Chaos, Solitons and Fractals 32 (2007) 1615–1624

www.elsevier.com/locate/chaos

On the Fibonacci k-numbers

Sergio Falcón *, Ángel Plaza
Department of Mathematics, University of Las Palmas de Gran Canaria (ULPGC), 35017-Las Palmas de Gran Canaria, Spain

Accepted 1 September 2006

Communicated by Prof. Marek-Crnjac

Abstract

We introduce a general Fibonacci sequence that generalizes, between others, both the classic Fibonacci sequence and
the Pell sequence. These general kth Fibonacci numbers fF k;n g1
n¼0 were found by studying the recursive application of
two geometrical transformations used in the well-known four-triangle longest-edge (4TLE) partition. Many properties
of these numbers are deduce directly from elementary matrix algebra.
 2006 Published by Elsevier Ltd.

1. Introduction

In the present days there is a huge interest of modern science in the application of the Golden Section and Fibonacci
numbers [1–19]. The Fibonacci numbers Fn are the terms of the sequence {0, 1 , 1, 2 , 3, 5 , . . .} wherein each term is the
sum of the two preceding terms, beginning with the values F0 = 0, and F1 = 1. On the other pffiffi hand the ratio of two con-
secutive Fibonacci numbers converges to the Golden Mean, or Golden Section, s ¼ 1þ2 5, which appears in modern
research in many fields from Architecture, Nature and Art [20–30] to physics of the high energy particles [31–33] or
theoretical physics [34–41].
As an example of the ubiquity of the Golden Mean in geometry we can think of the ratio between the length of a
diagonal and a side of a regular pentagon. The paper presented here was originated for the astonishing presence of the
Golden Section in a recursive partition of triangles in the context of the finite element method and triangular
refinements.

1.1. Grid generation and triangles

Grid generation and, in particular, the construction of ‘quality’ grids is a major issue in both geometric modeling
and engineering analysis [42–46]. Many of these methods employ forms of local and global triangle subdivision and
seek to maintain well shaped triangles. The four-triangle longest-edge (4TLE) partition is constructed by joining the
midpoint of the longest-edge to the opposite vertex and to the midpoints of the two remaining edges [47,48]. The
two subtriangles with edges coincident with the longest-edge of the parent are similar to the parent. The remaining

*
Corresponding author. Tel.: +34 928 45 88 27; fax: +34 928 45 87 11.
E-mail address: sfalcon@dma.ulpgc.es (S. Falcón).

0960-0779/$ - see front matter  2006 Published by Elsevier Ltd.

doi:10.1016/j.chaos.2006.09.022
1616 S. Falcón, Á. Plaza / Chaos, Solitons and Fractals 32 (2007) 1615–1624

two subtriangles form a similar pair that, in general, are not similar to the parent triangle. We refer to such new triangle
shapes as ‘dissimilar’ to those preceding. The iterative partition of obtuse triangles systematically improves the triangles
in the sense that the sequence of smallest angles monotonically increases, while the sequence of largest angles monoton-
ically decreases in an amount (at least) equal to the smallest angle of each iteration [44,48].
In this paper, we show the relation between the 4TLE partition and the Fibonacci numbers, as another example of
the relation between geometry and numbers. The use of the concept of antecedent of a (normalized) triangle is used to
deduce a pair of complex variable functions. These functions, in matrix form, allow us to directly and in an easy way,
present many of the basic properties of some of the best known recursive integer sequences, like the Fibonacci numbers
and the Pell numbers.

