Turk J Math
(2015) 39: 1004 – 1018
Turkish Journal of Mathematics
http://journals.tubitak.gov.tr/math/
c TÜBİTAK
⃝
doi:10.3906/mat-1503-80
Research Article
Dynamic behavior of a second-order nonlinear rational difference equation
Yacine HALIM1,2,∗, Nouressadat TOUAFEK3 , Yasin YAZLIK4
Department of Mathematics and Computer Sceince, Mila University Center, Mila, Algeria
2
LMAM Laboratory, Mohamed Seddik Ben Yahia University, Jijel, Algeria
3
LMAM Laboratory and Department of Mathematics, Mohamed Seddik Ben Yahia University, Jijel, Algeria
4
Department of Mathematics, Faculty of Science and Letters, Nevşehir University, Nevşehir, Turkey
1
Received: 30.03.2015
•
Accepted/Published Online: 16.07.2015
•
Printed: 30.11.2015
Abstract: This paper deals with the global attractivity of positive solutions of the second-order nonlinear difference
equation
axkn + b
k−1
∑
k−j
+ cxkn−1
xjn xn−1
j=1
xn+1 =
Axkn + B
k−1
∑
, k = 3, 4, ..., n = 0, 1, ...,
xjn xk−j
n−1
+
Cxkn−1
j=1
where the parameters a , b , c , A , B , C and the initial values x0 , x−1 are arbitrary positive real numbers.
Key words: Global stability, difference equations, local asymptotic stability, periodicity
1. Introduction and preliminaries
The study of difference equations is a very rich research field, and difference equations have been applied in
several mathematical models in biology, economics, genetics, population dynamics, medicine, and so forth.
Solving difference equations and studying the asymptotic behavior of their solutions has attracted the attention
of many authors, see for example [1, 3, 4, 6, 7, 8, 9, 10, 12, 14, 15, 16, 17, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29,
30, 31].
In [18, 19] the global stability of positive solutions of the difference equations
ax3n + bx2n xn−1 + cxn x2n−1 + dx3n−1
,
Ax3n + Bx2n xn−1 + Cxn x2n−1 + Dx3n−1
(1)
ax4n + bx3n xn−1 + cx2n x2n−1 + dxn x3n−1 + ex4n−1
,
Ax4n + Bx3n xn−1 + Cx2n x2n−1 + Dxn x3n−1 + Ex4n−1
(2)
xn+1 =
and
xn+1 =
∗Correspondence:
[email protected]
2010 AMS Mathematics Subject Classification: 39A10, 40A05.
1004
HALIM et al./Turk J Math
was investigated. These equations are special cases of the difference equation
axkn +
k−1
∑
k
bj xjn xk−j
n−1 + cxn−1
j=1
xn+1 =
Axkn
+
k−1
∑
.
Bj xjn xk−j
n−1
+
(3)
Cxkn−1
j=1
Here and motivated by the above-mentioned papers, we study the global character of positive solutions of the
difference equation Eq. (3) with bi = b, Bi = B , i = 1, 2, ..., k − 1 . That is the equation
axkn + b
k−1
∑
k
xjn xk−j
n−1 + cxn−1
j=1
xn+1 =
Axkn
+B
k−1
∑
, k = 3, 4, ..., n = 0, 1, ...,
xjn xk−j
n−1
+
(4)
Cxkn−1
j=1
where the parameters a , b, c , A , B , C and the initial values x0 , x−1 are arbitrary positive real numbers.
Now we recall some definitions and results that will be useful in our investigation; for more details see [11].
Let I be an interval of real numbers and let
F : I × I −→ I
be a continuous function. Consider the difference equation
xn+1 = F (xn , xn−1 ), n = 0, 1, ...,
(5)
with initial values x−1 , x0 ∈ I .
Definition 1 A point x ∈ I is called an equilibrium point of Eq.(5) if
x = F (x, x).
