Research article

Dynamic behaviors for reaction-diffusion neural networks with mixed delays

  • A class of reaction-diffusion neural networks with mixed delays is studied. We will discuss some important properties of the periodic mild solutions including existence and globally exponential stability by using exponential dissipation property of semigroup of operators and some analysis techniques. Finally, an example for the above neural networks is given to show the effectiveness of the results in this paper.

    Citation: Mei Xu, Bo Du. Dynamic behaviors for reaction-diffusion neural networks with mixed delays[J]. AIMS Mathematics, 2020, 5(6): 6841-6855. doi: 10.3934/math.2020439

    Related Papers:

    [1] Mingzhan Huang, Xiaohuan Yu . Dynamic analysis of a mosquito population model with a stage structure and periodic releases of sterile males. AIMS Mathematics, 2023, 8(8): 18546-18565. doi: 10.3934/math.2023943
    [2] Shouzong Liu, Yang Xu, Mingzhan Huang . Comparative analysis of sterile mosquito release strategies based on a population suppression model. AIMS Mathematics, 2024, 9(9): 23344-23367. doi: 10.3934/math.20241135
    [3] Zhongcai Zhu, Xue He . Rich and complex dynamics of a time-switched differential equation model for wild mosquito population suppression with Ricker-type density-dependent survival probability. AIMS Mathematics, 2023, 8(12): 28670-28689. doi: 10.3934/math.20231467
    [4] Folashade Agusto, Daniel Bond, Adira Cohen, Wandi Ding, Rachel Leander, Allis Royer . Optimal impulse control of West Nile virus. AIMS Mathematics, 2022, 7(10): 19597-19628. doi: 10.3934/math.20221075
    [5] Zongmin Yue, Yitong Li, Fauzi Mohamed Yusof . Dynamic analysis and optimal control of Zika virus transmission with immigration. AIMS Mathematics, 2023, 8(9): 21893-21913. doi: 10.3934/math.20231116
    [6] Liping Wang, Peng Wu, Mingshan Li, Lei Shi . Global dynamics analysis of a Zika transmission model with environment transmission route and spatial heterogeneity. AIMS Mathematics, 2022, 7(3): 4803-4832. doi: 10.3934/math.2022268
    [7] Puntipa Pongsumpun, Jiraporn Lamwong, I-Ming Tang, Puntani Pongsumpun . A modified optimal control for the mathematical model of dengue virus with vaccination. AIMS Mathematics, 2023, 8(11): 27460-27487. doi: 10.3934/math.20231405
    [8] Vikas Kumar, Nitu Kumari . Controlling chaos in three species food chain model with fear effect. AIMS Mathematics, 2020, 5(2): 828-842. doi: 10.3934/math.2020056
    [9] Lin Wu, Jianjun Jiao, Xiangjun Dai, Zeli Zhou . Dynamical analysis of an ecological aquaculture management model with stage-structure and nonlinear impulsive releases for larval predators. AIMS Mathematics, 2024, 9(10): 29053-29075. doi: 10.3934/math.20241410
    [10] Iftikhar Ahmad, Hira Ilyas, Muhammad Asif Zahoor Raja, Tahir Nawaz Cheema, Hasnain Sajid, Kottakkaran Sooppy Nisar, Muhammad Shoaib, Mohammed S. Alqahtani, C Ahamed Saleel, Mohamed Abbas . Intelligent computing based supervised learning for solving nonlinear system of malaria endemic model. AIMS Mathematics, 2022, 7(11): 20341-20369. doi: 10.3934/math.20221114
  • A class of reaction-diffusion neural networks with mixed delays is studied. We will discuss some important properties of the periodic mild solutions including existence and globally exponential stability by using exponential dissipation property of semigroup of operators and some analysis techniques. Finally, an example for the above neural networks is given to show the effectiveness of the results in this paper.


    As the vector of many diseases, mosquitoes are listed as one of the most lethal animals to humans in the world. With the development of people's understanding and research on mosquito-borne diseases, mosquito control is considered to be the key to suppress the spread of such diseases. In addition to the traditional physical and chemical means of mosquito control, in recent years, biological control methods have also received a lot of attention. For example, using the sterile insect technique or the Wolbachia driven mosquito control technique to release sterile mosquitoes or Wolbachia-infected mosquitoes into the field are common means of biological control [4,5,6,7,8,23,30].

    Mathematical models are widely used to study the transmission and control of mosquito-borne diseases. On the one hand, mathematical models are used to study the interaction between people and mosquitoes in the transmission process of different types of mosquito-borne diseases [10,11,21,28], and on the other hand, they are also used to explore the interaction between wild mosquitoes and the released sterile mosquitoes or Wolbachia-infected mosquitoes [1,9,13,14,15,18,19,29]. In recent years, some researchers have focused on the release strategy of sterile mosquitoes and Wolbachia-infected mosquitoes, considering the suppression effect of the release behavior on the wild mosquito population in different situations [12,16,17,20,22,24,25,26,27,31,32,33,34,35]. Some of the mosquito population suppression models currently studied [25,26,31,32] are based on an interaction model of two types of mosquitoes (wild mosquitoes and sterile mosquitoes) constructed by Cai et al. in [9] as follows

    {dw(t)dt=w(t)(aw(t)w(t)+g(t)μ1ξ1(w(t)+g(t))),dgdt=B()(μ2+ξ2(w(t)+g(t))), (1.1)

    where w(t) and g(t) are the number of the wild and sterile mosquitoes in the field at time t, respectively. a is the birth rate per wild mosquito and it follows logistic growth. ξi and μi,i=1,2 are the density dependent and independent death rate, respectively. Further, B() is the release rate of the sterile mosquitoes.