2. Normalized triangles, antecedents and complex valued functions

Since we were interested in the shape of the triangles, each triangle is scaled to have the longest-edge of unit length.
In this form, each triangle is represented for the three vertices: (0, 0), (1, 0) and z = (x, y). Since the two first vertices are
the extreme points of the longest-edge, the third vertex is located inside two bounding exterior circular arcs of unit
radius, as shown in Fig. 1. In the following, for any triangle t, the edges and angles will be respectively denoted in
decreasing order r1 P r2 P r3, and c P b P a.
Definition 1. The longest-edge (LE) partition of a triangle t0 is obtained by joining the midpoint of the longest-edge of t0
with the opposite vertex (Fig. 2(a)). The four-triangle longest-edge (4TLE) partition is obtained by joining the midpoint
of the longest-edge to the opposite vertex and to the midpoints of the two remaining edges (see Fig. 2(b)).
In the 4TLE scheme, subdivision leads to subtriangles that are similar to some previous parent triangles in the refine-
ment tree so generated. Other subtriangles may result that are not in such similarity classes yet and we refer to these as
new dissimilar triangles. We define the class Cn as the set of triangles for which the application of the 4TLE partition
produces exactly n dissimilar triangles.
Let us begin by describing a Monte Carlo computational experiment used to visually distinguish the classes of tri-
angles by the number of dissimilar triangles generated by the 4TLE partition. We proceed as follows: (1) Select a point
within the mapping domain comprised by the horizontal segment and by the two bounding exterior circular arcs. This
point (x, y) defines the apex of a target triangle. (2) For this selected triangle, 4TLE refinement is successively applied as
long as a new dissimilar triangle appears. This means that we recursively apply 4TLE and stop when the shapes of new
generated triangles are the same as those already generated in previous refinement steps. (3) The number of such refine-
ments to reach termination defines the number of dissimilar triangles associated with the initial triangle and this numer-
ical value is assigned to the initial point (x, y) chosen. (4) This process is progressively applied to a large sample of
triangles (points) uniformly distributed over the domain. (5) Finally, we graph the respective values of dissimilar trian-
gles in a corresponding color map to obtain the result in Fig. 3.

z=(a,b)

0 0

(0,0) (0.5,0) (1,0)

Fig. 1. Diagram for representing shape triangles.

 S. Falcón, Á. Plaza / Chaos, Solitons and Fractals 32 (2007) 1615–1624 1617

C
γ0 r2
r
3 t t1
1

β0 α0
t0 t0
A B
r
1
a b
Fig. 2. (a) LE partition of triangle t0, (b) 4TLE partition of triangle t0.

0.9 16

0.8 14

0.7
12
0.6
2 10
0.5
8
0.4
6
0.3

0.2 3 4

4 4
0.1 5 2
7
6 5 6
7
18 18
0
0 0.2 0.4 0.6 0.8 1

Fig. 3. Subregions for dissimilar triangle classes generated by Monte Carlo simulation.

Definition 2 (4TLE left and right antecedents). A given triangle tn+1, has two (different) triangles tn, denoted here as left
and right antecedents, whose 4TLE partition produces triangle tn+1.
As an example, triangle tn+1 in the diagram with vertices (0, 0), (1, 0) and z in Fig. 4(a), has left antecedent tn with
vertices z, (0, 0), and z + 1 in Fig. 4(b), and right antecedent tn with vertices z, (1, 0), and z  1 in Fig. 4(c).

t n+1
z z+1 z—1 z
(0,0) (1,0)
t n+1 t n+1
(a) Triangle tn+1
(0,0) (1,0) (0,0) (1,0)

(b) Left antecedent: t n (c) Right antecedent: t n

Fig. 4. Two antecedents for the 4TLE partition of triangle tn.

1618 S. Falcón, Á. Plaza / Chaos, Solitons and Fractals 32 (2007) 1615–1624

Theorem 3. The relation between the apex of a given triangle z in the right half of the diagram and the apices of its left and
1 1
right antecedents may be mathematically expressed by the maps fL ðzÞ ¼ zþ1 , and fR ðzÞ ¼ 2z , complex z.

1
Remark. Notice that for z having 2
6 ReðzÞ 6 1 then Re(fL(z)) 6 Re(fR(z)) (and hence the ‘left’/‘right’ terminology
given to these complex functions).

Theorem 4. The class separators determined experimentally in Fig. 3 may be generated mathematically as a recursive com-
position of left and right maps fL(z), and fR(z).
Function fR is a Moebius transformation (also homography or fractional linear transformation) [49,50], while
function fL is an anti-homography. Both may be considered as maps of the extended plane C into itself. fR is a con-
formal map, and hence it preserves angles in magnitude and direction, and straight lines and circles are transformed
into straight lines and circles. On the other hand, fL is not conformal, but angles are preserved in magnitude and
reversed in direction, as the complex conjugation. Also fL takes circles to circles (straight lines count as circles of
infinite radius).  
a b
In general a Moebius transformation hðzÞ ¼ azþb czþd
is defined by the matrix: H ¼ , whose elements are con-
c d
stant complex numbers and its determinant is not null to avoid the constant transformation. In similar way, an
zþb
anti-homography presents the form hðzÞ ¼ a czþd
and also has the same associated complex matrix.
   