Definition 2 Let x be an equilibrium point of Eq.(5).
i) The equilibrium x is called locally stable if for every ϵ > 0 , there exist δ > 0 such that for all x−1 , x0 ∈ I
with |x−1 − x| + |x0 − x| < δ , we have |xn − x| < ϵ , for all n ≥ −1.
ii) The equilibrium x is called locally asymptotically stable if it is locally stable, and if there exists γ > 0
such that if x−1 , x0 ∈ I and |x−1 − x| + |x0 − x| < γ then
lim xn = x.
n→+∞
iii) The equilibrium x is called a global attractor if for all x−1 , x0 ∈ I , we have
lim xn = x.
n→+∞
1005
HALIM et al./Turk J Math
iv) The equilibrium x is called global asymptotically stable if it is locally stable and a global attractor.
v) The equilibrium x is called unstable if it is not locally stable.
Suppose F is continuously differentiable in some open neighborhood of x, and let
p=
∂F
(x, x),
∂x
q=
∂F
(x, x).
∂y
Then the equation
yn+1 = pyn + qyn−1 , n = 0, 1, ...
(6)
is called the linearized equation of Eq. (5) about the equilibrium point x and
λ2 − pλ − q = 0
(7)
is called the characteristic equation of Eq. (6) about x.
The following theorem is very useful in establishing local stability.
Theorem 1 (Linearized stability)
1. If all the roots of Eq. (7) lie in the open unit disk |λ| < 1, then the equilibrium point x of Eq. (5) is
locally asymptotically stable.
2. If at least one root of Eq. (7) has absolute value greater than one, then the equilibrium point x of Eq. (5)
is unstable.
To establish convergence results, we need the following two theorems from [2].
Theorem 2 Let [a, b] be a closed and bounded interval of real numbers and let F ∈ C([a, b]k+1 , [a, b]) satisfy
the following conditions:
1. The function F (z1 , . . . , zk+1 ) is monotonic in each of its arguments.
2. For each m, M ∈ [a, b] and for each i ∈ {1, . . . , k + 1} , we define
{
Mi (m, M ) =
M, if F is increasing in zi
m, if F is decreasing in zi
and
mi (m, M ) = Mi (M, m)
and assume that if (m, M ) is a solution of the system:
{
then M = m .
1006
M = F (M1 (m, M ), . . . , Mk+1 (m, M ))
m = F (m1 (m, M ), . . . , mk+1 (m, M ))
HALIM et al./Turk J Math
Then there exists exactly one equilibrium x of the equation
xn+1 = F (xn , xn−1 , . . . , xn−k ), n = 0, 1, . . .
(8)
and every solution of Eq. (8) converges to x.
Theorem 3 Let I be a set of real numbers and let
F : I × I −→ I
be a function F (u, v), which decreases in u and increases in v . Then for every solution {xn }∞
n=−1 of the
equation
xn+1 = F (xn , xn−1 ), n = 0, 1, . . . ,
∞
the subsequences {x2n }∞
n=0 and {x2n+1 }n=−1 of even and odd terms of the solution do exactly one of the
following:
i) They are both monotonically increasing.
ii) They are both monotonically decreasing.
ii) Eventually, one of them is monotonically increasing and the other is monotonically decreasing.
The following results [5, 13] give the rate of convergence for solutions of a system of difference equations.
Let us consider the system of difference equations
Xn+1 = (A + Bn ) Xn , n ∈ N0 ,
(9)
where Xn is an m−dimensional vector, A ∈ C m×m is a constant matrix, and B : Z + → C m×m is a matrix
function satisfying
∥Bn ∥ → 0
(10)
as n → ∞.
Theorem 4 (Perron’s First Theorem) Suppose that condition (10) holds. If Xn is a solution of (9), then
either Xn = 0 for all large n or
ρ = lim
n→∞
∥Xn+1 ∥
∥Xn ∥
(11)
exists and is equal to the modulus of one of the eigenvalues of matrix A .
Theorem 5 (Perron’s Second Theorem) Suppose that condition (10) holds. If Xn is a solution of (9),
then either Xn = 0 for all large n or
1/n
ρ = lim (∥Xn ∥)
n→∞
(12)
exists and is equal to the modulus of one of the eigenvalues of matrix A .
1007
HALIM et al./Turk J Math
2. Stability of the positive solutions
Let f : (0, +∞)2 → (0, +∞) be the function defined by
axk + b
k−1
∑
xj y k−j + cy k
j=1
f (x, y) =
Axk + B
k−1
∑
.
j k−j
x y
+ Cy
k
j=1
In the following theorem we study the periodicity of the positive solutions.
