Professional Documents
Culture Documents
Research Article
Research Article
Research Article
Optimal Harvest Control in a Singular Prey-Predator Fishery
Model with Maturation Delay and Gestation Delay
Copyright © 2016 Chao Liu et al. This is an open access article distributed under the Creative Commons Attribution License, which
permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
This paper presents a singular prey-predator fishery model, where maturation delay for prey and gestation delay for predator are
considered. Fishing efforts are introduced to harvest prey and predator population, which are developed as control instruments to
investigate optimal utilization of fishery resource. By analyzing associated characteristic equation, local stability analysis is studied
due to combined variations of double time delays. Furthermore, Pontryagin’s maximum principle is utilized to characterize optimal
harvest control, and the optimality system is numerically solved based on an iterative method.
commercial harvesting. Hence, TC in (2) takes the following The rest of the sections of this paper are organized as fol-
form TC = 𝑐𝐸(𝑡). lows. Positivity of any solution of system (3) and uniform per-
Recently, many theoreticians and experimentalists have sistence of system (3) are investigated in the second section.
investigated complex dynamics of prey-predator fishery sys- Local stability analysis of system (3) around interior equi-
tem [16, 17]; it reveals that time delay may cause the loss of librium is discussed in the third section. The optimal system
stability and other complicated dynamical behavior such as is derived and solved numerically based on an iterative
the periodic structure and bifurcation phenomenon [15, 18, method with Runge-Kutta fourth order scheme. Optimal
19]. harvest control problems associated with maximizing total
Keeping these aspects in view, we will extend work in discounted net revenues from the fishery and minimizing
[5] by incorporating maturation delay for prey and gestation harvest cost are discussed in the fourth section. Numerical
delay for predator into system (1). In this paper, prey is simulations are carried out to support theoretical analysis in
considered to be delayed by maturation delay 𝜏1 ≥ 0 due to the fifth section. Finally, this paper ends with a conclusion.
population crowding [15, 16], and predator is assumed to be
delayed by gestation delay 𝜏2 ≥ 0 [16]. Based on (1) and (2), a 2. Positivity and Uniform Persistence
singular prey-predator fishery model is established as follows:
Theorem 1. Any solutions of system (3) with initial conditions
𝑥 (𝑡 − 𝜏1 ) 𝛽𝑥 (𝑡) 𝑦 (𝑡) (4) are positive.
𝑥̇ (𝑡) = 𝑟𝑥 (𝑡) (1 − )− − 𝐸 (𝑡)
𝐾 𝛼 + 𝑥 (𝑡)
Proof. Due to lemma in [20] and Theorem A.4 in [21],
⋅ 𝑥 (𝑡) , any solution of system (3) with initial conditions (4) exists
uniquely and each component of solution remains within
𝑚𝛽𝑥 (𝑡 − 𝜏2 ) interval [0, 𝑋0 ) for some 𝑋0 > 0. Standard and simple
𝑦̇ (𝑡) = 𝑦 (𝑡) [ − 𝑑] − 𝐸 (𝑡) 𝑦 (𝑡) ,
𝛼 + 𝑥 (𝑡 − 𝜏2 ) (3) arguments show that any solution of system (3) always exists
and stays positive.
0 = 𝐸 (𝑡)
Theorem 2. If V > 0, 𝜏1 and 𝜏2 are bounded, and the following
𝑎1 𝑎2
⋅ [𝑥 (𝑡) + 𝑦 (𝑡) − 𝑐] inequalities (6) hold, and then system (3) with initial conditions
𝑏1 + 𝐸 (𝑡) 𝑥 (𝑡) 𝑏2 + 𝐸 (𝑡) 𝑦 (𝑡) (4) is uniformly persistent:
− V, 𝑚𝛽𝐾𝑒𝑟𝜏1 > 𝛼𝑑,
where 𝑐 represents cost of commercial harvesting, V is
𝑎1 𝑏2 𝛽𝐾𝑒𝑟𝜏1 + 𝑎2 𝑏1 𝑟 (𝛼 + 𝐾𝑒𝑟𝜏1 ) > 𝑐𝑏1 𝑏2 𝛽,
assumed to be economic interest of commercial harvesting, (6)
𝐸(𝑡) represents commercial harvesting amount with respect 𝑐𝛼𝑏1 > 𝑎1 𝐾𝑒𝑟𝜏1 (𝑚𝛽𝐾𝑒𝑟𝜏1 − 𝛼𝑑) ,
to time 𝑡, and interpretations for other parameters and state
variables share the same interpretations introduced in system 𝑐𝑏2 𝛼𝛽 > 𝑎2 𝑟 (𝛼 + 𝐾𝑒𝑟𝜏1 ) (𝑚𝛽𝐾𝑒𝑟𝜏1 − 𝛼𝑑) .