0 1 0 1
In our case, let R ¼ , and L ¼ be the associated matrices to functions fR and fL, respectively. The
1 2 1 1
composition of two of such functions has as associated matrix the product of the matrices associated to the two initial
transformations. Similarly, any particular combination of transformations fR and fL is determined by the product of the
corresponding matrices in the same order. For instance, transformation ðfR  fL ðzÞÞ ¼ fR ðfL ðzÞÞ ¼ fR ðzþ1 1
Þ ¼ 21 1 ¼ 2zzþ1
þ1
      zþ1
0 1 0 1 1 1
could be given more easily by the matrix product R  L ¼  ¼ . Therefore, from now on, we
1 2 1 1 2 1
will substitute the use of the transformations fR and fL by the use of the associated matrices R and L.
Let us find the product Rk1L associated to the composition fRk1  fL which will be used below. It is easy to prove
       
k þ 2 k  1 k þ 2 k  1 0 1 k1 1
that Rk1 ¼ for all k P 1, and so Rk1  L ¼  ¼ .
k þ 1 k k þ 1 k 1 1 k 1

3. k-Fibonacci numbers

In this section, a new generalization of the Fibonacci numbers is introduced. It should be noted that the recurrence
formula of these numbers depend on one integral parameter instead of two parameters. We shall show that these num-
bers are related with the complex valued functions given above, and then, in some sense, with the 4TLE partition of
normalized triangles.
Definition 5. For any integer number k P 1, the kth Fibonacci sequence, say {Fk,n}n2N is defined recurrently by
F k;0 ¼ 0; F k;1 ¼ 1; and F k;nþ1 ¼ kF k;n þ F k;n1 for n P 1:
Particular cases of the previous definition are:

• If k = 1, the classic Fibonacci sequence is obtained:

F0 = 0, F1 = 1, and Fn+1 = Fn + Fn1 for n P 1:
{Fn}n2N = {0, 1, 1, 2, 3, 5, 8 , . . .}.
• If k = 2, the classic Pell sequence appears:
P0 = 0, P1 = 1, and Pn+1 = 2Pn + Pn1 for n P 1:
{Pn}n2N = {0, 1, 2, 5, 12, 29, 70 , . . .}.
• If k = 3, the following sequence appears:
H0 = 0, H1 = 1, and Hn+1 = 3Hn + Hn1 for n P 1:
{Hn}n2N = {0, 1, 3, 10, 33, 109 , . . .}.
The relation between matrix Rk1 Æ L and the kth Fibonacci sequence is given by the following proposition.
Proposition 6. For any integer n P 1 holds:
 S. Falcón, Á. Plaza / Chaos, Solitons and Fractals 32 (2007) 1615–1624 1619
 
F k;nþ1  F k;n F k;n
ðRk1  LÞn ¼ : ð1Þ
F k;nþ1  F k;n1 F k;n þ F k;n1

Proof (By induction). For n = 1:

   
k1 1 F k;2  F k;1 F k;1
Rk1  L ¼ ¼
k 1 F k;2  F k;0 F k;1 þ F k;0
since Fk,0 = 0, Fk,1 = 1, and Fk,2 = k.
Let us suppose that the formula is true for n  1:
 
F k;n  F k;n1 F k;n1
ðRk1  LÞn1 ¼ :
F k;n  F k;n2 F k;n1 þ F k;n2
Then,
   
F k;n  F k;n1 F k;n1 k1 1
ðRk1  LÞn ¼ ðRk1  LÞn1 ðRk1  LÞ ¼ 
F k;n  F k;n2 F k;n1 þ F k;n2 k 1
   
ðk  1ÞF k;n þ F k;n1 F k;n F k;nþ1  F k;n F k;n
¼ ¼ : 
kF k;n F k;n þ F k;n1 F k;nþ1  F k;n1 F k;n þ F k;n1
Particular cases are:

• If k = 1, the classic Fibonacci sequence is obtained: F0 = 0, F1 = 1, and Fn+1 = Fn + Fn1 for n P 1, so

 
F n1 F n
Ln ¼ . This formula shows the relation between the function fL and the classic Fibonacci sequence.
F n F nþ1  
n F2 F1
Note that matrix L is similar to the nth power of the Fibonacci Q matrix defined by Q ¼ , from where
  F1 F0
F nþ1 F n
Qn ¼ [51].
F n F n1
• If k = 2, we get the classic Pell sequence: P0 = 0, P1 = 1, and Pn+1 = 2Pn + Pn1 for n P 1, and then:
 
P n þ P n1 Pn
ðR  LÞn ¼ .
2P n P n þ P n1
• If k = 3, sequence {Hn}n2N is obtained, with H0 = 0, H1 = 1, and Hn+1 = 3Hn + Hn1, so:
 
2H n þ H n1 Hn
ðR2  LÞn ¼ .
3H n H n þ H n1

4. Properties from the determinant of matrix (Rk1 Æ L)n

For the shake of clarity we note in the sequel by T the matrix Rk1 Æ L. In this section, we shall study some properties
for the kth Fibonacci sequences which are directly obtained from the determinant of matrices Tn = (Rk1 Æ L)n, that is,
from the associated matrices to transformations fR and fL.
Proposition 7 (Catalan identity). F k;nþrþ1 F k;nþr1  F 2k;nþr ¼ ð1Þnþr .

Proof. If in Proposition 6 n is changed by n + r, the following matrix is obtained:

 
F k;nþrþ1  F k;nþr F k;nþr
ðRk1  LÞnþr ¼
F k;nþrþ1  F k;nþr1 F k;nþr þ F k;nþr1
and jðRk1  LÞnþr j ¼ F k;nþrþ1 F k;nþr1  F 2k;nþr . Since jRj = 1 and jLj = 1, we have j(Rk1 Æ L)n+rj = (1)n+r from where
the identity is obtained. h
Particular cases are:

• If k = 1 and r = 0, the Cassini’s identity or Simson formula for the classic Fibonacci sequence appears:
F nþ1 F n1  F 2n ¼ ð1Þn .
1620 S. Falcón, Á. Plaza / Chaos, Solitons and Fractals 32 (2007) 1615–1624

• If k = 2 and r = 0, for the Pell sequence it is obtained: P nþ1 P n1  P 2n ¼ ð1Þn .

• If k = 3 and r = 0, for the sequence {Hn}, we have: H nþ1 H n1  H 2n ¼ ð1Þn .

The relation between matrix Rk1 Æ L and the kth Fibonacci sequence is given by the following proposition.

5. Properties by summing up matrices (Rk1 Æ L)n

In this section, we shall show some properties for the sum of the terms of the kth Fibonacci sequences, obtained by
summing up the first n matrices (Rk1 Æ L)n.
Pn 1
Proposition 8. i¼1 F k;i ¼ k ðF k;nþ1 þ F k;n  1Þ.

Proof. Note that the term a12 in matrix Tn = (Rk1 Æ L)n is precisely Fk,n. Let Sn be the sum of the first n matrices
Tj = (Rk1 Æ L)j. That is, Sn = T + T2 +    + Tn. The argument here is the same that used in the proof of the sum of
the n first terms of a geometric numerical progression:
Since SnT = T2 + T3 +    + Tn + Tn+1, then Sn(T  I2) = Tn+1  T, where I2 is the 2 · 2 unit matrix. And,
therefore, Sn = (Tn+1  T)(T  I2)1.
Note, now, that
     
F k;nþ2  F k;nþ1 F k;nþ1 k1 1 F k;nþ2  F k;nþ1  k þ 1 F k;nþ1  1
T nþ1  T ¼  ¼ :
F k;nþ2  F k;n F k;nþ1 þ F k;n k 1 F k;nþ2  F k;n  k F k;nþ1 þ F k;n  1
On the other hand,
   
k2 1 1 0 1
T  I2 ¼ ) ðT  I 2 Þ1 ¼ :
k 0 k k 2k
Therefore,
   
1 F k;nþ2  F k;nþ1  k þ 1 F k;nþ1  1 0 1
Sn ¼ 
k F k;nþ2  F k;n  k F k;nþ1 þ F k;n  1 k 2k
Pn
and, finally, by obtaining the term a12 of the previous product, since this term is at the same time i¼1 F k;i , we get the
result. h
Particular cases:

• If k = 1, for the classic Fibonacci sequence, we obtain:

X
n
F i ¼ F nþ2  1:
i¼1

• If k = 2, for the Pell sequence we have

X
n
1
P i ¼ ðP nþ1 þ P n  1Þ:
i¼1
2

• If k = 3, the sum of the first elements of the sequence {Hn} is

X
n
1
H i ¼ ðH nþ1 þ H n  1Þ:
i¼1
3

By summing up the first n even terms of the kth Fibonacci sequence we obtain
Pn 1
Proposition 9. i¼1 F k;2i ¼ k ðF k;2nþ1  1Þ.

Proof. The proof is similar to the proof of Proposition 8, and we only show an outline of it. The sum is
S2n = T2 + T4 +    + T2n where T = Rk1 Æ L. By multiplying by T2 and, after some algebra, we get:
S2n = (T2n+2  T2)(T2  I2)1.
 S. Falcón, Á. Plaza / Chaos, Solitons and Fractals 32 (2007) 1615–1624 1621
 
1 1
Note that ðT 2  I 2 Þ1 ¼ 1 and since, the terms a12 of both sides are equal, the formula is
k k k1
obtained. h
Particular cases:

• If k = 1, for the classic Fibonacci sequence, we obtain:

X
n
F 2i ¼ F 2nþ1  1:
i¼1

• If k = 2, for the Pell sequence we have

X
n
1
P 2i ¼ ðP 2nþ1  1Þ:
i¼1
2

• If k = 3, the sum of the first even elements of sequence {Hn} is

X
n
1
H 2i ¼ ðH 2nþ1  1Þ:
i¼1
3

Now, considering Propositions 8 and 9 it is rightly obtained the sum of the first odd terms of the kth Fibonacci
sequence:
Pn 1
Corollary 10. i¼0 F k;2iþ1 ¼ k F k;2nþ2 .

Particular cases:

• If k = 1, for the classic Fibonacci sequence, we obtain:

X
n
F 2iþ1 ¼ F 2nþ2 :
i¼0

• If k = 2, for the Pell sequence we have

X
n
1
P 2iþ1 ¼ P 2nþ2 :
i¼0
2

• If k = 3, the sum of the first odd elements of sequence {Hn} is

X
n
1
H 2iþ1 ¼ H 2nþ2 :
i¼0
3

In a similar way, many formulas for partial sums of term of the kth Fibonacci sequence may be obtained and par-
ticularized for different values of k. For example:
Pn 1
Corollary 11. i¼0 F k;4iþ1 ¼ k F k;2nþ1 F k;2nþ2 .

Let p be a non-null real number. Next Proposition gives us the value for the sum of the first kth Fibonacci numbers
with weights pi:
Proposition 12. For each non-vanishing real number p:
X n  
F k;j p 1
¼ ðpF k;nþ1 þ F k;n Þ  1 : ð2Þ
j¼1
pj p2  kp  1 pnþ1

Pn 1 k1
j
Proof. The proof is similar to those given above, but now considering the matrix S n ¼ j¼1 p R L . h
Eq. (2) is known as Livio’s formula [2]. It should be noted that the denominator of the right-hand side of Livio’s
formula is precisely the characteristic kth Fibonacci polynomial. P F
Now, and also by using elementary matrix algebra, we will obtain a closed expression for limn!1 nj¼1 pk;jj .
1622 S. Falcón, Á. Plaza / Chaos, Solitons and Fractals 32 (2007) 1615–1624
pffiffiffiffiffiffiffiffi
2
Proposition 13. For each real number p, such that p > kþ 2k 4:
Xn
F k;j X 1
F k;j p
lim ¼ ¼ 2 : ð3Þ
n!1
j¼1
p j
j¼1
p j p  kp 1