Theorem 6 Let
p̆1
= aC − cA + aB + bC,
p̆2
= aC − cA + cC + aA − (k − 1)(bA + cB) + k(aB + bC),
q̆i
= aC − cA − i(bA + cB) + (i + 1)(aB + bC),
i = 1, 2, ..., k − 2.
Assume that p̆1 , p̆2 , q̆1 , q̆2 , ..., q̆k−2 ≥ 0 . Then Eq. (4) has no positive prime period-two solution.
Proof For the sake of contradiction, assume that there exist distinct positive real number α and β , such that
..., α, β, α, β, ...
is a period-two solution of Eq. (4). Then α, β satisfy the system
α = f (β, α), β = f (α, β).
Hence
(aβ k + b
k−1
∑
β j αk−j + cαk )
j=1
β.f (β, α) − α.f (α, β) = β.
Aβ k + B
k−1
∑
(aαk + b
β α
k−j
+ Cα
k
αj β k−j + cβ k )
j=1
− α.
j
k−1
∑
Aαk + B
j=1
k−1
∑
= 0,
j
α β
k−j
+ Cβ
j=1
which gives
(β − α).
F (β, α)
= 0,
K(β, α)
where
F (β, α) =
aC(α2k + β 2k ) + p̆1 βα(α2k−2 + β 2k−2 ) + q̆1 β 2 α2 (α2k−4 + β 2k−4)
+ q̆2 β 3 α3 (α2k−6 + β 2k−6 ) + ... + q̆k−2 β k−1 αk−1 (α2 + β 2 ) + p̆2 β k αk
k−1
∑
+ bB(
β j αk−j )2 ,
j=1
K(β, α) =
(Aβ k + B
k−1
∑
j=1
1008
β j αk−j + Cαk )(Aαk + B
k−1
∑
j=1
αj β k−j + Cβ k ).
k
HALIM et al./Turk J Math
F (β,α)
K(β,α)
Since
2
> 0, we get β = α , which is a contradiction.
In the sequel we need the following real numbers: r1 = aB − bA , r2 = aC − cA, r3 = bC − Bc .
Lemma 1
≥ max
a
A
1) Assume that
{b
c
B, C
}
and r3 ≥ 0 . Then f is increasing in x for each y and it is
decreasing in y for each x.
2) Assume that
a
A
≤ min
{b
c
B, C
}
and r3 ≤ 0 . Then f is decreasing in x for each y and it is increasing in
y for each x.
Proof
1. We have r3 ≥ 0 and it is easy to see that
≥ max
a
A
{b
c
B, C
}
implies
r1 , r2 ≥ 0.
Therefore, the result follows from the two formulas
r1
∂f
(x, y) =
∂x
k−1
∑
(k − j)xj+k−1 y k−j + r2 kxk−1 y k + r3
j=1
k−1
∑
jxj−1 y 2k−j
j=1
(Axk
+B
k−1
∑
j k−j
x y
,
k 2
+ Cy )
j=1
r1
∂f
(x, y) = −
∂y
k−1
∑
(k − j)xj+k y k−j−1 + r2 kxk y k−1 + r3
j=1
k−1
∑
jxj y 2k−j−1
j=1
(Axk + B
k−1
∑
.
xj y k−j + Cy k )2
j=1
2. The proof of 2) is similar and it will be omitted.
2
In the following result, we show that every positive solution of Eq. (4) is bounded.
Theorem 7 Let {xn }+∞
n=−1 be a positive solution of Eq. (4).
1) Assume that
a
A
≥ max{ Bb , Cc } and r3 ≥ 0. Then
c
a
≤ xn ≤
C
A
for all n ≥ 1 .
2) Assume that
a
A
≤ min{ Bb , Cc } and r3 ≤ 0 . Then
a
c
≤ xn ≤
A
C
for all n ≥ 1 .
1009
HALIM et al./Turk J Math
Proof
1. We have
−r1
xn+1 −
a
=
A
k−1
∑
k
xjn xk−j
n−1 − r2 xn−1
j=1
A(Axkn
+B
k−1
∑
,
xjn xk−j
n−1
+
xkn−1 )
j=1
r3
xn+1 −
c
=
C
k−1
∑
k
xjn xk−j
n−1 + r2 xn
j=1
C(Axkn + B
k−1
∑
.
k
xjn xk−j
n−1 + xn−1 )
j=1
Now using the fact that
a
A
≥ max{ Bb , Cc } , we get r1 , r2 ≥ 0 . Therefore, it follows that
a
c
≤ xn ≤ ,
C
A
for all n ≥ 1 .