(1) and (2). Furthermore, system (3) is investigated with initial
conditions Proof. By using Taylor series expansion [2], for 𝑥(𝑡) and 𝜏1 >
0, 𝜏2 > 0 we have
𝑥 (𝜃) ≥ 0,
𝑑
𝑦 (0) ≥ 0, 𝑥 (𝑡 − 𝜏1 ) = 𝑥 (𝑡) − 𝜏1 (𝑥 (𝑡) − 𝜏1 𝑥̇ (𝑡) + ⋅ ⋅ ⋅) ,
(4) 𝑑𝑡
(7)
𝐸 (0) ≥ 0, 𝑑
𝑥 (𝑡 − 𝜏2 ) = 𝑥 (𝑡) − 𝜏2 (𝑥 (𝑡) − 𝜏2 𝑥̇ (𝑡) + ⋅ ⋅ ⋅) .
𝜃 ∈ [−𝜏, 0] , 𝜏 = max {𝜏1 , 𝜏2 } . 𝑑𝑡
𝑥̇ (𝑡) 𝑥 (𝑡 − 𝜏1 ) ≤ 𝑥 (𝑡) ,
[ ] (8)
Ξ [𝑦̇ (𝑡)]
𝑥 (𝑡 − 𝜏2 ) ≤ 𝑥 (𝑡) .
[ 0 ]
By using (8) and the first equation of system (3), it gives
𝑥 (𝑡 − 𝜏1 ) 𝛽𝑥 (𝑡) 𝑦 (𝑡)
[ 𝑟𝑥 (𝑡) (1 − )− − 𝐸 (𝑡) 𝑥 (𝑡) ] (5) that
[ 𝐾 𝛼 + 𝑥 (𝑡) ]
[ ]
[
=[ 𝑦 (𝑡) [
𝑚𝛽𝑥 (𝑡 − 𝜏2 )
− 𝑑] − 𝐸 (𝑡) 𝑦 (𝑡)
]
],
𝑥 (𝑡) ≤ 𝑥 (𝑡 − 𝜏1 ) 𝑒𝑟𝜏1 ,
[ 𝛼 + 𝑥 (𝑡 − 𝜏2 ) ]
[ ] (9)
[ 𝑎1 𝑎2 ] 𝑥 (𝑡)
𝐸 (𝑡) [𝑥 (𝑡) + 𝑦 (𝑡) − 𝑐] − V 𝑥 (𝑡) ≤ 𝑟𝑥 (𝑡) (1 − ),
[ 𝑏1 + 𝐸 (𝑡) 𝑥 (𝑡) 𝑏2 + 𝐸 (𝑡) 𝑦 (𝑡) ] 𝐾𝑒𝑟𝜏1
where the third equation in system (3) does not contain any which derives that
differentiated variables; hence the third row in leading matrix
100 lim sup 𝑥 (𝑡) ≤ 𝐾𝑒𝑟𝜏1 fl Q1 . (10)
Ξ = [ 0 1 0 ] has a zero row. 𝑡→∞
000
Discrete Dynamics in Nature and Society 3
If 𝜏1 is bounded, then it is easy to show that 𝑄1 > 0 and is If the following two inequalities hold
bounded. According to (10), it can be derived that there exists
𝑇1 > 0 such that 𝑥(𝑡) ≤ 𝑄1 holds for 𝑡 > 𝑇1 + 𝜏1 . 𝑐𝛼𝑏1 > 𝑎1 𝐾𝑒𝑟𝜏1 (𝑚𝛽𝐾𝑒𝑟𝜏1 − 𝛼𝑑) ,
Based on the first equation of system (3) and practical (18)
interpretations of prey survival, it can be derived that there 𝑐𝑏2 𝛼𝛽 > 𝑎2 𝑟 (𝛼 + 𝐾𝑒𝑟𝜏1 ) (𝑚𝛽𝐾𝑒𝑟𝜏1 − 𝛼𝑑) ,
exists 𝑇2 > 𝑇1 + 𝜏1 such that 𝑦(𝑡) ≤ 𝑟[𝛼 + 𝑥(𝑡)]/𝛽, and
𝑟𝜏
𝑟 [𝛼 + 𝑥 (𝑡)] 𝑟 (𝛼 + 𝐾𝑒 1 ) then 𝑊1 > 0 and 𝑊2 > 0 and is bounded, and there exists
lim sup 𝑦 (𝑡) ≤ ≤ fl 𝑄2 . (11) 𝑇8 > 𝑇7 such that 𝑥(𝑡) ≥ 𝑊1 and 𝑦(𝑡) ≥ 𝑊2 hold for 𝑡 > 𝑇8 .