    It is noted that model (1.1) did not distinguish different stages of the mosquito's life cycle, and the growth of wild and sterile mosquitoes are both affected by intraspecific competition. However, we know that the intraspecific competition of mosquitoes mainly exists in the aquatic stage, while adult mosquitoes are little affected. Then, a revised model was developed in [19]

    {dw(t)dt=aw(t)w(t)+g(t)(1ξ1w(t))w(t)μ1w(t),dgdt=B()μ2g(t), (1.2)

    where 1ξ1w(t) describes the density dependent survival probability of wild mosquitoes. Based on model (1.2), a series of mosquito population suppression models were proposed and studied [20,35].

    It is widely known that a mosquito's life cycle consists of several stages which include egg, larva, pupa in an aquatic phase and adult in an aerial phase. In addition, the released mosquitoes are mainly sterile males or Wolbachia-infected males and they compete with wild male mosquitoes and occupying the opportunity to mate with wild female mosquitoes [16]. Therefore, in mathematical modelling, it is better to consider different growth stages of mosquitoes and distinguish male and female groups.

    In this work, we attempt to give and study a kind of mosquito population suppression model with sex and stage structure. The rest of this paper is arranged as follows: in Section 2, based on previous research, we give several mosquito population suppression models with life stage and sex structure. In Section 3, we study the dynamic behavior of the established models, and establish sufficient conditions for successful suppression and partial suppression of wild mosquito population. Then, in Section 4, a series of numerical examples are presented to illustrate the theoretical results we obtained. Finally, a brief conclusion is given in Section 5.

    According to the life cycle of mosquitoes, the following compartmental model was considered in [2].

    {dEdt=βEF(1EK)(νE+δE)E,dMdt=(1ν)νEEδMM,dFdt=ννEEMM+γsMsδFF,dMsdt=u()δsMs, (2.1)

    where E(t),M(t),F(t) and Ms(t) denote the density of mosquito in aquatic phase, the fertile adult males, the fertile adult females and the sterilized adult males at time t, respectively. u() stands for a releasing function. Model parameters are all positive constants, and their biological meaning can be referred to Table 1.

    Table 1.  Biological meaning of model parameter.
    βE the oviposition rate γs the mating competitiveness
    νE the transition rate to the adult phase δE the death rate of eggs
    δF the death rate of adult females δM the death rate of adult males
    δs the death rate of sterile mosquitoes ν the female proportion
    K the environmental capacity for eggs

     | Show Table
    DownLoad: CSV

    To reduce the dimension of the model, the authors of [2] assumed that the time dynamics of the mosquitoes in aquatic phase and the adult males compartments are fast, so the equations of E() and M() are at equilibrium. That is, E=βEFβEFK+νE+δE,M=(1ν)νEδME. Then, they deduced the following simplified version [2]

    {dF(t)dt=ν(1ν)β2Eν2EF2(βEFK+νE+δE)[(1ν)νEβEF+δMγsMs(βEFK+νE+δE)]δFFf(F(t),Ms(t)),dMsdt=u()δsMs. (2.2)

    The authors in [2] have demonstrated the reliability of the above simplified model through numerical analysis by comparing the dynamics of adult female mosquitoes in the original and simplified systems. The numerical solutions of fertile adult female mosquitoes were found to be very close, indicating that the simplified model is reliable.

    In the work [2], in addition to using the optimization method to study the release strategy, the authors studied the dynamical behavior of system (2.2) for the case u()=0, that is, sterile mosquitoes are released only once and no longer replenished. However, we know that the suppression of mosquito population requires a large number of sterile mosquitoes to be released several times, which is a relatively long-term continuous release behavior. In this paper, we will study a mosquito population suppression model incorporating stage and sex structure based on model (2.2). In addition, by employing the modeling ideas in [24], we consider Ms(t)0 as the density of sterile mosquitoes which are sexually active in the field at time t, then a non-autonomous mosquito population suppression model is obtained as follows

    dF(t)dt=f(F(t),Ms(t)). (2.3)

    A special case of model (2.3) is Ms(t)m. Yu et al. proposed the concept of the sexual lifespan of sterile mosquitoes in [25], [26], [31] and [32]. When the release period of sterile mosquitoes is equal to its sexual lifespan, the number of sexually active sterile mosquitoes in the field Ms(t) is a constant value m if its natural death is ignored. Then, the mosquito population suppression model (2.3) reduces to

    dF(t)dt=f(F(t),m)=ν(1ν)β2Eν2EF2(βEFK+νE+δE)[(1ν)νEβEF+δMγsm(βEFK+νE+δE)]δFF, (2.4)

    which is a autonomous equation.

    In addition, based on the work [2], we further investigate the suppression effect of long-term constant release of sterile mosquitoes on the wild mosquito population. Therefore, we let u()u, here u is a positive constant, then system (2.2) becomes another special case of model (2.3) as follows

    {dF(t)dt=f(F(t),Ms(t)),dMsdt=uδsMs(t)=g(Ms(t)). (2.5)

    In this paper, we will study dynamic behaviors of the non-autonomous mosquito population suppression model (2.3) and its two special cases (2.4) and (2.5), and mainly discuss the threshold conditions for successful suppression of mosquito population and the conditions for sustainable survival of mosquitoes.

    As in [16], to simplify the expression of the model, we let

    A=ν(1ν)β2Eν2E,B=βEK,C=νE+δE,D=(1ν)νEβE,H=δMγs,

    then model (2.3) becomes

    dF(t)dt=AF2(t)(BF(t)+C)[DF(t)+HMs(t)(BF(t)+C)]δFF(t). (3.1)

    If no biological control measure is introduced and there are only wild mosquitoes in the field, that is Ms(t)=0, then system (2.3) can be written as

    dF(t)dt=AFD(BF+C)δFF=BDδFFD(BF+C)[CB(ACDδF1)F]. (3.2)

    Obviously, when ACDδF, system (3.2) has no positive equilibrium and the extinction equilibrium F=0 is globally asymptotically stable.

    In this work we assume that

    A>CDδF,

    then system (3.2) has a unique positive equilibrium F=CB(ACDδF1) which is globally asymptotically stable.