2. The proof of 2) is similar and will be omitted.
2
The locally stability of the unique positive equilibrium point x =
a+(k−1)b+c
A+(k−1)B+C
of Eq. (4) is described in
the following theorem.
Theorem 8 Assume that
k (k − 1)(r1 + r3 ) + 2r2 < 1.
(a + (k − 1)b + c)(A + (k − 1)B + C)
Then the positive equilibrium point x =
a+(k−1)b+c
A+(k−1)B+C
of Eq. (4) is locally asymptotically stable.
Proof The linearized equation of Eq. (4) about x =
a+(k−1)b+c
A+(k−1)B+C
is
yn+1 = pyn + qyn−1 ,
where
r1
p=
∂f
(x, x) =
∂x
k−1
∑
j=1
(k − j)x2k−1 + r2 kx2k−1 + r3
k−1
∑
jx2k−1
j=1
,
)2
(
k−1
∑
Axk + B
xk + Cxk
j=1
[ k−1
k−1
k−1
∑
∑
∑ ]
k − r1
j + r2 k + r3
j
x2k−1 r1
=
j=1
j=1
(
)2
k−1
∑
1+C
x2k A + B
j=1
1010
j=1
,
HALIM et al./Turk J Math
=
k(k − 1)(r1 + r3 ) + 2r2 k
(
)2 ,
k−1
∑
2x A + B
1+C
j=1
=
k(k − 1)(r1 + r3 ) + 2kr2
.
2(a + (k − 1)b + c)(A + (k − 1)B + C)
Similarly, we have
q=
k(k − 1)(r1 + r3 ) + 2kr2
∂f
(x, x) = −
.
∂y
2 (a + (k − 1)b + c) (A + (k − 1)B + C)
The associated characteristic equation is
λ2 − pλ − q = 0.
Let h and g be the two functions defined by
h(λ) = λ2 , g(λ) = pλ + q.
We have
k (k − 1)(r1 + r3 ) + 2r2 < 1 = |h(λ)|, ∀λ ∈ C : |λ| = 1.
|g(λ)| ≤ |p| + |q| =
(a + (k − 1)b + c)(A + (k − 1)B + C)
Thus, by Rouché’s theorem, all zeros of λ2 − pλ − q = 0 lie in |λ| < 1 . Therefore, by Theorem (1), x is locally
asymptotically stable.
2
The following two theorems are devoted to the global stability of the positive equilibrium point x =
a+(k−1)b+c
A+(k−1)B+C
of Eq. (4).
Theorem 9 Let
p̆1
= aC − cA + aB + bC,
p̆2
= aC − cA + cC + aA − (k − 1)(bA + cB) + k(aB + bC),
q̆i
= aC − cA − i(bA + cB) + (i + 1)(aB + bC),
i = 1, 2, ..., k − 2.
1. Assume that
≤ min{ Bb , Cc } and r3 ≤ 0 .
•
a
A
•
k(k−1)(r1 +r3 )+2r2 (a+(k−1)b+c)(A+(k−1)B+C)
< 1.
• p̆1 , p̆2 , q̆i ≥ 0.
Then the positive equilibrium point x =
a+(k−1)b+c
A+(k−1)B+C
of Eq.(4) is globally asymptotically stable.
1011
HALIM et al./Turk J Math
Proof Let {xn }+∞
n=−1 be a positive solution of Eq. (4) with x−1 , x0 ∈ (0, +∞) . By Theorem (8) we need only
to prove that x is a global attractor, that is lim xn = x. By Lemma (1) part 2, we see that the function
n→∞
k−1
∑
axk + b
xj y k−j + cy k
j=1
f (x, y) =
Axk + B
k−1
∑
xj y k−j + Cy k
j=1
satisfies the Hypotheses of Theorem (3); also by Theorem (7) part 2 the solution is bounded. Hence, we get
lim x2n = l1 , lim x2n+1 = l2 , with l1 = f (l2 , l1 ), l2 = f (l1 , l2 ). Now, in view of Theorem (6) (and its proof),
n→∞
n→∞
Eq. (4) has no prime period-two solutions and we have
l1 = l2 = x =
a + (k − 1)b + c
.