𝑡→∞ 𝛽 𝛽
Based on the above analysis, it can be concluded that sys-
Similarly, based on the second equation of system (3) and tem (3) with initial conditions (4) is uniformly persistent.
practical interpretations of predator survival, it is shown that
there exists 𝑇3 > 𝑇2 such that
3. Local Stability Analysis
𝑚𝛽𝑥 (𝑡 − 𝜏2 )
𝐸 (𝑡) ≤ − 𝑑. (12) Firstly, interior equilibrium 𝑃∗ of system (3) in the case of V >
𝛼
0 is computed as follows: 𝑃∗ (𝑥∗ , 𝑦∗ , 𝐸∗ ) = (𝑥∗ , (𝛼 + 𝑥∗ )[𝐾𝑑 +
By using (8), if 𝜏2 is bounded then it follows that 𝐸(𝑡) ≤
𝑟(𝐾 − 𝑥∗ )]/𝐾𝛽 − 𝑚𝑥∗ , 𝑚𝛽𝑥∗ /(𝛼 + 𝑥∗ ) − 𝑑), and 𝑥∗ satisfy the
(𝑚𝛽𝑥(𝑡)/𝛼) − 𝑑, and
following equation:
𝑚𝛽𝐾𝑒𝑟𝜏1 − 𝛼𝑑
lim sup 𝐸 (𝑡) ≤ fl 𝑄3 . (13)
𝑡→∞ 𝛼 𝐴 0 𝑥∗6 + 𝐴 1 𝑥∗5 + 𝐴 2 𝑥∗4 + 𝐴 3 𝑥∗3 + 𝐴 4 𝑥∗2 + 𝐴 5 𝑥∗
𝑟𝜏1
If 𝑚𝛽𝐾𝑒 > 𝛼𝑑, then it can be derived that 𝑄3 > 0 and (19)
+ 𝐴 6 = 0,
is bounded. According to (13), it is shown that there exists
𝑇4 > 𝑇3 such that 𝐸(𝑡) ≤ 𝑄3 holds for 𝑡 > 𝑇4 .
By virtue of the third equation of system (3), when V > 0, where 𝐴 𝑘 (𝑘 = 0, 1, . . . , 6) are defined as follows:
it can be derived that there exists 𝑇5 > 𝑇4 such that
V 𝑎1 𝑎2 2
= 𝑥 (𝑡) + 𝑦 (𝑡) 𝐴 0 = 𝑟 (𝑑 − 𝑚𝛽) (𝑎1 + 𝑎2 − V + 𝑑 − 𝑚𝛽) ,
𝐸 (𝑡) 𝑏1 + 𝐸 (𝑡) 𝑥 (𝑡) 𝑏2 + 𝐸 (𝑡) 𝑦 (𝑡)
2
(14) 𝐴 1 = (𝑚𝛽 − 𝑑) [3𝛼𝑑𝑟
𝑎1 𝑥 (𝑡) 𝑎2 𝑦 (𝑡)
−𝑐≤ + − 𝑐,
𝑏1 𝑏2 + (𝐾𝑑 + 𝐾𝑟 − 𝛼𝑟 − 𝑚𝐾𝛽) (𝑚𝛽 − 𝑑 + V − 𝑎1 − 𝑎2 )
which implies that
− 𝑐𝑏1 𝑟] ,
lim inf 𝐸 (𝑡)
𝑡→∞
3 2
(15) 𝐴 2 = 𝐾𝛼 (𝑑 + 𝑟) (𝑚𝛽 − 𝑑) − 3𝛼𝑑 (𝑚𝛽 − 𝑑) (𝐾𝑑
V𝑏1 𝑏2 𝛽
≥ fl 𝑊3 .
