Processing math: 57%
Research article

A network model of social contacts with small-world and scale-free features, tunable connectivity, and geographic restrictions

  • Small-world networks and scale-free networks are well-known theoretical models within the realm of complex graphs. These models exhibit "low" average shortest-path length; however, key distinctions are observed in their degree distributions and average clustering coefficients: in small-world networks, the degree distribution is bell-shaped and the clustering is "high"; in scale-free networks, the degree distribution follows a power law and the clustering is "low". Here, a model for generating scale-free graphs with "high" clustering is numerically explored, since these features are concurrently identified in networks representing social interactions. In this model, the values of average degree and exponent of the power-law degree distribution are both adjustable, and spatial limitations in the creation of links are taken into account. Several topological metrics are calculated and compared for computer-generated graphs. Unexpectedly, the numerical experiments show that, by varying the model parameters, a transition from a power-law to a bell-shaped degree distribution can occur. Also, in these graphs, the degree distribution is most accurately characterized by a pure power-law for values of the exponent typically found in real-world networks.

    Citation: A. Newton Licciardi Jr., L.H.A. Monteiro. A network model of social contacts with small-world and scale-free features, tunable connectivity, and geographic restrictions[J]. Mathematical Biosciences and Engineering, 2024, 21(4): 4801-4813. doi: 10.3934/mbe.2024211

    Related Papers:

    [1] Morten Risstad, Mathias Holand . On the relevance of realized quarticity for exchange rate volatility forecasts. Data Science in Finance and Economics, 2024, 4(4): 514-530. doi: 10.3934/DSFE.2024021
    [2] Manveer Kaur Mangat, Erhard Reschenhofer, Thomas Stark, Christian Zwatz . High-Frequency Trading with Machine Learning Algorithms and Limit Order Book Data. Data Science in Finance and Economics, 2022, 2(4): 437-463. doi: 10.3934/DSFE.2022022
    [3] Kasra Pourkermani . VaR calculation by binary response models. Data Science in Finance and Economics, 2024, 4(3): 350-361. doi: 10.3934/DSFE.2024015
    [4] Obinna D. Adubisi, Ahmed Abdulkadir, Chidi. E. Adubisi . A new hybrid form of the skew-t distribution: estimation methods comparison via Monte Carlo simulation and GARCH model application. Data Science in Finance and Economics, 2022, 2(2): 54-79. doi: 10.3934/DSFE.2022003
    [5] Jin Zeng, Yijia Zhang, Yun Yin, Peter G Moffatt . The effect of the Covid pandemic on stock market volatility: Separating initial impact from time-to-recovery. Data Science in Finance and Economics, 2024, 4(4): 531-547. doi: 10.3934/DSFE.2024022
    [6] Maria Francesca Carfora, Albina Orlando . Application of the Gordon Loeb model to security investment metrics: a proposal. Data Science in Finance and Economics, 2024, 4(4): 601-614. doi: 10.3934/DSFE.2024025
    [7] Yue Yuin Lim, Sie Long Kek, Kok Lay Teo . Efficient state estimation strategies for stochastic optimal control of financial risk problems. Data Science in Finance and Economics, 2022, 2(4): 356-370. doi: 10.3934/DSFE.2022018
    [8] Dominic Joseph . Estimating credit default probabilities using stochastic optimisation. Data Science in Finance and Economics, 2021, 1(3): 253-271. doi: 10.3934/DSFE.2021014
    [9] Chong Li, Guoqiong Long, Shuai Li . Research on measurement and disequilibrium of manufacturing digital transformation: Based on the text mining data of A-share listed companies. Data Science in Finance and Economics, 2023, 3(1): 30-54. doi: 10.3934/DSFE.2023003
    [10] Nitesha Dwarika . Asset pricing models in South Africa: A comparative of regression analysis and the Bayesian approach. Data Science in Finance and Economics, 2023, 3(1): 55-75. doi: 10.3934/DSFE.2023004
  • Small-world networks and scale-free networks are well-known theoretical models within the realm of complex graphs. These models exhibit "low" average shortest-path length; however, key distinctions are observed in their degree distributions and average clustering coefficients: in small-world networks, the degree distribution is bell-shaped and the clustering is "high"; in scale-free networks, the degree distribution follows a power law and the clustering is "low". Here, a model for generating scale-free graphs with "high" clustering is numerically explored, since these features are concurrently identified in networks representing social interactions. In this model, the values of average degree and exponent of the power-law degree distribution are both adjustable, and spatial limitations in the creation of links are taken into account. Several topological metrics are calculated and compared for computer-generated graphs. Unexpectedly, the numerical experiments show that, by varying the model parameters, a transition from a power-law to a bell-shaped degree distribution can occur. Also, in these graphs, the degree distribution is most accurately characterized by a pure power-law for values of the exponent typically found in real-world networks.



    With the development of society, the number of deaths caused by influenza infection has had gradually increased. For example, the global pandemic in 1918 caused more deaths in the world than in the First World War[1,2]. According to statistical analysis, the global influenza pandemic is caused by H2N2 virus, which is called "Asian influenza" because it first occurred in Asia. The incidence rate was about 15–30% [3,4]. Recently, COVID-19 also has a great impact on human life. According to the latest real-time statistics of the World Health Organization, as of October 10, 2022, the cumulative number of confirmed cases worldwide exceeded 600 million, and the cumulative number of deaths was about 6.5 million[5]. Therefore, in order to better control and reduce the number of deaths, how to effectively and quickly control the spread of disease is worthy of further study.

    In the face of the global pandemic, many countries have responded to and mitigated the impact of influenza on human life through media reports, vaccination, disinfection, the use of protective equipment (such as masks), quarantine and other measures. In addition, some biological mathematicians have also established many mathematical models in order to better understand influenza infection and quantify the effectiveness of various control measures [6,7,8,9,10,11,12,13]. For example, considering social factors such as vaccination, media reports and protective measures, [14,15,16,17,18,19,20,21] established some different forms of disease models to analyze the impact of disease transmission. Practice shows that frequent hand washing, disinfection or wearing protective masks, as well as reducing or avoiding close contact with infected persons, can effectively reduce transmission [22,23,24,25,26,27,28,29]. However, media reports on influenza cases and deaths may have a great impact on the public, because the public will take some protective measures after media reports. In addition, quarantine and other measures will greatly reduce the effective contact rate between vulnerable people and infected people, thus reducing the spread of disease.

    Recently, many different control strategies have been proposed to apply to the control of some disease models. For example, by using multiple optimal control, Ndii and Adi [12] have studied the effects of individual awareness and vector controls on Malaria transmission dynamics. By using a non-smooth control strategy, Li et al.[13] have considered the bifurcations and dynamics of a plant disease system. By using the two thresholds control strategy, Li et al.[30] have considered the global dynamics of a Filippov predator-prey model. Chen et al. [31] have considered a two-thresholds policy for a Filippov model in combating influenza. Zhou et al. [32] have discussed a two-thresholds policy to interrupt transmission of West Nile Virus to birds. Meanwhile, based on the above mentioned transmission factors, some scholars have also done a lot of meaningful work. For example, Dong et al. [33] studied a kind of nonlinear incidence Filippov epidemic model to describe the impact of media in the process of epidemic transmission. They found that choosing an appropriate threshold value and control intensity can prevent the outbreak of infectious diseases, and media coverage can reduce the burden of disease outbreaks and shorten the duration of disease outbreaks. Can et al.[25] have studied a Filippov model describing the effects of media coverage and quarantine on the spread of human influenza. Xiao et al. [28] have discussed a media impact switching surface during an infectious disease outbreak. In real life, when the influenza started to spread, that is, when the epidemic was not serious, the public paid little attention to the influenza. Generally, only when the number of people who have already felt it reaches and exceeds a certain threshold level will the mass media begin to coverage in large numbers and the public begin take some countermeasures [11,17,19,20]. Currently, many control managements have been developed, including threshold [30], harvesting control management[34,35,36], impulsive control management[37,38] and other controls [39,40,41,42,43,44,45]. However, the implementation of this measure may cause some social impacts, such as socioeconomic decline and psychological impact on the quarantined people. It is important to consider when to take these control measures. Therefore, an appropriate threshold policy is needed to deal with the influenza outbreak, or at least reduce the number of infected people to an acceptable level. As far as we know, under logistic source and broken line control strategy, the sliding dynamics and bifurcations of a human influenza system have been seldom reported in existing work.

    Motivation and inspiration come from the discussion above. In this paper, we propose a human influenza system under logistic source and broken line control strategy. The main contributions of this paper include three points: First, choose two thresholds IT and ST as a broken line control strategy: Once the number of infected people exceeds IT, the media influence comes into play, and when the number of susceptible individuals is greater than ST, the control by quarantine of susceptible individuals control is open. Furthermore, by choosing different thresholds IT and ST, the existence and stability of all possible equilibria considered, and then the Filippov system tends to the pseudo-equilibrium on sliding mode domain or one endemic equilibrium or bistability endemic equilibria under some conditions. It is worth pointing out that in our paper there exists a new phenomenon of two real equilibrium points co-existing.

    In fact, from the perspective of control strategy, our paper puts forward a broken line control strategy. Different from the previous single-stage control strategy [30,34,35,37,38], our control strategy is divided into two stages and can control and simulate the real life situation well.

    This paper is structured as follows. In Section 2, we propose a non-smooth model under logistic source and broken line control strategy. By changing the infection threshold values and susceptibility threshold values, in Section 3, we consider the global dynamics of the system under Case 1: ST<S1<S2=S3. In Section 4, we study the global dynamics of the system under Case 2: S1<ST<S2=S3. Section 5 considers the global dynamics of the system under Case 3: S2=S3<ST.The regular/virtulal equilibrium bifurcations are given in Section 6. Finally, we summarize the main results of this paper and discuss some biological conclusions in Section 7.

    In [25], a Filippov human influenza model with effects of media coverage and quarantine was considered. Because a logistic source can better depict the real situation, motivated by the above discussion and the existing multiple threshold conditions [25,30,31,32,45], we consider the logistic source factor with the susceptible population, and then the new human influenza model can be described by

    {St=γS(1SK)βSI,It=βSI(d+δ+r)I,Rt=rIdI, (2.1)

    where S is susceptible individuals, I is infected individuals, and R is recovered individuals. γ is an intrinsic growth rate of susceptible individuals, and K is carrying capacity. β is the transmission rate. d is the natural death rate. δ is the death rate caused by disease, and r is the recovered rate.

    To better control the spread of the disease, we give the following broken line control strategy. If the infected populations is lower than IT, no control is required. When the infected populations is greater than IT, the media coverage control is open. That is, the mass media being coverage of information about the disease, including the route of transmission, the number of infected cases and the number of deaths.

    Therefore, the public realizes the harm of the disease, and they changed their behavior, leading to a decline in the contact rate β. In this paper, we consider that the positive number v1(0v11) is the reduction amount of contact rate. In addition, we consider the quarantine control as follows. If the number of susceptible individuals is less than ST, we do not quarantine susceptible individuals. If S>ST holds, the quarantine control is open, and this time we assume that the positive number v2(0v21) is the quarantine rate of susceptible individuals. Figure 1 shows the broken line control strategy. Based on system (2.1) and the broken line control strategy, in this paper, we propose a Filippov influenza system with media control and quarantine of susceptible populations as follows:

    {St=γS(1SK)β(1v1)SIv2S,It=β(1v1)SI(d+δ+r)I,Rt=rIdI (2.2)
    Figure 1.  Schematic diagram of the broken line control strategy.

    with

    (v1,v2)={(0,0), for I<IT,(p,0), for I>IT and S<ST,(p,q), for I>IT and S>ST. (2.3)

    Notice that the third equation R does not contain the variables S and I of (2.2), so we can not consider the third equation R of the system. Then, in this paper, we investigate a Filippov system

    {St=γS(1SK)β(1v1)SIv2S,It=β(1v1)SI(d+δ+r)I. (2.4)

    Then, the first quadrant is divided by the following five regions:

    Γ1={(S,I)R2+:I<IT},Γ2={(S,I)R2+:I>IT and S<ST},Γ3={(S,I)R2+:I>IT and S>ST},Π1={(S,I)R2+:I=IT},Π2={(S,I)R2+:I>IT and S=ST}.