Proof. The proof is based in the so-called Binet’s formula for the kth Fibonacci sequence:
pffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffi
kþ k 2 4 k k 2 4
Let rn1 ¼ 2 and r 2 ¼ 2 , where r1 and r2 are the fundamental roots of the kth Fibonacci sequence. Then,
r rn
F k;n ¼ r11 r22 . n n
P F
F ðrp1 Þ ðrp2 Þ
As a consequence, if p > r1, then limn!1 pk;n n ¼ limn!1 r1 r2 ¼ 0. And, therefore, limn!1 nj¼1 pk;jj ¼ p2 kp1
p
. h
Particular cases:
P Fj P1 F j 10
• If k = 1, for the classic Fibonacci sequence is obtained: 1 p
j¼1 pj ¼ p2 p1, which for p = 10 gives: j¼1 10j ¼ 89 [2].
P1 P j P
, which for p = 10 gives: 1
p Pj 10
• If k = 2, for the classic Pell sequence appears: j¼1 pj ¼ p2 2p1 j¼1 10j ¼ 79
.
P Hj P1 H j 10
• If k = 3, for sequence {Hn} results: 1 j¼1 pj ¼ p
p2 3p1
, which for p = 10 gives: j¼1 10j ¼ 69
.

6. Properties from the product of matrices (Rk1 Æ L)n

In this section, we shall prove some interesting properties of the kth Fibonacci sequences which may be easily
deduced from the product of matrices of the form (Rk1 Æ L)n. The first property is called convolution product:
Proposition 14.
F k;nþm ¼ F k;nþ1 F k;m þ F k;n F k;m1 : ð4Þ

Proof. Given the matrices (Rk1 Æ L)n, (Rk1 Æ L)m as Eq. (1), and considering the term a12 of the product
(Rk1 Æ L)n · (Rk1 Æ L)m, which is equal to the term a12 of matrix (Rk1 Æ L)n+m we get the result. h
Particular cases:

• If k = 1, for the classic Fibonacci sequence is obtained: Fn+m = Fn+1Fm + FnFm1 (Honsberger formula [2]).
• If k = 2, for the classic Pell sequence appears: Pn+m = Pn+1Pm + PnPm1.

Eq. (4) may be particularized in many ways. For example, if m = n we get: F k;2n ¼ ðF k;nþ1 þ F k;n1 ÞF k;n ¼
F k;nþ1 F k;n1
ðF k;nþ1 þ F k;n1 Þ k
, and, therefore,

1
F k;2n ¼ ðF 2k;nþ1  F 2k;n1 Þ; ð5Þ
k
which may be particularized as follows:

• If k = 1, for the classic Fibonacci sequence is obtained: F 2n ¼ F 2nþ1  F 2n1 .

• If k = 2, for the classic Pell results: P 2n ¼ 12 ðP 2nþ1  P 2n1 Þ.

On the other hand, if m = n + 1 in Eq. (4) we get

F k;2nþ1 ¼ F 2k;nþ1 þ F 2k;n : ð6Þ

By doing m = 2n in Eq. (4) we get

F k;3n ¼ F k;n ðF 2k;nþ1 þ F 2k;n þ F 2k;n1 Þ þ F k;nþ1 F k;n F k;n1 . Now, considering that Fk,n+1 = kFk,n + Fk,n1, we get
1
F k;3n ¼ ðF 3k;nþ1 þ kF 3k;n  F 3k;n1 Þ; ð7Þ
k
which, for k = 1 reads: F 3n ¼ F 3nþ1 þ F 3n  F 3n1 .
Similarly, we can deduce
 S. Falcón, Á. Plaza / Chaos, Solitons and Fractals 32 (2007) 1615–1624 1623

1
F k;4n ¼ ðF 4k;nþ1 þ 2k 2 F 4k;n  F 4k;n1 Þ þ 4F 3k;n F k;n1 ; ð8Þ
k
which, for k = 1 reads: F 4n ¼ F 4nþ1 þ 2F 4n  F 4n1 þ 4F 3n F n1 .
Remark. Notice that, if in the matrix product of the beginning of this section we would have considered the term a22
instead of the term a12, we would have obtained the following equation
F k;nþm þ F k;nþm1 ¼ F k;nþ1 F k;m þ F k;n F k;m þ F k;n F k;m1 þ F k;n1 F k;m1 ;
which, for k = 1 may be written as
F nþm þ F nþm1 ¼ F nþ2 F m þ F nþ1 F m1 ¼ F nþ1 F mþ1 þ F n F m :

7. Conclusions

New generalized kth Fibonacci sequences have been introduced and studied. Many of the properties of these
sequences are proved by simple matrix algebra. This study has been motivated by the arising of two complex valued
maps to represent the two antecedents in an specific four-triangle partition.

Acknowledgement

This work has been supported in part by CYCIT Project number MTM2005-08441-C02-02 from Ministerio de Edu-
cación y Ciencia of Spain.

References

[1] Hoggat VE. Fibonacci and Lucas numbers. Palo Alto (CA): Houghton-Mifflin; 1969.
[2] Livio M. The Golden ratio: The Story of Phi, the world’s most astonishing number. New York: Broadway Books; 2002.
[3] Horadam AF. A generalized Fibonacci sequence. Math Mag 1961;68:455–9.
[4] Shanon AG, Horadam AF. Generalized Fibonacci triples. Am Math Mon 1973;80:187–90.
[5] Hayashu K. Fibonacci numbers and the arctangent function. Math Mag 2003;76:214–5.
[6] Vajda S. Fibonacci and Lucas numbers, and the Golden Section. Theory and applications. Ellis Horwood Limited; 1989.
[7] Gould HW. A history of the Fibonacci Q-matrix and a higher-dimensional problem. Fibonacci Quart 1981;19:250–7.
[8] Kalman D, Mena R. The Fibonacci numbers – exposed. Math Mag 2003;76:167–81.
[9] Benjamin A, Quinn JJ. The Fibonacci numbers – exposed more discretely. Math Mag 2003;76:182–92.
[10] Stakhov A. The Golden section in the measurement theory. Comput Math Appl 1989;17(46):613–38.
[11] Stakhov A. The generalized principle of the golden section and its applications in mathematics, science, and engineering. Chaos,
Solitons & Fractals 2005;26:263–89.
[12] Stakhov A, Rozin B. The Golden shofar. Chaos, Solitons & Fractals 2005;26(3):677–84.
[13] Stakhov A, Rozin B. Theory of Binet formulas for Fibonacci and Lucas p-numbers. Chaos, Solitons & Fractals
2005;27(5):1163–77.
[14] Stakhov A. Fundamentals of a new kind of mathematics based on the golden section. Chaos, Solitons & Fractals
2006;27(5):1124–46.
[15] Stakhov A, Rozin B. The golden algebraic equations. Chaos, Solitons & Fractals 2006;27(5):1415–21.
[16] Stakhov A, Rozin B. The continuous functions for the Fibonacci and Lucas p-numbers. Chaos, Solitons & Fractals
2006;28:1014–25.
[17] Stakhov A. Fibonacci matrices, a generalization of the ‘Cassini formula’, and a new coding theory. Chaos, Solitons & Fractals
2006;30:56–66.
[18] Stakhov A. The golden section, secrets of the Egyptian civilization and harmony mathematics. Chaos, Solitons & Fractals
2006;30:490–505.
[19] Stakhov A. The generalized golden proportions, a new theory of real numbers, and ternary mirror-symmetrical arithmetic. Chaos,
Solitons & Fractals, in press, doi:10.1016/j.chaos.2006.01.028.
[20] Spinadel VW. In: Kim Williams, editor. The metallic means and design. Nexus II: architecture and mathematics. Edizioni
dell’Erba; 1998.
[21] Spinadel VW. The family of metallic means. Vis Math 1999;1(3). Available from: http://members.tripod.com/vismath/ .
[22] Spinadel VW. The metallic means family and forbidden symmetries. Int Math J 2002;2(3):279–88.
[23] Brousseau A. Fibonacci statistics in conifers. Fibonacci Quart 1969;7:525–32.
1624 S. Falcón, Á. Plaza / Chaos, Solitons and Fractals 32 (2007) 1615–1624

[24] Carson J. Fibonacci numbers and pineapple phyllotaxy the two-year. Coll Math J 1978;9(3):132–6.
[25] Caspar DLD, Fontano E. Five-fold symmetry in crystalline quasicrystal lattices. Proc Nat Acad Sci USA 1996;93(25):14271–8.
[26] Greller AM, Matzke EB. Organogenesis, aestivation, and anthesis in the flower of Lilium tigrinum. Bot Gaz 1970;131(4):304–11.
[27] Kirchoff BK, Rutishauser R. The phyllotaxy of costus (Costaceae). Bot Gaz 1990;151(1):88–105.
[28] Mitchison GJ. Phyllotaxis and the Fibonacci series. Sci, New Ser 1977;196(4287):270–5.
[29] Onderdonk PB. Pineapples and Fibonacci numbers. Fibonacci Quart 1979;8:507–8.
[30] Stein W. Modeling the evolution of stelar architecture in vascular plants. Int J Plant Sci 1993;154(2):229–63.
[31] El Naschie MS. Topological defects in the symmetric vacuum, anomalous positron production and the gravitational instanton. Int
J Mod Phys 2004;13(4):835–49.
[32] El Naschie MS. Anomaly cancellation and the mass spectrum of (1) Schrödinger’s cat. Chaos, Solitons & Fractals
2005;23:1089–90.
[33] El Naschie MS. Experimental and theoretical arguments for the number and mass of the Higgs particles. Chaos, Solitons &
Fractals 2005;23:1091–8.
[34] El Naschie MS. Quantum mechanics and the possibility of a Cantorian space-time. Chaos, Solitons & Fractals 1992;1:485–7.
[35] El Naschie MS. On dimensions on Cantor sets related systems. Chaos, Solitons & Fractals 1993;3:675–85.
[36] El Naschie MS. Statistical geometry of a cantor discretum and semiconductors. Comput Math Appl 1995;29(12):103–10.
[37] El Naschie MS. Deriving the essential features of the standard model from the general theory of relativity. Chaos, Solitons &
Fractals 2005;24:941–6.
[38] El Naschie MS. Non-Euclidean spacetime structure and the two-slit experiment. Chaos, Solitons & Fractals 2005;26:1–6.
[39] El Naschie MS. On the cohomology and instantons number in E-infinity Cantorian spacetime. Chaos, Solitons & Fractals
2005;26:13–7.
[40] El Naschie MS. Stability analysis of the two-slit experiment with quantum particles. Chaos, Solitons & Fractals 2005;26:291–4.
[41] El Naschie MS. Dead or alive: Desperately seeking Schrödinger’s cat. Chaos, Solitons & Fractals 2005;26:673–6.
[42] Carey GF. Computational grids: generation, refinement and solution strategies. Taylor and Francis; 1997.
[43] Knupp PM, Steinberg S. The fundamentals of grid generation. Boca Raton (FL): CRC Press; 1994.
[44] Plaza A, Suárez JP, Padrón MA, Falcón S, Amieiro D. Mesh quality improvement and other properties in the four-triangles
longest-edge partition. Comput Aided Geomet Des 2004;21(4):353–69.
[45] Plaza A, Suárez JP, Padrón MA. Fractality of refined triangular grids and space-filling curves. Eng Comput 2005;20(4):323–32.
[46] Plaza A, Suárez JP, Carey GF. A geometrical diagram for similarity classes in triangle subdivision. Comp Aided Geomet Des, in
press.
[47] Rivara MC. Algorithms for refining triangular grids suitable for adaptive and multigrid techniques. Inter J Num Method Eng
1984;20:745–56.
[48] Rivara MC, Iribarren G. The 4-triangles longest-side partition of triangles and linear refinement algorithms. Math Comput
1996;65(216):1485–502.
[49] Marsden JE, Hoffman MJ. Basic complex analysis. New York: W.H. Freeman; 1999.
[50] Schwerdtfeger H. Geometry of complex numbers. New York: Dover Publications, Inc.; 1979.
[51] Weisstein EW. Fibonacci Q-matrix. From MathWorld – A Wolfram Web Resource. Available from: http://mathworld.wol-
fram.com/FibonacciQ-Matrix.html.