
The boundedness character, persistent nature, and asymptotic conduct of non-negative outcomes of the system of three dimensional exponential form of difference equations were studied in this research:
xn+1=axn+byn−1e−xn, yn+1=cyn+dzn−1e−yn, zn+1=ezn+fxn−1e−zn,
where a, b, c, d, e and f are non-negative real values, and the initial values x−1, x0, y−1, y0, z−1, z0 are non-negative real values.
Citation: Abdul Khaliq, Haza Saleh Alayachi, Muhammad Zubair, Muhammad Rohail, Abdul Qadeer Khan. On stability analysis of a class of three-dimensional system of exponential difference equations[J]. AIMS Mathematics, 2023, 8(2): 5016-5035. doi: 10.3934/math.2023251
[1] | Junjie Li, Gurpreet Singh, Onur Alp İlhan, Jalil Manafian, Yusif S. Gasimov . Modulational instability, multiple Exp-function method, SIVP, solitary and cross-kink solutions for the generalized KP equation. AIMS Mathematics, 2021, 6(7): 7555-7584. doi: 10.3934/math.2021441 |
[2] | Siqin Tang, Hong Li . A Legendre-tau-Galerkin method in time for two-dimensional Sobolev equations. AIMS Mathematics, 2023, 8(7): 16073-16093. doi: 10.3934/math.2023820 |
[3] | Qian Cao, Xiaojin Guo . Anti-periodic dynamics on high-order inertial Hopfield neural networks involving time-varying delays. AIMS Mathematics, 2020, 5(6): 5402-5421. doi: 10.3934/math.2020347 |
[4] | Rashad M. Asharabi, Somaia M. Alhazmi . Accelerating the convergence of a two-dimensional periodic nonuniform sampling series through the incorporation of a bivariate Gaussian multiplier. AIMS Mathematics, 2024, 9(11): 30898-30921. doi: 10.3934/math.20241491 |
[5] | Zunyuan Hu, Can Li, Shimin Guo . Fast finite difference/Legendre spectral collocation approximations for a tempered time-fractional diffusion equation. AIMS Mathematics, 2024, 9(12): 34647-34673. doi: 10.3934/math.20241650 |
[6] | Kui Liu . Stability analysis for (ω,c)-periodic non-instantaneous impulsive differential equations. AIMS Mathematics, 2022, 7(2): 1758-1774. doi: 10.3934/math.2022101 |
[7] | Mouataz Billah Mesmouli, Cemil Tunç, Taher S. Hassan, Hasan Nihal Zaidi, Adel A. Attiya . Asymptotic behavior of Levin-Nohel nonlinear difference system with several delays. AIMS Mathematics, 2024, 9(1): 1831-1839. doi: 10.3934/math.2024089 |
[8] | Rana M. S. Alyoubi, Abdelhalim Ebaid, Essam R. El-Zahar, Mona D. Aljoufi . A novel analytical treatment for the Ambartsumian delay differential equation with a variable coefficient. AIMS Mathematics, 2024, 9(12): 35743-35758. doi: 10.3934/math.20241696 |
[9] | Lini Fang, N'gbo N'gbo, Yonghui Xia . Almost periodic solutions of a discrete Lotka-Volterra model via exponential dichotomy theory. AIMS Mathematics, 2022, 7(3): 3788-3801. doi: 10.3934/math.2022210 |
[10] | Wentao Wang . Positive almost periodic solution for competitive and cooperative Nicholson's blowflies system. AIMS Mathematics, 2024, 9(5): 10638-10658. doi: 10.3934/math.2024519 |
The boundedness character, persistent nature, and asymptotic conduct of non-negative outcomes of the system of three dimensional exponential form of difference equations were studied in this research:
xn+1=axn+byn−1e−xn, yn+1=cyn+dzn−1e−yn, zn+1=ezn+fxn−1e−zn,
where a, b, c, d, e and f are non-negative real values, and the initial values x−1, x0, y−1, y0, z−1, z0 are non-negative real values.