𝑎1 𝑏2 𝛽𝐾𝑒𝑟𝜏1 𝑟𝜏
+ 𝑎2 𝑏1 𝑟 (𝛼 + 𝐾𝑒 1 ) − 𝑐𝑏1 𝑏2 𝛽 + 𝐾𝑟 − 𝛼𝑟 − 𝑚𝐾𝛽) − 3𝛼2 𝑑2 𝑟 (𝑚𝛽 − 𝑑) + (𝑚𝛽
𝑟𝜏1 𝑟𝜏1
If 𝑎1 𝑏2 𝛽𝐾𝑒 + 𝑎2 𝑏1 𝑟(𝛼 + 𝐾𝑒 ) > 𝑐𝑏1 𝑏2 𝛽, then it is easy 2
to show that 𝑊3 > 0 and is bounded, and then there exists − 𝑑) [𝑐𝑏1 (𝐾𝑑 + 𝐾𝑟 − 𝛼𝑟 − 𝑚𝐾𝛽) + 𝐾𝛽𝑐𝑏2
𝑇6 > 𝑇5 such that 𝐸(𝑡) ≥ 𝑊3 holds for 𝑡 > 𝑇6 .
By virtue of the third equation of system (3), when V > 0, − 𝛼𝐾 (𝑑 + 𝑟) (𝑎1 + 𝑎2 − V)] + 𝛼 (𝑚𝛽 − 𝑑) (𝑚𝛽
it can be derived that there exists 𝑇7 > 𝑇6 such that
− 3𝑑) [(𝑎1 + 𝑎2 − V) (𝐾𝑑 + 𝐾𝑟 − 𝛼𝑟 − 𝑚𝐾𝛽)
𝑏 𝑎
𝐸 (𝑡) 𝑥 (𝑡) > 1 [𝑐 − 2 𝐸 (𝑡) 𝑦 (𝑡)] ,
𝑎1 𝑏2 − 𝑐𝑏1 𝑟] ,
(16)
𝑏2 𝑎 2
𝐸 (𝑡) 𝑦 (𝑡) > [𝑐 − 1 𝐸 (𝑡) 𝑥 (𝑡)] . 𝐴 3 = 𝛼𝐾 (𝑑 + 𝑟) (𝑐𝑏1 − 3𝛼𝑑) (𝑚𝛽 − 𝑑) + 𝑟𝛼3 𝑑3
𝑎2 𝑏1
According to (10), (11), and (13), it can be obtained that + 𝐾𝑐𝑏1 𝑏2 𝛽 (𝑚𝛽 − 𝑑) + V𝑏1 𝑏2 𝐾𝛽 + 𝛼 (𝑚𝛽 − 𝑑) (𝐾𝑑
lim inf 𝑥 (𝑡)
𝑡→∞ + 𝐾𝑟 − 𝛼𝑟 − 𝑚𝐾𝛽) [3𝛼𝑑2 + 𝑐𝑏1 (𝑚𝛽 − 3𝑑)] + (V
𝑐𝑏2 𝛼𝛽 − 𝑎2 𝑟 (𝛼 + 𝐾𝑒𝑟𝜏1 ) (𝑚𝛽𝐾𝑒𝑟𝜏1 − 𝛼𝑑)
≥ fl 𝑊1 , − 𝑎1 ) 𝑏2 𝐾𝛽 (𝑚𝛽 − 𝑑) + 𝑟𝑑𝛼2 (3𝑑 − 2𝑚𝛽)
𝛼𝛽
(17)
⋅ [(𝑎1 + 𝑎2 − V) (𝐾𝑑 + 𝐾𝑟 − 𝛼𝑟 − 𝑚𝐾𝛽) − 𝑐𝑏1 𝑟]
𝑏 [𝛼𝑐𝑏1 − 𝑎1 𝐾𝑒𝑟𝜏1 (𝑚𝛽𝐾𝑒𝑟𝜏1 − 𝛼𝑑)]
lim inf 𝑦 (𝑡) ≥ 2
𝑡→∞ 𝑎2 𝑏1 (𝑚𝛽𝐾𝑒𝑟𝜏1 − 𝛼𝑑) + 𝛼 (𝑚𝛽 − 𝑑) (𝑚𝛽 − 3𝑑) [𝐾𝛽𝑐𝑏2
fl 𝑊2 . − 𝛼𝐾 (𝑑 + 𝑟) (𝑎1 + 𝑎2 − V)] ,
4 Discrete Dynamics in Nature and Society
2 2
𝛽𝑥∗ 𝑦∗ 𝑎2 𝑏2 𝐸∗ 𝑦∗ (𝑏1 + 𝐸∗ 𝑥∗ ) + 𝑎1 𝑏1 𝐸∗ 𝑥∗ (𝑏2 + 𝐸∗ 𝑦∗ )
𝐵1 = − − ,
(𝛼 + 𝑥∗ )2 𝑎1 𝑏1 𝑥∗ (𝑏2 + 𝐸∗ 𝑦∗ )2 + 𝑎2 𝑏2 𝑦∗ (𝑏1 + 𝐸∗ 𝑥∗ )2 − 𝑐 (𝑏1 + 𝐸∗ 𝑥∗ )2 (𝑏2 + 𝐸∗ 𝑦∗ )2
2 2