    The non-smooth system in region Γi for i=1,2,3 is described by

    (StIt)=(γS(1SK)βSIβSI(d+δ+r)I)=F1(S,I),(S,I)Γ1;(StIt)=(γS(1SK)β(1p)SIβ(1p)SI(d+δ+r)I)=F2(S,I),(S,I)Γ2;(StIt)=(γS(1SK)β(1p)SIqSβ(1p)SI(d+δ+r)I)=F3(S,I), (S,I)Γ3. (2.5)

    The normal vector of the system is defined as n1=(0,1)T in Π1, while the normal vector of the system is n2=(1,0)T in Π2. Denote the right-hand side of system (2.4) by f. The following definitions are necessary in this paper[14,46,47].

    Definition 1. [14] A point E is called a real equilibrium of system (2.4) if there exists i{1,2,3} such that Fi(E)=0 and EΓi, denoted by ER.

    Definition 2. [14] A point E is called a virtual equilibrium of system (2.4) if there exists i{1,2,3} such that Fi(E)=0 and E¯Γi, where ¯Γi is the closure of Γi, denoted by EV.

    Denote the equations that describe the sliding mode dynamics on the sliding mode domain iΠi by Fsi(S,I),i{1,2}.

    Definition 3. [14] A point E is called a pseudo-equilibrium of system (2.4) if point E is an equilibrium of Fsi(S,I) on sliding mode domain . That is, Fsi(E)=0, and point EΠi,i{1,2}.

    Lemma 1. The solution (S(t),I(t)) of system (2.4) with the positive initial value S(0) and I(0) satisfy S(t)>0 and I(t)>0 for t[0,+).

    Proof. First, we prove that S(t)>0. If not, we assume that there is a time t1 satisfying the solution S(t1)0. Then we know that there exists the other time t>0 such that the solution S(t)=0 and S(t)>0 for t[0,t). From the first equation of system (2.4), one has

    dSdt=γS(1SK)β(1v1)SIv2S=S[r(1SK)β(1v1)Iv2].

    Further, we obtain

    S(t)=S(0)et0[r(1SK)β(1v1)Iv2]ds>0,

    which is a contradiction with S(t)=0. Thus, S(t)>0, for all t>0.

    Next, we prove that I(t)>0. If not, we assume there is a time t2 satisfying the solution I(t2)0. Then, there exists the other time ˜t>0 such that the solution I(˜t)=0, and I(t)>0 for time t[0,˜t). From the second equation of system (2.4), one gets

    dIdt=I[(β(1v1)S(d+δ+r)](d+δ+r)I,t[0,˜t].

    Then, if t=˜t, we have

    I(˜t)I0e(d+δ+r)˜t>0

    which is a contradiction with I(˜t)=0. Thus, I(t)>0, for all t>0. In a word, we have that (S(t),I(t)) of (2.4) with the positive initial value S(0) and I(0) satisfies S(t)>0 and I(t)>0 for t[0,+). The proof is finished.

    Lemma 2. The solution (S,I) of system (2.4) is bounded.

    Proof. Notice the first equation of system (2.4). Then,

    dSdtx=K=β(1v1)KIv2S<0,anddSdtx>K<0.

    Hence, there is a positive number T such that the solution S(t)<K for tT.

    Let N=S+I, and then for tT, we obtain

    dNdt=γS(1SK)v2S(d+δ+r)IγK(1SK)v2S(d+δ+r)IγKmin{γ+v2,d+δ+r}N.

    Moreover,

    lim suptNγKmin{γ+v2,d+δ+r}.

    Therefore, under Lemma 1, we know that the solution of (2.4) is bounded. The proof is finished.

    Next, we give the basic reproduction number R0i,i=1,2,3, of the system. For example, in region Γ1, the basic reproduction number is defined as R01=Kβd+δ+r. In region Γ2, the basic reproduction number is defined as R02=Kβ(1p)d+δ+r. In region Γ3, the basic reproduction number is defined as R03=(γp)Kβ(1p)γ(d+δ+r).

    For system (2.4), clearly, in region Γ1, (2.4) has three equilibria, i.e., E10=(0,0), E11=(K,0) and E1=(S1,I1)=(d+δ+rβ,γβ(1d+δ+rKβ)), which is a stable node if R01>1.

    In region Γ2, system (2.4) has three equilibria, i.e., E20=(0,0), E21=(K,0) and E2=(S2,I2)=(d+δ+rβ(1p),γβ(1p)(1d+δ+rKβ(1p))), which is a stable node (focus) if R02>1.

    In region Γ3, system (2.4) has three equilibria, i.e., E30=(0,0), E31=(Kγqγ,0) and E3=(S3,I3)=(d+δ+rβ(1p),γβ(1p)(1d+δ+rKβ(1p)qr)), which is a stable node (focus) if R03>1.

    Theorem 1. Suppose that R0i<1, and the disease free equilibrium Ei1(i=1,2,3) of (2.4) is globally asymptotically stable. In addition, the endemic equilibrium Ei(i=1,2,3) of (2.4) is globally asymptotically stable if R0i>1.

    Proof. Since R0i<1,i=1,2,3, a Lyapunov function is considered as

    L(t)=SSi1Si1lnSSi1+I.

    Applying LaSalle's invariance principle[13,14], we know that Ei1(i=1,2,3) of system (2.4) is globally asymptotically stable.

    When R0i>1,i=1,2,3, the following Lyapunov function is considered:

    L(t)=SSiSilnSSi+IIiIilnIIi.

    Using LaSalle's invariance principle [13,14], we know that points Ei(i=1,2,3) of system (2.4) areglobally asymptotically stable. The proof is finished.

    This paper only considers the global dynamics of (2.4) under case R0i>1 (i=1,2,3). Next, we aim to address the richness of the possible equilibria and sliding modes on Π1 and Π2 that the system with can exhibit.

    From (2.4), when S1<S2=S3 and I3<I2, we consider the following three cases: ST<S1,S1<ST<S2=S3, and S2=S3<ST with varied IT. Further, according to the dynamics in each case, the biological phenomena of (2.4) are described in this section.

    Throughout the paper, the S-nullclines and I-nullclines of (2.4) are represented by the dashed curves and dash-dot lines, respectively. Thus, S=S1,S2 and S3 are the I-nullclines of F1,F2 and F3, denoted by L12,L22 and L32, respectively. That is, the curves

    L12:={(S,I)Γ1:γS(1SK)βSI=0},
    L22:={(S,I)Γ2:γS(1SK)β(1p)SI=0}

    and

    L32:={(S,I)Γ3:γS(1SK)β(1p)SIqS=0}

    are the S-nullclines of systems F1,F2 and F3, denoted by L11,L21 and L31, respectively.

    In this part, we first consider sliding mode dynamics of (2.4) on Π1 under Case 1: ST<S1<S2=S3. Second, the sliding mode dynamics on Π2 are also given under Case 1: ST<S1<S2=S3. In addition, we investigate the bifurcations of (2.4) under Case 1: ST<S1<S2=S3. Finally, some numerical simulations are displayed to confirm the results.

    This part investigates the existence of the sliding mode region on Π1. Based on Definition 3, if n1,F1>0 and n1,F3<0 hold, then we know that there is the sliding mode region 1, which is expressed as

    1={(S,I)Π1:S1<S<S3}.

    Using the Filippov convex method [13,14], we obtain

    (StIt)=λ1F1(S,I)+(1λ1)F3(S,I),where λ1=n1,F3n1,F3F1.

    So, we have the differential equations describing the sliding mode dynamics along the manifold 1 for system (2.4):

    (StIt)=(γS(1SK)qpS(d+δ+r)IT+(d+δ+r)qβp0). (3.1)

    Next, we analyze the existence of the positive equilibriums on 1 of (3.1). Let

    Δ1=(γqp)24γK[(d+δ+r)IT(d+δ+r)qβp],IT=qβp+(γqp)2K4γ(d+δ+r).

    Proposition 1. For varied IT, we have the following results.

    If IT>IT, then system (3.1) has no equilibrium.

    If IT>IT>qβp, system (3.1) has two positive equilibria E±s1=(S±s1,IT), where S±s1=(γqp)±Δ12γK.

    If IT<qβp, then system (3.1) has a unique positive equilibrium Es2=(Ss2,IT), where Ss2=(γqp)+Δ12γK.

    In addition, when the sliding mode 1 has a pseudo-equilibrium, we have

    ST=γqp2γK. (3.2)

    Proposition 2. Under the condition ST<S1, Es11, and the following results are given.

    If IT<I3, we have E+s11.

    If I3<IT<I1, we have E+s11.

    If IT>I1, we have E+s11.

    Proposition 3. Under the condition S1<ST<S3, the following results hold.

    (1) Assume that I1<I3, and then

    if I1<IT<I3, we have Es11, E+s11;

    if I3<IT<IT, we have Es11, E+s11.

    (2) Assume that I1>I3, and then

    if I3<IT<I1, we have Es11, E+s11;

    if I1<IT<IT, we have Es11, E+s11.

    Proposition 4. Under the condition ST>S3, E+s11, and the following conclusions hold.

    If IT<I1, we have Es11.

    If I1<IT<I3, we have Es11.

    If IT>I3, we have Es11.

    Theorem 2. If IT>IT>qβp, then a stable pseudo-equilibriums E+s1 is located on the sliding mode 1, and the unstable pseudo-equilibriums Es1 is located on the sliding mode 1.

    Proof. Notice that

    S(rS(1SK)qpS(d+δ+r)IT+(d+δ+r)qβp)|S±s1=Δ1.

    Therefore, the point E+s1 is attracting, and the point Es1 is repelling.

    Proposition 5. Assume that γqpγK>d+δ+rβ(1p), and then the pseudo-equilibrium Es2 is not located on 1.

    Proof. From Proposition 1, Eq (3.2) and the condition of Proposition 5, one has

    Ss2=(γqp)+Δ12γK>2ST=2γqpγK>d+δ+rβ(1p)=S3,

    which implies that Ss22ST>S3. Based on the definition of sliding mode region 1, we know that Ss21. Then, 1 does not have a pseudo-equilibrium Es2. The proof is completed.

    Let

    J1=γβ(1p)(1STKqγ),J2=γβ(1p)(1STK).

    Clearly J1<J2. A subset of Π2 is a sliding mode domain if F2,n2F3,n2<0. If IT>J2, there is no sliding mode domain on Π2. If IT<J2, then the sliding mode domain of (2.4) on Π2 is given as

    2={(S,I)Π2:max{IT,J1}<I<J2}. (3.3)

    Using the Filippov convex method[13], we have the differential equations describing the sliding mode dynamics along the manifold 2 for system (2.4) with (2.3):

    (StIt)=(0β(1p)STI(d+δ+r)I). (3.4)

    Clearly, there is not a positive equilibrium. Thus, if there exists a sliding domain 2 on Π2, we know the system does not have a pseudo-equilibrium.

    Due to ST<S1<S2=S3, we know that E2Γ2 is a virtual equilibrium, denoted by EV2. However, point E1 and point E3 may exist depending on IT, and we have the following three Propositions:

    Proposition 6. Assume that ST<S1, and the following assertions hold.

    If IT<I3, we have Es11,E+s11,E1Γ1,E3Γ3.

    If I3<IT<I1, we have Es11,E+s11,E1Γ1,E3Γ3.

    If I1<IT<IT, we have Es11,E+s11,E1Γ1,E3Γ3.

    If IT>IT, we have Es1andE+s1don't exist,E1Γ1,E3Γ3.

    Proposition 7. Under the condition S1<ST<S3, the following assertions hold.

    (1) Assume that I1<I3, and further,

    if IT<I1, we have Es11,E+s11,E1Γ1,E3Γ3;

    if I1<IT<I3, we have Es11,E+s11,E1Γ1,E3Γ3;

    if I3<IT<IT, we have Es11, E+s11,E1Γ1,E3Γ3;

    if IT>IT, we have Es1andE+s1don't exist,E1Γ1,E3Γ3.

    (2) Assume that I1>I3, and further,

    if IT<I3, we have Es11,E+s11,E1Γ1,E3Γ3;

    if I3<IT<I1, we have Es11,E+s11,E1Γ1,E3Γ3;

    if I1<IT<IT, we have Es11, E+s11,E1Γ1,E3Γ3;

    if IT>IT, we have Es1andE+s1don't exist,E1Γ1,E3Γ3.