A + (k − 1)B + C
2
Theorem 10 Let
p1
= cA − aC + bA + cB,
p2
= cA − aC + aA + cC − (k − 1)(bC + aB) + k(bA + cB),
qi
= cA − aC − i(bC + aB) + (i + 1)(bA + cB), i = 1, 2, ..., k − 2.
1. Assume that
a
b c
≥ max{ , } and r3 ≥ 0 .
A
B C
k (k − 1)(r1 + r3 ) + 2r2 •
< 1.
(a + (k − 1)b + c)(A + (k − 1)B + C)
•
• p1 , p2 , qi ≥ 0.
Then the positive equilibrium point x =
Proof
a+(k−1)b+c
A+(k−1)B+C
of Eq. (4) is globally asymptotically stable.
c a
Let {xn }+∞
n=−1 be a positive solution of Eq. (4) with x−1 , x0 ∈ [ C , A ]. By Theorem (8) we need only
to prove that x is a global attractor, that is lim xn = x. Consider again the function
n→∞
axk + b
k−1
∑
xj y k−j + cy k
j=1
f (x, y) =
Axk
+B
k−1
∑
.
j k−j
x y
+ Cy
k
j=1
a
a
Suppose that (m, M ) ∈ [ Cc , A
] × [ Cc , A
] is a solution of the system
M = f (M1 (m, M ), M2 (m, M )), m = f (m1 (m, M ), m2 (m, M )).
1012
(13)
HALIM et al./Turk J Math
By Lemma (1) part 1, we have
M1 (m, M ) = M, M2 (m, M ) = m, m1 (m, M ) = m, m2 (m, M ) = M.
(14)
From (13) and (14), we get
aM k + b
k−1
∑
M j mk−j + cmk
j=1
M=
AM k
+B
k−1
∑
k−1
∑
amk + b
j=1
,m=
j
k−j
M m
+ Cm
k
mj M k−j + cM k
Amk
+B
j=1
k−1
∑
.
j
m M
k−j
+ CM
k
j=1
Hence,
amk + b
k−1
∑
mj M k−j + cM k
aM k + b
j=1
M.
Amk
+B
k−1
∑
m M
k−j
+ CM
k
M j mk−j + cmk
j=1
− m.
j
k−1
∑
AM k
+B
j=1
k−1
∑
= 0,
j
M m
k−j
+ Cm
k
j=1
which can be written as
(M − m).
L(m, M )
= 0,
K(m, M )
where
L(m, M ) = cA(M 2k +m2k )+p1 mM (M 2k−2 +m2k−2 )+q1 m2 M 2 (M 2k−4 +m2k−4 )+q2 m3 M 3 (M 2k−6 +m2k−6 )+
∑k−1
... + qk−2 mk−1 M k−1 (M 2 + m2 ) + p2 mk M k + bB( j=1 mj M k−j )2 , and
K(m, M ) = (Amk + B
k−1
∑
mj M k−j + CM k )(AM k + B
j=1
Since
L(m,M )
K(m,M )
k−1
∑
M j mk−j + Cmk ).
j=1
> 0, we get M = m. From this fact and by Lemma (1) part 1, all Hypotheses of Theorem (2)
2
are satisfied and we have lim xn = x.