𝑎2 𝑏2 𝛽𝑥∗ 𝑦∗2 𝐸∗ (𝑏1 + 𝐸∗ 𝑥∗ ) + 𝑎1 𝑏1 𝛽𝑥∗ 𝑦∗ 𝐸∗ (𝛼 + 𝑥∗ ) (𝑏2 + 𝐸∗ 𝑦∗ )
𝐵2 = 2 2 2 2
,
(𝛼 + 𝑥∗ )2 [𝑎1 𝑏1 𝑥∗ (𝑏2 + 𝐸∗ 𝑦∗ ) + 𝑎2 𝑏2 𝑦∗ (𝑏1 + 𝐸∗ 𝑥∗ ) − 𝑐 (𝑏1 + 𝐸∗ 𝑥∗ ) (𝑏2 + 𝐸∗ 𝑦∗ ) ]
𝑟𝑥∗
𝐵3 = , (24)
𝐾
2
𝑟𝑎2 𝑏2 𝑥∗ 𝑦∗ 𝐸∗ (𝑏1 + 𝐸∗ 𝑥∗ )
𝐵4 = − 2 2 2 2
,
𝐾 [𝑎1 𝑏1 𝑥∗ (𝑏2 + 𝐸∗ 𝑦∗ ) + 𝑎2 𝑏2 𝑦∗ (𝑏1 + 𝐸∗ 𝑥∗ ) − 𝑐 (𝑏1 + 𝐸∗ 𝑥∗ ) (𝑏2 + 𝐸∗ 𝑦∗ ) ]
𝑚𝛼𝛽𝑥∗ 𝑦∗ 𝛽 𝑎2 𝑏2 𝐸∗ (𝑏1 + 𝐸∗ 𝑥∗ )
𝐵5 = [ − ].
(𝛼 + 𝑥∗ )2 𝛼 + 𝑥∗ 𝑎1 𝑏1 𝑥∗ (𝑏2 + 𝐸∗ 𝑦∗ )2 + 𝑎2 𝑏2 𝑦∗ (𝑏1 + 𝐸∗ 𝑥∗ )2 − 𝑐 (𝑏1 + 𝐸∗ 𝑥∗ )2 (𝑏2 + 𝐸∗ 𝑦∗ )2
3.1. Case I: 𝜏1 > 0, 𝜏2 = 0. When 𝜏1 > 0 and 𝜏2 = 0, (23) can According to Routh-Hurwitz criterion [21], a simple suffi-
be rewritten as follows: cient condition (𝐵2 + 𝐵5 )2 − 𝐵42 < 0 guarantees existence of
positive root 𝜔1∗ of (27). Hence, (25) has a pair of purely imag-
𝜆2 + 𝐵1 𝜆 + 𝐵2 + 𝐵5 + (𝐵3 𝜆 + 𝐵4 ) 𝑒−𝜆𝜏1 = 0. (25)
inary roots of the form ±𝑖𝜔1∗ .
By substituting 𝜆 = 𝑖𝜔1 (𝜔1 > 0) into (25) and separating By eliminating sin(𝜔1 𝜏1 ) from (26), it can be calculated
∗
real and imaginary parts, it gives that that 𝜏1𝑘 corresponding to 𝜔1∗ is as follows:
𝜔12 − (𝐵2 + 𝐵5 ) = 𝐵4 cos (𝜔1 𝜏1 ) + 𝐵3 𝜔1 sin (𝜔1 𝜏1 ) ,
(26) 1 (𝐵 − 𝐵1 𝐵3 ) 𝜔1∗2 − (𝐵2 + 𝐵5 ) 𝐵4
∗
𝜏1𝑘 = arccos { 4 }
𝐵1 𝜔1 = −𝐵3 𝜔1 cos (𝜔1 𝜏1 ) + 𝐵4 sin (𝜔1 𝜏1 ) , 𝜔1∗
𝐵42 + 𝐵32 𝜔1∗2
(28)
which derives that 2𝑘𝜋
2 + ∗,
𝜔14 + [𝐵12 − 2 (𝐵2 + 𝐵5 ) − 𝐵32 ] 𝜔12 + (𝐵2 + 𝐵5 ) − 𝐵42 𝜔1
(27)
= 0. where 𝑘 = 0, 1, . . ..
Discrete Dynamics in Nature and Society 5
It follows from Butler’s lemma [22] that system (3) is − 𝐵3 𝐵5 𝜎1 cos (𝜎1 𝜏̃2 )
locally stable around 𝑃∗ when 0 < 𝜏1 < 𝜏10
∗
, 𝜏2 = 0.
Based on the above analysis, local stability analysis of − 𝐵4 𝐵5 sin (𝜎1 𝜏̃2 ) ,
system (3) around 𝑃∗ is concluded in Theorem 3.
𝑓12 (𝜎1 , 𝜏̃2 ) = 𝐵42 + 𝐵32 𝜎12 .
Theorem 3. If (𝐵2 + 𝐵5 )2 − 𝐵42 < 0, then system (3) is locally (33)
stable around 𝑃∗ when 0 < 𝜏1 < 𝜏10
∗ ∗
, 𝜏2 = 0, and 𝜏10 is defined
in (28). Based on 𝑓1𝑘 (𝜎1 , 𝜏̃2 ), 𝑘 = 0, 1, 2, defined in (32), we define
2 2 2
𝐹1 (𝜎1 , 𝜏̃2 ) = 𝑓10 (𝜎1 , 𝜏̃2 ) + 𝑓11 (𝜎1 , 𝜏̃2 ) − 𝑓12 (𝜎1 , 𝜏̃2 )
3.2. Case II: 𝜏1 = 0, 𝜏2 > 0. When 𝜏1 = 0, 𝜏2 > 0, by (34)
using similar arguments in Section 3.1, local stability analysis = 0.
of system (3) around 𝑃∗ can be concluded in the following
Theorem 4. It is difficult to discuss root properties of transcendental
equation (34) due to its complicated form. Without investi-
Theorem 4. If (𝐵2 + 𝐵4 )2 − 𝐵52 < 0, then system (3) is locally gating the properties of roots of (34) in detail, it follows from
stable around 𝑃∗ when 𝜏1 = 0, 0 < 𝜏2 < 𝜏20
∗ ∗
, and 𝜏20 is defined simple computations that (23) has a pair of purely imaginary
as follows: roots when (34) has at least two roots 0 < 𝜎10 < 𝜎11 < ⋅ ⋅ ⋅ .
By denoting 𝜎1𝑤 = max{𝜎1𝑘 , 𝑘 = 0, 1, . . .} and 𝜏1 which
∗ 1 𝜔2∗2 − 𝐵2 − 𝐵4 2𝑘𝜋 is considered to be bifurcation parameter, we can obtain the
𝜏2𝑘 = arccos { }+ ∗ , (29)
𝜔2∗ 𝐵5 𝜔2 corresponding critical value 𝜏1𝑤 :
where 𝑘 = 0, 1, . . . and 𝜔2∗ is a positive root of the following 𝜓1𝑤 + 2𝑘𝜋
equation: 𝜏1𝑘 = ,
𝜎1𝑤 (35)
2 2
𝜔24 + [(𝐵1 + 𝐵3 ) − 2 (𝐵2 + 𝐵4 )] 𝜔22 + (𝐵2 + 𝐵4 ) − 𝐵52 𝜏1𝑤 = min {𝜏1𝑘 , 𝑘 = 0, 1, . . .} ,
(30)
= 0. where 𝑘 = 0, 1, . . . , 𝜓1𝑐 ∈ [0, 2𝜋) satisfies the following
equations:
∗
3.3. Case III: 𝜏1 > 0, 𝜏2 ∈ (0, 𝜏20 ), 𝜏1 ≠ 𝜏2 . In this subsection,
𝜏1 is considered to be bifurcation parameter and 𝜏2 is regarded 𝑓10 (𝜎1𝑤 , 𝜏̃2 )
∗ ∗
cos 𝜓1𝑤 = ,
as fixed value 𝜏̃2 ∈ (0, 𝜏20 ), where 𝜏20 is determined in 𝑓12 (𝜎1𝑤 , 𝜏̃2 )
Theorem 4. (36)
Let 𝜆 = 𝑖𝜎1 (𝜎1 > 0) be the root of (23), and two tran- 𝑓 (𝜎 , 𝜏̃ )
sin 𝜓1𝑤 = 11 1𝑤 2 .
scendental equations are obtained as follows: 𝑓12 (𝜎1𝑤 , 𝜏̃2 )
𝜎12 − (𝐵2 + 𝐵5 cos (𝜎1 𝜏̃2 )) Based on the above analysis, local stability analysis
around 𝑃∗ due to dynamical variation of bifurcation param-
= 𝐵4 cos (𝜎1 𝜏1 ) + 𝐵3 𝜎1 sin (𝜎1 𝜏1 ) , eter 𝜏1 can be summarized as follows.
