fibonacci advanced

Upload: coeus-apollo

Post on 03-Jun-2018

238 views

Category:

Documents


1 download

TRANSCRIPT

  • 8/12/2019 fibonacci advanced

    1/10

    ADVANCED PROBLEMS AND SOLUTIONS

    Edited by

    RAYMO ND E. WHI TNEY

    Lock Haven State College, Lock Haven, Pennsylvania

    Se nd a l l c om mun ic a t ions c onc e rn ing Adva nc e d Pr ob le m s a nd So lu tions to R a ymond E .

    W hi tne y , M a the m a t i c s De p a r tm e n t , Loc k Ha ve n S ta t e C o l l e ge , Loc k Ha ve n , Pe nn s y lva n ia

    17745 . Th i s de p a r tm e n t e s pe c ia l ly we lc o me s p rob le ms be l i e ve d to be new o r e x te nd ing o ld

    re s u l t s . P r op os e r s s hou ld s ubm i t s o lu t ions o r o the r in fo rma t ion tha t w i l l a s s i s t t he e d i to r .

    To fa c i l i t a t e t he i r c ons ide ra t ion , s o lu t ions s hou ld be s ubmi t t e d on s e p a ra t e s igne d s h e e t s

    wi th in two mon ths a f t e r pub l i c a t ion of the p rob le m s .

    H-195 Proposed by Verner

    Hogg att, Jr.,SanJose State College,SanJose, California

    Consider the array indicated below:

    2

    5

    3

    34

    89

    2

    4

    9

    22

    56

    45

    3

    7

    6

    38

    4

    27

    65

    5

    6

    6

    22

    (i) Show tha t the row sum s ar e F , n ^ 2.

    ( i i) Show tha t the r is i ng diagona l sum s ar e the convolut ion of

    (

    F

    2 n - l ^

    = 0

    a n d

    { ^ 2 , 2 ) } ;

    = 0

    ,

    t he ge n e ra l i z e d nu mb e rs of H a r r i s a nd S ty le s .

    H-196 Proposed by J. B. Roberts, Reed College, Portland, Oregon.

    (a) Let A

    0

    be the s e t o f i n t e g ra l pa r t s of t he pos i t i ve in t e g r a l mu l t ip l e s o f r , whe r e

    1 + ^ 5

    T =

    413

  • 8/12/2019 fibonacci advanced

    2/10

    414 ADVANCED PRO BLE MS AND SOLUTIONS [Oct .

    a nd l e t A - , m = 0 , 1 , 2 , , be the s e t o f i n t e g ra l pa r t s o f t he num be r s n r

    2

    for n 61 A . Pr ov e tha t the col lec t ion of Z of a l l pos i t ive in teg ers is the d is jo in t

    union of the A..

    J

    (b) Ge ner a l iz e the pro pos i t ion in (a ) .

    H-197 Proposed byLawrence

    Somer

    Universityof Hi n o s, Urbana, Illinois.

    n

    s ion re l a t ions h ip :

    Le t { i r } _ be the t -F ib ona c c i s e que n c e s w i th pos i t i ve e n t r i e s s a t is fy ing the re c u r -

    F ind

    (t)

    u

    n

    . l i m

    n

    t

    Vu .

    Z

    i

    n - i

    i= l

    (t)

    u

    n+1

    u

    oo n

    SOLUTIONS

    HYPER

    -TENSION

    H-185 Proposedby L

    Carlitz

    DukeUniversity Durham, North Carolina.

    Show that

    (1 - 2x)

    n

    = ( - D

    n

    "

    k

    (

    n

    2

    +

    k

    k

    ) (

    2

    k

    k

    ) (1 - v )

    n

    -

    k

    2 F l

    [-k; n + k + 1; k + 1; x]

    k=0 ^ / \ /

    w h e r e

    2

    Fi [a ,b ; c ; x ] de no te s the hyp e rge om e t r i c func tion .

    SolutionbytheProposer.