Difference equations have a wide range of applications in math, physics and engineering, as well as in business and other professions. See [1,2,3,4] and the references given therein for a list of publications and books on difference equations theory and applications. The qualitative features of difference equations of exponential form have recently attracted a lot of attention [5,6,7,8]. The authors of [9] explored the difference equation's boundedness, asymptotic nature, periodicity of the solutions, and the stability of the non-negative equilibrium:
un+1=αun+βun−1e−un, n=0,1,⋯, |
wherein α and β are non-negative constants, and the initial values u−1, u0 are non-negative numbers. Because it comes from models that investigate the amount of litter in perennial grasslands, this equation can be called a biological model. The authors of [10] looked at similar conclusions for a system of difference equations:
un+1=αvn+βun−1e−vn, vn+1=γun+δvn−1e−un, |
wherein α, β, γ and δ are non-negative constants, and the initial values u−1, u0, v−1, v0 are non-negative numbers. In addition, the researcher examines the character of boundedness, persistence, and asymptotic nature of the non-negative solutions of the subsequent exponential difference equations in [1]:
un+1=αun+βvn−1e−un, vn+1=γvn+δun−1e−vn, |
wherein α, β, γ and δ are non-negative constants, and the initial values u−1, u0, v−1, v0 are also non-negative numbers. We explore the character of boundedness, persistence, and the convergence rate of the non-negative outcomes of (1.1) to the unique positive equilibrium point of the subsequent exponential difference equations, motivated by the studies mentioned above:
xn+1=axn+byn−1e−xn, yn+1=cyn+dzn−1e−yn,zn+1=ezn+fxn−1e−zn, | (1.1) |
where a, b, c, d, e and f are non-negative real numbers, and the initial values x−1, x0, y−1, y0, z−1, z0 are also non-negative real numbers.
We look at the existence and uniqueness of the non-negative equilibrium point of the system (1.1) in first theorem.
Theorem 2.1. The foregoing claims are valid.
(i) Assume that
a, b, c, d, e, f∈(0, 1), θ=bdf(1−a)(1−c)(1−e)>1. | (2.1) |
This leads to a unique equilibrium (¯x, ¯y, ¯z) for the system (1.1). However,
ln(θ)−¯y1+f1−e≤¯x≤lnθ,ln(θ)−¯z1+b1−a≤¯y≤lnθ, ln(θ)−¯x1+d1−c≤¯z≤lnθ. | (2.2) |
(ii) Assume that a, b, c, d, e, f are positive real values such that
a, b, c, d, e, f∈(0, 1), θ≤1. | (2.3) |
The zero equilibrium (0, 0, 0) is the unique equilibrium solution of system (1.1).
Proof. (i) Assume the following set of algebraic equations:
x=ax+bye−x,y=cy+dze−y,z=ez+fxe−z, |
or equivalently,
(1−a)x=bye−x,(1−c)y=dze−y,(1−e)z=fxe−z. | (2.4) |
Multiplying Eq (2.4),
(1−a)(1−c)(1−e)xyz=bdfxyze−(x+y+z),(1−a)(1−c)(1−e)bdf=e−(x+y+z), |
then,
e(x+y+z)=bdf(1−a)(1−c)(1−e),x+y+z=lnbdf(1−a)(1−c)(1−e). |
Then, from Eq (2.4), if x≠0, y≠0 and z≠0, we get
x+y+z=lnθ. | (2.5) |
From Eqs (2.4) and (2.5),
(1−a)xexb=y,(1−c)yeyd=z,(1−e)zezf=x. |
Put value of y=(1−a)xexb in Eq (2.5), we get
x=ln(θ)−z1+1−abex. |
Now, put value of (1−c)yeyd=z in Eq (2.5), we get
y=ln(θ)−x1+1−cdey. |
Similarly, put (1−e)zezf=x in Eq (2.5), we get
z=ln(θ)−y1+1−efez. |
We consider the function:
F(x)=x−ln(θ)−z1+1−abex. |
So, from (2.1), we get that F(0)<0 and Limx→∞ F(x)=∞. Then, there exist a ¯x∈(0, ∞) such that
¯x=ln(θ)−¯z1+1−abe¯x. | (2.6) |
Similarly, we can prove that there exists a ¯y∈(0, ∞) and ¯z∈(0, ∞) such that
¯y=ln(θ)−¯x1+1−cde¯y, | (2.7) |
¯z=ln(θ)−¯y1+1−efe¯z. | (2.8) |
To find z from z=ez+fxe−z as
z=f1−exe−z. | (2.9) |
So, from (2.5),
x+y+z=lnθ, |
x+y+f1−exe−z=lnθ, |
x[1+f1−ee−z]=ln(θ)−y, |
x=ln(θ)−y1+f1−ee−z. |
Now, we will find ¯x at z=0:
¯x=ln(θ)−¯y1+f1−e. | (2.10) |
Similarly, we will prove that
¯y=ln(θ)−¯z1+b1−a, | (2.11) |
¯z=ln(θ)−¯x1+d1−c. | (2.12) |
Therefore, from (2.1), (2.5) and combining with Eqs (2.6) and (2.10), we obtained
ln(θ)−¯y1+f1−e≤ln(θ)−¯z1+1−abe¯x=¯x≤lnθ. |
In similar way, we obtained
ln(θ)−¯z1+b1−a≤ln(θ)−¯x1+1−cde¯y=¯y≤lnθ,ln(θ)−¯x1+d1−c≤ln(θ)−¯y1+1−efe¯z=¯z≤lnθ. |
And thus (2.2) holds. To demonstrate uniqueness, we suppose that another non-negative equilibrium (¯x1, ¯y1, ¯z1) of (1.1) exists. We can assume that ¯x<¯x1 without losing generality. Then we obtain the following from (2.6):
¯x=lnθ−¯z1+1−abe¯x<¯x1=lnθ−¯z1+1−abe¯x1, |
and so e¯x1≤e¯x, which is a contradiction.
So,
¯x=¯x1, |
similarly,
¯y=¯y1 and ¯z=¯z1. |
The proof is now completed.
Proof. (ii) Since (2.3) is still valid, then we can deduce from (2.5) that x+y+z≤0, implying that (0, 0, 0) is the only non-negative equilibrium point. The proof is now finished.
We explore the boundedness and persistence of the non-negative solutions of (1.1) in the next proposition.
Theorem 3.1. There are valid arguments for the following:
(i) Assume that
a, b, c, d, e, f∈(0,1). | (3.1) |
Then, every positive solution of (1.1) is bounded.
(ii) Suppose that (3.1) holds. Suppose also that
b1−a>1,d1−c>1,f1−e>1. | (3.2) |
Then, every positive solution of (1.1) is bounded and persists.
Proof. Suppose that (xn, yn, zn) be an arbitrarily solution to (1.1).
(i) We will assume M is positive, such that
M≥max{x−1, y−1, z−1, x0, y0, z0, ln(11−a), ln(11−c), ln(11−e)}. | (3.3) |
The following function is considered:
h(x)=Me−x+ax, x∈[0, M]. |
There is also that
h′(x)=−Me−x+a, h′′(x)=Me−x>0. |
In light of this, it follows
h(x)≤max{h(0), h(M)}, x∈[0, M]. | (3.4) |
Furthermore, we can deduce the following from (3.3):
h(0)=M, h(M)=Me−M+aM<Me−ln(11−a)+aM=M(1−a)+aM=M. | (3.5) |
From (3.4) and (3.5), we get that
h(x)≤M, x∈[0, M]. | (3.6) |
As a result of relations (1.1), (3.1), (3.3) and (3.6), it follows that
x1=ax0+by−1e−x0≤ax0+Me−x0=h(x0)≤M. |
So, x1≤M.
Now, consider the function
K(y)=cy+Me−y, y∈[0, M],K′(y)=c−Me−y,K′′(y)=Me−y>0. |
Therefore, it holds that
K(y)≤max{K(0),K(M)}, y∈[0, M]. | (3.7) |
Now, from (3.3),
K(0)=M,K(M)=cM+Me−M<cM+Me−ln(11−c)=cM+M(1−c)=M. | (3.8) |
From (3.7) and (3.8),
K(y)≤M, y∈[0, M]. |
Therefore, relations (1.1), (3.1), (3.3) and (3.8),
y1=cy0+dz−1e−y0≤cy0+Me−y0=K(y0)≤M, y1≤M. |
Similarly, if
g(z)=ez+Me−z, |
then, using the same logic as before, we can show that
z1≤M. |
As a result of our inductive reasoning, we can demonstrate
xn≤M, n=1,2,3,⋯,yn≤M, n=1,2,3,⋯,zn≤M, n=1,2,3,⋯. |
So, we conclude from above results (xn, yn, zn) is bounded.
Proof. (ii) We can show that (xn, yn, zn) persists. We look at the numbers for this.
R=ln(b/(1−a)), S=ln(d/(1−c)), T=ln(f/(1−e)). | (3.9) |
Let
m=min{x−1, x0, y−1, y0, z−1, z0, R, S, T}. |
Then, using (3.2) and arguing as in the proof of (3.1) of [10], we get the following:
If x0≤R, then,
x1≥min{x0, y−1}. |
In addition, if x0>R, y−1≤R, we take
x1>y−1. |
Finally, if x0>R, y−1>R, we get
x1>R. |
So, here is what we've got:
x1≥m. |
In a similar manner, we can demonstrate that
y1≥m, z1≥m. |
We may prove the following by arguing like we did earlier and using the induction method:
xn≥m, yn≥m, zn≥m. |
This completes the proof.
We evaluate the convergence rate of a system (1.1) for all initial values that converge at equilibrium E(¯x, ¯y, ¯z) in this segment by existing theory [11]. For various three-dimensional systems, the convergence rate of solutions that converge to an equilibrium has been determined.
Theorem 4.1. Assume systems (3.1) and (3.2) hold and
max{f1−e, d1−c, b1−a}<min{eef, ecd, eab}. | (4.1) |
Then, each non-negative solution of (1.1) tends to the unique non-negative equilibrium of (1.1).
Proof. Consider (xn, yn, zn) be an arbitrary solution of (1.1). From Theorem 3.1 we get that
l1=Limn→∞infxn>0, L1=Limn→∞supxn<∞,l2=Limn→∞infyn>0, L2=Limn→∞supyn<∞,l3=Limn→∞infzn>0, L3=Limn→∞supzn<∞. | (4.2) |
Then, from (4.2) and for every ϵ>0, there exist an n0(ϵ) such that n≥n0(ϵ),
l1−ϵ≤xn≤L1+ϵ,l2−ϵ≤yn≤L2+ϵ,l3−ϵ≤zn≤L3+ϵ, | (4.3) |
and so from (1.1) and (4.3) we have for n≥n0 that
xn+2=axn+1+byne−xn+1≤axn+1+b(L2+ϵ)e−xn+1=gL2+ϵ(xn+1), | (4.4) |
where gL2+ϵ(x)=ax+b(L2+ϵ)e−x.
But we have that
g′L2+ϵ(x)=a−b(L2+ϵ)e−x,g′′L2+ϵ(x)=b(L2+ϵ)e−x>0. |
Therefore, for x∈[l1−ϵ, L1+ϵ], we get that
gL2+ϵ(x)≤max{gL2+ϵ(l1−ϵ), gL2+ϵ(L1+ϵ)}. |
Then, from (4.4) we take the following:
xn+2≤gL2+ϵ(xn+1)≤max{gL2+ϵ(l1−ϵ), gL2+ϵ(L1+ϵ)}, |
which implies that
L1≤max{gL2+ϵ(l1−ϵ), gL2+ϵ(L1+ϵ)}. |
So, for ϵ⟶0,
L1≤max{gL2(l1), gL2(L1)}. | (4.5) |
Similarly, from (1.1) and (4.3) we have for n≥n0 that
yn+2=cyn+1+dzne−yn+1≤cyn+1+d(L3+ϵ)e−yn+1=hL3+ϵ(yn+1), | (4.6) |
where hL3+ϵ(y)=cy+d(L3+ϵ)e−y.