(31)
𝐵1 𝜎1 − 𝐵5 sin (𝜎1 𝜏̃2 ) Theorem 5. If (34) has at least two roots 0 < 𝜎10 < 𝜎11 < ⋅ ⋅ ⋅ ,
and there exists 𝜏1𝑤 defined in (35), then system (3) is locally
= 𝐵4 sin (𝜎1 𝜏1 ) − 𝐵3 𝜎1 cos (𝜎1 𝜏1 ) , stable around 𝑃∗ when 0 < 𝜏1 < 𝜏1𝑤 , 𝜏2 = 𝜏̃2 .
where 𝐵𝑘 , 𝑘 = 1, 2, . . . , 5, are defined in (23).
∗
By eliminating 𝜏̃2 from (31), it can be derived that 3.4. Case IV: 𝜏2 > 0, 𝜏1 ∈ (0, 𝜏10 ), 𝜏1 ≠ 𝜏2 . In this
subsection, 𝜏2 is considered to bifurcation parameter and 𝜏1 is
𝑓10 (𝜎1 , 𝜏̃2 ) regarded as fixed value 𝜏̃1 ∈ (0, 𝜏10∗ ∗
), where 𝜏10 is determined
cos (𝜎1 𝜏1 ) = ,
𝑓12 (𝜎1 , 𝜏̃2 ) in Theorem 3. By using similar arguments in Theorem 5 of
(32) this paper, local stability analysis around 𝑃∗ can be concluded
𝑓 (𝜎 , 𝜏̃ ) in Theorem 6.
sin (𝜎1 𝜏1 ) = 11 1 2 ,
𝑓12 (𝜎1 , 𝜏̃2 ) Firstly, we define following equation:
2 2 2
where 𝑓1𝑘 (𝜎1 , 𝜏̃2 ), 𝑘 = 0, 1, 2, are defined as follows: 𝐹2 (𝜎2 , 𝜏̃1 ) = 𝑓20 (𝜎2 , 𝜏̃1 ) + 𝑓21 (𝜎2 , 𝜏̃1 ) − 𝑓22 (𝜎2 , 𝜏̃1 )
(37)
𝑓10 (𝜎1 , 𝜏̃2 ) = (𝐵4 − 𝐵1 𝐵3 ) 𝜎12 − 𝐵2 𝐵4 = 0,
− 𝐵4 𝐵5 cos (𝜎1 𝜏̃2 ) where 𝑓2𝑘 (𝜎2 , 𝜏̃2 ), 𝑘 = 0, 1, 2, are defined as follows:
+ 𝐵3 𝐵5 𝜎1 sin (𝜎1 𝜏̃2 ) , 𝑓20 (𝜎2 , 𝜏̃1 ) = 𝜎22 − 𝐵2 − 𝐵4 cos (𝜎2 𝜏̃1 )
𝑓21 (𝜎2 , 𝜏̃1 ) = 𝐵1 𝜎2 + 𝐵3 𝜎2 cos (𝜎2 𝜏̃1 ) The Hamiltonian function 𝐻(𝑡) associated with system
(3) with 𝜏̂1 and 𝜏̂2 is
− 𝐵4 sin (𝜎2 𝜏̃1 ) ,
𝑎1
𝑓22 (𝜎2 , 𝜏̃1 ) = 𝐵5 . 𝐻 (𝑡) = 𝑒−𝛿𝑡 [𝐸 (𝑡) 𝑥 (𝑡)
𝑏1 + 𝐸 (𝑡) 𝑥 (𝑡)
(38)
𝑎2
+ 𝐸 (𝑡) 𝑦 (𝑡) − 𝑐𝐸 (𝑡)] + 𝜆 1 (𝑡)
Theorem 6. If (37) has at least two roots 0 < 𝜎20 < 𝜎21 < 𝑏2 + 𝐸 (𝑡) 𝑦 (𝑡)
(43)
⋅ ⋅ ⋅ , there exists a critical delay 𝜏2𝑤 , and then system (3) is 𝑥𝜏̂1
𝛽𝑥 (𝑡) 𝑦 (𝑡)
locally stable around 𝑃∗ when 0 < 𝜏2 < 𝜏2𝑤 , 𝜏1 = 𝜏̃1 , where ⋅ [𝑟𝑥 (𝑡) (1 − )− − 𝐸 (𝑡) 𝑥 (𝑡)]