    We s tar t wi th the ident i ty

    r s

    E

    (2r + 3s) (y - z) z _

    r s ( r + 2s ) ^

    +

    .2r +3 s+l 1 - y - z

    r , s = 0 y*

    Now pu t y = u + v, z = v, so tha t

    fr

    v V

    2 r

    +

    3 s ) L

    u v

    = 1

    1 ;

    JL J r l s l ( r + 2s ) l ,-

    +

    ,2r +3 s+l 1 - u - 2v

    r . s = 0 *

    U v

    '

  • 8/12/2019 fibonacci advanced

    3/10

    1972] ADVANCED PRO BLEM S AND SOLUTIONS 415

    The right-hand side of (*) is equal to

    n=0

    while the le f t -h and s ide

    oo oo

    E

    (2r + 3s)l r s V

    1

    / -x\

    k

    f

    2r

    + 3 s + k \ , ^

    r l s l ( r

    +

    2s)l

    U V

    ^

    {

    ~

    1}

    { k J

    ( U + V )

    r , s = 0 k= 0

    It follow s th at

    , , . ,n \" ^ , , ,k v ^ .

    n

    x k / 2 r + 3s + k \ (2r + 3s ) l r s

    =

    L - < -

    1 ) r + t u

    s

    P +

    t

    ( m o d

    P>"

    FIBONACCI IS A SQUARE

    H-187 Proposed by Ira

    Gessel

    Harvard University, Cambridge,

    Massachusetts.

    Pro b le m: Show tha t

    a

    pos i t i ve in t e ge r

    n is a

    F i b o n a c c i n u m b e r

    if

    and only

    if

    e i the r

    5n

    2

    +

    4 o r

    5n

    2

    - 4 is a

    s q u a r e .

    Solution by the Proposer.

    L e t

    F

    0

    = 0, F

    4

    = 1, F - = F

    +

    F _-

    be the F ibona c c i s e r i e s a nd L

    0

    = 2 , =

    1 ,

    L

    ,- = L

    +

    L

    n

    be the Lu cas ser ie s . I t i s wel l known tha t

    r+ 1

    r r - 1

    (1) ( -I)"

    +

    F*

    r

    =

    F

    r + 1

    F

    r

    _

    1

    (2)

    L

    r

    =

    F

    r + 1

    +

    F

    r

    _

    x

    Sub t ra c t ing fou r t ime s the f i r s t f rom the s qua re of the secon d equat io n, we have

    whe nc e

    L

    2

    _

    4 ( - l )

    r

    -

    4 F

    2

    = (F _,, - F -

    )

    2

    = F

    2

    ,

    r

    r r+1 r - 1 r

    5 F

    2

    +

    4 ( - l )

    r

    = L

    2

    .

    r

    r

  • 8/12/2019 fibonacci advanced

    6/10

    418 ADVANCED PR OB L EM S

    AND

    SOLUTIONS

    [Oct.

    T h u s if n is a F i b o n ac c i n u m b e r , e i t h e r 5n

    2

    + 4 o r 5n

    2

    - 4 is a s q u a r e .

    I have

    two

    p r o o f s

    of the

    c o n v e r s e .

    F i r s t

    Proof. We us e the

    t h e o r e m ( H a r dy

    and

    W r i g h t ,

    An

    In t roduc t ion

    to the

    T h e o r y

    of

    N u m b e r s ,

    p. 153)

    tha t

    if p and q ar e

    i n t e g e r s ,

    x is a

    r e a l n u m b e r ,

    and

    | (p /q)

    - x | 2, so by the two t h e o r e m s

    quoted above , k /n = F - / F for s o m e r, and s inc e bo th f ra c t ions a re r e d u c e d , n = F .

    Second Proof. A s s u m e 5n

    2

    4 = m

    2

    . T h e n m

    2

    - 5n

    2

    = 4, so

    m + \/ 5n m - N/5n _

    n

    "

    2 * 2

    l

    '

    and s ince

    m and n

    ha ve

    the

    s a m e p a r i t y ,

    m

    + \

    r

    5n , m - V5n

    _ _ and

    a r e i n t e g e r s

    in Q ( N / 5 ) ,

    w h e r e

    Q is the

    r a t i o n a l s ,

    and

    s inc e the i r p roduc t

    is 1,

    the y

    are

    u n i t s . It is

    wel l known (Hardy

    and

    W r igh t ,

    p. 221)

    t h a t

    the

    on ly in t e g ra l un i t s

    of

    Q(\/I>)

    are

    of the fo rm

    x~

    , w h e r e x = (1 + \/"5)/2.

    T h e n

    we

    have

    (m

    +

    ^ 5 n ) / 2

    = x

    r

    = \ (x

    r

    + y

    r

    ) + f L ^ ^ L N/5 ,

    2

    L ^ J

    w h e r e y = -1 /x . Now x. + y = L and

  • 8/12/2019 fibonacci advanced

    7/10

    1972] ADVANCED PROB LEM S AND SOLUTIONS 419

    (

    X

    r

    - y

    r

    ) / ^ 5 = F

    r

    (Hardy and Wri ght , p . 148) . Thu s

    | ( m + N/5n) = | ( L

    r

    + ^ 5 F

    r

    > ,

    so n = F .

    S U M S E R IE S

    H-189 Proposedby L Carlitz DukeUniversity Durham, NorthCarolina Corrected).

    Show that

    (2r + 3s ) l (a - b y )

    r

    b

    S

    y

    r + 2 s

    ^ r l s ( r + 2 s ) l

    H

    _, ,2r +3 s+l , ,

    2

    r 9 S = 0

    ' (1 + ay) 1 - ay - by

    2

    Solution

    b y

    theProposer.

    P u t

    i - ar

    so tha t

    1 - ax - bx

    2

    A

    m=0

    m

    x

    m

    1 = E *

    (1 - ax - bx

    2

    )(l - y)

    n

    J

    m , n = 0

    v

    ni n

    x

    y

    R e p la c ing y by x" y th i s be c o me s

    1 V ^

    m

    ~

    n n

    z /

    J

    G x y

    m , n = 0

    1 - ax - bx

    2

    )( l - x y)

    Hence tha t par t of the expans ion of

    (1 - ax - bx

    2

    ) ( l - x V )

    tha t i s independen t of x is equal to

  • 8/12/2019 fibonacci advanced

    8/10

    420 ADVANCED PRO BLE MS AND SOLUTIONS [Oct .

    (*)

    1 - ay - by

    2

    On the oth er hand, s ince

    (1 - ax - bx

    2

    )( l - x~ y) = (1 + ay) - x(a - by) - bx

    2

    - x " y ,

    we have

    (1 - a x - b x ^ U - x - V ) "

    1

    = [ ^ a - b ^ + b x ^ + x - V l

    v T S t

    E

    (r + s + t ) l (a - by) b y r+ 2s -t

    r ls l t l , - ^

    x

    r + s + t + l

    x

    r,s,t=0

    ( 1 + a

    ^

    The pa rt of th is sum tha t i s independent of x is obta ined by taking t = r + 2s . We ge t

    E

    ( 2 r + 3 s) l ( a - b y ) V y

    r 2 s

    r

    T

    .sI(r + 2s)l ,.. , v2 r+3 s+i

    r , s = 0

    ( 1 + a y )

    Since th is i s equal to (*) , we have proved the s ta ted ident i ty .

    I T S A M O D W O R L D

    H-190 ProposedbyH. H.Ferns Victoria, ritish Columbia.

    P r o v e t h e f o l l o w i n g

    2

    r

    F = n (mod 5)

    2

    r

    L = 1 (mod 5) ,

    whe re F a nd L a r e the n F ibona c c i a nd n Luc a s nu m be rs , r e s p e c t iv e ly , a nd r i s

    the lea s t res idu e of n - 1 (mod 4) .

    Solutionb ytheProposer.

