Professional Documents
Culture Documents
Dynamics and Simulations of Discrete Caputo-Conformable Fractional Order Lotka-Voltera Models
Dynamics and Simulations of Discrete Caputo-Conformable Fractional Order Lotka-Voltera Models
Research Article
Open Access. © 2022 Feras Yousef et al., published by De Gruyter. This work is licensed under the Creative Commons Attribution 4.0
International License.
Discretized Caputo-conformable fractional-order Lotka–Volterra models 101
d d a ⎝⎣ h ⎦ ⎠⎝ ⎝⎣ h ⎦ ⎠ ⎝⎣ h ⎦ ⎠⎠
(6)
⎨ t t
Hence, we are lead to work with the following non- ⎪ D αy (t ) = y⎛⎡ ⎤h⎞⎛−q + x ⎛⎡ ⎤h⎞⎞ ,
⎜ ⎟
dimensionalized system: ⎩ ⎝ ⎣ h ⎦ ⎠⎝ ⎝⎣ h ⎦ ⎠⎠
t
⎧ dx (t ) = rx (t ) − px 2 (t ) − qx (t )y (t ) , where [ h ] denotes the integer part of t ∈ [nh , (n + 1)h),
⎪ dt n = 0, 1, … , and h > 0 is a discretization parameter.
(2)
⎨ d y (t ) The nth iterative solution of system (6) is given by:
⎪ dt = x (t )y (t ) − qy (t ) ,
⎩ α
a2 b a ⎧ xn + 1(t ) = xn(nh) + (t − nh) (xn(nh)(r − pxn(nh)
where r = cd , p = ce , and q = c . ⎪ α Γ (α )
Thus, the following fractional-order system is obtained ⎪
− qyn(nh))) , (7)
by changing the first-order time derivatives by the deriva- ⎨
⎪ (t − nh)α
tives of fractional order. ⎪ yn + 1(t ) = yn(nh) + αΓ(α) ( yn(nh)( −q + xn(nh))) .
α 2 ⎩
⎧ D x (t ) = rx (t ) − px (t ) − qx (t )y (t ) , (3)
⎨ α If we let t → (n + 1)h in system (7), then we will
⎩ D y (t ) = x (t )y (t ) − qy (t ) ,
achieve the discretized version of the model (4):
where 0 < α < 1 is the fractional-order derivative para-
α
meter and t > 0. It is easily verified that the equilibrium ⎧ xn + 1 = xn + h (xn(r − pxn − qy )) ,
n
⎪ α Γ(α )
points of system (3) are the trivial state E0 = (0, 0), the (8)
⎨ h α
2 The dynamics of the discretized Proof. The Jacobian matrix computed at the point of equi-
librium E0 for the linearization of system (8) is given by
Caputo fractional-order α
⎛1 + h r ⎞
prey–predator system ⎜ α Γ (α )
0
⎟
J (E0) = ⎜ hα ⎟
In this section, we consider the Caputo fractional-order 0 1− q
⎜ α Γ (α ) ⎟
version of model (3) as follows: ⎝ ⎠
102 Feras Yousef et al.
hα hα
and has eigenvalues λ1 = 1 + αΓ(α) r and λ2 = 1 − αΓ(α) q . Proof. The Jacobian matrix computed at the point of
Since r > 0, then ∣λ1∣ > 1. Now, since q > 0, we have the equilibrium E2 for the linearization of system (8) is given
following cases: by
2αΓ(α)
(i) If 0 < h < α , then ∣λ2 ∣ < 1 and E0 is a saddle α q2hα ⎞
q ⎛ 1 − pqh −
point. ⎜ α Γ (α ) α Γ(α ) ⎟
2αΓ(α)
J (E2 ) = ⎜ α ⎟.
(ii) If h > α , then ∣λ2 ∣ > 1 and E0 is a source. h ⎛ r − pq ⎞
q ⎜ ⎜ 1 ⎟
⎟
2αΓ(α) ⎝ α Γ(α ) ⎝ q ⎠ ⎠
(iii) If h = α , then ∣λ2 ∣ = 1 and E0 is a nonhyper-
q
Next, the trace and determinant of J (E2 ) are com-
bolic. □
puted as follows:
⎝ α Γ (α ) ⎝ p ⎠⎠
hα
(q, r − pq
q ) forfeits its stability through a
and has eigenvalues λ1 = 1 − and r λ2 = 1 + (i) Transcritical bifurcation if r = pq ;
αΓ(α)
hα
αΓ(α) ( r
p )
− q . Now, we have the following cases:
(ii) Flip bifurcation if r =
2pqhα
α Γ( α )
−4
+ pq ;
(i) If q − ( 2αΓ(α)
hα )p < r < min ( 2αΓ(α)
hα
, )
pq , then ∣λ1∣ < 1 and q ( ) hα 2
α Γ( α )
(q, r − pq
) is local asymptotically stable if and only if q ( )
α Γ( α )
q
⎛ 2pqhα ⎞
(iii) Det(J (E2 )) = 1 when r = p ( αΓ(α)
hα )
+ q . Hence, the
αΓ(α)
−4 α Γ (α ) point of equilibrium E2 forfeits its stability through
max ⎜ 2
+ pq , pq ⎟ < r < p⎛ α + q ⎞ .