x→∞
3. Rate of convergence and numerical examples
Now we estimate the rate of convergence of a solution that converges to the positive equilibrium point x =
a+(k−1)b+c
A+(k−1)B+C
of Eq. (4). First, we will find a system that satisfies the error terms. Hence, we consider the
quantity
axkn + b
k−1
∑
j=1
xn+1 − x =
Axkn + B
k−1
∑
j=1
k
xjn xk−j
n−1 + cxn−1
−
k
xjn xk−j
n−1 + Cxn−1
a + (k − 1) b + c
A + (k − 1) B + C
1013
HALIM et al./Turk J Math
(
k−1
∑
r1 (k − 1) xkn + r2 xkn + r3
j=1
(
=
Axkn + B
)
xjn xk−j
n−1
k−1
∑
j=1
(
r1
(k −
1) xkn
−
(
=
k−1
∑
j=1
Axkn + B
j=1
r1
j=1
k
xjn xk−j
n−1 + Cxn−1
+
k−1
∑
−
k−1
∑
)
xjn xk−j
n−1
+ r2 xkn−1 + r3 (k − 1) xkn−1
)
)
xjn xk−j
n−1
(
(
r2 xkn
−
xkn−1
)
(A + (k − 1) B + C)
(
+ r3
j=1
)
xjn xk−j
n−1
k−1
∑
)
xjn xk−j
n−1
− (k −
1) xkn−1
(A + (k − 1) B + C)
+ Cxkn−1
for n ∈ N0 . The last equality can be written as follows:
xn+1 − x
r1
=
[( k
) ( k
)
( k
)]
2 k−2
k−1
x − xn xk−1
n−1 + xn − xn xn−1 + · · · + xn − xn xn−1
(n
)
k−1
∑ j k−j
Axkn + B
xn xn−1 + Cxkn−1 (A + (k − 1) B + C)
j=1
(
)
k−1
r2 (xn − xn−1 ) xk−1
+ xk−2
n
n xn−1 + · · · + xn−1
)
+(
k−1
∑ j k−j
k
k
Axn + B
xn xn−1 + Cxn−1 (A + (k − 1) B + C)
j=1
+
r3
[(
)]
)
( k−1
) (
k
k
xn xk−1
− xkn−1 + x2n xk−2
n−1 − xn−1 + · · · + xn xn−1 − xn−1
(n−1
)
.
k−1
∑
j
k−j
Axkn + B
xn xn−1 + Cxkn−1 (A + (k − 1) B + C)
j=1
After some computations, we get
[
k−2
∑
r1 (xn − xn−1 )
xn+1 − x
j=0
(
=
k−j−2
xj+1
n xn−1
Axkn + B
k−1
∑
j=1
j=0
k−1
∑
j=0
+(
Axkn + B
k−1
∑
j=1
r3 (xn − xn−1 )
(
0
∑
j=0
j+k−1
x−j
n xn−1
Axkn + B
k−1
∑
j=1
+ ··· +
j=0
)
]
xj+k−1
x−j
n
n−1
(A + (k − 1) B + C)
+
(A + (k − 1) B + C)
1
∑
j=0
j+k−2
x1−j
n xn−1
k
xjn xk−j
n−1 + Cxn−1
+ ··· +
)
k−2
∑
j=0
]
xk−j−2
xj+1
n
n−1
(A + (k − 1) B + C)
or
xn+1 − x = (r1 S1 + r2 S2 + r3 S3 ) (xn − xn−1 ) ,
1014
0
∑
xjn xk−j−1
n−1
)
k
xjn xk−j
n−1 + Cxn−1
[
k−j−3
xj+2
n xn−1
k
xjn xk−j
n−1 + Cxn−1
r2 (xn − xn−1 )
+
+
k−3
∑
HALIM et al./Turk J Math
where
k−2
∑
S1 =
k−j−2
xj+1
+
n xn−1
j=0
(
Axkn + B
k−1
∑
j=1
k−3
∑
0
∑
k−j−3
xj+2
+ ··· +
xj+k−1
x−j
n xn−1
n
n−1
j=0
j=0
)
,
k
xjn xk−j
n−1 + Cxn−1
k−1
∑
j=0
S2 = (
Axkn + B
k−1
∑
j=1
(A + (k − 1) B + C)
xjn xk−j−1
n−1
)
k
xjn xk−j
n−1 + Cxn−1
(A + (k − 1) B + C)
and
0
∑
j=0
j+k−1
x−j
+
n xn−1
S3 = (
Axkn + B
k−1
∑
j=1
1
∑
k−2
∑ k−j−2 j+1
j+k−2
x1−j
+ ··· +
xn
xn−1
n xn−1
j=0
j=0
)
.
k
xjn xk−j
n−1 + Cxn−1
(A + (k − 1) B + C)
Hence, we get the equation
xn+1 − x = (r1 S1 + r2 S2 + r3 S3 ) (xn − x) − (r1 S1 + r2 S2 + r3 S3 ) (xn−1 − x) .