But we have that
h′L3+ϵ(y)=c−d(L3+ϵ)e−y,h′′L3+ϵ(y)=d(L3+ϵ)e−y>0. |
Therefore, for y∈[l2−ϵ, L2+ϵ], we get that
hL3+ϵ(y)≤max{hL3+ϵ(l2−ϵ), hL3+ϵ(L2+ϵ)}. |
Then, from (4.6) we take
yn+2≤hL3+ϵ(yn+1)≤max{hL3+ϵ(l2−ϵ), hL3+ϵ(L2+ϵ)}, |
which implies that
L2≤max{hL3+ϵ(l2−ϵ), hL3+ϵ(L2+ϵ)}. |
So, for ϵ⟶0,
L2≤max{hL3(l2), hL3(L2)}. | (4.7) |
Similarly, from (1.1) and (4.3) we have for n≥n0 that if
KL1+ϵ(z)=ez+f(L1+ϵ)e−z, |
we can prove that
L3≤max{KL1(l3), KL1(L3)}. | (4.8) |
We claim that
l1>ln(bL2a),l2>ln(dL3c),l3>ln(fL1e). | (4.9) |
Suppose on contrary, that either
l1≤ln(bL2a) | (4.10) |
or
l2≤ln(dL3c) | (4.11) |
or
l3≤ln(fL1e). | (4.12) |
Suppose first that (4.10) valid. Then, since g′L2(x)=a−bL2e−x, we have that gL2 is a non-increasing for x≤ln(bL2a) and consequently we obtained from (4.10) that
gL2(l1)≤gL2(0)=bL2<L2. | (4.13) |
Then, from (4.5) and (4.13) we have that
L1≤max{L2, gL2(L1)}. | (4.14) |
Since it hold that h′L3(y)=c−dL3e−y, we conclude that hL3 is non-increasing function for y≤ln(dL3c) and non-decreasing for y≥ln(dL3c). Then, if l2≥ln(dL3c), we have that
hL3(l2)<hL3(L2). | (4.15) |
If l2≤ln(dL3c), we get that
hL3(l2)<hL3(0)=dL3<L3. | (4.16) |
Relations (4.7), (4.15) and (4.16) imply that
L2≤max{L3, hL3(L2)}. | (4.17) |
Similarly, if (4.12) holds, then,
L3≤max{L1, KL1(L3)}. | (4.18) |
Suppose now that L1≤L2≤L3. Then, from (4.18), we get that
L3≤KL1(L3)=eL3+fL1e−L3,L3≤eL3+fL3e−L3, |
which implies that
L3≤ln(f1−e). | (4.19) |
Since (4.10) holds. We get that
l1≤ln(bL2a),l1≤ln(bL3a), ∵L2≤L3,el1≤bL3a,1+l1≤el1≤bL3a,1+l1≤bL3a, |
and so (4.19) implies that
l1≤bL3a−1≤baln(f1−e)−1. | (4.20) |
We get the following from (4.1):
f1−e<eab, |
and so
baln(f1−e)−1<0. |
Then, from (4.20) we have that l1<0, which is a contradiction. So, (4.10) is not true if L1≤L2≤L3.
Suppose now that L1≤L3≤L2. Then from (4.17) we take that
L2≤hL3(L2)=cL2+dL3e−L2,L2≤cL2+dL2e−L2, |
which implies that
L2≤ln(d1−c). | (4.21) |
Since (4.10) holds. We get that
l1≤ln(bL2a),el1≤bL2a,1+l1≤el1≤bL2a,1+l1≤bL2a, |
and so (4.21) implies that
l1≤bL2a−1≤baln(d1−c)−1. | (4.22) |
Moreover, from (4.1) we get
d1−c<eab, |
so,
baln(d1−c)−1<0. |
Then, from (4.22) we have that l1<0, which is a contradiction. So, (4.10) is not true if L1≤L3≤L2.
Now again suppose that L2≤L3≤L1. Then, from (4.14) we take that
L1≤gL2(L1)=aL1+bL2e−L1,L1≤aL1+bL1e−L1, |
and so
L1≤ln(b1−a). | (4.23) |
Since (4.10) holds, we get
l1≤ln(bL2a),l1≤ln(bL1a), ∵L2≤L1,el1≤bL1a,1+l1≤el1≤bL1a, |
and so (4.23) implies that
l1≤bL1a−1≤baln(b1−a)−1. | (4.24) |
We get from (4.1) that
b1−a<eab, |
and so
baln(b1−a)−1<0. |
Then, from (4.24) we have that l1<0, which is a contradiction. So, (4.10) is not true if L2≤L3≤L1.
Now again suppose that L2≤L1≤L3. Then, from (4.18) we take that
L3≤KL1(L3)=eL3+fL1e−L3,L3≤eL3+fL3e−L3, |
and so
L3≤ln(f1−e). | (4.25) |
Since (4.10) holds. We get
l1≤ln(bL2a),l1≤ln(bL3a), ∵L2≤L3,el1≤bL3a,1+l1≤el1≤bL3a, |
and so (4.25) implies that
l1≤bL3a−1≤baln(f1−e)−1. | (4.26) |
We get from (4.1) that
f1−e<eab, |
and so
baln(f1−e)−1<0. |
Then, from (4.26) we have that l1<0, which is a contradiction. So, (4.10) is not true if L2≤L1≤L3.