⎜q
⎝ ( )
h α
αΓ(α)
⎟
⎠
⎝ h ⎠ a Neimark–Sacker bifurcation if r = p ( αΓ(α)
hα
+q . ) □
Discretized Caputo-conformable fractional-order Lotka–Volterra models 103
Figure 1: Stable dynamical behavior of system (8) subject to the initial condition (x (0), y (0)) = (0.25, 0.2) for the parameters
p = 0.5, q = 1, h = 0.15, and r = 1 : (a) α = 0.95, (b) α = 0.75, (c) α = 0.6, and (d) α = 0.5.
104 Feras Yousef et al.
Figure 2: Stable dynamical behavior of system (22) subject to the initial condition (x (0), y (0)) = (0.25, 0.2) for the parameters
p = 0.5, q = 1, h = 0.15, and r = 1 : (a) α = 0.95, (b) α = 0.75, (c) α = 0.6, and (d) α = 0.5.
f (t + ε (t − a)1 − α) − f (t ) Applying the rule in Eq. (12) to the first equation in the
Daαf (t ) = lim , 0 < α < 1. (11)
ε→0 ε system (13), for h > 0 and t ∈ [nh , (n + 1)h), n = 0, 1, … ,
From the aforementioned definition, it has been evinced gives the following Bernoulli differential equation:
in ref. [40] the following necessary fact: d x (t )
(t − nh)1 − α + (qy (nh) − r )x (t ) = −px 2 (t ) . (14)
Daαf (t ) = (t − a)1 − αf ′(t ) . (12) dt
We obtain by simplifying this equation
In the following, we will adopt piecewise-constant
approximation to discretize the model (10). x′(t ) (r − qy (nh)) p
− + = . (15)
x 2 (t ) x (t )(t − nh)1 − α (t − nh)1 − α
⎧ D αx (t ) = x (t )⎛⎜r − px (t ) − qy⎛⎡ t ⎤h⎞⎞⎟,
⎪ a
⎝ ⎝⎣ h ⎦ ⎠⎠
(13)
⎨ α t
⎪ Da y (t ) = y (t )⎛⎜−q + x ⎛⎡ ⎤h⎞⎞⎟ .
⎩ ⎝ h
⎝⎣ ⎦ ⎠⎠
Discretized Caputo-conformable fractional-order Lotka–Volterra models 105
Figure 3: Stable dynamical behavior of system (8) subject to the initial condition (x (0), y (0)) = (0.25, 0.2) for the parameters
p = 0.5, q = 1, α = 0.95, and r = 1 : (a) h = 0.15, (b) h = 0.35, (c) h = 0.55, and (d) h = 0.75.
Figure 4: Stable dynamical behavior of system (22) subject to the initial condition (x (0), y (0)) = (0.25, 0.2) for the parameters
p = 0.5, q = 1, α = 0.95, and r = 1 : (a) h = 0.15, (b) h = 0.35, (c) h = 0.55, and (d) h = 0.75.
For t → (n + 1)h in Eq. (20) and replacing y (nh) with Theorem 3.1. The point of equilibrium E0 = (0, 0) is a
yn provides saddle point.
hα
yn + 1 = yne (xn − q) α . (21)
Proof. The Jacobian matrix computed at the point of
Consequently, the discretized version of the model equilibrium E0 for the linearization of system (22) is given
(13) is derived as follows: by
⎧x = xn(r − qyn) hα
⎪ n+1 hα
, ⎛er α 0 ⎞
pxn + (r − qyn − pxn)e−(r − qyn) α (22) J (E0) = ⎜ hα
⎨ 0 e qα⎟
−
α
⎪ yn + 1 = yne (xn − q) hα . ⎝ ⎠
⎩
hα hα
and has eigenvalues λ1 = e r α and λ2 = e−q α . Thus, E0 is a
We next investigate the local stability and bifurcation
saddle point since ∣λ1∣ > 1 and ∣λ2 ∣ < 1. □
to the points of equilibrium for system (22).