(15)
Note that
lim S1
=
lim S2
=
n→∞
n→∞
lim S3 =
n→∞
k (k − 1)
,
2 (a + (k − 1) b + c) (A + (k − 1) B + C)
k
,
(a + (k − 1) b + c) (A + (k − 1) B + C)
since xn → x as n → ∞. Let
en = xn − x.
Then Eq. (15) becomes
en+1 = (p + ϵ1 (n))en + (q + ϵ2 (n))en−1
(16)
where ϵ1 (n), ϵ2 (n) → 0 as n → ∞. Clearly Eq. (16) can be written in the matrix form
(
en
en+1
)
(
=
0 1
q p
)(
en−1
en
(
and the characteristic equation of the matrix
linearized equation at x =
a+(k−1)b+c
A+(k−1)B+C
)
0 1
q p
(
+
0
ϵ2 (n)
0
ϵ1 (n)
)(
en−1
en
)
)
is the same as the characteristic equation of the
. Using Perron’s Theorems, we have the following result.
Theorem 11 Let x be the positive equilibrium point and the sequence (xn )n≥1 be a positive solution of Eq.
(
)
en
(4). Then the error vector En =
of every solution of Eq. (4) satisfies both of the asymptotic relations
en−1
ρ = lim
n→∞
∥En+1 ∥
1/n
, ρ = lim (∥En ∥)
,
n→∞
∥En ∥
1015
HALIM et al./Turk J Math
where en = xn − x and ρ is equal to the modulus of one of the roots of the characteristic equation.
For confirming the results of this paper, we consider the following numerical examples.
Example 1 If we take k = 3 , a = 2 , b = 1 , c = 2 , A = 5 , B = 2 , C = 3 we obtain the equation
xn+1 =
2x3n + x2n xn−1 + xn x2n−1 + 2x3n−1
.
5x3n + 2x2n xn−1 + 2xn x2n−1 + 3x3n−1
(17)
Let x−1 = 0.45 , x0 = 0.55 , and we have x = 0.5 . All conditions of Theorem (9) are satisfied and
lim xn = x.
n→+∞
(See Figure 1).
0.52
0.72
X(n)
X(n)
0.715
0.515
0.71
0.51
0.705
0.505
0.7
X(n)
X(n)
0.5
0.695
0.495
0.69
0.49
0.685
0.485
0.68
0.48
0.475
0.675
0
10
20
30
40
0.67
50
0
10
20
n
30
40
50
n
Figure 1. This figure shows the global attractivity of the
Figure 2. This figure shows the global attractivity of the
equilibrium point ( x = 0.5 ) of Eq. (17).
equilibrium point ( x = 0.707... ) of Eq. (18).
Example 2 If we take k = 4 , a = 2 , b = 2 , c = 6 , A = 1.3, B = 3 , C = 9.5 we obtain the equation
xn+1 =
2x4n + 2x3n xn−1 + 2x2n x2n−1 + 2xn x3n−1 + 6x4n−1
.
1.3x4n + 3x3n xn−1 + 3x2n x2n−1 + 3xn x3n−1 + 9.5x4n−1
(18)
Let x−1 = 0.95, x0 = 0.75, and we have x = 0.707 · · · . All conditions of Theorem (10) are satisfied and
lim xn = x. (See Figure 2).
n→+∞
Acknowledgment
The authors would like to thank the referee for careful reading and for his/her comments and suggestions, which
greatly improved an earlier version of this paper.
References
[1] Camouzis E, Ladas G, Voulov H D. On the dynamics of xn+1 =
α+γxn−1 +δxn−2
A+xn−2
. J Difference Equ Appl 2003; 9:
731–738.
[2] Camouzis E, Ladas G. Dynamics of Third-Order Rational Difference Equations with Open Problems and Conjectures. Boca Raton, FL, USA: Chapman & Hall/CRC, 2008.
1016
HALIM et al./Turk J Math
[3] Din Q, Qureshi MN, Khan A Q. Dynamics of a fourth-order system of rational difference equations. Adv Difference
Equ 2012; 2012: 215.
[4] Elabbasy EM, Elsayed EM. On the global attractivity of difference equation of higher order. Carpathian J Math
2008; 24: 45–53.