Now again suppose that L3≤L2≤L1. Then, from (4.14) we take that
L1≤gL2(L1)=aL1+bL2e−L1,L1≤aL1+bL1e−L1, |
and so
L1≤ln(b1−a). | (4.27) |
Since (4.10) holds. We get
l1≤ln(bL2a),l1≤ln(bL1a), ∵L2≤L1,el1≤bL1a,1+l1≤el1≤bL1a. |
Therefore, (4.27) implies that
l1≤bL1a−1≤baln(b1−a)−1. | (4.28) |
We get from (4.1) that
b1−a<eab, |
and so
baln(b1−a)−1<0. |
Then, from (4.28) we have that l1<0, which is a contradiction. So, (4.10) is not true if L3≤L2≤L1.
Now again suppose that L3≤L1≤L2. Then, from (4.17) we take that
L2≤hL3(L2)=cL2+dL3e−L2,L2≤cL2+dL2e−L2. |
Therefore,
L2≤ln(d1−c). | (4.29) |
Since (4.10) holds, we get
l1≤ln(bL2a),el1≤bL2a,1+l1≤el1≤bL2a, |
and so (4.29) implies that
l1≤bL2a−1≤baln(d1−c)−1. | (4.30) |
We get from (4.1) that
d1−c<eab, |
and so
baln(d1−c)−1<0. |
Then, from (4.30) we have that l1<0, which is a contradiction. So, (4.10) is not true if L3≤L1≤L2.
Working in a similar manner and using (4.1), we can prove that (4.11) and (4.12) are not true for each:
L1≤L2≤L3,L1≤L3≤L2,L2≤L3≤L1,L2≤L1≤L3,L3≤L2≤L1,L3≤L1≤L2. |
So relations (4.9) are satisfied.
Since relations (4.9) hold, gL2 is an increasing function for x≥ln(bL2a), hL3 is an increasing function for y≥ln(dL3c), and KL1 is also an increasing function for z≥ln(fL1e). We then obtain
gL2(l1)≤gL2(L1),hL3(l2)≤hL3(L2),KL1(l3)≤KL1(L3). | (4.31) |
So, from (4.5), (4.7), (4.8) and (4.31) we have that
L1≤gL2(L1),L2≤hL3(L2),L3≤KL1(L3). | (4.32) |
Then relations (4.32) imply that
(1−a)L1eL1b≤L2,(1−c)L2eL2d≤L3,(1−e)L3eL3f≤L1, |
as a result of (2.4), we can simply deduce
F(L1)≤0=F(¯x). |
As F is a non-decreasing function, we obtain
L1≤¯x. | (4.33) |
The following can be proved in a similar way:
G(L2)≤0=G(¯y),H(L3)≤0=H(¯z), |
where
H(z)=(1−a)(1−c)(1−e)ez+s(z)+r(x)bdf−1, r(x)=(1−a)xexb, s(z)=(1−e)zezf. |
Due to the fact that G is a non-decreasing function, we obtain
L2≤¯y. | (4.34) |
Similarly, H is a non-decreasing function, we get
L3≤¯z. | (4.35) |
We can now demonstrate that
¯x<l1<L1 , ¯y<l2<L2, ¯z<l3<L3. | (4.36) |
We derive the following from (1.1) and (4.3):
xn+1≥axn+b(l2−ϵ)e−xn, n≥no(ϵ). | (4.37) |
We look at the following function:
gl2−ϵ(x)=ax+b(l2−ϵ)e−x,g′l2−ϵ(x)=a−b(l2−ϵ)e−x. |
We have that gl2−ϵ is non-decreasing for x≥ln(b(l2−ϵ)a). In addition, since (4.9) valid, then there exists ϵ>0 such that
l1−ϵ>ln(bL2a)>ln(b(L2−ϵ)a). | (4.38) |
Then, from (4.3) and (4.33) we get
xn≥l1−ϵ>ln(b(L2−ϵ)a)≥ln(b(l2−ϵ)a), n≥n0(ϵ). | (4.39) |
As a result, relations (4.37) and (4.39) indicate the following:
xn+1≥a(l1−ϵ)+b(l2−ϵ)e−(l1−ϵ), n≥n0(ϵ). |
And so,
l1≥a(l1−ϵ)+b(l2−ϵ)e−(l1−ϵ). |
For ϵ⟶0, we get
l1≥al1+bl2e−l1. | (4.40) |
Similarly, using (1.1) and (4.9) and arguing as above, we get
l2≥cl2+dl3e−l2, | (4.41) |
and
l3≥el3+fl1e−l3. | (4.42) |
Therefore, from relations (4.33)–(4.35) and (4.40)–(4.42), we have that
l1=L1=¯x, l2=L2=¯y, l3=L3=¯z. |
This completes the proof.