    Proposition 8. Under the condition ST>S3, the following assertions hold.

    If IT<I1, we have Es11,E+s11,E1Γ1,E3Γ3.

    If I1<IT<I3, we have Es11,E+s11,E1Γ1,E3Γ3.

    If I3<IT<IT, we have Es11,E+s11,E1Γ1,E3Γ3.

    If IT>IT, we have Es1andE+s1don't exist,E1Γ1,E3Γ3.

    Based on Propositions 6–8, we have the following summary:

    B1. Let Es11,E+s11,E1Γ1,E3Γ3, and the value Υ=(ST,IT) belongs to the set B11. Then, we conclude that there does not exist a pseudo-equilibrium, and all trajectories of the system (2.4) will converge to ER3, as shown in Figure 2(a), where

    B11={ΥR2+:ST<S1,IT<min{I1,I3}}.
    Figure 2.  The trajectory of (2.4) under Case 1: ST<S1<S2=S3, d=0.01,δ=0.01,r=0.01,γ=0.2,K=3, (a) where β=0.7,p=0.2,q=0.2,ST=1.5,IT=1; (b) where β=0.7,p=0.2,q=0.2,ST=1.5,IT=2.8; (c) where β=0.7,p=0.2,q=0.2,ST=1.5,IT=3.5; (d) where β=1.5,p=0.5,q=0.2,ST=1,IT=2.8; (e) where β=1.2,p=0.5,q=0.2,ST=1,IT=3.

    B2. Let Es11,E+s11,E1Γ1,E3Γ3, and the value Υ=(ST,IT) belongs to the set B21. Then, we know that E+s11Π1 is a stable pseudo-equilibrium, and all solutions of the system will approach E+s1, as shown in Figure 2(b), where

    B21={ΥR2+:ST<S1,I3<IT<I1}.

    B3. Let Es11,E+s11,E1Γ1,E3Γ3, and the value Υ=(ST,IT) belongs to the set B31B32B33. Then, we conclude that the system (2.4) does not have a pseudo-equilibrium, and all trajectories of the system (2.4) will converge to the equilibrium point ER1, as shown in Figure 2(c), where

    B31={ΥR2+:ST<S1,I1<IT<IT, if ST<S1},B32={ΥR2+:ST<S1,IT>IT, if S1<ST<S3},B33={ΥR2+:ST<S1,I3<IT<IT, if ST>S3}.

    B4. Let Es11,E+s11,E1Γ1,E3Γ3, and the value Υ=(ST,IT) belongs to the set B41. Then, we know that Es11Π1 is an unstable pseudo-equilibrium, and all solutions will approach ER1 or ER3. The result of this numerical simulation is shown in Figure 2(d), where

    B41={ΥR2+:ST<S1,I1<IT<I3}.

    B5. Let Es11,E+s11,E1Γ1,E3Γ3, and the value Υ=(ST,IT) belongs to the set B51. Then, we conclude that Es11Π1 is an unstable pseudo-equilibrium, and all solutions of the system (2.4) will tend to ER1 or E+s1. The result of this numerical simulation is shown in Figure 2(e), where

    B51={ΥR2+:ST<S1,max{I1,I3}<IT<IT, if S1<ST<S3}.

    For the B1 of case 1, when system (2.4) has a unique equilibrium ER3, we have the following.

    Theorem 3. If ST<S1<S3=S2 and IT<min{I1,I3}, then the point ER3 of system (2.4) is globally asymptotically stable.

    Proof. Suppose that the system (2.4) has a closed orbit U (shown in Figure 3(a)) that surrounds the real equilibrium ER32 and the sliding mode 1. Define U=U1+U2+U3, where Ui=UΓi,i=1,2,3. Let Ω be the bounded region delimited by U and Ωi=ΩΓi for i=1,2,3. Considering the Dulac function D=1SI. Three Steps are given as follows:

    Figure 3.  The schematic diagram of limit cycle.

    Step 1: System (2.4) does not have a closed orbit in region Γi,i=1,2,3.

    U((Df1)S+(Df2)I)dSdI=3i=1Ui((DFi1)S+(DFi2)I)dSdI=3γK3i=1Ui1IdSdI<0,

    where f1 is the first component of f, and f2 is the second component of f.Fi1 is the first component of Fi and Fi2 is the second component of Fi,i=1,2,3. Let ˜Ωi be the region bounded by ˜Ui,Pi and Qi, where ˜Ωi and ˜Ui depend on ϵ and converge to Ωi and Ui as ϵ approaches 0.

    Step 2: System (2.4) does not have a closed trajectory in region Γ.

    We can get

    Ωi((DFi1)S+(DFi2)I)dSdI=limϵ0˜Ωi((DFi1)S+(DFi2)I)dSdI.

    Since dS=F11dt and dI=F12dt along ˜U1 and dI=0 along P1, by using Green's theorem, for region ˜Ω1, we have

    ˜Ω1((DF11)S+(DF12)I)dSdI=˜Ω1DF11dIDF12dS=˜U1DF11dIDF12dS+P1DF11dIDF12dS=P1DF12dS. (3.5)

    Similarly, we obtain

    ˜Ω2((DF21)S+(DF22)I)dSdI=P2DF22dS+Q2DF21dI (3.6)

    and

    ˜Ω3((DF31)S+(DF32)I)dSdI=P3DF32dS+Q3DF31dI. (3.7)

    From Eqs (3.5)–(3.7), we have

    0>3i=1Ωi((DFi1)S+(DFi2)I)dSdI=limϵ03i=1˜Ωi(DFi1S+DFi2I)dSdI=limϵ0(P1DF12dSP2DF22dS+Q2DF21dIP3DF32dS+Q3DF31dI). (3.8)

    Denote the intersection points of the closed trajectory U and the line I=IT by A1 and A2 and the intersection point of U and line S=ST if I>IT by A3. In addition, denote the intersection point of the line I=IT and the line S=ST by ET. Note that A1S<ST<A2S and A3I>IT. Then, Eq (3.6) becomes

    0>A1SA2S(βd+δ+rS)dSSTA1S(β(1p)d+δ+rS)dS+A3IIT(γ(1SK)Iβ(1p))dIA2SST(β(1p)d+δ+rS)dS+ITA3I(γ(1SK)qIβ(1p))dI=A2SA1SpdS+ITA3I(qI)dI=p(A2SA1S)+qln(A3IIT)>0, (3.9)

    which is a contradiction. Therefore, we know that there does not have a closed orbit U surrounding the sliding mode 1 and the real equilibrium ER3.

    Step 3: System (2.4) does not have a closed trajectory in regions Γi and Γj(ij). With a similar proof procedure to Step 2, it is easy to that there is no closed trajectory in Γ1 and Γ2 (see Figure 3(d)), Γ1 and Γ3 (see Figure 3(b)), Γ2 and Γ3 (see Figure 3(c)), respectively.

    Therefore, based on Steps 1–3, we know that the point ER3 of (2.4) is globally asymptotically stable if ST<S1<S3=S2 and IT<min{I1,I3}. This completes this theorem.

    With a similar proof procedure to Theorem 3, we have the following.

    Theorem 4. If ST<S1<S2=S3 and the conditions of (B3) hold, then the point ER1 of system (2.4) is globally asymptotically stable.

    In this part, we first consider sliding mode dynamics of (2.4) on Π1 under Case 2: S1<ST<S2=S3. Second, we study the sliding mode dynamics on Π2 under Case 2: S1<ST<S2=S3. In addition, the bifurcations of (2.4) are investigated under Case 2: S1<ST<S2=S3. Finally, some numerical simulations are displayed to confirm the results.

    For Case 2, S1<ST<S2=S3, two sliding domains on Π1 are given as

    3={(S,I)Π1:S1<S<ST},4={(S,I)Π1:ST<S<S3}.

    The dynamics on 3 are governed by

    (StIt)=(γS(1SK)(d+δ+r)IT0). (4.1)

    Now, we investigate the existence of a positive equilibrium on 3 of system (4.1). Let

    Δ2=γ24γ(d+δ+r)ITK,IT=γK4(d+δ+r).

    Proposition 9. For varied IT, we have the following results.

    (1) If IT>IT, then system (4.1) does not have an equilibrium;

    (2) If 0<IT<IT, system (4.1) has two positive equilibria E±s3=(S±s3,IT), where S±s3=γ±Δ22γK.

    Next, we find the conditions of the pseudo-equilibrium on the sliding mode 3. Let

    H1=γKd+δ+rST2+γd+δ+rST,ST=K2.

    Proposition 10. Under the condition ST<S1<ST, Es33, and the following assertions hold.

    (1) If IT>I1, we have E+s33;

    (2) If H1<IT<I1, we have E+s33;

    (3) If IT<H1, we have E+s33.

    Proposition 11. Under the condition S1<ST<ST, we have the following results.

    (1) If IT<min{I1,H1}, we have E±s33.

    (2) If min{I1,H1}<IT<max{I1,H1}, we have

    Es33,E+s33ifI1<H1;

    Es33,E+s33ifI1>H1.

    (3) If max{I1,H1}<IT<IT, we have E±s33.

    Proposition 12. Under the condition S1<ST<ST, E+s33, and the following assertions hold.

    (1) If IT<I1, we have Es33;

    (2) If I1<IT<H1, we have Es33;

    (3) If IT>H1, we have Es33.

    Theorem 5. If 0<IT<IT, then a stable pseudo-equilibrium E+s3 of (2.4) is located on the sliding mode 3, and the unstable pseudo-equilibrium Es3 of (2.4) is located on the sliding mode 3.

    Proof. Notice that

    S(γS(1SK)(d+δ+r)IT)|E±s3=Δ2.

    Therefore, the point E+s3 is attracting, and the point Es3 is repelling.

    The dynamics on region 4 are described by (3.1). Let

    H2=qβpγK(d+δ+r)ST2+(γd+δ+rqp(d+δ+r))ST.

    From Proposition 1, we have the following.

    Proposition 13. Under the condition ST<ST<S3, Es14, and the following assertions hold.

    (1) If IT<I3, we have E+s14;

    (2) If I3<IT<H2, we have E+s14;

    (3) If IT>H2, we have E+s14.

    Proposition 14. Under the condition ST<ST<S3, we have the following results.

    (1) If IT<min{I3,H2}, we have E±s14.

    (2) If min{I3,H2}<IT<max{I3,H2}, we have

    Es14,E+s14ifI3>H2;

    Es14,E+s14ifI3<H2.

    (3) If max{I3,H2}<IT<IT, we have E±s14.

    Proposition 15. Under the condition ST<S3<ST, E+s14, and the following assertions hold.

    (1) If IT<H2, we have Es14;

    (2) If H2<IT<I3, we have Es14;

    (3) If IT>I3, we have Es14.

    Theorem 6. If 0<IT<IT, then a stable pseudo-equilibrium E+s1 is located on 4, and the unstable pseudo-equilibrium Es1 is located on 4.

    If S1<ST<S2=S3, the sliding mode dynamics on region Π2 are the same as Section 3.2. We omit it here.

    For this case, we conclude that the point E2 is a virtual equilibrium. E1 and E3 are changeable depending on IT, and we have the following five situation.

    Proposition 16. Under the conditions ST<S1<ST and ST<ST<S3, the following assertions hold.

    (1) If IT<I3, we have E+s33,E+s14,E1Γ1,E3Γ3;

    (2) If I3<IT<H2, we have E+s33,E+s14,E1Γ1,E3Γ3;

    (3) If H2<IT<H1, we have E+s33,E+s14,E1Γ1,E3Γ3;

    (4) If H1<IT<I1, we have E+s33,E+s14,E1Γ1,E3Γ3;

    (5) If I1<IT<IT, we have E+s33,E+s14,E1Γ1,E3Γ3.

    Proposition 17. Under the conditions S1<ST<ST and ST<ST<S3, the following assertions hold.

    (1) Assume that I1>H1, and further,

    if IT<I3, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if I3<IT<H2, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if H2<IT<H1, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if H1<IT<I1, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if I1<IT<IT, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3.