[5] Elaydi S. An Introduction to Difference Equations. Undergraduate Texts in Mathematics. New York, NY, USA:
Springer, 1999.
[6] Elsayed EM. Behavior and expression of the solutions of some rational difference equations. J Comput Anal Appl
2013; 15: 73–81.
[7] Elsayed EM. On the dynamics of a higher order rational recursive sequence. Commun Math Anal 2012; 12: 117–133.
[8] Halim Y. Global character of systems of rational difference equations. Electron J Math Analysis Appl 2015; 3:
204–214.
[9] Halim Y, Touafek N, Elsayed EM. Closed form solution of some systems of rational difference equations in terms
of Fibonacci numbers. Dynam Cont Dis Ser A 2014; 21: 473–486.
[10] Ibrahim TF, Touafek N. On a third order rational difference equation with variable coefficients. Dynam Cont Dis
Ser B 2013; 20: 251–264.
[11] Grove EA, Ladas G. Periodicities in Nonlinear Difference Equations. Boca Raton, FL, USA: Chapman and Hall/CRC
Press, 2004.
[12] Kulenović MRS, Nurkanović M. Asymptotic behavior of a competitive system of linear fractional difference equations. Adv Difference Equ 2006; 2006: 19756.
[13] Pituk M. More on Poincaré’s and Peron’s theorems for difference equations. J Difference Equ Appl 2002; 8: 201–216.
[14] Taskara T, Uslu K, Tollu DT. The periodicity and solutions of the rational difference equation with periodic
coefficients. Comput Math Appl 2011; 62: 1807–1813.
[15] Tollu DT, Yazlik Y, Taskara N. On the solutions of two special types of Riccati difference equation via Fibonacci
numbers. Adv Difference Equ 2013; 2013: 174.
[16] Tollu DT, Yazlik Y, Taskara N. The solutions of four Riccati difference equations associated with Fibonacci numbers.
BALKAN J M 2014; 2: 163–172.
[17] Tollu DT, Yazlik Y, Taskara N. On fourteen solvable systems of difference equations. Appl Math Comput 2014;
233: 310–319.
[18] Touafek N. On a second order rational difference equation. Hacet J Math Stat 2012; 41: 867–874.
[19] Touafek N, Halim Y. Global attractivity of a rational difference equation. Math Sci Lett 2013; 2: 161–165.
[20] Touafek N, Halim Y. On max type difference equations: expressions of solutions. Int J Nonlinear Sci 2011; 11:
396–402.
[21] Touafek N, Elsayed EM. On a third order rational systems of difference equations. An Stiint Univ Al I Cuza Iasi
Mat; In press DOI: 10.2478/aicu-2014-004.
[22] Touafek N, Elsayed EM. On the solutions of systems of rational difference equations. Math Comput Modelling 2012;
55: 1987–1997.
[23] Touafek N, Elsayed EM. On the periodicity of some systems of nonlinear difference equations. Bull Math Soc Sci
Math Roum Nouv Sér 2012; 55: 217–224.
[24] Van Khuong V, Nam Phong M. On a system of two difference equations of exponential form. Int J Difference Equ
2013; 8: 215–223.
[25] Yalçinkaya I, Çinar C. On the dynamics of the difference equation xn+1 =
axn−k
p
b+cxn
. Fasc Math 2009; 42: 141–148.
[26] Yazlik Y. On the solutions and behavior of rational difference equations. J Comput Anal Appl 2014; 17: 584–594.
1017
HALIM et al./Turk J Math
[27] Yazlik Y, Tollu DT, Taskara N. On the solutions of a max-type difference equation system. Math Meth Appl Sci
2015; DOI: 10.1002/mma.3377.
[28] Yazlik Y, Tollu DT, Taskara N. On the behaviour of solutions for some systems of difference equations. J Comput
Anal Appl 2015; 18: 166–178.
[29] Yazlik Y, Elsayed EM, Taskara N. On the behaviour of the solutions of difference equation system. J Comput Anal
Appl 2014; 16: 932–941.
[30] Zhang Q, Zhang W. On a system of two high-order nonlinear difference equations. Adv Math Phys 2014; 2014:
729273.
[31] Zhang Q, Yang L, Liu J. Dynamics of a system of rational third-order difference equation. Adv Difference Equ 2012;
2012: 136.
1018