In an effort to our theoretical dialogue, we take into account several interesting numerical examples on this segment. These examples constitute distinct varieties of qualitative conduct of solutions to the system (1.1) of nonlinear difference equations. The first example indicates that positive equilibrium of system (1.1) is unstable with suitable parametric choices. Moreover, from the remaining examples it is clear that unique positive equilibrium point of system (1.1) is globally asymptotically stable with different parametric values. All plots on this segment are drawn with the help of MATLAB.
Example 5.1. Let a=0.9, b=27, c=0.5, d=94, e=0.3, f=67. Then the system (1.1) can be written as
xn+1=0.9xn+27yn−1e−xn, yn+1=0.5yn+94zn−1e−yn, zn+1=0.3zn+67xn−1e−zn, | (5.1) |
with initial conditions x−1=8, x0=7, y−1=6, y0=5, z−1=4, z0=3. In this case, the positive equilibrium point of the system (5.1) is unstable. Moreover, in Figure 1, the graphs of xn, yn and zn are shown in Figure (1a), (1b) and (1c) respectively, and XY, YZ and ZX attractors of the system (5.1) are shown in Figure (1d), (1e) and (1f) respectively. Also the combined graph of all respective phase portrait of system (5.1) is shown in Figure (1g).
Example 5.2. Let a=0.009, b=0.3, c=0.005, d=0.9, e=0.003, f=0.7. Then the system (1.1) can be written as
xn+1=0.009xn+0.yn−1e−xn, yn+1=0.005yn+0.9zn−1e−yn, zn+1=0.003zn+0.7xn−1e−zn, |
with initial conditions x−1=0.008, x0=0.007, y−1=0.006, y0=0.005, z−1=0.004, z0=0.03. In this case, the positive equilibrium point of the system (5.2) is given by (¯x, ¯y, ¯z)=(9.404×10−11, 4.993×10−10, 1.276×10−10). Moreover, in Figure 2, the graphs of xn, yn and zn are shown in Figure (2a), (2b) and (2c) respectively, and XY, YZ and ZX attractors of the system (5.1) are shown in Figure (2d), (2e) and (2f) respectively. Also the combined graph of all respective phase portrait of system (5.2) is shown in Figure (2g).
In this work, we analyze the qualitative behavior of a system of exponential difference equations. Using our model (1.1), we have demonstrated that a positive steady state exists and is unique. We verify the bounds of positive solutions as well as their persistence. We have also established that the positive equilibrium point of system (1.1) under certain parametric conditions is asymptotically stable locally as well as globally. In dynamical structures theory, the goal is to look at a system's global behavior through knowledge of its current state. It is possible to determine what parametric conditions result in these long-term behaviors by determining the possible global behavior of the system. Further, the convergence rate of positive solutions of (1.1) that converge to a unique point of positive equilibrium is determined.
In our future work, we will study some more qualitative properties such as bifurcation analysis, chaos control, and Maximum Lyapunov exponent of the said model. Some interesting numerical simulations with the help of Mathematica presenting bifurcation and chaos control are also part of our future goal.
The authors declare that they have no conflicts of interest regarding the publication of this paper.
[1] |
H. El-Metwally, E. A. Grove, G. Ladas, R. Levins, M. Radin, On the difference equation xn+1=α+βxn−1e−xn, Nonlinear Anal., 47 (2001), 4623–4634. https://doi.org/10.1016/S0362-546X(01)00575-2 doi: 10.1016/S0362-546X(01)00575-2
![]() |
[2] | E. A. Grove, G. Ladas, N. R. Prokup, R. Levins, On the global behavior of solutions of a biological model, Commun. Appl. Nonlinear Anal, 7(2000), 33–46. |
[3] |
G. Papaschinopoulos, G. Ellina, K. B. Papadopoulos, Asymptotic behavior of the positive solutions of an exponential type system of difference equations, Appl. Math. Comput., 245 (2014), 181–190. https://doi.org/10.1016/j.amc.2014.07.074 doi: 10.1016/j.amc.2014.07.074
![]() |
[4] |
W. J. Wang, H. Feng, On the dynamics of positive solutions for the difference equation in a new population model, J. Nonlinear Sci. Appl., 9 (2016), 1748–1754. http://dx.doi.org/10.22436/jnsa.009.04.30 doi: 10.22436/jnsa.009.04.30
![]() |
[5] | G. Stefanidou, G. Papaschinopoulos, C. J. Schinas, On a system of two exponential type difference equations, Commun. Appl. Nonlinear Anal., 17 (2010), 1–13. |
[6] |
G. Papaschinopoulos, M. A. Radin, C. J. Schinas, On the system of two difference equations of exponential form: xn+1=a+bxn−1e−yn, yn+1=c+dyn−1e−xn, Math. Comput. Model., 54 (2011), 2969–2977. https://doi.org/10.1016/j.mcm.2011.07.019 doi: 10.1016/j.mcm.2011.07.019
![]() |
[7] |
G. Papaschinopoulos, M. A. Radin, C. J. Schinas, Study of the asymptotic behavior of the solutions of three systems of difference equations of exponential form, Appl. Math. Comput., 218 (2012), 5310–5318. https://doi.org/10.1016/j.amc.2011.11.014 doi: 10.1016/j.amc.2011.11.014
![]() |
[8] |
I. Ozturk, F. Bozkurt, S. Ozen, On the difference equation yn+1=(α+βe−yn)/(γ+yn−1), Appl. Math. Comput., 181 (2006), 1387–1393. https://doi.org/10.1016/j.amc.2006.03.007 doi: 10.1016/j.amc.2006.03.007
![]() |
[9] |
G. Papaschinopoulos, C. J. Schinas, On the dynamics of two exponential type systems of difference equations, Comput. Math. Appl., 64 (2012), 2326–2334. https://doi.org/10.1016/j.camwa.2012.04.002 doi: 10.1016/j.camwa.2012.04.002
![]() |
[10] |
G. Papaschinopoulos, N. Fotiades, C. J. Schinas, On a system of difference equations including negative exponential terms, J. Differ. Equ. Appl., 20 (2014), 717–732. https://doi.org/10.1080/10236198.2013.814647 doi: 10.1080/10236198.2013.814647
![]() |
[11] |
M. Pituk, More on Poincaré's and Peron's theorems for difference equations, J. Differ. Equ. Appl., 8 (2002), 201–216. https://doi.org/10.1080/10236190211954 doi: 10.1080/10236190211954
![]() |
1. | Abdul Khaliq, Tarek F. Ibrahim, Abeer M. Alotaibi, Muhammad Shoaib, Mohammed Abd El-Moneam, Dynamical Analysis of Discrete-Time Two-Predators One-Prey Lotka–Volterra Model, 2022, 10, 2227-7390, 4015, 10.3390/math10214015 | |
2. | Abdul Khaliq, Irfan Mustafa, Tarek F. Ibrahim, Waleed M. Osman, Bushra R. Al-Sinan, Arafa Abdalrhim Dawood, Manal Yagoub Juma, Stability and Bifurcation Analysis of Fifth-Order Nonlinear Fractional Difference Equation, 2023, 7, 2504-3110, 113, 10.3390/fractalfract7020113 | |
3. | Abdul Khaliq, Stephen Sadiq, Hala M. E. Ahmed, Batul A. A. Mahmoud, Bushra R. Al-Sinan, Tarek Fawzi Ibrahim, The Dynamical Behavior of a Three-Dimensional System of Exponential Difference Equations, 2023, 11, 2227-7390, 1808, 10.3390/math11081808 | |
4. | Hashem Althagafi, Dynamics of difference systems: a mathematical study with applications to neural systems, 2025, 10, 2473-6988, 2869, 10.3934/math.2025134 |