Discretized Caputo-conformable fractional-order Lotka–Volterra models 107
Figure 5: Bifurcation diagram of system (8) as a function of r for the parameters p = 0.5, q = 1, and h = 0.15: (a) α = 0.95 and (b) α = 0.5.
Figure 6: Bifurcation diagram of system (22) as a function of r for the parameters p = 0.5, q = 1 , and h = 0.15: (a) α = 0.95 and (b) α = 0.5.
108 Feras Yousef et al.
Figure 7: Maximum Lyapunov exponents corresponding to: (a) Figure 5(a) and (b) Figure 6(a).
Det(J (E2 )) =
pαe
−pqhα
α − h α (r (
− pq ) −1 + e
−pqhα
α
). (23)
(q, r − pq
q ) forfeits its stability through a Neimark–Sacker
pα bifurcation if r = p ( α
hα )
+q .
Figure 8: Chaotic attractor for the parameters p = 0.5, q = 1 , α = 0.95, h = 0.15, and r = 4: (a) system (8) and (b) system (22).
Discretized Caputo-conformable fractional-order Lotka–Volterra models 109
Figure 9: Time series plot of system (8) with respect to Figure 5(a): (a) asymptotically stable for r = 2 and (b) chaotic for r = 4, and system
(22) with respect to Figure 6(a): (c) asymptotically stable for r = 2 and (d) chaotic for r = 4.
Proof. From Eq. (23), we have Det(J (E2 )) = 1 when r = be fixed and vary α , h, and r . Suppose that the initial state
of systems (8) and (22) is (0.25, 0.2). Figures 1 and 2
p( α
hα )
+ q . Hence, the point of equilibrium E2 forfeits
demonstrate the local stable dynamic behaviors for the
its stability through a Neimark–Sacker bifurcation if
two-dimensional discrete systems (8) and (22), respec-
r=p ( α
hα
+q .) □ tively, at the positive point of equilibrium E2 with les-
sening the fractional-order parameter α . We note that
lessening the fractional-order parameter α and fixing the
discretization parameter h lead to destabilize the two-
4 Numerical simulations dimensional discrete systems and chaotic behavior occurs.
Figures 3 and 4 show the local stable dynamic behaviors
Theoretical studies cannot be verified without numerical for the two-dimensional discrete systems (8) and (22),
investigation of the obtained results. In the present study, respectively, at the positive point of equilibrium E2 with
numerical computations have been accomplished through rising the discretization parameter h. We note that rising
the use of MATLAB-R2020a software. Let p = 0.5 and q = 1 the discretization parameter h along with fixed fractional-
110 Feras Yousef et al.
order parameter α leads to destabilize the two-dimensional equilibrium E2 . While, the discrete conformable-system
discrete systems and chaotic behavior occurs. undergoes only a Neimark–Sacker bifurcation.
The bifurcation diagrams in Figures 5 and 6 demonstrate
that rising the values of r may destabilize the point of equili- On conclusion, these fractional derivatives act to
brium E2 through a Neimark–Sacker bifurcation. The maxi- some extent the same function in importing some of the
mum Lyapunov exponents corresponding to Figures 5(a) inherited properties of the time fractional to the time-
and 6(a) are given in Figure 7. The chaotic attractor for integer Lotka–Volterra prey–predator models.
system (8) and system (22) is presented in Figure 8. The
chaotic behavior exists when lessening the fractional-order Funding information: The authors state no funding
parameter α and rising the discretization parameter h. involved.
[13] Naik PA, Eskandari Z, Shahraki HE. Flip and generalized flip [27] Alquran M, Alsukhour M, Ali M, Jaradat I. Combination of
bifurcations of a two-dimensional discrete-time chemical Laplace transform and residual power series techniques to
model. Math Model Numer Simu Appl. 2021;1(2):95–101. solve autonomous n-dimensional fractional nonlinear sys-
[14] Naik PA, Yavuz M, Qureshi S, Zu J, Townley S. Modeling and tems. Nonlinear Eng. 2021;10(1):282–92.
analysis of COVID-19 epidemics with treatment in fractional [28] Hammouch Z, Yavuz M, Özdemir N. Numerical solutions and
derivatives using real data from Pakistan. Eur Phys J Plus. synchronization of a variable-order fractional chaotic system.
2020;135(10):1–42. Math Model Numer Simu Appl. 2021;1(1):11–23.
[15] Yavuz M, Coşar FÖ, Günay F, Özdemir FN. A new mathematical [29] Kumar P, Erturk VS. Dynamics of cholera disease by using two
modeling of the COVID-19 pandemic including the vaccination recent fractional numerical methods. Math Model Numer Simu
campaign. Open J Model Simu. 2021;9(3):299–321. Appl. 2021;1(2):102–11.