    For the case Ms(t)m>0, (2.4) is a autonomous equation, and we can also discuss its equilibria. To get positive equilibria of (2.4), we need to consider the following algebraic equation

    AF(BF+C)[DF+mH(BF+C)]δF=0,

    which is equivalent to

    δFB(D+mHB)(F+CB)2+(A+δFCD)(F+CB)ACBB(F+CB)[D(F+CB)+mHB(F+CB)CDB]=0.

    Thus, the existence and the number of positive equilibria of model (2.4) are equivalent to that of the following quadratic equation

    G(F,m)A1(m)(F+CB)2A2(F+CB)+A3=0, (3.3)

    where A1(m)=δFB(D+mHB),A2=A+δFCD,A3=ACB.

    Using the discriminant (m)=A224A1(m)A3, we introduce a threshold of the release amount

    m=(ACDδF)24ACDδFDHB,

    which satisfies (m)=0.

    According to [16], we have the following conclusions.

    Lemma 1. When A>CDδF, the nonnegative equilibria of model (2.4) can be classified as follows:

    (i) If 0<m<m, then model (2.4) has three equilibria: F=0,

    F1(m)=A22A1(m)[11A1(m)A1(m)]CB,
    F2(m)=A22A1(m)[1+1A1(m)A1(m)]CB.

    (ii) If m=m, then model (2.4) has two equilibria: F=0 and

    F(m)=A22A1(m)CB=ACDδFA+CDδFK(νE+δE)βE.

    (iii) If m>m, then the extinction equilibrium F=0 is the unique equilibrium of model (2.4).

    Remark 1. Since A1(m) is strictly monotonically increasing with respect to m, while F2(m) is strictly monotonically decreasing with respect to m in (0,m), we have

    F2(m)<F2(0)=F,and0<F1(m)<F2(m)<F,m(0,m).

    We first study the dynamics of (2.3) for the case when the number of sexual active sterile mosquitoes in the field, Ms(t), is a given nonnegative continuous function.

    Let F(t,Ms,t0,F0) be a solution of (2.3) through (t0,F0), then F(t,Ms,t0,F0)C1([t0,),[0,)) satisfies (2.3) on [t0,) and F(t0)=F0.

    According to equation (2.3), it is clear that dF(t)dt|F=0=0, and

    dF(t)dtAFD(BF+C)δFF=BDδFFD(BF+C)(FF),

    then we get the following conclusion.

    Lemma 2. For any t00 and F0>0, the solution F(t,Ms,t0,F0) of (2.3) is positive and lim suptF(t)F.

    In the following, we investigate the stability of the extinction equilibrium F=0. For convenience, let

    m0=inft(0,)Ms(t).

    Theorem 1. If A>CDδF, m0>0 and Ms(t)m,t0, then the extinction equilibrium F=0 of model (2.3) is locally asymptotically stable, and limtF(t,Ms,t0,F0)=0 if 0<F0<F1(m0). In addition, F(t)=F(t,Ms,t0,F0)F(m),tt0 if F0F(m).

    Proof. We first prove that the extinction equilibrium F=0 of model (2.3) is locally stable. Since m0>0, according to Lemma 1 and the properties of G(F,m) defined in (3.3), we know that G(F,m0)>0 for all F(0,F1(m0)). Then, for any δ(0,F1(m0)), the following inequality

    AF(BF+C)[DF+m0H(BF+C)]δF<0,0<Fδ (3.4)

    holds.

    We claim that

    0<F(t0)<δF(t)=F(t,Ms,t0,F0)<δ,tt0. (3.5)

    If it does not hold, then there must exist t1>t0 such that F(t1)=δ and 0<F(t)<δ,t0t<t1, thus F(t1)0. However, according to (2.3) and (3.4), we have

    0F(t1)F(t1)[AF(t1)(BF(t1)+C)[DF(t1)+m0H(BF(t1)+C)]δF]=δ[Aδ(Bδ+C)[Dδ+m0H(Bδ+C)]δF]<0.

    A contradiction occurs. Therefore, F=0 is locally stable.

    We further prove the attractivity of F=0. From (3.5), we have that for any 0<F(t0)<δ, ¯F=lim suptF(t,Ms,t0,F0)δ. By the fluctuation lemma, there exists a time sequence {ωk} such that limkF(ωk)=¯F,limkF(ωk)=0. Taking the limit in (2.3) along {ωk} gives

    A¯F2(B¯F+C)[D¯F+H(lim suptMs(ωk))(B¯F+C)]=δF¯F, (3.6)

    and consequently

    0¯F[A¯F(B¯F+C)[D¯F+m0H(B¯F+C)]δF],

    which leads to ¯F=0 or A¯F(B¯F+C)[D¯F+m0H(B¯F+C)]δF. According to (3.4), the second case cannot occur, then ¯F=0 and limtF(t,Ms,t0,F0)=0. Thus, F=0 is locally asymptotically stable. Noting the arbitrariness of δ, which can be sufficiently close to F1(m0), we can get limtF(t,Ms,t0,F0)=0 if 0<F0<F1(m0).

    Next, we prove that: F0F(m)F(t)=F(t,Ms,t0,F0)F(m),tt0. If it does not hold, then there must exist ˜t>t0 and ˉt(t0,˜t) such that F(˜t,Ms,t0,F0)<F(m) and F(ˉt,Ms,t0,F0)=F(m). We consider the following comparison equation of (2.3)

    {dF(t)dt=AF2(t)(BF(t)+C)[DF(t)+mH(BF(t)+C)]δFF(t)=f(F(t),m)δFF,F(ˉt)=F(m). (3.7)

    Since 0<m0Ms(t)m,t0, we have f(F(t),Ms(t))δFFf(F(t),m)δFF. We also know that F(m) is an equilibrium of (3.7), so the solution of (3.7) passing through (ˉt,F(m)) is ˜F(t,m,ˉt,F(m))=F(m). According to the comparison principle, we obtain that the solution of (2.3) passing through (ˉt,F(m)) satisfies

    F(t,Ms,t0,F0)=F(t,Ms,ˉt,F(m))˜F(t,m,ˉt,F(m))=F(m),tˉt,

    which contradicts to F(˜t,Ms,t0,F0)<F(m) and ˉt(t0,˜t). The proof is completed.

    Remark 2. Theorem 1 shows that when the initial population size of wild mosquitoes is not large (F0<F1(m0)), even if the release intensity of sterile mosquitoes is not strong (Ms(t)m), the mosquito population can be successfully suppressed. However, if the initial size of mosquito population is large (F0F(m)), the same release intensity will lead to the failure of mosquito population suppression.

    In the following, we try to determine a release intensity to ensure successful suppression of mosquito population for large initial population size. Let

    ¯Ms=lim suptMs(t),Ms_=lim inftMs(t).

    Theorem 2. If A>CDδF, m0>0 and Ms_>m, then the extinction equilibrium F=0 of model (2.3) is globally asymptotically stable.

    Proof. Since m0>0, the extinction equilibrium F=0 of model (2.3) is locally asymptotically stable from the proof of Theorem 1. We only need to show that F=0 is globally attractive. According to Lemma 1, for any t00 and F0>0, ¯F=lim suptF(t,Ms,t0,F0) exists.

    We now show that ¯F=0. If it does not hold, then ¯F>0. By the fluctuation lemma, there exists a time sequence {τk} such that limkF(τk)=¯F,limkF(τk)=0. In addition, since m0>0 and MS_>m, there must be a subsequence {ˉτk}{τk} such that limkMs(ˉτk)>m. Taking the limit in (2.3) along {ˉτk} gives

    ¯F[A¯F(B¯F+C)[D¯F+mH(B¯F+C)]δF]>0,

    which leads to G(¯F,m)<0. However, from Lemma 1 and the properties of G(F,m) defined in (3.3), we know that G(F,m)0 for F(0,), then a contradiction occurs. Therefore, ¯F=0, and for any t00 and F0>0, limtF(t,Ms,t0,F0)=0 holds. The proof is completed.

    Remark 3. Theorem 2 shows that when we increase the release intensity of sterile mosquitoes so that m0>0 and Ms_>m, the mosquito population can always be successfully suppressed. This result can be conveniently applied in practice.

    In this subsection, we investigate two special cases of the autonomous models (2.4) and (2.5), and mainly study the dynamic behavior.

    For model (2.4), the release period of the sterile mosquito is equal to its sexual lifespan, and the number of sexually active sterile mosquitoes in the field remains unchanged, that is, Ms(t)m. We have the following results for (2.4).

    Theorem 3. If A>CDδF and 0<m<m, then the dynamics of model (2.4) has the following conclusions:

    (i) F=0 is uniformly stable and limtF(t,m,t0,F0)=0 if 0<F0<F1(m).

    (ii) F=F1(m) is unstable.

    (iii) F=F2(m) is uniformly stable, and limtF(t,m,t0,F0)=F2(m) if F0>F1(m).

    Proof. From Lemma 1, when 0<m<m, model (2.4) has three equilibria: F=0, F=F1(m) and F=F2(m).

    (ⅰ) This conclusion can be verified by slightly modifying the proof of the same result in Theorem 1. In fact, if we replace m0 and MS by m, then we can deduce the conclusion (ⅰ).

    (ⅱ) The instability of F1(m) follows directly from the attractivity of F0=0.

    (ⅲ) We first show that for any t00 and δ(0,F2(m)F1(m)), if |F2(m)F0|<δ, then

    |F(t,m,t0,F0)F2(m)|<δ,tt0. (3.8)

    If it does not hold, then there must exist t1>t0 such that |F(t1,m,t0,F0)F2(m)|=δ and |F(t,m,t0,F0)F2(m)|<δ for t0t<t1. There are two possible cases: F(t1)=F2(m)+δ and F(t1)=F2(m)δ. If F(t1)=F2(m)+δ, then we have F(t1)0 and F(t)<F2(m)+δ for t0t<t1. Substitute t1 into (2.4), and we can obtain

    AF2(t1)(BF(t1)+C)[DF(t1)+mH(BF(t1)+C)]δF(F2(m)+δ), (3.9)

    and consequently

    A(F2(m)+δ)2(B(F2(m)+δ)+C)[D(F2(m)+δ)+mH(B(F2(m)+δ)+C)]δF(F2(m)+δ),

    which can be further reduced to G(F2(m)+δ,m)0. However, G(F,m)>0 for all F>F2(m), which is a contradiction. For the second case, if F(t1)=F2(m)δ and F(t)>F2(m)δ for t0t<t1, through similar discussions, we can get G(F2(m)δ,m)0. However, G(F,m)<0 for all F1(m)<F<F2(m), which is also a contradiction. Thus, F2(m) is uniformly stable.

    Now we further prove its attractivity. To this end, we only need to show

    limtF(t,m,t0,F0)=F2(m)for anyt00,F0(F1(m),). (3.10)

    In fact, by (3.8), for any δ(0,F2(m)F1(m)) and F0>F1(m)+δ, we can get F(t,m,t0,F0)>F1(m)+δ,tt0. Then, F_=lim inftF(t,0,t0,F0)F1(m)+δ, and there must be a monotonic time sequence {tk} satisfying limktk=, such that limkF(tk)=F_ and limkF(tk)=0. Taking limit on both sides of (2.4) along {tk}, and we can get

    AF_2(BF_+C)[DF_+mH(BF_+C)]=δFF_,

    which can be further reduced to G(F_,m)=0. Since F_F1(m)+δ>F1(m), and G(F,m)=0 if and only if F=F2(m) for F>F1(m), we can deduce F_=F2(m). In addition, according to Lemma 2, we know ¯F=lim suptF(t,m,t0,F0) exists and is finite. By similar discussion, we can derive ¯F=F2(m). Then, we have F_=¯F=F2(m) which verifies (3.10). The proof is completed.

    As m increases to m, these two positive equilibria F=F1(m) and F=F2(m) coincide as the unique positive equilibrium F(m). When m>m, the extinction equilibrium F=0 is the only equilibrium of (2.4). By discussions similar to the above, we can obtain the following conclusions.

    Theorem 4. (1) If A>CDδF and m=m, then the extinction equilibrium F=0 of (2.4) is uniformly stable, and limtF(t,m,t0,F0)=0 if 0<F0<F(m),t00. In addition, the unique positive equilibrium F=F(m) is globally asymptotically stable from the right-side, and limtF(t,m,t0,F0)=F(m) if F0>F(m),t00.

    (2) If m>m, then the unique equilibrium F=0 of (2.4) is globally asymptotically stable.

    Remark 4. According to Theorem 3 and 4, as the value of m changes, the attraction domain of the stable extinction equilibrium F=0 is different. This indicates that under a specific release intensity, the initial size of mosquito population determines whether the mosquito population can be finally suppressed. The larger the m, the larger the corresponding attraction area of F=0, and the more likely the mosquito population will be suppressed.

    Next, we investigate the effect of long-term constant releases of sterile mosquitoes on the suppression of wild mosquito population based on system (2.5).

    Clearly, for any t00, if F(t0)>0,Ms(t0)>0, then F(t)>0,Ms(t)>0 for all tt0. In addition, F(t)max{F(t0),F},MS(t)max{Ms(t0),uδs} hold for tt0.

    By choosing the Dulac function B(F,Ms)=1F2, we can deduce

    (fB)F+(gB)Ms=A[2BDF+CD+2BHMs(BF+C)][DF(BF+C)+HMs(BF+C)2]2+δFδsF, (3.11)

    where f and g are the functions at the right end of equations in system (2.5).

    Noting that if δF<δs (this assumption is consistent with the actual situation), then (fB)F+(gB)Ms<0 for (F,Ms)R+2, thus system (2.5) has no closed orbits in R+2.

    It is easy to know that system (2.5) has a boundary equilibrium E0(0,Ms), where Ms=uδs. To study positive equilibria of (2.4), we consider the algebraic equation

    AF(BF+C)[DF+uδsH(BF+C)]δF=0,

    which is equivalent to

    δFB(D+uδsHB)(F+CB)2+(A+δFCD)(F+CB)ACBB(F+CB)[D(F+CB)+uδsHB(F+CB)CDB]=0.

    Therefore, the existence and number of positive equilibria of model (2.5) are equivalent to the existence and number of positive roots of the following quadratic equation

    G1(F,u)~A1(u)(F+CB)2A2(F+CB)+A3=0, (3.12)

    where ~A1(u)=δFB(D+uδsHB), and A2,A3 are the same as in Lemma 1.

    By introducing a threshold for the release rate

    u=(ACDδF)24ACDδFDδsHB=δsm,

    we can get the following conclusion by similar discussion in Lemma 1.

    Lemma 3. When A>CDδF, the nonnegative equilibria of system (2.5) can be classified as follows:

    (i) If 0<u<u, then system (2.5) has three equilibria: E0(0,Ms), E1(F1(u),Ms) and E2(F2(u),Ms), where

    F1(u)=A22~A1(u)[11~A1(u)~A1(u)]CB,F2(u)=A22~A1(u)[1+1~A1(u)~A1(u)]CB.

    (ii) If u=u, then system (2.5) has two equilibria: E0(0,Ms) and E(F(u),Ms), where

    F(u)=A22~A1(u)CB=ACDδFA+CDδFK(νE+δE)βE.

    (iii) If u>u, then the boundary equilibrium E0(0,Ms) is the unique equilibrium of system (2.5).

    We now investigate the stability of these equilibria. The Jacobian matrix of system (2.5) has the form

    JE=(J11J120δs),

    where

    J11=ACF[DF(BF+C)+HMs(BF+C)2]2(DF+2HMs(BF+C))δF,
    J12=AHF2(BF+C)2[DF(BF+C)+HMs(BF+C)2]2.

    Obviously, J11(E0)=δF, thus the boundary equilibrium E0(0,Ms) of system (2.5) is a node which is locally asymptotically stable. In particular, If u>u, then the unique equilibrium E0(0,Ms) is globally asymptotically stable.

    When 0<u<u, we have AFi[DFi(BFi+C)+HMs(BFi+C)2]=δF,i=1,2, where Fi=Fi(u). It is clear that J12(Ei)<0. Let

    Θi=DFi(BFi+C)+HMs(BFi+C)2,

    then

    J11(Ei)=ACFiΘ2i[DFi+2HMs(BFi+C)]AFiΘi=AFiΘ2i{C[DFi+2HMs(BFi+C)](BFi+C)[DFi+HMs(BFi+C)]}=AFiΘ2i{[CD(Fi+CB)+2CHBMs(Fi+CB)C2DB]B(Fi+CB)[D(Fi+CB)+BHMs(Fi+CB)CDB]}=AFiδFΘ2i{~A1(u)(Fi+CB)2+2CB~A1(u)(Fi+CB)C2DδFB}=AFiδFΘ2i{A3A2(Fi+CB)+2CB~A1(u)(Fi+CB)C2DδFB}=AFiδFΘ2i{A3C2DδFB(A22CB~A1(u))(Fi+CB)}. (3.13)

    Since Fi+CB=A22~A1(u)[11~A1(u)~A1(u)], A2=A+δFCD<2A,A3=ACB, ~A1(u)=A224A3=A22B4AC and ~A1(u)<~A1(u), we can further get

    J11(Ei)=AFiδFΘ2i{2ACBA222~A1(u)±(A222~A1(u)A2CB)1~A1(u)~A1(u)}=AFiδFΘ2i×A22B2A2C~A1(u)2~A1(u)B{4AC~A1(u)A22BA22B2A2C~A1(u)±1~A1(u)~A1(u)}=AFi[A22B2A2C~A1(u)]2~A1(u)BδFΘ2i{4AC~A1(u)A22B2A2C~A1(u)(1~A1(u)~A1(u))±1~A1(u)~A1(u)}. (3.14)

    Noting that

    A22B2A2C~A1(u)>B{A222A2CB~A1(u)}>B{A224A3~A1(u)}=0,

    then we can easily get J11(E2)<0, and the positive equilibrium E2(F2(u),Ms) is a locally asymptotically stable node.

    Let

    TE1(u)=4AC~A1(u)A22B2A2C~A1(u)(1~A1(u)~A1(u))+1~A1(u)~A1(u),

    thus

    J11(E1)=AF1[A22B2A2C~A1(u)]2~A1(u)BδFΘ21TE1(u). (3.15)

    In addition, by ~A1(u)=A224A3=A22B4AC, we get

    TE1(u)=1~A1(u)~A1(u)4AC4AC2A2C~A1(u)~A1(u)(1~A1(u)~A1(u))=1~A1(u)~A1(u)2A2AA2~A1(u)~A1(u)(1~A1(u)~A1(u)). (3.16)

    Because ~A1(u)=δFB(D+uδsHB)<~A1(u) for 0<u<u, we have ~A1(u)~A1(u)(0,1). Let y=1~A1(u)~A1(u) and N=AδFCD, then N>1,

    0<y<1δFBDA22B4AC=14N(N+1)2=N1N+1<N1N+1

    and

    TE1(u)=y2A2A+A2(y21)y2=y[12NyN1+(N+1)y2].

    Consider w(y)=2NyN1+(N+1)y2, and by direct calculation, we can get that w(y)=2N(N1)2N(N+1)y2[N1+(N+1)y2]2=0 has only one positive root y=N1N+1, and w(y) is strictly monotonically increasing on (0,N1N+1] with w(N1N+1)=1. Then, w(y)<1 for any 0<y<N1N+1 and TE1(u)>0. The positive equilibrium E1(F1(u),Ms) is an unstable saddle.

    When u increases to u=u, these two positive equilibria E1(F1(u),Ms) and E2(F2(u),Ms) coincide as the unique positive equilibrium E(F(u),Ms), which is a saddle node.

    In summary, we have the following results.

    Theorem 5. If A>CDδF and δF<δs, then we have the following conclusions for model (2.5).

    (i) If 0<u<u, then the boundary equilibrium E0(0,Ms) is a locally asymptotically stable node, E1(F1(u),Ms) is an unstable saddle, while E2(F2(u),Ms) is a locally asymptotically stable node.

    (ii) If u=u, then the unique positive equilibrium E(F(u),Ms) is an unstable saddle node.

    (iii) If u>u, then the unique equilibrium E0(0,Ms) is globally asymptotically stable.

    In this section, some numerical examples are presented to confirm the theoretical results we obtained in last section. Most parameter values are taken from [2], [3] and [23] (see Table 2).

    Table 2.  Model parameter values from [2], [3] and [23].
    Parameters Value Unit Parameters Value Unit
    βE 10 day1 γs 1 -
    νE 0.05 - δE 0.03 day1
    δF 0.04 day1 δM 0.1 day1
    δs 0.12 day1 ν 0.49 -

     | Show Table
    DownLoad: CSV

    We apply the calculation method in [2] and [3] for the environmental capacity of eggs K, where an island of 74 ha (hectares) was considered. The male population was estimated to be 69 ha1, then they assumed that the number of adult wild males was M=F=69×74=5106 when the population reaches a stable state. By direct calculation, they got the eggs number at the stable state was E=δMM(1ν)νE2.0023×104. Because F=ννEEδF=(νE+δE)EβE(1E/K) holds, the approximate value of environmental capacity can be deduced as

    K=E1(νE+δE)δFβEννE2.0289×104.

    It is easy to verify that A=0.0625>δFCD0.0008 and m=(ACDδF)24ACDδFDHB9.5123×104. Without releasing sterile mosquitoes, the number of wild adult female ones will stabilize at F=CB(ACDδF1)1.2094×104.

    For the general non-autonomous case, we first take Ms(t)=4×104×(1.3+sin(πt/8)) for model (2.3). By direct calculation, we get m0=inft(0,)Ms(t)=1.2×104 and Ms(t)9.2×104<m,t0. From Figure 1(a), we can see that the extinction equilibrium F=0 is locally asymptotically stable (which is consistent with the conclusion of Theorem 1), and there is also a positive periodic solution that is locally stable. That is, the model presents bistability phenomenon. When increasing the release intensity of sterile mosquitoes to Ms(t)=3.2×105×(1.3+sin(πt/8)), we have MS_=lim inftMS(t)=9.6×104>m,t0. According to Theorem 2, the extinction equilibrium F=0 is globally asymptotically stable (see Figure 1(b)).

    Figure 1.  Profiles produced by model (2.3) with Ms(t)m or MS_>m. (a) Bistability phenomenon with Ms(t)=4×104×(1.3+sin(πt/8)); (b) Global stability of the extinction equilibrium F=0.

    Comparing the constraints for the function Ms(t) in Theorem 1 (Ms(t)m) and Theorem 2 (MS_>m), we note that there are other cases of Ms(t). To this end, we consider special forms for Ms(t) such that its function values distributed on both sides of the threshold m. We choose two forms: Ms(t)=1.2×105×(1.3+sin(πt/8)) and Ms(t)=6×104×(1.3+sin(πt/8)), which both satisfy MS_<m<¯MS. From Figure 2, we can see that for the former one, F=0 is globally asymptotically stable (see Figure 2(a)), while bistability phenomenon is present for the latter form (see Figure 2(b)). This also shows that the conditions in Theorem 1 and 2 are sufficient but not necessary.

    Figure 2.  Profiles produced by model (2.3) with MS_<m<¯MS. (a) Global stability of the extinction equilibrium F=0; (b) Bistability phenomenon.

    For the autonomous cases, we first verify the results we get for model (2.4). Select 0<m=6×104<m9.5123×104, from Figure 3(a), we find that the extinction equilibrium F=0 is locally asymptotically stable. In addition, there are also two positive equilibria and the larger one is locally stable. As the release intensity increases to m=10×104>m, the extinction equilibrium F=0 becomes globally asymptotically stable (see Figure 3(b)). These are consistent with the conclusions in Theorem 3 and 4.

    Figure 3.  Profiles produced by model (2.4). (a) Bistability phenomenon with m=6×104; (b) Global stability of the extinction equilibrium F=0.

    Then for system (2.5), we get the threshold for the release rate u=mδs1.1415×104. By taking u=1×104<u, Figure 4(a) also shows bistability phenomenon, and both the wild mosquito-extinction equilibrium E0(0,Ms) and the larger positive equilibrium E2(F2,Ms) are locally stable. The wild mosquito-extinction equilibrium E0(0,Ms) becomes globally asymptotically stable when the release rate increases to u=1.2×104>u (see Figure 4(b)).

    Figure 4.  Profiles produced by system (2.5). (a) Bistability phenomenon with u=10×104; (b) Global stability of the wild mosquito-extinction equilibrium E0(0,Ms).

    In addition to the two modes described by models (2.4) and (2.5), there is another common release mode of sterile mosquitoes in practice, namely, impulsive release, which means that the release function u() in (2.2) is a sum of Dirac deltas. Now let's study this situation numerically. We assume that sterile mosquitoes are released with amount of σ at discrete times t=kT,k=0,1,2,, where T is the release period, then the number of sterile mosquitoes in the environment satisfies

    {dMs(t)dt=δsMs(t),fortkT,Ms(kT+)=Ms(kT)+σ,k=0,1,2,. (4.1)

    Obviously, system (4.1) has a unique positive periodic solution ~Ms(t) which is globally asymptotically stable, where

    ~Ms(t)=σexp(δs(tkT))1exp(δsT),t(kT,(k+1)T],k=0,1,2,,~Ms(kT+)=σ1exp(δsT).

    In addition, the expression for the solution of system (4.1) is Ms(t)=(Ms(0+)~Ms(0+))exp(δst)+~Ms(t), then we can deduce limtMs(t)=~Ms(t). Since what we want to study is the asymptotic behavior of the population development system of wild mosquitoes, we let Ms(t)=~Ms(t).

    Fix a release period T=5. When we select σ=8×104, by simple calculating, we can get ~Ms(t)>m,t0. Then we can see that the extinction equilibrium is globally asymptotically stable (see Figure 5(a)). When choose σ=4.2×104, then we can get lim inft~Ms(t)<m, and we find that in addition to the extinction equilibrium, there is also a locally stable positive periodic solution(see Figure 5(b)).

    Figure 5.  Profiles produced by systems (2.3) and (4.1). (a) Global stability of the extinction equilibrium with σ=8×104; (b) Bistability phenomenon with σ=4.2×104.

    In this paper, we studied a kind of mosquito population suppression model, taking into account the stage structure of mosquito growth as well as the sex structure. First, we constructed a non-autonomous mosquito population suppression model (2.3) by taking the number of sexually active sterile mosquitoes released into the field as a given function Ms(t). Then we considered a special case, that is, the release period of the sterile mosquito is equal to its sexual lifespan. The non-autonomous model (2.3) is transformed into a one-dimensional ordinary differential equation model (2.4). Finally, continuing the work in [2], we considered another special case and investigated a two-dimensional mosquito population suppression system (2.5).

    For the non-autonomous mosquito population suppression model (2.3), we determined a critical value m for Ms(t). If m0>0 and Ms(t)m,t0, then the extinction equilibrium F=0 is locally but not globally asymptotically stable (Theorem 1). While when m0>0 and MS_>m, then the extinction equilibrium F=0 becomes globally asymptotically stable (Theorem 2). For the one-dimensional ordinary differential equation model (2.4), if 0<m<m, both the extinction equilibrium F=0 and the larger positive equilibrium F2(m) are locally stable (Theorem 3). If m=m, the unique positive equilibrium F=F(m) is semi-stable. While if m>m, then the unique equilibrium F=0 becomes globally asymptotically stable (Theorem 4). For the two-dimensional mosquito population suppression system (2.5), we also found a critical value u of the release rate u. Through dynamic analysis, we finally obtained conclusions similar to model (2.4) (Theorem 5). Finally, we provided a series of numerical examples to verify our theoretical results.

    In this work, we just gave sufficient conditions for successful suppression and persistence of mosquito population for the non-autonomous model (2.3). In the numerical example, we also found that mosquito populations can be eliminated with a relatively weak release function Ms(t). In the future research, we will try new research methods to obtain sufficient and necessary conditions for mosquito population suppression.

    In addition, to reduce the difficulty of mathematical analysis, when constructing model (5), we omitted the equations for the aquatic stage population and the adult male mosquito population under specific assumptions, and only used the model parameters to characterize the stage and sex structure of the wild mosquito population. This may make the characterization of these structures less precise. Therefore, in future research, we will use high-dimensional models to accurately depict the stage and gender structures of mosquitoes and study various release strategies.

    This work is supported by the National Natural Science Foundation of China (11901502), Program for Innovative Research Team (in Science and Technology) in University of Henan Province (21IRTSTHN014), Natural Science Foundation of Henan Province(222300420016) and Nanhu Scholars Program for Young Scholars of XYNU.

    The authors declare no conflicts of interest in this paper.



    [1] S. Mohamad, Exponential stability in Hopfield-type neural networks with impulses, Chaos Soliton. Fract., 32 (2007), 456-467. doi: 10.1016/j.chaos.2006.06.035
    [2] J. Lu, D. W. C. Ho, J. Cao, A unified synchronization criterion for impulsive dynamical networks, Automatica, 46 (2010), 1215-1221. doi: 10.1016/j.automatica.2010.04.005
    [3] P. B. Watta, K. Wang, M. H. Hassoun, Recurrent neural nets as dynamical boolean systems with application to associative memory, IEEE T. Neural Networ., 8 (1997), 1268-1280. doi: 10.1109/72.641450
    [4] H. Yin, B. Du, X. Cheng, Stochastic patch structure Nicholson's blowfies system with mixed delays, Adv. Differ. Equ., 386 (2020), 1-11.
    [5] S. Mou, H. Gao, J. Lam, et al. A new criterion of delay dependent asymptotic stability for Hopfield neural networks with time delay, IEEE T. Neural Networ., 19 (2008), 532-535. doi: 10.1109/TNN.2007.912593
    [6] C. Liu, W. Liu, Z. Yang, et al. Stability of neural networks with delay and variable-time impulses, Neurocomputing, 171 (2016), 1644-1654. doi: 10.1016/j.neucom.2015.07.007
    [7] S. Hu, J. Wang, Global stability of a class of discrete-time recurrent neural networks, IEEE Trans. Circuits Syst. I, 49 (2017), 1104-1117.
    [8] B. Du, Anti-periodic solutions problem for inertial competitive neutral-type neural networks via Wirtinger inequality, J. Inequal. Appl., 2019 (2019), 1-15. doi: 10.1186/s13660-019-1955-4
    [9] Q. Zhu, J. Cao, Exponential stability of stochastic neural networks with both Markovian jump parameters and mixed time delays, IEEE T. Syst. Man Cy. B, 41 (2011), 341-353.
    [10] J. H. Park, O. Kwon, Design of state estimator for neural networks of neutral-type, Appl. Math. Comput., 202 (2008), 360-369.
    [11] T. Zhou, B. Du, H. Du, Positive periodic solution for indefinite singular Lienard equation with p-Laplacian, Adv. Differ. Equ., 2019 (2019), 1-17. doi: 10.1186/s13662-018-1939-6
    [12] Z. Gui, W. Ge, X. Yang, Periodic oscillation for a Hopfield neural networks with neutral delays, Phys. Lett. A, 364 (2007), 267-273. doi: 10.1016/j.physleta.2006.12.013
    [13] K. Wang, Z. Teng, H. Jiang, Adaptive synchronization in an array of linearly coupled neural networks with reaction-diffusion terms and time delays, Commun. Nonlinear Sci., 17 (2012), 3866-3875. doi: 10.1016/j.cnsns.2012.02.020
    [14] Y. Wu, L. Liu, J. Hu, et al. Adaptive Antisynchronization of Multilayer Reaction-Diffusion Neural Networks, IEEE T. Neur. Net. Lear., 29 (2018), 807-818. doi: 10.1109/TNNLS.2017.2647811
    [15] J. Wang, H. Wu, Synchronization and adaptive control of an array of linearly coupled reactiondiffusion neural networks with hybrid coupling, IEEE T. Cybernetics, 44 (2013), 1350-1361.
    [16] L. Duan, L. Huang, Z. Guo, et al. Periodic attractor for reaction-diffusion high-order Hopfield neural networks with time-varying delays, Comput. Math. Appl., 73 (2017), 233-245. doi: 10.1016/j.camwa.2016.11.010
    [17] B. Lisena, Average criteria for periodic neural networks with delay, Discrete Cont. Dyn. B, 19 (2014), 761-773.
    [18] Z. Wang, L. Liu, Q. Shan, et al. Stability criteria for recurrent neural networks with time-varying delay based on secondary delay partitioning method, IEEE T. Neur. Net. Lear., 26 (2015), 2589-2595. doi: 10.1109/TNNLS.2014.2387434
    [19] M. Wang, Semigroup of Operators and Evolutionary Equations, Bei Jing: Science Press, 2006.
    [20] H. Yin, B. Du, Q. Yang, et al. Existence of Homoclinic orbits for a singular differential equation involving p-Laplacian, J. Funct. Space., 2020 (2020), 1-7.
    [21] J. Cao, Global stability conditions for delayed CNNs, IEEE Trans. Circuits Syst. I, 48 (2001), 1330-1333. doi: 10.1109/81.964422
    [22] X. Wang, M. Jiang, S. Fang, Stability analysis in Lagrange sense for a non-autonomous CohenGrossberg neural network with mixed delays, Nonlinear Anal-Theor., 70 (2009), 4294-4306. doi: 10.1016/j.na.2008.09.019
    [23] A. V. Rezounenko, J. Wu, A non-local PDE model for population dynamics with state-selective delay: Local theory and global attrators, J. Comput. Appl. Math., 190 (2006), 99-113. doi: 10.1016/j.cam.2005.01.047
    [24] J. Li, J. H. Huang, Uniform attractors for non-autonomous parabolic equations with delays, Nonlinear Anal-Theor., 71 (2009), 2194-2209. doi: 10.1016/j.na.2009.01.053
    [25] L. Evans, Partial Differential Equations (Graduate Studies in Mathematics, Vol. 19), American Mathematical Society, Providence, Rhode, 1998.
  • This article has been cited by:

    1. Yanyuan Xing, Qian Li, Yunfeng Liu, Stability and backward bifurcation in a malaria-transmission model with sterile mosquitoes, 2025, 2025, 2731-4235, 10.1186/s13662-025-03880-5
  • Reader Comments
  • © 2020 the Author(s), licensee AIMS Press. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0)
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Metrics

Article views(3860) PDF downloads(133) Cited by(5)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog