä suomentanut

Helena Kesäniemi

Hämeenlinnassa, Arvi A. Karisto Osakeyhtiö, 1926.

SISÄLLYS:

Edellinen osa. Henkilöitä.

I. Valoa ja varjoa

II. Matkatoverit

III. Koti

IV. Mrs Flintwinch näkee unta

V. Perheasioita

VI. Marshalsean isä

VII. Marshalsean lapsi

VIII. Telkien takana

IX. Pikku äiti

X. Sisältää koko hallitsemistaidon

XI. Vapaa

XII. Bleeding Heart Yard

XIII. Patriarkallisuutta

XIV. Pikku Dorrit kutsuissa

XV. Mrs Flintwinch näkee taas unta

XVI. Ei kenenkään heikkous

XVII. Ei kenenkään kilpailija

XVIII. Pikku Dorritin ihailija

XIX. Marshalsean isä parissa kolmessa tilanteessa

XX. Seuraelämässä

XXI. Mr Merdlen tauti

XXII. Pulmallinen kysymys

XXIII. Koneet käynnissä

XXIV. Ennustuksia

XXV. Salajuonien punojia ja muita

XXVI. Ei kenenkään mielentila

XXVII. Viisikolmatta

XXVIII. Ei kenenkään häviäminen

XXIX. Mrs Flintwinch uneksii edelleen

XXX. Herrasmiehen sana

XXXI. Itsetuntoa

XXVII. Lisää ennustuksia

XXVIII. Mrs Merdlen valitus

XXXIV. Barnacleja iso liuta

HENKILÖITÄ.

Bangham, pesijätär ja mrs Dorritin hoitajatar Marshalsean velkavankilassa. Barnacle, Tite, ylhäinen virkamies verukevirastossa. — Clarence, »Barnacle nuorempi», hänen poikansa, virkamies verukevirastossa. — loordi Decimus Tite, ensinmainitun setä, myös verukevirastossa. — Ferdinand, loordin yksityissihteeri. Beadle, Harriet, perhepiirissä »Tattycoram», löytölasten kodista otettu tyttö, Minnie Meaglesin kamarineito. Bob, Pikku Dorritin kummi, Marshalsean vankilan vartija. Casby, Christopher, Bleading Heart Yardin alueen isäntä. Cavalletto, John Baptist, Rigaudin vankilatoveri Marseillesissa; myöhemmin Arthur Clennamin palveluksessa. Chivery, John, Marshalsean vartija, joka ei asu siellä. — mrs, hänen vaimonsa. — nuorempi, John, heidän poikansa. Clennam, mrs, leskirouva. — Arthur, edellisen oletettu poika, mutta todellisuudessa vain poikapuoli; nai Pikku Dorritin. Cripples, iltakoulun opettaja. Master, edellisen poika. Daives, hoitajatar. Dorrit, William, velkavanki Marshalseassa. — Amy, lisänimeltään »Pikku Dorrit», edellisen tytär; menee naimisiin Arthur Clennamin kanssa. — Edward, perhepiirin »Tip», Pikku Dorritin vanhempi veli. — Fanny, Pikku Dorritin vanhempi sisar; menee naimisiin Edmund Sparklerin kanssa. — Frederick, William Dorritin veli. Doyce, Daniel, koneseppä; ottaa liiketoverikseen Arthur Clennamin. Finching, Flora, varakas

leskirouva, Christopher Casbyn tytär. Flintwinch, Jeremiah, mrs Clennamin palvelija ja myöhemmin liiketoveri. — Affery, edellisen vaimo, mrs Clennamin palvelijatar. — Ephraim, ensinmainitun kaksoisveli. General, leskirouva, Dorritin tyttärien seuranainen. Gowan, Henry, taiteilija; nai Minnie Meaglesin. — mrs, hänen äitinsä. Haggage, lääkäri, velkavanki Marshalseassa. Jenkinson, verukeviraston lähetti. Maggy, Banghamin tyttärentytär, Pikku

Dorritin suojatti. Marvon, kapteeni, Edward Dorritin velkoja. Meagles, entinen pankkiiri. — mrs, hänen rouvansa. — Minnie, perhepiirissä »Pet», heidän tyttärensä, myöhemmin Henry Gowanin puoliso. Merdle, lontoolainen pankkiiri; menee vararikkoon ja surmaa itsensä. — mrs, hänen rouvansa, Fanny Dorritin anoppi. Mr F:n Iäti, Flora Finchingin huollettavaksi joutunut miesvainajan vanha sukulainen. Nandy, John Edward, mrs Plornishin isä. Pancks, Casbyn vuokrien kerääjä. Pet, kts. Meagles, Minnie. Plornish, rappari, Casbyn vuokralaisia. — mrs, hänen vaimonsa. Rigaud, muilta nimiltään Lagnier ja Blandois, roisto, joka kiristää rahoja mrs Clennainilta. Rugg, asioitsija, jonka luona Pancks asuu. — Anastasia, hänen tyttärensä. Sparkler, Edmund, mrs Merdlen poika ensimmäisestä avioliitosta; nai Fanny Dorritin. Sliltstalking, loordi Lancaster, verukeviraston virkamies, Englannin edustaja ulkomailla. Tattycoram, kts. Beadle, Harriet. Tickit, emännöitsijä Meaglesin perheessä. Tinkier, William Dorritin palvelija. Tip, kts. Dorrit, Edward. Wade, katkeroitunut neiti, joka houkuttelee Tattycoramin pois Meaglesin perheestä. Wobbler, kirjuri verukevirastossa.

Alkulause.

Pikku Dorrit ilmestyi ensin kahtenakymmenenä niteenä vuosina 1855—57; kuhunkin oli taiteilija-nimimerkki Phiz piirtänyt pari kuvaa. Alkuansa Dickens oli tällä kertaa suunnitellut vain suppeampaa kertomusta, jonka nimeksi tulisi »Nobody's Fault» (Ei kenenkään syy), mutta kun käsikirjoitusta oli valmistunut neljään niteeseen, hylkäsi hän mainitun nimen, sillä hän oli jo silloin kehittänyt mielessään kertomukseen sopivia uusia aatteita ja laajemman sarjan eri luonteita ja vaiheita.

Se aihepiiri, jonka hän oli ensin ottanut käsiteltäväkseen, olikin epäilemättä liian niukka kelvatakseen suurteoksen pohjaksi. Siinä näet oli päähenkilönä mies, joka oli syynä kaikkiin kertomuksessa esitettäviin onnettomuuksiin, mutta sysäsi ne sallimuksen niskoille ja huomautti jokaisen uuden kommelluksen tultua: »Onpa sentään hyvä, ettei tarvitse ketään syyttää!» Tällainen ura oli liian kaita, jotta hän olisi voinut sitä pitkin panna liikkeelle vilkkaan mielikuvituksensa monilukuiset ja toimeliaat olennot. Niinpä sitten kävikin, että kirjailijan piti levittää eteensä laajempi taulu kuvatakseen siinä melkoista vaihtelevien luonteiden ryhmää köyhyyden ja rikkauden vastakkaisissa oloissa.

Tämän romaanin kirjoittaminen ei aluksi sujunut Dickensiltä helposti. Ensimmäiset niteet tuottivat hänelle montakin levotonta, tuskastuttavaa hetkeä, sillä vaikka hänestä hyvällä syyllä voi sanoa, että hän usein ihaili omaa kykyään ja tyyliään lämpimästi ja vilpittömästi, saattoi hän taas toisinaan nousta itsensä perin vaateliaaksi arvostelijaksi. Teos oli aloitettu Tavistock Housessa toukokuussa 1855 (hänen kirjeissään on tosin mainittu lokakuu ja Pariisi käsikirjoituksen alun valmistumisajaksi ja -paikaksi), ja hän kirjoitti niinä päivinä Wilkie Collinsille m.m.: »En osaa kuvata sitä rauhatonta mielentilaa, jossa kävelen edestakaisin huoneessani

Turn static files into dynamic content formats.

Create a flipbook
Issuu converts static files into: digital portfolios, online yearbooks, online catalogs, digital photo albums and more. Sign up and create your flipbook.