𝐾 𝛼 + 𝑥 (𝑡)
𝜏2𝑤 = min{(𝜓2𝑤 + 2𝑘𝜋)/𝜎2𝑤 , 𝑘 = 0, 1, . . .}, 𝜎2𝑤 = max{𝜎2𝑘 , 𝑘 =
0, 1, . . .} and 𝜓2𝑤 ∈ [0, 2𝜋) satisfies the following equations: 𝑚𝛽𝑥𝜏̂2
+ 𝜆 2 (𝑡) [𝑦 (𝑡) ( − 𝑑) − 𝐸 (𝑡) 𝑦 (𝑡)] ,
𝛼 + 𝑥𝜏̂2
𝑓20 (𝜎2𝑤 , 𝜏̃1 )
cos 𝜓2𝑤 = , where 𝑥𝜏̂1 fl 𝑥(𝑡 − 𝜏̂1 ), 𝑥𝜏̂2 fl 𝑥(𝑡 − 𝜏̂2 ), and 𝜆 1 (𝑡), 𝜆 2 (𝑡) are
𝑓22 (𝜎2𝑤 , 𝜏̃1 )
(39) adjoint functions.
𝑓 (𝜎 , 𝜏̃ )
sin 𝜓2𝑤 = 21 2𝑤 1 . Theorem 7. There exists an optimal harvest 𝐸𝛿 which is deter-
𝑓22 (𝜎2𝑤 , 𝜏̃1 ) mined by (44); furthermore, there exist adjoint functions 𝜆 1 (𝑡),
𝜆 2 (𝑡) satisfying system (47) with 𝜆 𝑘 (𝑡𝑓 ) = 0, 𝑘 = 1, 2.
4. Optimal Control Problem
𝑎1 𝑏1 𝑥𝛿 𝑎2 𝑏2 𝑦𝛿
In this section, 𝜏1 and 𝜏2 are considered to be a fixed value 2
+ 2
∗ (𝑏1 + 𝐸𝛿 𝑥𝛿 ) (𝑏2 + 𝐸𝛿 𝑦𝛿 ) (44)
𝜏̂1 ∈ (0, 𝜏1𝑤 ) and 𝜏̂2 ∈ (0, 𝜏20 ), respectively; 𝜏̂1 ≠ 𝜏̂2 , 𝜏1𝑤 , and
∗
𝜏20 have been determined in Theorems 5 and 4, respectively. = 𝑐 + [𝑥𝛿 𝜆 1 (𝑡) + 𝑦𝛿 𝜆 2 (𝑡)] 𝑒 . 𝛿𝑡
It follows from Theorem 5 that system (3) with 𝜏̂1 and 𝜏̂2
is locally stable around 𝑃∗ , and we will investigate optimal Proof. By differentiating 𝐻(𝑡), the adjoint system can be
control 𝐸(𝑡) with maximizing total discounted net revenues obtained as follows:
and minimizing commercial harvesting cost during [𝑡0 , 𝑡𝑓 ].
The optimal harvest control problem is formulated as follows: 𝑑𝜆 1 𝜕𝐻 𝜕𝐻
=− (𝑡) − 𝜒1[𝑡0 ,𝑡𝑓 −̂𝜏1 ] (𝑡) (𝑡 + 𝜏̂1 )
𝑑𝑡 𝜕𝑥 𝜕𝑥𝜏̂1
𝑡𝑓
𝑎1
𝐽 (𝐸) = ∫ 𝑒−𝛿𝑡 [𝐸 (𝑡) 𝑥 (𝑡) 𝜕𝐻
𝑡0 𝑏1 + 𝐸 (𝑡) 𝑥 (𝑡) − 𝜒2[𝑡0 ,𝑡𝑓 −̂𝜏2 ] (𝑡) (𝑡 + 𝜏̂2 ) , (45)
(40) 𝜕𝑥𝜏̂2
𝑎2
+ 𝐸 (𝑡) 𝑦 (𝑡) − 𝑐𝐸 (𝑡)] 𝑑𝑡, 𝑑𝜆 2 𝜕𝐻
𝑏2 + 𝐸 (𝑡) 𝑦 (𝑡) =− (𝑡) ,
𝑑𝑡 𝜕𝑦
International
Journal of Journal of
Mathematics and
Mathematical
Discrete Mathematics
Sciences