    (2) Assume that H2<I1<H1, and further,

    if IT<I3, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if I3<IT<H2, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if H2<IT<I1, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if I1<IT<H1, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if H1<IT<IT, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3.

    (3) Assume that I3<I1<H2, and

    if IT<I3, then Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if I3<IT<I1, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if I1<IT<H2, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if H2<IT<H1, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if H1<IT<IT, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3.

    (4) Assume that I1<I3, and further,

    if IT<I1, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if I1<IT<I3, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if I3<IT<H2, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if H2<IT<H1, we have Es33,E+s33,E+s14,E1Γ1,E3Γ3;

    if H1<IT<IT, we have Es33E+s33,E+s14,E1Γ1,E3Γ3.

    Proposition 18. Under the conditions S1<ST<ST and ST<ST<S3, we have the following results.

    (1) Assume that I1>H2, and further,

    if IT<I3, we have Es33,E+s14,E1Γ1,E3Γ3;

    if I3<IT<H2, we have Es33,E+s14,E1Γ1,E3Γ3;

    if H2<IT<I1, we have Es33,E+s14,E1Γ1,E3Γ3;

    if I1<IT<H1, we have Es33,E+s14,E1Γ1,E3Γ3;

    if H1<IT<IT, we have Es33,E+s14,E1Γ1,E3Γ3.

    (2) Assume that I3<I1<H2, and further,

    if IT<I3, we have Es33,E+s14,E1Γ1,E3Γ3;

    if I3<IT<I1, we have Es33,E+s14,E1Γ1,E3Γ3;

    if I1<IT<H2, we have Es33,E+s14,E1Γ1,E3Γ3;

    if H2<IT<H1, we have Es33,E+s14,E1Γ1,E3Γ3;

    if H1<IT<IT, we have Es33,E+s14,E1Γ1,E3Γ3.

    (3) Assume that I1<I3, and further,

    if IT<I1, we have Es33,E+s14,E1Γ1,E3Γ3;

    if I1<IT<I3, we have Es33,E+s14,E1Γ1,E3Γ3;

    if I3<IT<H2, we have Es33,E+s14,E1Γ1,E3Γ3;

    if H2<IT<H1, we have Es33,E+s14,E1Γ1,E3Γ3;

    if H1<IT<IT, we have Es33,E+s14,E1Γ1,E3Γ3.

    Proposition 19. Under the conditions S1<ST<ST and ST<ST<S3, we have the following results.

    (1) Assume that I3>H1, I1<H2<H1<I3, and

    if IT<I1, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

    if I1<IT<H2, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

    if H2<IT<H1, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

    if H1<IT<I3, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

    if I3<IT<IT, we have Es33,Es14,E+s14,E1Γ1,E3Γ3.

    (2) Assume that H2<I3<H1, I1<H2<I3<H1, and further,

    if IT<I1, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

    if I1<IT<H2, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

    if H2<IT<I3, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

    if IT>I3, then

      ◇ if IT>H1, and further,

        ◇ if I3<IT<H1, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

        ◇ if H1<IT<IT, we have Es33,Es14,E+s14,E1Γ1,E3Γ3.

      ◇ if IT<H1, and

        ◇if I3<IT<IT, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

        ◇ if IT<IT<H1, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

        ◇ if H1<IT<IT, we have Es33,Es14,E+s14,E1Γ1,E3Γ3.

    (3) Assume that I1<I3<H2, I1<I3<H2<H1, and

    if IT<I1, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

    if I1<IT<I3, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

    if I3<IT<H2, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

    if IT>H2,

      ◇ if IT>H1, and further,

        ◇ if H2<IT<H1, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

        ◇ if H1<IT<IT, we have Es33,Es14,E+s14,E1Γ1,E3Γ3.

      ◇ if IT<H1, and further,

        ◇ if H2<IT<IT, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

        ◇ if IT<IT<H1, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

        ◇ if H1<IT<IT, we have Es33,Es14,E+s14,E1Γ1,E3Γ3.

    (4) Assume that I3<I1, further, I3<I1<H2<H1, and

    if IT<I3, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

    if I3<IT<I1, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

    if I1<IT<H2, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

    if IT>H2, and further,

      ◇ if IT>H1, and

        ◇if H2<IT<H1, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

        ◇if H1<IT<IT, we have Es33,Es14,E+s14,E1Γ1,E3Γ3.

      ◇ if IT<H1, and further

        ◇ if H2<IT<IT, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

        ◇ if IT<IT<H1, we have Es33,Es14,E+s14,E1Γ1,E3Γ3;

        ◇ if H1<IT<IT, we have Es33,Es14,E+s14,E1Γ1,E3Γ3.

    Proposition 20. Suppose S1<ST<ST and ST<S3<ST, E+s33 and E+s14Π1, we have the following results.

    (1) Assume that I3>H1, and further,

    if IT<I1, we have Es33,Es14,E1Γ1,E3Γ3;

    if I1<IT<H2, we have Es33,Es14,E1Γ1,E3Γ3;

    if H2<IT<H1, we have Es33,Es14,E1Γ1,E3Γ3;

    if H1<IT<I3, we have Es33,Es14,E1Γ1,E3Γ3;

    if I3<IT<IT, we have Es33Es14,E1Γ1,E3Γ3.

    (2) Assume that I3<H1, and further,

    if IT<I1, we have Es33,Es14,E1Γ1,E3Γ3;

    if I1<IT<H2, we have Es33,Es14,E1Γ1,E3Γ3;

    if H2<IT<I3, we have Es33,Es14,E1Γ1,E3Γ3;

    if I3<IT<H1, we have Es33,Es14,E1Γ1,E3Γ3;

    if H1<IT<IT, we have Es33,Es14,E1Γ1,E3Γ3.

    Based on Propositions 16–20, the following summary is given.

    C1. Let Es33,E+s33,Es14,E+s14,E1Γ1,E3Γ3, and the value Υ=(ST,IT) belongs to the set C11. Then, we conclude that the system (2.4) does not have a pseudo-equilibrium, and all trajectories of the system (2.4) will converge to the equilibrium point ER3. The result of this numerical simulation is shown in Figure 4(a), where

    C11={ΥR2+:S1<ST<S3,IT<min{I1,I3}}.
    Figure 4.  The trajectory of system (2.4) under Case 2: S1<ST<S2=S3, d=0.1,δ=0.1,r=0.1,γ=0.2,K=1, (a) where β=0.7,p=0.2,q=0.2,ST=3,IT=1; (b) where β=0.7,p=0.2,q=0.2,ST=3,IT=2.8; (c) where β=0.7,p=0.2,q=0.2,ST=3,IT=3.1; (d) where β=0.7,p=0.2,q=0.2,ST=3,IT=3.21; (e) where β=0.7,p=0.2,q=0.2,ST=3,IT=3.4; (f) where β=1.8,p=0.6,q=0.4,ST=2.3,IT=2.5.

    C2. Let Es33,E+s33,Es14,E+s14,E1Γ1,E3Γ3, and the value Υ=(ST,IT) belongs to the set C21. Then, we know that E+s14Π1 is a stable pseudo-equilibrium, and all solutions of the system (2.4) will approach the point E+s1, as shown in Figure 4(b), where

    C21={ΥR2+:S1<ST<S3,I3<IT<min{H2,I1}}.

    C3. Let E_{s3}^{-} \notin \ell_3, E_{s3}^{+} \notin \ell_3, E_{s1}^{-} \notin \ell_4, E_{s1}^{+} \notin \ell_4, E_{1}\notin \Gamma_1, E_{3}\notin \Gamma_3 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set C_{3-1} . Then, we conclude that the system (2.4) does not have a pseudo-equilibrium, and all trajectories of the system (2.4) will converge to the point E_T = (S_T, I_T) . The result of this numerical simulation is shown in Figure 4(c), where

    \begin{equation*} \begin{array}{l} C_{3-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, H_{2} < I_T < {\text{min}}\{H_{1}, I_1^*\}\right\}. \end{array} \end{equation*}

    C4. Let E_{s3}^{-} \notin \ell_3, E_{s3}^{+} \in \ell_3, E_{s1}^{-} \notin \ell_4, E_{s1}^{+} \notin \ell_4\subset \Pi_1, E_{1}\notin \Gamma_1, E_{3}\notin \Gamma_3 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set C_{4-1} . Then, we know that the point E_{s3}^{+}\in \ell_3 is a stable pseudo-equilibrium, and all solutions of the system (2.4) will approach the point E_{s3}^{+} , as shown in Figure 4(d), where

    \begin{equation*} \begin{array}{l} C_{4-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, H_{1} < I_T < I_1^*\right\}. \end{array} \end{equation*}

    C5. Let E_{s3}^{-} \notin \ell_3, E_{s3}^{+} \notin \ell_3, E_{s1}^{-} \notin \ell_4, E_{s1}^{+} \notin \ell_4, E_{1}\in \Gamma_1, E_{3}\notin \Gamma_3 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set C_{5-1} \cup C_{5-2} \cup C_{5-3}\cup C_{5-4} . Then, we know that the system (2.4) does not have a pseudo-equilibrium, and all trajectories of the system (2.4) will converge to the point E_{1}^R . The result of this numerical simulation is shown in Figure 4(e), where

    \begin{equation*} \begin{array}{l} C_{5-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, I_1^* < I_T < I_T^*, \text { if } S_{T}^{*'} < S_{1}^{*} < S_{T} {\text{ and }} S_{T}^{*} < S_T < S_{3}^{*}\right\}, \\ C_{5-2} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, H_{1} < I_T < I_T^{*'}, \text { if }S_{1}^{*} < S_T < S_{T}^{*'} {\text{ and }} S_{T}^{*} < S_T < S_{3}^{*}\right\}, \\ C_{5-3} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, H_{1} < I_T < I_T^{*'}, \text { if }S_{1}^{*} < S_T < S_{T}^{*'} {\text{ and }} S_{T} < S_{T}^{*} < S_{3}^{*}, I_3^* < H_{1}\right\}, \\ C_{5-4} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, {\text{max}}\{H_{1}, I_3^*\} < I_T < I_T^{*'}, \text { if }S_{1}^{*} < S_T < S_{T}^{*'} {\text{ and }}S_T < S_{3}^{*} < S_{T}^{*}\right\}. \end{array} \end{equation*}

    C6. Let E_{s3}^{-} \in \ell_3, E_{s3}^{+} \notin \ell_3, E_{s1}^{-} \notin \ell_4, E_{s1}^{+} \notin \ell_4, E_{1}\in \Gamma_1, E_{3}\in \Gamma_3 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set C_{6-1} . Then, we conclude that E_{s3}^{-}\in \ell_3 is an unstable pseudo-equilibrium, and the solution of the system (2.4) will approach the point E_{1}^R or E_{3}^R or E_T . The result of this numerical simulation is shown in Figure 4(f), where

    \begin{equation*} \begin{array}{l} C_{6-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, I_1^* < I_T < {\text{min}}\{H_2, I_3^*\}\right\}. \end{array} \end{equation*}

    C7. Let E_{s3}^{-} \in \ell_3, E_{s3}^{+} \notin \ell_3, E_{s1}^{-} \notin \ell_4, E_{s1}^{+} \in \ell_4, E_{1}\in \Gamma_1, E_{3}\notin \Gamma_3 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set C_{7-1} . Then, we know that E_{s3}^{-}\in \ell_3 is an unstable pseudo-equilibrium, where E_{s1}^{+} \in \ell_4 is stable. The solution of the system (2.4) will tend to E_{s1}^{+} or E_{1}^R or E_T . The result of this numerical simulation is shown in Figure 5(a), where

    \begin{equation*} \begin{array}{l} C_{7-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, {\text{max}}\{I_1^*, I_3^*\} < I_T < H_{2}\right\}. \end{array} \end{equation*}
    Figure 5.  The trajectory of system (2.4) under Case 2: S_1^* < S_T < S_{2}^{*} = S_{3}^{*} , d = 0.1, \delta = 0.1, r = 0.1, \gamma = 0.2, K = 3 , (a) where \beta = 1.8, p = 0.6, q = 0.4, S_T = 2.3, I_T = 2.8 ; (b) where \beta = 1.8, p = 0.6, q = 0.4, S_T = 2.3, I_T = 3 ; (c) where \beta = 1.2, p = 0.5, q = 0.4, S_T = 3, I_T = 3.2 ; (d) where \beta = 1.5, p = 0.5, q = 0.4, S_T = 1.46, I_T = 2.67 ; (e) where \beta = 1.5, p = 0.5, q = 0.4, S_T = 1.31, I_T = 2.64 ; (f) where \beta = 1.5, p = 0.5, q = 0.4, S_T = 1.7, I_T = 2.8 .