    In an unpub l ished pape r by the pr op os er , i t i s shown tha t

    ^\ E

    2k

    n

    +1

    )

    5

    k

    k=0

  • 8/12/2019 fibonacci advanced

    9/10

    1 9 7

    2 ] ADVANCED PROB LEM S AND SOLUTIONS 421

    Hence

    IT]

    k = l

    x

    '

    Thus

    5

    k

    .

    2

    n 1

    F

    n

    = n (mod 5)

    Let n - 1 = 4m + r , wh ere 0 < r < 4 . Then

    But

    Hence

    2 F = 2 F = 2 2 F = n (mod 5) .

    2

    4 m

    = ( 2

    4

    }

    m ^

    1 ( m o d 5 )

    ^

    To p rove

    use

    2

    r

    F = n (mod 5) .

    2

    r

    L = 1 (mo d 5) ,

    M

    k= 0

    %

    '

    (which is der ived in the same paper) and proceed as above*

    JU S T S O MA N Y T W O S

    H-192 Proposed byRonaldAlter UniversityofKentucky Lexington, Kentucky.

    I f

    3n+l

    c

    n

    3=0

    M1+J.

    . s( J H

    prove tha t

    c

    =

    2

    6 n + 3

    . N

    s

    (N odd , n >0 ) .

    n

    olution

    by

    the

    Proposer

    In the s eque nce

    b

    k

    = b

    k - l "

    3 b

    k - 2 '

    ( k

    "

    l j b

    i =

    b

    2 = D >

    it is ea sy to show tha t 2 is the hi gh est pow er of 2 tha t div ide s b. if and only if k = 3

    (mod 6) , Als o , by der iv in g the ap pro pr ia te Bine t for mu la , i t fo l lows tha t

  • 8/12/2019 fibonacci advanced

    10/10

    422 ADVANCED PROB LEM S AND SOLUTIONS [Oct .

    Thus

    ^ i m - c - ^ i - -

    k ^ 1

    2

    j=o

    The de s i re d re s u l t fo l lows by obs e rv ing

    6n+3 6n+2 n *

    Ed i t o r i a l No te : P l e a s e s ubmi t s o lu t ions fo r a ny o f t he p rob le m p r opo s a l s . W e ne e d f re s h

    blood

    A GOLDEN SECTION SEARCH PROBLEM

    R E X H . S H U D D E

    Waval Post Graduate School, Monterey, California

    Afte r t i r i ng of u s ing num e rou s qua d ra t i c func t ions a s ob je c t ive func t ions fo r e xa m ple s

    in my ma the ma t i c a l p rog ra m min g c ou rs e , I pos e d the fo llowing p ro b le m fo r

    myself:

    De s ign

    a un imoda l funct ion ove r t he (0,1) i n t e rva l wh ic h i s c onc a ve , h a s a ma x im um in the in t e r io r

    o f (0 ,1 ) , a nd i s no t a qua d ra t i c func t ion . The pu rp os e w a s to de m ons t ra t e num e r i c a l ly the

    golden sec t ion search.*

    My f i r s t thoughts we re to add two funct ions which ar e concav e o ver the (0 ,1) in te rva l

    with the pr op er ty tha t one goes to -o a t 0 and the o th er goes to -o a t 1 . My two in i t ia l

    c ho ic e s we r e log x a nd l / (x - 1 ) . The go lde n s e c t ion s e a rc h s t a r t s a t t he two po in t s x

    t

    =

    1 - (1/0) and x

    2

    = l/(p wh ere 0 = (1 +

    N / 5 ) / 2 .

    Af t e r s e a rc h ing w i th 8 po in t s , I no t i c e d tha t

    the in te rva l of un cer ta in ty s t i l l conta in ed the f i r s t se ar ch point so I thought i t abou t t im e to

    f ind the loca t io n of the ma xim um an alyt i ca l ly . I wa s dumfounded to d is co ve r tha t i f I con t in-

    ued indefin i te ly wi th the se ar ch m y in ter va l of un cer ta in ty would s t i l l conta in the in i t ia l sea rc h

    point .

    * Doug la s J . W i lde , Op t imum Se e k ing M e tho ds , P re n t i c e Ha l l , Inc . (1964) .