[16] Allegretti S, Bulai IM, Marino R, Menandro MA, Parisi K. [30] Maayah B, Yousef F, Arqub OA, Momani S, Alsaedi A.
Vaccination effect conjoint to fraction of avoided contacts for a Computing bifurcations behavior of mixed type singular time-
Sars-Cov-2 mathematical model. Math Model Numer Simu fractional partial integrodifferential equations of Dirichlet
Appl. 2021;1(2):56–66. functions types in Hilbert space with error analysis. Filomat.
[17] Özköse F, Yavuz M. Investigation of interactions between 2019;33(12):3845–53.
COVID-19 and diabetes with hereditary traits using real data: [31] Jaradat I, Alquran M, Katatbeh Q, Yousef F, Momani S,
a case study in Turkey. Comp Bio Med. 2022;141:105044. Baleanu D. An avant-garde handling of temporal-spatial frac-
[18] Joshi H, Jha BK. Chaos of calcium diffusion in Parkinsonas tional physical models. Int J Nonlinear SciNumer Simu.
infectious disease model and treatment mechanism via Hilfer 2020;21(2):183–94.
fractional derivative. Math Model Numer Simu Appl. [32] Jaradat I, Alquran M, Yousef F, Momani S, Baleanu D. On (2+1)-
2021;1(2):84–94. dimensional physical models endowed with decoupled spatial
[19] Özköse F, Senel MT, Habbireeh R. Fractional-order mathema- and temporal memory indices. Eur Phys J Plus.
tical modelling of cancer cells-cancer stem cells-immune 2019;134(7):360.
system interaction with chemotherapy. Math Model Numer [33] Momani S, Arqub OA, Maayah B, Yousef F, Alsaedi A. A reliable
Simu Appl. 2021;1(2):67–83. algorithm for solving linear and nonlinear Schrödinger equa-
[20] Naik PA, Owolabi KM, Yavuz M, Zu J. Chaotic dynamics of a tions. Appl Comput Math. 2018;17(2):151–60.
fractional order HIV-1 model involving AIDS-related cancer [34] Yousef F, Alkam O, Saker I. The dynamics of new motion styles
cells. Chaos Solit Fract. 2020;140:110272. in the time-dependent four-body problem: weaving periodic
[21] Yavuz M, Özdemir N. Analysis of an epidemic spreading model solutions. Eur Phys J Plus. 2020;135(9):742.
with exponential decay law. Math Scie Appl E-Notes. [35] Yousef F, Momani S, Abdalmohsen R. Analytic solution of
2020;8(1):142–54. spatial-temporal fractional Klein-Gordon equation arising in
[22] Gurcan F, Kaya G, Kartal S. Conformable fractional order lotka- physical models. Proc Int Conf Frac Differ Appl. 2018:1–4.
volterra predator-prey model: discretization, stability and doi: 10.2139/ssrn.3277393.
bifurcation. J Comput Nonlinear Dynam. 2019;14(11):111007. [36] Kartal S, Gurcan F. Discretization of conformable fractional
[23] Wang Z, Xie Y, Lu J, Li Y. Stability and bifurcation of a delayed differential equations by a piecewise constant approximation.
generalized fractional-order prey–predator model with inter- Int J Comput Math. 2019;96(9):1849–60.
specific competition. App Math Comp. 2019;347:360–9. [37] Kot M. Elements of mathematical ecology. Cambridge:
[24] Ahmed E, Elgazzar AS. On fractional order differential Cambridge University Press; 2001.
equations model for nonlocal epidemics. Phys A. [38] Elaydi S. Discrete Chaos: with Applications in Science and
2007;379:607–14. Engineering. Boca Raton: Chapman and Hall/CRC; 2008.
[25] Ahmed E, El-Sayed AMA, El-Saka HAA. Equilibrium points, [39] Khalil R, AlHorani M, Yousef A, Sababheh M. A new definition
stability and numerical solutions of fractional-order preda- of fractional derivative. J Compu Appl Math. 2014;264:65–70.
tor-prey and rabies models. J Math Anal Appl. [40] Abdeljawad T. On conformable fractional calculus. J Comput
2007;325:542–53. Appl Math. 2015;279:57–66.
[26] Jaradat I, Alquran M, Sulaiman TA, Yusuf A. Analytic simulation [41] Teodoro GS, TenreiroMachado JA, De Oliveira EC. A review of
of the synergy of spatial-temporal memory indices with pro- definitions of fractional derivatives and other operators.
portional time delay. Chaos Solit Fract. 2022;156:111818. J Comput Phys. 2019;338:195–208.