    C8. Let E_{s3}^{-} \in \ell_3, E_{s3}^{+} \notin \ell_3, E_{s1}^{-} \notin \ell_4, E_{s1}^{+} \notin \ell_4, E_{1}\in \Gamma_1, E_{3}\notin \Gamma_3 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set C_{8-1} \cup C_{8-2} \cup C_{8-3}\cup C_{8-4} . Then, we conclude that E_{s3}^{-}\in \ell_3 is an unstable pseudo-equilibrium. The solution of the system (2.4) will approach the equilibrium point E_{1}^R or E_T , as shown in Figure 5(b), where

    \begin{equation*} \begin{array}{l} C_{8-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, {\text{max}}\{I_1^*, H_{2}\} < I_T < H_{1}, \text { if }S_{1}^{*} < S_{T}^{*'} < S_{T} {\text{ and }} S_{T}^{*} < S_T < S_{3}^{*}\right\}, \\ C_{8-2} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, {\text{max}}\{I_1^*, H_{2}\} < I_T < H_{1}, \text { if }S_{1}^{*} < S_T < S_{T}^{*'} {\text{ and }} S_{T}^{*} < S_T < S_{3}^{*}\right\}, \\ C_{8-3} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, I_T^* < I_T < H_{1}, \text { if }S_{1}^{*} < S_T < S_{T}^{*'} {\text{ and }} S_{T} < S_{T}^{*} < S_{3}^{*}\right\}, \\ C_{8-4} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, I_3^* < I_T < H_{1}, \text { if }S_{1}^{*} < S_T < S_{T}^{*'} {\text{ and }}S_T < S_{3}^{*} < S_{T}^{*}\right\}. \end{array} \end{equation*}

    C9. Let E_{s3}^{-} \in \ell_3, E_{s3}^{+} \in \ell_3, E_{s1}^{-} \notin \ell_4, E_{s1}^{+} \notin \ell_4, E_{1}\in \Gamma_1, E_{3}\notin \Gamma_3 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to C_{9-1} . Then, we conclude that E_{s3}^{-}\in \ell_3 is a unstable pseudo-equilibrium, where E_{s3}^{+} \in \ell_3 is stable. The solution of the system (2.4) will tend to the equilibrium point E_{s3}^{+} or E_{1}^R or E_T . The results of this numerical simulation are shown in Figure 5(c), where

    \begin{equation*} \begin{array}{l} C_{9-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, H_{1} < I_T < I_{T}^{*'}, \text { if }S_{1}^{*} < S_{T}^{*'} < S_{T} {\text{ and }} S_{T}^{*} < S_T < S_{3}^{*}\right\}. \end{array} \end{equation*}

    C10. Let E_{s3}^{-} \in \ell_3, E_{s3}^{+} \notin \ell_3, E_{s1}^{-} \in \ell_4, E_{s1}^{+} \notin \ell_4, E_{1}\in \Gamma_1, E_{3}\in \Gamma_3 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set C_{10-1} \cup C_{10-2} . Then, we conclude that E_{s3}^{-}\in \ell_3 is an unstable pseudo-equilibrium. The solution of the system (2.4) will converge to E_{1}^R or E_{3}^R or E_T , as shown in Figure 5(d), where

    \begin{equation*} \begin{array}{l} C_{10-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, H_2 < I_T < {\text{min}}\{H_1, I_3^*\}, \text { if }S_{1}^{*} < S_T < S_{T}^{*'} {\text{ and }} S_{T} < S_{T}^{*} < S_{3}^{*}\right\}, \\ C_{10-2} = \left\{\Upsilon\in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, H_2 < I_T < {\text{min}}\{H_1, I_3^*\}, \text { if }S_{1}^{*} < S_T < S_{T}^{*'} {\text{ and }}S_T < S_{3}^{*} < S_{T}^{*}\right\}. \end{array} \end{equation*}

    C11. Let E_{s3}^{-} \notin \ell_3, E_{s3}^{+} \notin \ell_3, E_{s1}^{-} \in \ell_4, E_{s1}^{+} \notin \ell_4, E_{1}\in \Gamma_1, E_{3}\in \Gamma_3 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set C_{11-1} \cup C_{11-2} . Then, we show that E_{s1}^{-}\in \ell_4 is an unstable pseudo-equilibrium. The solution of the system (2.4) will tend to E_{1}^R or E_{3}^R or E_T , as shown in Figure 5(e), where

    \begin{equation*} \begin{array}{l} C_{11-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, H_{1} < I_T < I_3^*, \text { if }S_{1}^{*} < S_T < S_{T}^{*'} {\text{ and }} S_{T} < S_{T}^{*} < S_{3}^{*}\right\}, \\ C_{11-2} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, H_{1} < I_T < I_3^*, \text { if }S_{1}^{*} < S_T < S_{T}^{*'} {\text{ and }}S_T < S_{3}^{*} < S_{T}^{*}\right\}. \end{array} \end{equation*}

    C12. Let E_{s3}^{-} \in \ell_3, E_{s3}^{+} \notin \ell_3, E_{s1}^{-} \in \ell_4, E_{s1}^{+} \in \ell_4, E_{1}\in \Gamma_1, E_{3}\notin \Gamma_3 , \text { if }S_{1}^{*} < S_T < S_{T}^{*'} {\text{ and }} S_{T} < S_{T}^{*} < S_{3}^{*} , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set C_{12-1} . Then, we know that the point E_{s3}^{-}\in \ell_3 and the value E_{s1}^{-} \in \ell_4 are unstable pseudo-equilibriums, where E_{s1}^{+} \in \ell_4 is stable. The solution of the system (2.4) will approach E_{s1}^{+} or E_{1}^R or E_T . The result of this numerical simulation is shown in Figure 5(f), where

    \begin{equation*} \begin{array}{l} C_{12-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, {\text{max}}\{I_3^*, H_{2}\} < I_T < {\text{min}}\{H_1, I_T^*\}\right\}. \end{array} \end{equation*}

    C13. Let E_{s3}^{-} \notin \ell_3, E_{s3}^{+} \notin \ell_3, E_{s1}^{-} \in \ell_4, E_{s1}^{+} \in \ell_4, E_{1}\in \Gamma_1, E_{3}\notin \Gamma_3 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set C_{13-1} . Then, we conclude that E_{s1}^{-}\in \ell_4 is an unstable pseudo-equilibrium, where E_{s1}^{+} \in \ell_4\subset \Pi_1 is stable. The solution of the system (2.4) will converge to E_{s1}^{+} or E_{1}^R or E_T , as shown in Figure 6, where

    \begin{equation*} \begin{array}{l} C_{13-1} = \left\{\Upsilon\in \mathbb{R}_{+}^{2}:S_1^* < S_T < S_{3}^{*}, {\text{max}}\{I_3^*, H_{1}\} < I_T < I_T^*, \text { if }S_{1}^{*} < S_T < S_{T}^{*'} {\text{ and }} S_{T} < S_{T}^{*} < S_{3}^{*}\right\}. \end{array} \end{equation*}
    Figure 6.  Dynamics of system (2.4) under Case 2: S_1^* < S_T < S_{2}^{*} = S_{3}^{*} . Here, the parameter values are \beta = 1.5, d = 0.01, \delta = 0.01, r = 0.01, \gamma = 0.2, K = 3, p = 0.5, q = 0.4, S_T = 1.48, I_T = 2.8 and d+\delta+r = 0.03 .

    In this part, we first consider sliding mode dynamics of (2.4) on \Pi_1 under Case 3: S_{2}^{*} = S_{3}^{*} < S_T . Second, the sliding mode dynamics on \Pi_2 are given under Case 3: S_{2}^{*} = S_{3}^{*} < S_T . In addition, we investigate the bifurcations of (2.4) under Case 3: S_{2}^{*} = S_{3}^{*} < S_T . Finally, some numerical simulations are displayed to confirm the results.

    If \langle n_1, F_1\rangle > 0 and \langle n_1, F_2\rangle < 0 on \ell_5 , then \ell_5 is described as

    \begin{equation*} \label{eq5.1} \ell_5 = \{(S, I)\in \Pi_1:S_1^* < S < S_2^* \}. \end{equation*}

    Next, the conditions of a pseudo-equilibrium on the sliding mode \ell_5 are given as follows.

    Proposition 21. Under the condition S_{T}^{*'} > S_{2}^{*} , E_{s 3}^{+} \notin \ell_5 and the following assertions hold.

    (1) If I_T < I_1^* , we have E_{s 3}^{-} \notin \ell_5 ;

    (2) If I_{1}^{*} < I_T < I_{2}^{*} , we have E_{s 3}^{-} \in \ell_5 ;

    (3) If I_T > I_{2}^{*} , we have E_{s 3}^{-} \notin \ell_5 .

    Proposition 22. Under the condition S_{1}^{*} < S_{T}^{*'} < S_{2}^{*} , the following assertions hold.

    (1) Assume that I_{1}^* < I_{2}^* , and further,

    if I_{1}^* < I_T < I_{2}^* , we have E_{s 3}^{-} \in \ell_5 , E_{s 3}^{+} \notin \ell_5 ;

    if I_{2}^* < I_T < I_{T}^{*'} , we have E_{s 3}^{-} \in \ell_5 , E_{s 3}^{+} \in \ell_5 .

    (2) Assume that I_{1}^* > I_{2}^* , and further,

    if I_{2}^* < I_T < I_{1}^* , we have E_{s 3}^{-} \notin \ell_5 , E_{s 3}^{+} \in \ell_5 ;

    if I_{1}^* < I_T < I_{T}^{*'} , we have E_{s 3}^{-} \in \ell_5 , E_{s 3}^{+} \in \ell_5 .

    Proposition 23. Under the condition S_{T}^{*'} < S_{1}^{*} , E_{s 3}^{-} \notin \ell_5 and the following assertions hold.

    (1) If I_T < I_{2}^{*} , we have E_{s 3}^{+} \notin \ell_5 ;

    (2) If I_{2}^{*} < I_T < I_{1}^{*} , we have E_{s 3}^{+} \in \ell_5 ;

    (3) If I_T > I_1^* , we have E_{s 3}^{+} \notin \ell_5 .

    Theorem 7. If 0 < I_T < I_T^{*'} , the sliding mode \ell_5 has a stable pseudo-equilibrium E_{s 3}^{+} , and the sliding mode \ell_5 E_{s 3}^{-} has an unstable pseudo-equilibrium.

    When S_{2}^{*} = S_{3}^{*} < S_T , the sliding mode dynamics on \Pi_2 are the same as Section 3.2. We omit it here.

    For this Case 3, the point E_{3} is a virtual equilibrium, denoted by E_{3}^V . Points E_{1} and E_{2} are changeable depending on I_T , and then we have the following.

    Proposition 24. Under the condition S_{T}^{*'} > S_{2}^{*} , the following assertions hold.

    (1) If I_T < I_{1}^* , we have E_{s 3}^{-} \notin \ell_5, E_{s 3}^{+} \notin \ell_5, E_{1}\notin \Gamma_1, E_{2}\in \Gamma_2 ;

    (2) If I_{1}^* < I_T < I_{2}^* , we have E_{s 3}^{-} \in \ell_5, E_{s 3}^{+} \notin \ell_1, E_{1}\in \Gamma_1, E_{2}\in \Gamma_2 ;

    (3) If I_{2}^* < I_T < I_{T}^{*'} , we have E_{s 3}^{-} \notin \ell_5, E_{s 3}^{+} \notin \ell_5, E_{1}\in \Gamma_1, E_{2}\notin \Gamma_2 ;

    (4) If I_T > I_{T}^{*'} , we have that E_{s 3}^{-} \mathit{{\text{and}}} E_{s 3}^{+} \mathit{{\text{do not exist}}}, E_{1}\in \Gamma_1, E_{2}\notin \Gamma_2 .

    Proposition 25. Under the condition S_{1}^{*} < S_{T}^{*'} < S_{2}^{*} , the following assertions hold.

    (1) Assume that I_{1}^* < I_{2}^* , and further,

    if I_T < I_{1}^* , we have E_{s 3}^{-} \notin \ell_5, E_{s 3}^{+} \notin \ell_5, E_{1}\notin \Gamma_1, E_{2}\in \Gamma_2 ;

    if I_{1}^* < I_T < I_{2}^* , we have E_{s 3}^{-} \in \ell_5, E_{s 3}^{+} \notin \ell_5, E_{1}\in \Gamma_1, E_{2}\in \Gamma_2 ;

    if I_{2}^* < I_T < I_{T}^{*'} , we have E_{s 3}^{-} \in \ell_1 , E_{s 3}^{+} \in \ell_5, E_{1}\in \Gamma_1, E_{2}\notin \Gamma_2 ;

    if I_T > I_{T}^{*'} , we have that E_{s 1}^{-} \mathit{{\text{and}}} E_{s 1}^{+} \mathit{{\text{do not exist}}}, E_{1}\in \Gamma_1, E_{2}\notin \Gamma_2 .

    (2) Assume that I_{1}^* > I_{2}^* , and further,

    if I_T < I_{2}^* , we have E_{s 3}^{-} \notin \ell_5, E_{s 3}^{+} \notin \ell_5, E_{1}\notin \Gamma_1, E_{2}\in \Gamma_2 ;

    if I_{2}^* < I_T < I_{1}^* , we have E_{s 3}^{-} \notin \ell_5, E_{s 3}^{+} \in \ell_5, E_{1}\notin \Gamma_1, E_{2}\notin \Gamma_2 ;

    if I_{1}^* < I_T < I_{T}^{*'} , we have E_{s 3}^{-} \in \ell_5 , E_{s 3}^{+} \in \ell_5\, E_{1}\in \Gamma_1, E_{2}\notin \Gamma_2 ;

    if I_T > I_{T}^{*'} , we have that E_{s 3}^{-} \mathit{{\text{and}}} E_{s 3}^{+} \mathit{{\text{do not exist}}}, E_{1}\in \Gamma_1, E_{2}\notin \Gamma_2 .

    Proposition 26. Under the condition S_{T}^{*'} < S_{1}^{*} , the following assertions hold.

    (1) If I_T < I_{2}^* , we have E_{s 3}^{-} \notin \ell_5, E_{s 3}^{+} \notin \ell_5, E_{1}\notin \Gamma_1, E_{2}\in \Gamma_2 ;

    (2) If I_{2}^* < I_T < I_{1}^* , we have E_{s 3}^{-} \notin \ell_5, E_{s 3}^{+} \in \ell_5, E_{1}\notin \Gamma_1, E_{2}\notin \Gamma_2 ;

    (3) If I_{1}^* < I_T < I_{T}^{*'} , we have E_{s 3}^{-} \notin \ell_5, E_{s3}^{+} \notin \ell_5, E_{1}\in \Gamma_1, E_{2}\notin \Gamma_2 ;

    (4) If I_T > I_{T}^{*'} , we have that E_{s 3}^{-} \mathit{{\text{and}}} E_{s 3}^{+} \mathit{{\text{do not exist}}}, E_{1}\in \Gamma_1, E_{2}\notin \Gamma_2 .

    Based on Propositions 24–26, the following summary is given.

    D1. Let E_{s3}^{-} \notin \ell_5, E_{s3}^{+} \notin \ell_5, E_{1}\notin \Gamma_1, E_{2}\in \Gamma_2 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set D_{1-1} . Then, we conclude that the system (2.4) does not have a pseudo-equilibrium, and all trajectories of the system (2.4) will converge to E_{2}^R . The result of this numerical simulation is shown in Figure 7(a), where

    \begin{equation*} \begin{array}{l} D_{1-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_{2}^{*} < S_T, I_T < {\text{min}}\{I_{1}^*, I_{2}^*\}\right\}. \end{array} \end{equation*}
    Figure 7.  The trajectory of the system (2.4) under Case 3: S_{2}^{*} = S_{3}^{*} < S_T , d = 0.01, \delta = 0.01, r = 0.01, \gamma = 0.2, K = 3 , (a) where \beta = 0.7, p = 0.2, q = 0.2, S_T = 3.8, I_T = 1 ; (b) where \beta = 1.5, p = 0.5, q = 0.2, S_T = 3.8, I_T = 2.8 ; (c) where \beta = 0.7, p = 0.2, q = 0.2, S_T = 3.8, I_T = 3.5 ; (d) where \beta = 1.2, p = 0.5, q = 0.2, S_T = 3.8, I_T = 3.15 ; (e) where \beta = 0.7, p = 0.2, q = 0.2, S_T = 3.8, I_T = 3 .

    D2. Let E_{s3}^{-} \in \ell_5, E_{s3}^{+} \notin \ell_5, E_{1}\in \Gamma_1, E_{2}\in \Gamma_2 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set D_{2-1} . Then, we show that E_{s3}^{-}\in \ell_5 is an unstable pseudo-equilibrium. The solution of the system (2.4) will approach E_{1}^R or E_{2}^R , as shown in Figure 7(b), where

    \begin{equation*} \begin{array}{l} D_{2-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_{2}^{*} < S_T, I_{1}^* < I_T < I_{2}^* \right\}. \end{array} \end{equation*}

    D3. Let E_{s3}^{-} \notin \ell_5, E_{s3}^{+} \notin \ell_5, E_{1}\in \Gamma_1, E_{2}\notin \Gamma_2 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set D_{3-1}\cup D_{3-2}\cup D_{3-3} . Then, we know that the system (2.4) does not have a pseudo-equilibrium, and all trajectories of the system (2.4) will converge to E_{1}^R , as shown in Figure 7(c), where

    \begin{equation*} \begin{array}{l} D_{3-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_{2}^{*} < S_T, I_{2}^* < I_T < I_{T}^{*'}, \text { if } S_{T}^{*'} > S_{2}^{*} \right\}, \\ D_{3-2} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_{2}^{*} < S_T, I_T > I_{T}^{*'} {\text{ if }}S_{1}^{*} < S_{T}^{*'} < S_{2}^{*} \right\}, \\ D_{3-3} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_{2}^{*} < S_T, I_{1}^* < I_T < I_{T}^{*'}, \text { if } S_{T}^{*'} < S_{1}^{*}\right\}. \end{array} \end{equation*}

    D4. Let E_{s3}^{-} \in \ell_5, E_{s3}^{+} \in \ell_5, E_{1}\in \Gamma_1, E_{2}\notin \Gamma_2 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set D_{4-1} . Then, we show that E_{s3}^{-}\in \ell_5 is an unstable pseudo-equilibrium, and the solution of system (2.4) will approach E_{1}^R or E_{s3}^{+} , as shown in Figure 7(d), where

    \begin{equation*} \begin{array}{l} D_{4-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_{2}^{*} < S_T, {\text{max}}\{I_{1}^*, I_{2}^*\} < I_T < I_{T}^{*'}, \text { if }S_{1}^{*} < S_{T}^{*'} < S_{2}^{*}\right\} . \end{array} \end{equation*}

    D5. Let E_{s3}^{-} \notin \ell_5, E_{s3}^{+} \in \ell_5, E_{1}\notin \Gamma_1, E_{2}\notin \Gamma_2 , and the value \Upsilon = \left(S_{T}, I_{T}\right) belongs to the set D_{5-1} . Then, we conclude that E_{s3}^{+}\in \ell_5 is a stable pseudo-equilibrium. All solutions of the system (2.4) will tend to E_{s3}^{+} . The result of this numerical simulation is shown in Figure 7(e), where

    \begin{equation*} \begin{array}{l} D_{5-1} = \left\{\Upsilon \in \mathbb{R}_{+}^{2}:S_{2}^{*} < S_T, I_{2}^* < I_T < I_{1}^* \right\}. \end{array} \end{equation*}

    Remark 1. For smooth system (2.1), we have discussed the three equilibrium points, that is, (0, 0) , the disease free equilibrium point E_{i1}(i = 1, 2, 3) and the endemic equilibrium point E_{i}(i = 1, 2, 3) of (2.4). By constructing a Lyapunov function, we obtain the global stability of system (2.1) in Theorem 1. For the non-smooth system (2.4), we investigate the non-smooth system (2.4) with two threshold control strategies. Using a Filippov analysis method, Green's formula, the comparison theorem and numerical simulation method, the rich dynamics of the system are given, such as the bistability phenomenon, the globally stable pseudo-equilibrium and the regular/virtulal equilibrium bifurcations. Through the two control strategies, we can control the disease individuals to the appropriate balance. In particular, Theorems 3, 6 and 7 in this paper cannot appear in the smooth system (2.1); please see Figure 4(b), (c). There is bistability in the system (2.4).

    Remark 2. In [25], a Filippov model describing the effects of media coverage and quarantine on the spread of human influenza was considered, and the threshold conditions for stability switches were obtained analytically. The discontinuous system (2.4) considered in our paper is a logistic source, and [25] considered a linear source. Second, the dynamics are different. Our paper employs the Green's theorem and a Dulac function. Then, we show that two real equilibria occur simultaneously in our paper. Using numerical simulation methods, the sliding dynamics and bifurcations of a human influenza system under logistic source and broken line control strategy are given. The results of this paper are new with respect to[25].

    Based on the previous discussion, it is shown that system (2.4) will exhibit multiple equilibriums and sliding modes. In order to better construct the bifurcation diagram, we choose \gamma and I_{T} as bifurcation parameters, and the other parameters are fixed as shown in Figure 8. With the expressions of equilibria found in Section 2.3, the lines to divide the relevant parameter plane are given as follows:

    l_{1}: = \left\{(\gamma , I_{T})| I_{T} = I_{1}^{*} = \frac{\gamma }{\beta}(1-\frac{d+\delta+r}{K\beta})\right\},
    l_{2}: = \left\{(\gamma , I_{T})| I_{T} = I_{2}^{*} = \frac{\gamma}{\beta(1-p)}(1-\frac{d+\delta+r}{K\beta(1-p)})\right\},
    l_{3}: = \left\{(\gamma , I_{T})| I_{T} = I_{3}^{*} = \frac{\gamma}{\beta(1-p)}(1-\frac{d+\delta+r}{K\beta(1-p)}-\frac{q}{r})\right\}.
    Figure 8.  The regular/virtulal equilibrium bifurcations where \beta = 0.7, p = 0.2, q = 0.2, d = 0.01, \delta = 0.01, r = 0.01, K = 3 .

    The three solid lines l_{1} , l_{2} and l_{3} divide the \gamma-I_{T} two-dimensional plane space into four regions in the first quadrant. Suppose that the control value I_{T} satisfies I_{3}^{*} < I_{T} < I_{2}^{*} and I_{1}^{*} < I_{3}^{*} (that is, regions \Omega_{2}^{*} and region \Omega_{3}^{*} ; see Figure 8), the points E_{2} and E_{3} are virtual equilibria points (denoted by E_{2}^{v} , and E_{3}^{v} , respectively), and E_{s1}^{-} exists with the sliding mode domain. If the control value I_{T} satisfies I_{T} > I_{2}^{*} (that is, \Omega_{1}^{*} , as shown Figure 8), E_{2} is a regular equilibrium, while the point E_{3} is a virtual equilibrium (denoted by the equilibrium E_{2}^{R} and the equilibrium E_{3}^{v} , respectively), and point E_{s1}^{-} does not exist with the sliding mode domain. If I_{T} < I_{3}^{*} (that is, \Omega_{4}^{*} ; see Figure 8), E_{3} is a regular equilibrium, while point E_{2} is a virtual equilibrium (denoted by E_{3}^{R} and E_{2}^{v} ), and point E_{s1}^{-} does not exist with the sliding mode domain.

    Next, we choose q and I_{T} as bifurcation parameters, and the other parameters are fixed. From Proposition 1 in this paper, the lines to divide the relevant parameter plane are given as follows:

    l_{4}: = \left\{(q, I_{T})| I_{T} = I_T^* = \frac{q}{\beta p}+\frac{(\gamma-\frac{q}{p})^2 K}{4\gamma(d+\delta+r)}\right\},
    l_{5}: = \left\{(q, I_{T})| I_{T} = \frac{q}{\beta p}\right\}.

    The two solid lines l_{4} and l_{5} divide the q-I_{T} two-dimensional plane space into three regions in the first quadrant \mathbb{R}^{+} . Suppose that the control value I_{T} satisfies I_T > I_T^* (that is, region \Omega_{7} ; see Figure 9(a)), then system (3.1) has no equilibrium. If the control value I_{T} satisfies I_T^* > I_T > \frac{q}{\beta p} (that is, region \Omega_{6} , as shown in Figure 9(a)), system (3.1) has two positive equilibria E_{s1}^{+} = (S_{s1}^{+}, I_T) and E_{s1}^{-} = (S_{s1}^{-}, I_T) , where S_{s1}^{\pm} = \frac{(\gamma-\frac{q}{p})\pm\sqrt \Delta_1}{2\gamma}K . When the control value I_{T} satisfies 0 < I_T < \frac{q}{\beta p} (that is, region \Omega_{5} ; see Figure 9(a)), system (3.1) has a unique positive equilibrium E_{s2} = (S_{s2}, I_T) , where S_{s2} = \frac{(\gamma-\frac{q}{p})+\sqrt \Delta_1}{2\gamma}K .

    Figure 9.  The pseudo-equilibrium bifurcations, (a) where \beta = 0.7, p = 0.2, d = 0.01, \delta = 0.01, r = 0.01, \gamma = 0.2, K = 3 ; (b) where \beta = 0.7, p = 0.2, d = 0.01, \delta = 0.01, r = 0.01, q = 0.2, K = 3 .

    We choose \gamma and I_{T} as bifurcation parameters, and the other parameters are fixed. With Proposition 9 in this paper, the line to divide the relevant parameter plane is given as

    l_{6}: = \left\{(\gamma, I_{T})|I_{T} = I_{T}^{*'} = \frac{\gamma K}{4(d+\delta+r)}\right\}.

    The solid line l_{6} divides the \gamma-I_{T} two-dimensional plane space into two regions in the first quadrant \mathbb{R}^{+} . Suppose that the control value I_{T} satisfies I_T > I_T^{*'} (that is, region \Omega_{8} , as shown in Figure 9(b)), and then system (4.1) does not have an equilibrium. If the control value I_{T} satisfies 0 < I_T < I_T^{*'} (that is, region \Omega_{9} , as shown in Figure 9(b)), system (4.1) has two positive equilibria E_{s3}^{+} = (S_{s3}^{+}, I_T) and E_{s3}^{-} = (S_{s3}^{-}, I_T) , where S_{s3}^{\pm} = \frac{\gamma\pm\sqrt \Delta_2}{2\gamma}K .

    Remark 3. Notice that[30] considered the global dynamics of a Filippov predator-prey model with two thresholds for integrated pest management. By using Filippov theory, the sliding mode dynamics and global dynamics were established. Different from [30], our paper shows the dynamic behavior of the Filippov model with respect to all possible equilibria. It is shown that the Filippov system tends to the pseudo-equilibrium on sliding mode domain or one endemic equilibrium or two endemic equilibria under some conditions. Second, although both this paper and [30] discuss the global dynamics of a Filippov model with two thresholds, this paper first gives different control strategies. In particular, the two real equilibria occur simultaneously using methods such as Green's theorem and a Dulac function.

    In this paper, we have established a non-smooth system to determine whether it is necessary to adopt the control strategy of media coverage and quarantine of susceptible individuals according to the number of infected and susceptible individuals. Media coverage changes the transmission mode of influenza. Further, in order to reduce the spread of influenza, when the number of cases exceeds the larger infection threshold I_T , and the number of susceptible individuals is greater than S_T , we will quarantine the susceptible individuals. It is worth noting that there are two difficulties in this paper. First, the traditional continuity theory cannot be applied due to the non-smooth system with the broken line control strategy. For example, when proving the global stability of discontinuous systems, the traditional Lyapunov function cannot be similarly constructed. Second, Green's formula of continuous systems cannot be used to prove the existence of global stability of the pseudo equilibriums in discontinuous systems. In this paper, by choosing different thresholds I_{T} and S_{T} and using Filippov theory, we study the dynamic behavior of the Filippov model with respect to all possible equilibria. The regular/virtulal equilibrium bifurcations are given. It is shown that the Filippov system tends to the pseudo-equilibrium on sliding mode domain or one endemic equilibrium or two endemic equilibria under some conditions.

    Next, we summarize the corresponding biological results of Tables 13. These results show that the choice of values of I_T and S_T is very important, and it determines whether to adopt control strategies.

    Table 1.  Dynamics of Filippov model (2.4).
    Condition 1 Condition 2 Result
    S_T < S_1^* (S_T, I_T)\in B_{1-1} Figure 2(a)
    S_1^* < S_T < S_2^* (S_T, I_T)\in C_{1-1} Figure 4(a)
    S_2^* < S_T (S_T, I_T)\in D_{1-1} Figure 7(a)

     | Show Table
    DownLoad: CSV
    Table 2.  Dynamics of Filippov model (2.4).
    Condition 1 Condition 2 Result
    S_T < S_1^* (S_T, I_T)\in B_{2-1} Figure 2(b)
    (S_T, I_T)\in B_{3-1}\cup B_{3-2}\cup B_{3-3} Figure 2(c)
    S_1^* < S_T < S_2^* (S_T, I_T)\in C_{2-1} Figure 4(b)
    (S_T, I_T)\in C_{3-1} Figure 4(c)
    (S_T, I_T)\in C_{4-1} Figure 4(d)
    (S_T, I_T)\in C_{5-1}\cup C_{5-2}\cup C_{5-3}\cup C_{5-4} Figure 4(e)
    S_2^* < S_T (S_T, I_T)\in D_{3-1}\cup D_{3-2}D_{3-3} Figure 7(c)
    (S_T, I_T)\in D_{5-1} Figure 7(e)

     | Show Table
    DownLoad: CSV
    Table 3.  Dynamics of Filippov model (2.4).
    Condition 1 Condition 2 Result
    S_T < S_1^* (S_T, I_T)\in B_{4-1} Figure 2(d)
    (S_T, I_T)\in B_{5-1} Figure 2(e)
    S_1^* < S_T < S_2^* (S_T, I_T)\in C_{6-1} Figure 4(f)
    (S_T, I_T)\in C_{7-1} Figure 5(a)
    (S_T, I_T)\in C_{8-1}\cup C_{8-2}\cup C_{8-3}\cup C_{8-4} Figure 5(b)
    (S_T, I_T)\in C_{9-1} Figure 5(c)
    (S_T, I_T)\in C_{10-1}\cup C_{10-2} Figure 5(e)
    (S_T, I_T)\in C_{11-1}\cup C_{11-2} Figure 5(e)
    (S_T, I_T)\in C_{12-1} Figure 5(f)
    (S_T, I_T)\in C_{13-1} Figure 6
    S_2^* < S_T (S_T, I_T)\in D_{2-1} Figure 7(b)
    (S_T, I_T)\in D_{4-1} Figure 7(d)

     | Show Table
    DownLoad: CSV

    ● In Table 1, we know that the infection threshold value I_T is chosen to be small enough, i.e., I\ll I_{T} , and then the number of infected individuals will reach the equilibrium E_1^R of system (2.4).

    ● From Table 2, system (2.4) has a unique globally asymptotically stable pseudo-equilibrium E_{s1}^+ or E_{s3}^+ if I = I_T or admits a unique globally asymptotically stable equilibrium when I < I_T . Our control goal can be achieved finally, and there is no need to adjust the threshold strategy.

    ● In Table 3, the solution of system (2.4) will converge to a locally asymptotically stable equilibrium if I < I_{T} or tends to a locally asymptotically stable equilibrium E_{3}^{R} when I > I_{T} or pseudo-equilibrium E_{s1}^+, E_{s3}^+ if I = I_{T} . We show that it may be necessary to adjust the threshold policy according to the initial number of susceptible individuals and infected individuals. The results obtained have certain guiding significance for choosing thresholds and designing a corresponding threshold strategy.

    Next, we consider the effect of key parameters in the subsystem on the basic regeneration number R_{0i} as follows.

    The three-dimensional diagram of the parameter space (\delta, d, R_{01}) is shown in Figure 10(a) under the parameter values of K = 4 , \beta = 0.5 , r = 0.2 . It observe when the parameter \delta = 0.8, d increase from 0.83 to 1 , the basic reproduction number R_{01} decreases correspondingly and is less than unit 1. The trajectory of the subsystem (2.4) will converge globally to the free equilibrium (see Theorem 1), implying that the infected individuals extinct and then a stable free steady state occurs.

    Figure 10.  (a) Variation of the basic reproduction number R_{01} with the effect of parameters d and \delta . (b) Variation of R_{02} with the effect of parameters \beta and p . (c) Variation of R_{01} with the effect of parameters \beta and \delta . (d) Variation of R_{03} with the effect of parameters \beta and p .

    The three-dimensional diagram of the parameter space (\beta, p, R_{02}) is shown in Figure 10(b) under the parameter values of K = 2 , d = 0.1 , r = 0.05 , \delta = 0.05 . It is easy to observe when fixing the p = 0.5 , \beta increase from 0.78 to 1 , R_{02} decreases correspondingly and is less than unit 1. By using Theorem 1, the infected individuals persist and the trajectory of the subsystem (2.4) will converge globally to the endemic steady state.

    The three-dimensional diagram of the parameter space (\beta, \delta, R_{01}) is shown in Figure 10(c) under the parameter values of K = 2 , d = 0.1 , r = 0.1 . We observe when the parameters \delta, \beta increase from 0.6 to 1 , R_{01} also increases correspondingly and is greater than the unit 1. By using Theorem 1, the infected individuals persist and the trajectory of the subsystem (2.4) will converge globally to the endemic steady state.

    The three-dimensional diagram of the parameter space (\beta, p, R_{03}) is shown in Figure 10(d) under the parameter values of K = 1 , \gamma = 1.8 , r = \frac{1}{27} , \delta = \frac{1}{27} , d = \frac{1}{27} . It is easy to observe when fixing the parameter p = 0.2, with a transmission rate \beta increase from 0.81 to 1 , R_{03} increases correspondingly and is greater than the unit 1. By using Theorem 1, the infected individuals persist, and the trajectory of the subsystem (2.4) will converge globally to the endemic steady state. When fixing the parameters p = 0.6, transmission rate \beta increase from 0.81 to 1 , R_{03} decreases correspondingly and is less than the unit 1. The trajectory of the subsystem (2.4) will converge globally to the free equilibrium (see Theorem 1 of our paper), implying that the infected individuals become extinct, and then a stable free steady state occurs.

    In addition, this model has not been validated by actual influenza data, and we only have analyzed theoretically. In the next stage, we will verify and simulate the validity of the conclusions in actual time from some websites and statistics of health departments. However, the paper has studied the non-smooth system of two threshold control strategies and has validated the correctness of the theory through numerical simulation. Due to the serious lack of current influenza data from SARS-CoV-2 infections, the verification of the work is extremely difficult. However, the theory of this paper can provide appropriate guidance for the current influenza by SARS-CoV-2 infection. In this paper, we only consider the dynamics of the system (2.4) if the basic reproduction number R_{0i} > 1 . However, under the saturation rate \frac{\beta SI}{1+I} , and the conditions R_{02} < R_{01} < 1 and R_{02} < 1 < R_{01} , the dynamical behaviors of system (2.4) and the method of proving global stability are not yet fully clear and would be our further topic.

    We sincerely thank the anonymous referees for their very detailed and helpful comments on which improved the quality of this paper. This work is supported in part by the Yunnan Fundamental Research Projects (No: 202101BE070001-051).

    The authors have no conflict of interest to declare in carrying out this research work.



