Transcript

Complex Fibonacci Numbers and Fibonacci QuaternionsAuthor(s): A. F. HoradamSource: The American Mathematical Monthly, Vol. 70, No. 3 (Mar., 1963), pp. 289-291Published by: Mathematical Association of AmericaStable URL: http://www.jstor.org/stable/2313129 .

Accessed: 27/08/2013 10:22

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Mathematical Association of America is collaborating with JSTOR to digitize, preserve and extend access toThe American Mathematical Monthly.

http://www.jstor.org

This content downloaded from 128.103.149.52 on Tue, 27 Aug 2013 10:22:14 AMAll use subject to JSTOR Terms and Conditions

1963] MATHEMATICAL NOTES 289

COMPLEX FIBONACCI NUMBERS AND FIBONACCI QUATERNIONS

A. F. HORADAM, University of New England, Armidale, Australia

1. Introduction. In [1], [2] and [3], properties of the generalised Fibonacci sequence H1 H2 H3 H4 Hs H6 . . . Hn, . . . (1) H(_HXp): p p+q 2p+q 3p+2q 5p+3q 8p+5q (p-q)Fn+qFn+l*

were discussed (p, q arbitrary integers), where F. is the nth term of the classical Fibonacci sequence F( H1o). The Lucas sequence is G-H12.

Here, we extend the results of [1 ] and [2 ] to the complex case. Because of the various types of sequences dealt with, it is convenient to distinguish the cor- responding p, q, 1, m, e of [1] by an appropriate suffix. Thus, in (1), we must now write PH, * - - , eH for p, * * * ,e.

2. Generalised complex Fibonacci sequence. Define the nth generalised com- plex Fibonacci number

(2) D. = Hn + iHn+l = (PH - qH + iqH)Fn + (qH + ipH)Fn+l

using (1), whence

(3) PD = (1 + i)PH + iqH, D = qH + ipH

i.e., the generalised complex Fibonacci sequence is D -H(1+i)pH+iqff,q+ipH. Special cases: (a) PH=1, qH=O: complex Fibonacci sequence C-H1+i,i

(b) PH = 1, qH = 2: complex Lucas sequence H1+3i,2+i. From (a), we have

(4) ec = 2pa-qc = 2( + i)-i=2 + i,

and from [1], (3) and (4),

eD = { (1 + i)PH + iqH}2 { (1 + i)PH + iqH} { qH + ipH} - { qH + iPH}2

= (2 + i){pH - pHqH - qH} = eCeH.

Another sequence, among many, for which ec =2 +i is H1,-i implying Hn =-iCn-1. The conjugate C,n of Cn = F,, +iFn+l leads to the sequence C-H_i,_ or, alternatively, H1,i for which Hn = iCn-,. Also Fc = ec.

Parallelling [1], we have, for C, 2

Cn-lCn+l - Cn = (-1)nec 2 2

Cn, + Cn,+1 = iecF2n+2 2 2

Cn+1-Cn-1 = iecF2n+l (5) ~ ~~2 32= 22

(2Cn+lCn+2) + (CnCn+3) = (2Cn+lCn+2 + Cn) (Pythagorean theorem) Cn,,+ + (_ 1)rC,_,

= Fr+1 + Fr-1 = a, + bT (independent of n). C,,

This content downloaded from 128.103.149.52 on Tue, 27 Aug 2013 10:22:14 AMAll use subject to JSTOR Terms and Conditions

290 MATHEMATICAL NOTES [March

Also, for moduli, writing I Cn4 2 = cn, we find

2 2

Jcn = Fn + Fn+l = F2n+l (6) {2f + 2~ Cn Cn-1 =Fn+l-Fn_l= F2n

i.e., terms F1, F2, F3s, F4, *, F2n, F2n+l, * * * of F are expressible as co, c1-co, c1, c2 - c1, . . . i , cn-Cl C* *, respectively. Results (5) and (6) may be ex- tended to the general sequence without any difficulty, though some of the formulae begin to look a bit awkward. In the case of jDn| 2 =H + H = (2PH-qH)H2,+1-eHF2n+1, we see that there is no simple analogue to the moduli-sequence above for F.

Whereas [3 ] the geometrical meaning of eH is the area of a certain parallelo- gram, in the complex cases ec and eD are merely complex numbers. Notice that arg Cn = tan-' (Fn+1/Fn) -> tan-' (1+V/5)/2 _ 58?17' and -r/4 < arg Cn <tan' 2( 63?26%), i.e., Cn moves in a very restricted region of the complex plane. Similarly for Dn.

3. Mixed complex Fibonacci number. It is tempting to experiment with a complex number Fn+iGn+l but nothing very interesting seems to develop. However, we may define the mixed complex Fibonacci number

(7) Mn = Fn + iGn = (1 - i)Fn + 2iFn+1

whence PM=l+i(=pc), qM=2i(=2qc), em=6, i.e., M--Hl+i,2i. Note that arg Mn= tan 1(Gn/Fn) ->tan-'1S5.

Generalising for Hn, Hn' (nth terms of Hpq, Hrs respectively) we have

(8) Hn = Hn + iJHn

whence N=H,+ir,q+is and eN=eH-eH1+i(2pr-2qs-qr-ps), i.e., eN is purely real if 2 (pr-qs)=qr+ps and purely imaginary if eH = eH'. In particular, if r=q, s=p, then eN= 2(p2-q2) -i(p2+q2) =p2jc-q2ec.

4. Fibonacci quaternions. Define the nth Fibonacci quaternion to be

(9) Qn = Fn + iFn+1 + jFn+2 + kFn+3,

(where i2 =j2 = k2 1, ij = -ji = k, jk = -kj = i, ki = -ik =j) with conjugate quaternion Qn = Fn-iFn+1-jFn+2 kFn+3, so that from [1 ],

3 2

(10) QnQn= 1 Fni = 3F2n+3

and 2

(1) Qn= 2FnQn -QnQn

which, on division by Qni 0, yields the obviously true relation 2 Fn = Qn +Q.

This content downloaded from 128.103.149.52 on Tue, 27 Aug 2013 10:22:14 AMAll use subject to JSTOR Terms and Conditions

1963] MATHEMATICAL NOTES 291

Furthermore, we find 3

(12) EI Qn+i = Gn+3 + iGn+4 + jGn+5 + kGn+6 i=O

which is the quaternion Q,+3 for the Lucas sequence. Generalising, we define the nth generalised Fibonacci quaternion

(13) Pn- =n + iHn+l + jHn?+2+ k1n+3

yielding, in particular, PnPn = 3 { (2PH - qH) H2n+3-eHF2n+3 }

It is not intended to carry the theory any further at this stage. Our definitions of complex Fibonacci sequences and quaternions do lead to

some reasonably neat relationships. However, the question remains: are there other definitions which are more effective?

References 1. A. F. Horadam, A generalized Fibonacci sequence, this MONTHLY, 68 (1961) 455-459. 2. , Fibonacci number triples, this MONTHLY, 68 (1961) 751-753. 3. , Fibonacci numbers and a geometrical paradox, Math, Mag., 35 (1962) 1-11.

THE INVERSION OF A CONVOLUTION TRANSFORM WHOSE KERNEL IS A LAGUERRE POLYNOMIAL

D. V. WIDDER, Harvard University

Recently Ta Li [2] has shown that a certain convolution transform whose kernel involves a Tchebycheff polynomial can be inverted by another similar convolution. R. G. Buschman [1 ] has proved a similar result for Legendre poly- nomials. The present author has given new proofs of these two inversions by the method of the Laplace transform. That method reveals how similar formulas may be invented, and we apply it here to invert the transform

(1) f(x) = Ln(- t)g(t)dt,

where Ln(t) is the Laguerre polynomial

1 Ln(X) = - Dn(e-xXn).

ni

We prove the following theorem.

THEOREM. If g(x) C C1 for 0 ?x < oo, g(0) = 0, and f(x) is defined by equation (1), then f(x) E C2 for 0 <x < oo, f(O) =f'(O) = 0, and ( x

(2) g (x) Ln+ (t -x)ex-'(D -1) 2f(t) dt. IL

This content downloaded from 128.103.149.52 on Tue, 27 Aug 2013 10:22:14 AMAll use subject to JSTOR Terms and Conditions