    [1] A. L. Barabási, R. Albert, H. Jeong, Scale-free characteristics of random networks: the topology of the World-Wide Web, Physica A, 281 (2000), 69–77. https://doi.org/10.1016/S0378-4371(00)00018-2 doi: 10.1016/S0378-4371(00)00018-2
    [2] R. Cohen, K. Erez, D. ben-Avraham, S. Havlin, Resilience of the Internet to random breakdowns, Phys. Rev. Lett., 85 (2000), 4626–4628. https://doi.org/10.1103/PhysRevLett.85.4626 doi: 10.1103/PhysRevLett.85.4626
    [3] H. A. Herrmann, J. M. Schwartz, Why COVID-19 models should incorporate the network of social interactions, Phys. Biol., 17 (2020), 065008. https://doi.org/10.1088/1478-3975/aba8ec doi: 10.1088/1478-3975/aba8ec
    [4] G. S. Hartnett, E. Parker, T. R. Gulden, R. Vardavas, D. Kravitz, Modelling the impact of social distancing and targeted vaccination on the spread of COVID-19 through a real city-scale contact network, J. Complex Netw., 9 (2021), cnab042. https://doi.org/10.1093/comnet/cnab042 doi: 10.1093/comnet/cnab042
    [5] D. Camacho, A. Panizo-LLedot, G. Bello-Orgaz, A. Gonzalez-Pardo, E. Cambria, The four dimensions of social network analysis: An overview of research methods, applications, and software tools, Inf. Fusion, 63 (2020), 88–120. https://doi.org/10.1016/j.inffus.2020.05.009 doi: 10.1016/j.inffus.2020.05.009
    [6] R. Pastor-Satorras, A. Vespignani, Epidemic spreading in scale-free networks, Phys. Rev. Lett., 86 (2001), 3200–3203. https://doi.org/10.1103/PhysRevLett.86.3200 doi: 10.1103/PhysRevLett.86.3200
    [7] D. Brockmann, D. Helbing, The hidden geometry of complex, network-driven contagion phenomena, Science, 342 (2013), 1337–1342. https://doi.org/10.1103/10.1126/science.1245200 doi: 10.1103/10.1126/science.1245200
    [8] E. N. Gilbert, Random graphs, Ann. Math. Statist., 30 (1959), 1141–1144. https://doi.org/10.1214/aoms/1177706098 doi: 10.1214/aoms/1177706098
    [9] P. Erdös, A. Rényi, On random graphs Ⅰ, Publ. Math. Debrecen, 6 (1959), 290–297.
    [10] P. Erdös, A. Rényi, On the evolution of random graphs, Publ. Math. Inst. Hungar. Acad. Sci., 5 (1960), 17–61.
    [11] R. Albert, A. L. Barabási, Statistical mechanics of complex networks, Rev. Mod. Phys., 74 (2002), 47–97. https://doi.org/10.1103/RevModPhys.74.47 doi: 10.1103/RevModPhys.74.47
    [12] M. E. J. Newman, The structure and function of complex networks, SIAM Rev., 45 (2003), 167–256. https://doi.org/10.1137/S003614450342480 doi: 10.1137/S003614450342480
    [13] S. Boccaletti, V. Latora, Y. Moreno, M. Chavez, D. U. Hwanga, Complex networks: Structure and dynamics, Phys. Rep., 424 (2006), 175–308. https://doi.org/10.1016/j.physrep.2005.10.009 doi: 10.1016/j.physrep.2005.10.009
    [14] F. Battiston, G. Cencetti, I. Iacopini, V. Latora, M. Lucas, A. Patania, et al., Networks beyond pairwise interactions: Structure and dynamics, Phys. Rep., 874 (2020), 1–92. https://doi.org/10.1016/j.physrep.2020.05.004 doi: 10.1016/j.physrep.2020.05.004
    [15] D. J. Watts, S. H. Strogatz, Collective dynamics of 'small-world' networks, Nature, 393 (1998), 440–442. https://doi.org/10.1038/30918 doi: 10.1038/30918
    [16] S. H. Strogatz, Exploring complex networks, Nature, 410 (2001), 268–276. https://doi.org/10.1038/35065725 doi: 10.1038/35065725
    [17] A. L. Barabási, R. Albert, Emergence of scaling in random networks, Science, 286 (1999), 509–512. https://doi.org/10.1126/science.286.5439.509 doi: 10.1126/science.286.5439.509
    [18] G. U. Yule, A mathematical theory of evolution, based on the conclusions of Dr. J. C. Willis, Philos. Trans. R. Soc. London Ser. B, 213 (1925), 21–87. https://doi.org/10.1098/rstb.1925.0002 doi: 10.1098/rstb.1925.0002
    [19] H. A. Simon, On a class of skew distribution functions, Biometrika, 42 (1955), 425–440. https://doi.org/10.1093/biomet/42.3-4.425 doi: 10.1093/biomet/42.3-4.425
    [20] D. J. S. Price, A general theory of bibliometric and other cumulative advantage processes, J. Amer. Soc. Inform. Sci., 27 (1976), 292–306. https://doi.org/10.1002/asi.4630270505 doi: 10.1002/asi.4630270505
    [21] G. Lima-Mendez, J. van Helden, The powerful law of the power law and other myths in network biology, Mol. Biosyst., 5 (2009), 1482–1493. https://doi.org/10.1039/b908681a doi: 10.1039/b908681a
    [22] W. Willinger, D. Alderson, J. C. Doyle, Mathematics and the Internet: A source of enormous confusion and great potential, Not. Am. Math. Soc., 56 (2009), 586–599.
    [23] A. D. Broido, A. Clauset, Scale-free networks are rare, Nat. Commun., 10 (2019), 1017. https://doi.org/10.1038/s41467-019-08746-5 doi: 10.1038/s41467-019-08746-5
    [24] C. Kasper, B. Voelkl, A social network analysis of primate groups, Primates, 50 (2009), 343–356. https://doi.org/10.1007/s10329-009-0153-2 doi: 10.1007/s10329-009-0153-2
    [25] S. Hennemann, B. Derudder, An alternative approach to the calculation and analysis of connectivity in the world city network, Environ. Plan. B-Plan. Des., 41 (2014), 392–412. https://doi.org/10.1068/b39108 doi: 10.1068/b39108
    [26] Y. L. Chuang, T. Chou, M. R. D'Orsogna, A network model of immigration: Enclave formation vs. cultural integration, Netw. Heterog. Media, 14 (2019), 53–77. https://doi.org/10.3934/nhm.2019004 doi: 10.3934/nhm.2019004
    [27] R. Munoz-Cancino, C. Bravo, S. A. Rios, M. Grana, On the combination of graph data for assessing thin-file borrowers' creditworthiness, Expert Syst. Appl., 213 (2023), 118809. https://doi.org/10.1016/j.eswa.2022.118809 doi: 10.1016/j.eswa.2022.118809
    [28] A. N. Licciardi Jr., L. H. A. Monteiro, A complex network model for a society with socioeconomic classes, Math. Biosci. Eng., 19 (2022), 6731–6742. https://doi.org/10.3934/mbe.2022317 doi: 10.3934/mbe.2022317
    [29] L. H. A. Monteiro, D. C. Paiva, J. R. C. Piqueira, Spreading depression in mainly locally connected cellular automaton, J. Biol. Syst., 14 (2006), 617–629. https://doi.org/10.1142/S0218339006001957 doi: 10.1142/S0218339006001957
    [30] P. H. T. Schimit, B. O. Santos, C. A. Soares, Evolution of cooperation in Axelrod tournament using cellular automata, Physica A, 437 (2015), 204–217. https://doi.org/10.1016/j.physa.2015.05.111 doi: 10.1016/j.physa.2015.05.111
    [31] P. H. T. Schimit, L. H. A. Monteiro, On the basic reproduction number and the topological properties of the contact network: An epidemiological study in mainly locally connected cellular automata, Ecol. Model., 220 (2009), 1034–1042. https://doi.org/10.1016/j.ecolmodel.2009.01.014 doi: 10.1016/j.ecolmodel.2009.01.014
    [32] H. A. L. R. Silva, L. H. A. Monteiro, Self-sustained oscillations in epidemic models with infective immigrants, Ecol. Complex., 17 (2014), 40–45. https://doi.org/10.1016/j.ecocom.2013.08.002 doi: 10.1016/j.ecocom.2013.08.002
    [33] L. H. A. Monteiro, D. M. Gandini, P. H. T. Schimit, The influence of immune individuals in disease spread evaluated by cellular automaton and genetic algorithm, Comput. Meth. Programs Biomed., 196 (2020), 105707. https://doi.org/10.1016/j.cmpb.2020.105707 doi: 10.1016/j.cmpb.2020.105707
    [34] K. Klemm, V. M. Eguiluz, Growing scale-free networks with small-world behavior, Phys. Rev. E, 65 (2002), 057102. https://doi.org/10.1103/PhysRevE.65.057102 doi: 10.1103/PhysRevE.65.057102
    [35] P. Holme, B. J. Kim, Growing scale-free networks with tunable clustering, Phys. Rev. E, 65 (2002), 026107. https://doi.org/10.1103/PhysRevE.65.026107 doi: 10.1103/PhysRevE.65.026107
    [36] Z. Z. Zhang, L. L. Rong, B. Wang, S. G. Zhou, J. H. Guan, Local-world evolving networks with tunable clustering, Physica A, 380 (2007), 639–650. https://doi.org/10.1016/j.physa.2007.02.045 doi: 10.1016/j.physa.2007.02.045
    [37] H. X. Yang, Z. X. Wu, W. B. Du, Evolutionary games on scale-free networks with tunable degree distribution, EPL, 99 (2012), 10006. https://doi.org/10.1209/0295-5075/99/10006 doi: 10.1209/0295-5075/99/10006
    [38] E. R. Colman, G. J. Rodgers, Complex scale-free networks with tunable power-law exponent and clustering, Physica A, 392 (2013), 5501–5510. https://doi.org/10.1016/j.physa.2013.06.063 doi: 10.1016/j.physa.2013.06.063
    [39] L. Wang, G. F. Li, Y. H. Ma, L. Yang, Structure properties of collaboration network with tunable clustering, Inf. Sci., 506 (2020), 37–50. https://doi.org/10.1016/j.ins.2019.08.002 doi: 10.1016/j.ins.2019.08.002
    [40] C. P. Warren, L. M. Sander, I. M. Sokolov, Geography in a scale-free network model, Phys. Rev. E, 66 (2002), 056105. https://doi.org/10.1103/PhysRevE.66.056105 doi: 10.1103/PhysRevE.66.056105
    [41] J. M. Kumpula, J. P. Onnela, J. Saramäki, K. Kaski, J. Kertész, Emergence of communities in weighted networks, Phys. Rev. Lett., 99 (2007), 228701. https://doi.org/10.1103/PhysRevLett.99.228701 doi: 10.1103/PhysRevLett.99.228701
    [42] Y. Murase, J. Török, H. H. Jo, K. Kaski, J. Kertész, Multilayer weighted social network model, Phys. Rev. E, 90 (2014), 052810. https://doi.org/10.1103/PhysRevE.90.052810 doi: 10.1103/PhysRevE.90.052810
    [43] S. Wolfram, Cellular automata and complexity: Collected papers, Westview Press, Boulder, 1994.
    [44] A. Landherr, B. Friedl, J. Heidemann, A critical review of centrality measures in social networks, Bus. Inf. Syst. Eng., 2 (2010), 371–385. https://doi.org/10.1007/s12599-010-0127-3 doi: 10.1007/s12599-010-0127-3
    [45] L. Ljung, System identification: Theory for the user, Prentice-Hall, Upper Saddle River, 1998.
    [46] A. Clauset, C. R. Shalizi, M. E. J. Newman, Power-law distributions in empirical data, SIAM Rev., 51 (2009), 661–703. https://doi.org/10.1137/070710111 doi: 10.1137/070710111
    [47] F. Liljeros, C. R. Edling, L. A. N. Amaral, H. E. Stanley, Y. Aberg, The web of human sexual contacts, Nature, 411 (2001), 907–908. https://doi.org/10.1038/35082140 doi: 10.1038/35082140
  • This article has been cited by:

    1. Wenjie Li, Yajuan Guan, Jinde Cao, Fei Xu, Global dynamics and threshold control of a discontinuous fishery ecological system, 2024, 182, 09600779, 114817, 10.1016/j.chaos.2024.114817
  • Reader Comments
  • © 2024 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(1470) PDF downloads(186) Cited by(1)

Article outline

Figures and Tables

Figures(4)  /  Tables(3)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog