Processing math: 100%
Research article Special Issues

Revisiting the classical target cell limited dynamical within-host HIV model - Basic mathematical properties and stability analysis


  • In this article, we reconsider the classical target cell limited dynamical within-host HIV model, solely taking into account the interaction between CD4+ T cells and virus particles. First, we summarize some analytical results regarding the corresponding dynamical system. For that purpose, we proved some analytical results regarding the system of differential equations as our first main contribution. Specifically, we showed non-negativity and boundedness of solutions, global existence in time and global uniqueness in time and examined stability properties of two possible equilibria. In particular, we demonstrated that the virus-free equilibrium and the plateau-phase equilibrium are locally asymptotically stable using the Routh–Hurwitz criterion under appropriate conditions. As our second main contribution, we underline our theoretical findings through some numerical experiments with standard Runge–Kutta time stepping schemes. We conclude this work with a summary of our main results and a suggestion of an extension for more complex dynamical systems with regard to HIV-infection.

    Citation: Benjamin Wacker. Revisiting the classical target cell limited dynamical within-host HIV model - Basic mathematical properties and stability analysis[J]. Mathematical Biosciences and Engineering, 2024, 21(12): 7805-7829. doi: 10.3934/mbe.2024343

    Related Papers:

    [1] Churni Gupta, Necibe Tuncer, Maia Martcheva . Immuno-epidemiological co-affection model of HIV infection and opioid addiction. Mathematical Biosciences and Engineering, 2022, 19(4): 3636-3672. doi: 10.3934/mbe.2022168
    [2] Nara Bobko, Jorge P. Zubelli . A singularly perturbed HIV model with treatment and antigenic variation. Mathematical Biosciences and Engineering, 2015, 12(1): 1-21. doi: 10.3934/mbe.2015.12.1
    [3] Zhaohui Yuan, Xingfu Zou . Global threshold dynamics in an HIV virus model with nonlinear infection rate and distributed invasion and production delays. Mathematical Biosciences and Engineering, 2013, 10(2): 483-498. doi: 10.3934/mbe.2013.10.483
    [4] A. M. Elaiw, N. H. AlShamrani . Analysis of an HTLV/HIV dual infection model with diffusion. Mathematical Biosciences and Engineering, 2021, 18(6): 9430-9473. doi: 10.3934/mbe.2021464
    [5] Cameron Browne . Immune response in virus model structured by cell infection-age. Mathematical Biosciences and Engineering, 2016, 13(5): 887-909. doi: 10.3934/mbe.2016022
    [6] Hongbin Guo, Michael Yi Li . Global dynamics of a staged progression model for infectious diseases. Mathematical Biosciences and Engineering, 2006, 3(3): 513-525. doi: 10.3934/mbe.2006.3.513
    [7] Andrew Omame, Sarafa A. Iyaniwura, Qing Han, Adeniyi Ebenezer, Nicola L. Bragazzi, Xiaoying Wang, Woldegebriel A. Woldegerima, Jude D. Kong . Dynamics of Mpox in an HIV endemic community: A mathematical modelling approach. Mathematical Biosciences and Engineering, 2025, 22(2): 225-259. doi: 10.3934/mbe.2025010
    [8] Yuyi Xue, Yanni Xiao . Analysis of a multiscale HIV-1 model coupling within-host viral dynamics and between-host transmission dynamics. Mathematical Biosciences and Engineering, 2020, 17(6): 6720-6736. doi: 10.3934/mbe.2020350
    [9] Shilian Xu . Saturated lysing efficiency of CD8+ cells induced monostable, bistable and oscillatory HIV kinetics. Mathematical Biosciences and Engineering, 2024, 21(10): 7373-7393. doi: 10.3934/mbe.2024324
    [10] Archana N. Timsina, Yuganthi R. Liyanage, Maia Martcheva, Necibe Tuncer . A novel within-host model of HIV and nutrition. Mathematical Biosciences and Engineering, 2024, 21(4): 5577-5603. doi: 10.3934/mbe.2024246
  • In this article, we reconsider the classical target cell limited dynamical within-host HIV model, solely taking into account the interaction between CD4+ T cells and virus particles. First, we summarize some analytical results regarding the corresponding dynamical system. For that purpose, we proved some analytical results regarding the system of differential equations as our first main contribution. Specifically, we showed non-negativity and boundedness of solutions, global existence in time and global uniqueness in time and examined stability properties of two possible equilibria. In particular, we demonstrated that the virus-free equilibrium and the plateau-phase equilibrium are locally asymptotically stable using the Routh–Hurwitz criterion under appropriate conditions. As our second main contribution, we underline our theoretical findings through some numerical experiments with standard Runge–Kutta time stepping schemes. We conclude this work with a summary of our main results and a suggestion of an extension for more complex dynamical systems with regard to HIV-infection.



    Today, approximately 40 million people are infected with the human immunodeficiency virus (HIV) and nearly 5.5 million people are unaware of it [1,2]. For that reason, research on this infectious disease without treatment still can be regarded as an important topic from a biological and clinical point of view.

    Since HIV was found to be the main reason for the acquired immune deficiency syndrome (AIDS), many modeling approaches have been explored over the course of the last decades to simulate its time development. At the end of the twentieth century, different approaches such as CD4+ T cell subpopulations [3], experimental data [4] or simpler fundamental models [5,6,7,8,9] were applied to better understand the dynamics of primary HIV infections. Reviews and fundamental models were proposed at the beginning of the twenty-first century [10,11,12]. Afterward, some works on global stability of fundamental models on viral dynamics were published [13,14,15]. Later, main fundamental models were reviewed in [16,17,18]. Furthermore, different aspects such as drug therapy or treatment can be implemented to obtain realistically dynamical models [19,20,21,22,23]. Furthermore, agent-based models can be applied as an alternative [24,25]. In this work, we specifically consider the well-known, classical target cell within-host HIV model

    dT(t)dt=rβV(t)T(t)dT(t)=:f1(t,T(t),Ti(t),V(t)),dTi(t)dt=βV(t)T(t)δTi(t)=:f2(t,T(t),Ti(t),V(t)),dV(t)dt=πTi(t)cV(t)=:f3(t,T(t),Ti(t),V(t)),T(0)=T0>0,Ti(0)=Ti,00,V(0)=V0>0,} (1.1)

    where all model parameters, also called constant problem parameters, and all variables, also known as solution components, are described in Table 1; quantities with index 0 represent initial conditions. We briefly want to remark that the state variable Ti(t) indicates infected CD4+ T cells and the index i reflects this infected state.

    Table 1.  Explanation of all problem constants and all solution components of (1.1).
    Symbol Meaning Unit
    t time day
    r constant production rate of target CD4+ T cells cells(μL)1(day)1
    β constant infection rate of target CD4+ T cells with HIV viral particles μL(virionsday)1
    d constant clearance rate of target CD4+ T cells (day)1
    δ constant clearance rate of infected target CD4+ T cells (day)1
    c constant elimination rate of HIV viral particles (day)1
    π constant replication rate of HIV viral particles virions(cellsday)1
    T(t) number of target CD4+ T cells (density in blood) cells(μL)1
    Ti(t) number of infected target CD4+ T cells (density in blood) cells(μL)1
    V(t) number of HIV viral particles (density in blood) virions(μL)1

     | Show Table
    DownLoad: CSV

    As later explained, this model accurately describes viral load during primary HIV infection in the acute phase. Additionally, our ideas for proofs of basic mathematical properties, given in the later part of this work, can be transferred to more complex models. For these reasons, we choose to mainly concentrate our mathematical examination on model (1.1), although we are aware of dynamical models including more tissue and mechanisms, as presented in our previous discussion.

    Throughout this article, all model parameters are assumed to be positive. Units are taken from the work by Alizon and Magnus [17]. Further, we want to shed some light on the dynamical system's structure. r represents a constant production rate of target CD4+ T cells while the linear term dT(t) stands for the constant elimination of CD4+ T cells due to non-disease natural reasons. The main non-linear term of this first-order, non-linear dynamical system is given by βT(t)V(t), which models the reduction or target CD4+ T cells by being infiltrated by virus particles. Consequently, δTi(t) and cV(t) model death or elimination processes of infected CD4+ T cells or of virions. Hence, system (1.1) describes a non-linear dynamical system of first order. Average lifetimes of infected CD4+ T cells and of virus particles are represented by 1d and 1c, as described by Nowak and Bangham [6]. For further details regarding biological interpretations, we refer interested readers to [5,6,7].

    Let us present a concise history on mathematical modeling in primary HIV infection and related fields. In 1994, Essunger and Perelson proposed a model of HIV infection of CD4+ T cell subpopulations [3]. Their main mathematical interest was the possible stability of steady states, also known as equilibrium points. In 1996, Kirschner presented one simplified model for primary HIV infection similar to model (1.1), but modified by a Michaelis-Menten mechanism [5]. In her work, she mainly numerically investigated steady states or linear growth of virus particles in time. In the same year, Nowak and Bingham suggested different models, including (1.1) in [6] and they also mainly investigated plateau-phase equlibrium points. For example, they extended our basic, i.e., very fundamental, model by immune responses to virus particles. One year later, Bonhoeffer and co-authors discussed model (1.1) and modified versions in which they emphasized the discussion of equilibria and numerical simulations [7] (compare page 6971 and especially the section about "A Basic Model" in this reference). Finzi and Siliciano also discussed equilibria for a simplified primary HIV-infection model [8]. In 1998, De Boer and Perelson investigated different models similar to (1.1) in [9]. However, they were mainly concerned with steady states. One year later, Perelson and Nelson primarly summarized analytical investigations on (1.1) regarding stability of equilibria [10]. In 2000, Stafford and co-workers used model (1.1) for parameter estimation problems [11]. Two years later, Perelson reviewed some classical mathematical models of primary HIV infection [12]. In 2004, Korobeinikov introduced different Lyapunov functions for global stability of simple epidemiological and virus dynamical models [13,14]. Additionally, Wang and Li proved global stability with respect to a modified mathematical model of primary HIV infection [15]. One year later, Ribeiro investigated the basic mathematical model (1.1) numerically [16]. In 2012 and 2013, Alizon, Magnus, Perelson and Ribeiro reviewed basic and more sophisticated mathematical models on dynamics of primary HIV infections including, for example, immune responses as seen in [6] or different virus strains. In recent years, stability analysis of more sophisticated models on HIV infections was developed [20,21,22]. Recently, Xu modified the basic model (1.1) by a constant CD8+ T cell density in blood and examined stability properties of the suggested model [23].

    The aforementioned articles mainly examined stability of the plateau-phase equilibrium states of mathematical models of HIV infections and the modeling of infection over time. Hence, we aim to provide proofs of existence and uniqueness of solutions of (1.1) globally in time as these are essential properties for biologically plausible models. Additionally, since this model (1.1) properly describes primary infection during the acute phase of HIV [11], it seems crucial to present thorough proofs of certain basic mathematical properties. For that reason, we need statements on boundedness and non-negativity of (1.1). Additionally, we want to derive all possible equilibria in a thorough manner. To the best of our knowledge, existence and uniqueness were not proven in the aforementioned articles. Furthermore, different properties such as boundedness, non-negativity, or derivation of the equilibrium points were only mentioned in the aforementioned works. Hence, our main goal is to collect these important properties with mathematical derivations.

    Although system (1.1) is one of the simplest models for primary HIV infection, many studies have focused on applications and modelling with respect to the disease's time development [4,5,10,12,18]. For that reason, we want to thoroughly investigate and re-examine system (1.1), since a detailed analysis can be seen as a preparatory step for future research. In previous investigations [26,27,28,29,30], different systems of differential equations were first analyzed and then numerical algorithms were developped and applied for its numerical solution.

    Our two main contributions can be summarized as follows:

    1) In Section 2, we mainly prove analytical results with respect to the dynamical system (1.1). Here, we begin with the non-negativity of possible solutions for all t0. Afterward, we demonstrate that all solution components remain bounded for all t0. To establish these results, we need to examine subsystems of (1.1), which is done in the proof of Lemma 3. In addition, we provide results for existence and uniqueness globally in time for all t0 in Sections 2.3 and 2.4. We conclude this section with a stability result for the virus-free equilibrium point and the plateau-phase disease equilibrium point based on the Routh–Hurwitz criterion. Here, the basic reproduction number occurs as a by-product of this criterion. In addition, we obtain the basic reproduction number by considering the approach of van den Driessche [31,32,33].

    2) By applying typical Runge–Kutta time stepping schemes in Section 3, we investigate our theoretical findings using numerical simulations and thus underline our results.

    To summarize, we aim to provide thorough proofs of analytical results in order to stress the biological usefulness of model (1.1).

    In this section, we prove some analytical results with respect to (1.1). Time-continuous solution components are assumed because of the dynamical system's structure, since all problem constants are positive and, as a consequence, no jumps in problem parameters exist. For additional interpretation, we refer our readers back to the introduction.

    We want to note that all right-hand side functions fj(t,T(t),Ti(t),V(t)) of (1.1) for each index j{1,2,3} are continuous with respect to all state variables. Since (1.1) can be equivalently written as an integral equation, all state variables are continuously differentiable functions with respect to time.

    Here, we discuss the non-negativity of possible solutions for system (1.1). This is of importance since only non-negative solution components of (1.1) have biological relevance.

    Lemma 2.1. All solution components of (1.1) remain non-negative for all t0.

    Proof. Let us assume that there might be a time where at least one solution component becomes negative. Due to the continuity of all solution components, there exists a time point t00 where T(t0)=0, Ti(t0)=0 or V(t0)=0 hold. Here, it is important to keep in mind that our initial conditions need to be non-negative as stated in (1.1).

    Let us first assume that T(t0)=0 holds while all other solution components are non-negative due to continuity. Then

    T(t0)=rβV(t0)T(t0)=0dT(t0)=0=r>0

    holds and this implies T(t0)>0.

    Now, let us assume that Ti(t0)=0 holds while all other solution components are non-negative due to continuity. Then

    Ti(t0)=βV(t0)T(t0)δTi(t0)=0=βV(t0)T(t0)0

    holds and this implies Ti(t0)0.

    Let us assume that V(t0)=0 holds while all other solution components are non-negative due to continuity. Then

    V(t0)=πTi(t0)cV(t0)=0=πTi(t0)0

    holds and this implies V(t0)0.

    Inductively, for later time points where at least one solution component is zero, we can apply the same argument. This means that no state variable can become negative. In conclusion, all solution components remain non-negative for all t0 due to continuity and non-negativity of initial conditions, which finishes our proof.

    In this subsection, we investigate the boundedness of all solution components of system (1.1). For this proof, we need one comparison principle from differential equations, stated in [34, Lemma 3.4].

    Lemma 2.2. Consider the scalar differential equation

    u(t)=f(t,u(t)),u(t0)=u0

    where f(t,u) is continuous in t and locally Lipschitz in u for all tt0 and all uJR. Let [t0,T) (T could be infinity) be the maximal interval of existence of the solution u(t). Suppose u(t)J for all t[t0,T). Let v(t) be a continuous functions whose upper right-hand derivative D+v(t) satisfies the differential inequality

    D+v(t)f(t,v(t)),v(t0)u0

    with v(t)J for all t[t0,T). Then, it holds v(t)u(t) for all t[t0,T).

    Now, we are able to prove the boundedness of all states variables of (1.1).

    Lemma 2.3. All solution components of (1.1) remain bounded for all t0.

    Proof. 1) Let us first consider T(t)=rβV(t)T(t)dT(t). Due to Lemma 2.1, we conclude that

    T(t)=rβV(t)T(t)dT(t)=r(βV(t)+d)T(t)rdT(t)

    holds. Consequently, by application of the comparison principle from Lemma 2.2 and by non-negativity, we notice that

    0T(t)rd+(T0rd)exp(dt)

    is valid for all t0 because all assumptions of Lemma 2.2 are fulfilled in this situation. Finally, this implies

    0T(t)max{rd,T0}

    for all t0.

    2) By investigating the subsystem

    T(t)=rβV(t)T(t)dT(t),Ti(t)=βV(t)T(t)δTi(t)} (2.1)

    of (1.1) and defining

    M(t)=T(t)+Ti(t) (2.2)

    as the complete number of target CD4+ T cells, we obtain the following differential equation

    M(t):=(T(t)+Ti(t))=T(t)+Ti(t)=(rβV(t)T(t)dT(t))+(βV(t)T(t)δTi(t))=rdT(t)δTi(t)rmin{d,δ}(T(t)+Ti(t))=rmin{d,δ}M(t)

    with initial condition M0:=T0+Ti,0. Hence, it follows

    0M(t)rmin{d,δ}+(M0rmin{d,δ})exp(min{d,δ}t) (2.3)

    for all t0, which implies the validity of

    0M(t)max{rmin{d,δ},M0} (2.4)

    for all t0.

    3) Due to the boundedness of T(t) and Ti(t) for all t0, we obtain the following differential inequality

    V(t):=πTi(t)cV(t)πmax{rmin{d,δ},M0}cV(t)

    and we can conclude

    0V(t)πcmax{rmin{d,δ},M0}+(V0πcmax{rmin{d,δ},M0})exp(ct) (2.5)

    for all t0, which has

    0V(t)max{πcmax{rmin{d,δ},M0},V0} (2.6)

    for all t0 as a consequence.

    This proves our assertion that all solution components of (1.1) remain bounded for all t0. This finishes our proof.

    Here, we want to give one statement from [35, Theorem 4.7.1] or [36, Theorem 2.2]. We consider a general initial-value problem

    z(t)=G(t,z(t)),z(0)=z0} (2.7)

    where z(t)=(z1(t),,zn(t))T denotes our solution vector, the vectorial function is given by G(t,z(t))=(g1(t,z(t)),,gn(t,z(t)))T, and initial conditions are given by z0Rn. By Rn, we denote a suitable vector norm on Rn. Here, we apply the supremum norm

    F(x):=supj=1,,nsupxRn|fj(x)|

    on the space of bounded, continuous functions on Rn since this leads to a Banach space.

    Theorem 2.1. If G:[0,)×RnRn is locally Lipschitz-continuous in both time and state variables and if there are non-negative real functions D:[0,)[0,) and K:[0,)[0,) such that

    G(t,z(t))RnK(t)z(t)Rn+D(t)

    holds for all z(t)Rn, then the solution of the initial-value problem (2.7) exists for all t0. Moreover, for every finite T0, we have

    z(t)Rnz0Rnexp(Kmax|t|)+DmaxKmax(exp(Kmax|t|)1)

    for all t[0,T] where

    Dmax=max0sT|D(s)|andKmax=max0sT|K(s)|

    are described.

    Now, we are able to show the global existence of all solution components for our dynamical system in (1.1) for all t0. Existence is one main property that reliable models in natural sciences should fulfill. This also holds for uniqueness, later analyzed in this work.

    Theorem 2.2. There exists a solution of (1.1) globally in time for all t0.

    Proof. We define

    G(t,T(t),Ti(t),V(t)):=(rβV(t)T(t)dT(t)βV(t)T(t)δTi(t)πTi(t)cV(t))=:(g1(t,T(t),Ti(t),V(t))g2(t,T(t),Ti(t),V(t))g3(t,T(t),Ti(t),V(t)))

    for our vectorial function of (1.1).

    1) At first, we prove Lipschitz-continuity locally for g1(t,T(t),Ti(t),V(t)) because the other component functions are estimated in a similar manner. By the boundedness of all solution components by Lemma 2.3, we can assume that

    Tsup:=supt0|T(t)|;Ti,sup:=supt0|Ti(t)|andVsup:=supt0|V(t)|

    exist. We obtain

    g1(t,T1(t),T1,i(t),V1(t))g1(t,T2(t),T2,i(t),V2(t))=(βV1(t)T1(t)dT1(t))(βV2(t)T2(t)dT2(t))dT1(t)T2(t)+βV1(t)T1(t)V2(t)T2(t)=dT1(t)T2(t)+βV1(t)T1(t)V1(t)T2(t)+V1(t)T2(t)V2(t)T2(t)dT1(t)T2(t)+βV1(t){T1(t)T2(t)}+βT2(t){V1(t)V2(t)}.

    This result implies

    g1(t,T1(t),T1,i(t),V1(t))g1(t,T2(t),T2,i(t),V2(t))dT1(t)T2(t)+βV1(t){T1(t)T2(t)}+βT2(t){V1(t)V2(t)}dT1(t)T2(t)+βVsupT1(t)T2(t)+βTsupV1(t)V2(t)(d+β{Vsup+Tsup})(T1(t)T2(t),T1,i(t)T2,i(t),V1(t)V2(t)).

    As a consequence, we conclude that our defined vectorial function of (1.1) is locally Lipschitz-continuous.

    2) By our assumptions, we obtain

    g1(t,T(t),Ti(t),V(t))=rβV(t)T(t)dT(t)r+βV(t)max{rd,T0}+dmax{rd,T0}(r+dmax{rd,T0})=:A1+(βmax{rd,T0})=:B1(T(t),Ti(t),V(t))=A1+B1(T(t),Ti(t),V(t))

    for the first vectorial component,

    g2(t,T(t),Ti(t),V(t))=βV(t)T(t)δTi(t)βmax{rd,T0}V(t)+δTi(t)(βmax{rd,T0}+δ)=:B2(T(t),Ti(t),V(t))

    for the second vectorial component and

    g3(t,T(t),Ti(t),V(t))=πTi(t)cV(t)πTi(t)+cV(t)(π+c)=:B3(T(t),Ti(t),V(t))

    for the final vectorial component. Set

    A:=A1andB:=max{B1,B2,B3}.

    Furthermore, we see that

    G(t,T(t),Ti(t),V(t))A+B(T(t),Ti(t),V(t))

    is valid for all t0. This inequality implies the existence of all solution components globally in time for all t0 due to the application of Theorem 2.1.

    Hence, our proof is complete.

    To show the uniqueness of system (1.1), we need Banach's fixed point theorem; compare [37, Theorem Ⅴ.18].

    Theorem 2.3. Let (X,ϱ) be a complete metric space with metric ϱ:X×X[0,). Let T:XX be a strict contraction, i.e., there exists a constant K[0,1) such that ϱ(Tx,Ty)Kϱ(x,y) holds for all x,yX. Then, the mapping T has a unique fixed point in X.

    Now, we are able to formulate our statement on uniqueness of system (1.1) globally in time.

    Theorem 2.4. System (1.1) possesses a unique solution globally in time for all t0.

    Proof. We first define the equivalent system of integral equations

    T(t)=T0+t0{rβV(s)T(s)dT(s)}ds,Ti(t)=Ti,0+t0{βV(s)T(s)δTi(s)}ds,V(t)=V0+t0{πTi(s)cV(s)}ds} (2.8)

    to system (1.1) for the application of Banach's fixed point theorem.

    1) The function space of bounded, continuous functions on the interval [0,) is a complete metric space with the supremum norm.

    2) First, we estimate

    T1(τ1)T2(τ1)βτ10{V2(s)T2(s)V1(s)T1(s)}ds+dτ10{T2(s)T1(s)}dsβτ1TsupV1(s)V2(s)+βτ1VsupT1(s)T2(s)+dτ1T1(s)T2(s)τ1(βTsup+βVsup+d)(T1(s)T2(s),T1,i(s)T2,i(s),V1(s)V2(s))

    by the boundedness of all solution components. If we choose τ112(βTsup+βVsup+d), we obtain

    T1(τ1)T2(τ1)12(T1(s)T2(s),T1,i(s)T2,i(s),V1(s)V2(s)).

    3) Second, we see that

    T1,i(τ2)T2,i(τ2)=τ20{βV1(s)T1(s)δT1,i(s)}dsτ20{βV2(s)T2(s)δT2,i(s)}dsβτ2TsupV1(s)V2(s)+βτ2VsupT1(s)T2(s)+δτ2T1,i(s)T2,i(s)τ2(βTsup+βVsup+δ)(T1(s)T2(s),T1,i(s)T2,i(s),V1(s)V2(s))

    holds by the boundedness of all solution components. If we choose τ212(βTsup+βVsup+δ), we obtain

    T1,i(τ2)T2,i(τ2)12(T1(s)T2(s),T1,i(s)T2,i(s),V1(s)V2(s)).

    4) Third, we notice that

    V1(τ3)V2(τ3)πτ3T1,i(s)T2,i(s)+cτ3V1(s)V2(s)τ3(π+c)(T1(s)T2(s),T1,i(s)T2,i(s),V1(s)V2(s))

    is valid by the boundedness of all solution components. If we choose τ312(π+c), we obtain

    V1(τ3)V2(τ3)12(T1(s)T2(s),T1,i(s)T2,i(s),V1(s)V2(s)).

    5) If we choose τmin{τ1,τ2,τ3}, we get

    (T1(s)T2(s),T1,i(s)T2,i(s),V1(s)V2(s))12(T1(s)T2(s),T1,i(s)T2,i(s),V1(s)V2(s))

    as an estimate. Hence, system (1.1) has a unique fixed point on [0,τ].

    Inductively, we can conclude that this fixed point is unique on every interval [kτ,(k+1)τ] for all k{0}N. This implies the uniqueness of a solution of (1.1) globally in time for all t0.

    Denote possible equilibrium states by (T,Ti,V). From (1.1), we obtain the system of equations

    rβVTdT=0,βVTδTi=0,πTicV=0} (2.9)

    for possible equilibrium states.

    Lemma 2.4. The two possible equilibrium points read

    (T1,T1,i,V1)=(rd,0,0)and(T2,T2,i,V2)=(cδβπ,βπrcdδβδπ,βπrcdδβcδ). (2.10)

    Proof. We want to split our proof into two parts.

    1) We can easily check by plugging

    (T1,T1,i,V1)=(rd,0,0)

    into (2.9) that this point is definitely one possible equilibrium state.

    2) The third equation of (2.9) yields

    V=πcTiorTi=cπV. (2.11)

    Considering the second equation of (2.9), we obtain

    βVTδTi=0βπcTiTδTi=0Ti{βπcTδ}=0

    and this implies

    T=cδβπ. (2.12)

    Looking at the first equation of (2.9), we get

    rβVTdT=0r=Vβcδβπ+cdδβπrcdδβπ=Vcδπ(rcdδβπ)πcδ=Vπrcδdβ=V

    and it follows

    V=βπrcdδβcδ. (2.13)

    As a consequence, it holds

    Ti=cπV=cπ(βπrcdδβcδ)=βπrcdδβδπ

    and this results in

    Ti=βπrcdδβδπ. (2.14)

    Hence, the second possible equilibrium state reads

    (T2,T2,i,V2)=(cδβπ,βπrcdδβδπ,βπrcdδβcδ)

    This proves our proposition.

    We note that these two equilibrium points are the same as solely mentioned in [11]. Here, we give one statement for locally asymptotic stable equilibria of an autonomous dynamical system

    x(t)=G(x(t))

    from [35, Theorem 6.1.1].

    Theorem 2.5. Suppose that b is an equilibrium point for x(t)=G(x(t)) where GC1(U) with a domain URd. Furthermore, we assume that

    (λj(DG))<0

    holds for all j{1,2,,d} where DG is the Jacobian of G. Then, there is a neighborhood V of b in Rd such that, for any initial data bV, the initial value problem

    x(t)=G(x(t))withx(0)=b

    has a solution for all t0 and limtx(t)=b.

    We consider a special case of the Routh–Hurwitz criterion; compare [38].

    Lemma 2.5. Consider the characteristic equation

    a0λ3+a1λ2+a2λ+a3=0

    of a corresponding Jacobian of the linearization of a dynamical system. Its eigenvalues all have negative real parts and it is locally asymptotically stable if and only if

    aj>0for allj{0,1,2,3}anda1a2a0a3>0. (2.15)

    Hence, we can prove that the virus-free equilibrium state (T1,T1,i,V1) and the plateau-phase equilibrium state (T2,T2,i,V2) from Lemma 2.4 are locally asymptotically stable. For that reason, we define the basic reproduction number R0:=βπrcdδ of primary HIV infections, which can be regarded as a transition point from the virus-free to the plateau-phase equilibrium state. In [6], this threshold was only defined. Here, we want to give a derivation based on an approach by van den Driessche [32]. Fore more details, we refer interested readers to that article. We mainly follow concise ideas from [33] for a within-host model of COVID-19.

    We reorganize (1.1) as follows

    Ti(t)=βV(t)T(t)δTi(t),V(t)=πTi(t)cV(t),T(t)=rβV(t)T(t)dT(t)

    where we consider the subsystem

    Ti(t)=βV(t)T(t)δTi(t),V(t)=πTi(t)cV(t)

    of infected and viral particles. We define two vectors

    F(Ti(t),V(t))=(βV(t)T(t)0)andV(Ti(t),V(t))=(δTi(t)cV(t)πTi(t))

    such that

    (Ti(t)V(t))=F(Ti(t),V(t))V(Ti(t),V(t))

    holds. For the approach by van den Driessche, one needs the virus-free equilibrium state

    (T1,T1,i,V1)=(rd,0,0)

    as later given in Theorem 2.6. By computing both Jacobians

    F=(0βT100)andV=(δ0πc)

    of these vectorial functions at this virus-free equilibrium state, the basic reproduction number is given by

    R0=ϱ(F(V)1).

    Here, ϱ defines the spectral radius of the considered matrix. Since we obtain

    (V)1=1cδ(c0πδ)andF(V)1=(βπcδT1βcT100),

    this yields

    R0=ϱ(F(V)1)=βπcδT1=βπrcδd.

    This basic reproduction number helps us to determine and distinguish the stability of equilibrium states. More specifically, if R0<1 holds, the disease's progress settles to the virus-free equilibrium state while it settles in the disease-plateau-phase equilibrium state if R0>1 holds. This threshold is also an important number in mathematical epidemiology for a disease's spread [13,14]. Additionally, although Korobeinikov proved the global stability [13,14], we want to show locally asymptotic stability by using the Routh–Hurwitz criterion, which, to the best of our knowledge, cannot be found in the aforementioned articles.

    Theorem 2.6. (i) Suppose that

    βπrcdδ<0R0<1

    holds. The virus-free equilibrium state

    (T1,T1,i,V1)=(rd,0,0)

    of our dynamical system

    x(t):=(T(t)Ti(t)V(t))=(rβV(t)T(t)dT(t)βV(t)T(t)δTi(t)πTi(t)cV(t))=:G(x(t))

    from (1.1) is locally asymptotically stable.

    (ii) Suppose that

    βπrcdδ>0R0>1

    holds. The plateau-phase equlibrium state

    (T2,T2,i,V2)=(cδβπ,βπrcdδβδπ,βπrcdδβcδ)=(cδβπ,cdδ(R01)βδπ,cdδ(R01)βcδ)

    of our dynamical system

    x(t):=(T(t)Ti(t)V(t))=(rβV(t)T(t)dT(t)βV(t)T(t)δTi(t)πTi(t)cV(t))=:G(x(t))

    from (1.1) is locally asymptotically stable.

    Proof. We divide our proof into multiple steps.

    1)$ Let

    G(T,Ti,V)=(rβVTdTβVTδTiπTicV).

    Its Jacobian reads

    DG(T,Ti,V)=(βVd0βTβVδβT0πc).

    2) We compute the characteristic equation of the previous Jacobian. At first, we obtain

    det(DG(T,Ti,V)λI3×3)=det(βVdλ0βTβVδλβT0πcλ)=(βVdλ){(δλ)(cλ)πβT}βT{βVπ}=(λ+βV+d)(λ+δ)(λ+c)+(βV+d+λ)πβTβTβVπ=(λ+βV+d)(λ+δ)(λ+c)+(λ+d)πβT=0.

    This yields

    λ3+λ2{c+δ+βV+d}+λ{cδ+βV(c+δ)+(c+δ)dπβT}+{βVcδ+cdδβdπT}=0.

    3)ⅰ. Let us first consider the virus-free equilibrium state (T1,T1,i,V1) and let R0<1. We can define

    a0:=1>0,a1:=c+δ+βV1+d=c+δ+d>0,a2:=cδ+βV1(c+δ)+(c+δ)dπβT1,a3:=βV1cδ+cdδβdπT1.

    Obviously, a0 and a1 are positive. By plugging in the definition of the virus-free equilibrium point, we conclude that

    a2:=cδ+(c+δ)dπβrd=cδdπβrd+(c+δ)d=cdδ(1R0)d>0+(c+δ)d>0,a3:=cdδβπr=cdδ(1R0)>0>0,

    and

    a1a2a0a3:=(c+d+δ)((c+δ)d+cdδβπrd)(cdδβπr)=(c+d+δ)(c+δ)d+(c+d+δ)cdδ(1R0d)+βπrcdδ(c+d+δ)cdδ(1R0d)>0+βπr>0

    hold. Hence, if R0<1 is valid, we have locally asymptotic stability by the Routh–Hurwitz criterion.

    3)ⅱ. Here, we consider the plateau-phase equilibrium state (T2,T2,i,V2) and let R0>1. Now, we can again define

    a0:=1>0,a1:=c+δ+βV2+d>0,a2:=cδ+βV2(c+δ)+(c+δ)dπβT2,a3:=βV2cδ+cdδβdπT2.

    Obviously, a0 and a1 are positive. By plugging in the definition of the plateau-phase equilibrium point, we conclude that

    a2=cδ+βV2(c+δ)+(c+δ)dπβT2=cδ+β(βπrcdδβcδ)(c+δ)+(c+δ)dπβcδβπ=β(βπrcdδβcδ)(c+δ)+(c+δ)d=βcdδ(R01βcδ)>0(c+δ)+(c+δ)d>0

    and

    a3=βV2cδ+cdδβdπT2=β(βπrcdδβcδ)cδ+cdδβdπcδβπ=β(βπrcdδβcδ)cδ=βcdδ(R01βcδ)>0cδ>0

    hold. Hence, all coefficients aj are positive for all j{0,1,2,3}. Since we want to apply the Routh–Hurwitz criterion from Lemma 2.5, we still need to show

    a1a2a0a3>0.

    We obtain

    a1a2a0a3=(c+δ+β(βπrcdδβcδ)+d)(β(βπrcdδβcδ)(c+δ)+(c+δ)d)1(β(βπrcdδβcδ)cδ)=(c+β(βπrcdδβcδ)+d)(β(βπrcdδβcδ)(c+δ)+(c+δ)d)+δ(β(βπrcdδβcδ)(c+δ)+(c+δ)d)(β(βπrcdδβcδ)cδ)>(c+β(βπrcdδβcδ)+d)(β(βπrcdδβcδ)(c+δ)+(c+δ)d)>0

    because

    (βπrcdδβcδ)=cdδ(R01)βcδ>0

    holds, which shows this assertion.

    Since all assumptions of the Routh–Hurwitz criterion from Lemma 2.5 are fulfilled, its application for both cases yields the desired stability results and finishes our proof.

    We give two numerical examples to provide evidence for our theoretical findings. First, we consider the case R0>1 with all model parameters taken from [11,17], which means that we obtain the plateau-phase equilibrium point. Additionally, we also provide the case R0<1, which corresponds to the virus-free equilibrium state.

    In this subsection, we apply the following initial conditions and constant problem parameters, taken from [11,17] and presented in Table 2. We want to remark again that all constant problem parameters are assumed to be positive. In addition, we note that the initial conditions of (1.1) are non-negative for numerical simulations. Taking all values of Table 2, we can compute the basic reproduction number R0, which reads

    R0:=βπrcdδ=6.51048500.1730.010.398.0278>1

    in this case. Hence, R0>1 holds and we expect an plateau-phase equilibrium state.

    Table 2.  Values for initial conditions and constant problem parameters for numerical simulation of (1.1) where time t is measured in days and R08.0278>1.
    Constant Value
    T0 10cellsμL
    Ti,0 0cellsμL
    V0 106virionsμL
    r 0.17cellsμLday
    β 6.5104μLvirionsday
    d 0.011day
    δ 0.391day
    π 850virionscellsday
    c 31day

     | Show Table
    DownLoad: CSV

    Here, we use the standard function ode45 of GNU Octave [39]. For further information regarding these modified Runge-Kutta time stepping methods, we refer interested readers to [40,41]. Our GNU Octave code can be found in the supplementary file for reproducibility. We must note that we shortened all time vectors and all solution components vectors for the plotting of the figure due to representation problems on the author's computer.

    Using the given initial conditions and problem parameters from Table 2, we obtain

    T1=2.1176,T1,i=0.3816,V1=108.12

    for the coordinates of the plateau-phase equilibrium point. Simulation results of (1.1) with parameters from Table 2 can be seen in Figure 1. Here, we can see that the model of primary HIV infection converges towards the plateau-phase disease equilibrium point after a certain amount of time. This shows that system (1.1) is especially appropriate for the acute and asymptotic phase of HIV infections [5,17] and it settles into the correct equilibrium state as we can see in Figure 1. Additionally, we also notice that our theoretical results of boundedness and non-negativity for all solution components of system (1.1) hold in numerical simulations. However, we address that preservation of boundedness or non-negativity is not intrinsically fulfilled by explicit time integration methods; compare with [30].

    Figure 1.  Simulation results of model (1.1) with initial conditions and constant problem parameters taken from Table 2. Here, it holds that R0>1. Dashed lines represent corresponding equilibrium variables T, Ti, and V.

    Here, we present a case for R0<1. Let us take the following model parameters as presented in Table 3. It holds that

    R0:=βπrcdδ=6.51048500.1730.10.390.8028<1

    and we notice that the graphs of all solution components converge to the correct equilibrium point. Using all model parameters from Table 3, we obtain

    (T2,T2,i,V2)=(1.7000,0,0)
    Table 3.  Values for initial conditions and constant problem parameters for numerical simulation of (1.1) where time t is measured in days and R0<1.
    Constant Value
    T0 10cellsμL
    Ti,0 0cellsμL
    V0 106virionsμL
    r 0.17cellsμLday
    β 6.5104μLvirionsday
    d 0.11day
    δ 0.391day
    π 850virionscellsday
    c 31day

     | Show Table
    DownLoad: CSV

    as the correct virus-free equilibrium state, as seen in Figure 2.

    Figure 2.  Simulation results of model (1.1) with initial conditions and constant problem parameters taken from Table 3. Here, R00.8028<1. Dashed lines represent corresponding equilibrium variables T, Ti, and V.

    In this work, we examined and re-investigated analytical properties of the classical target cell limited dynamical within-host HIV model (1.1). Here, we focused on important facts such as non-negativity, boundedness, global existence and global uniqueness in time for all solution components. These are all important properties regarding biological relevance of model (1.1). Furthermore, we showed that the virus-free equilibrium point and the plateau-phase disease equilibrium point of (1.1) are locally asymptotically stable. We proved that they can be distinguished by the basic reproduction number R0:=βπrcdδ. Finally, we highlighted our thereotical findings by numerical simulations, based on Runge–Kutta time stepping methods, and demonstrated that our results hold true in both cases R0<1 and R0>1.

    We want to remark that our basic model (1.1) does not include more complex factors of HIV infections such as different virus strains, immune responses, or other cells such as CD8+ T cells. Hence, it could be of interest to examine the analytical properties of more complex models [5,17] that include additional aspects of HIV infections. The model in the work by Kirschner [5] is better suited for long-time modeling of HIV infection, since it also includes the rapid decline of CD4+ T cells approximately ten years after infection. It is also worth noting that classical explicit time stepping schemes do not preserve boundedness or non-negativity [30,42]. These methods are based on a methodology proposed by Mickens [43]. Hence, the development of non-negativity-preserving time integration methods for model (1.1) might be a future research direction, adding to this research article, where we mainly focused on analytical aspects.

    The author declares that he has not used Artificial Intelligence (AI) tools in the creation of this article.

    The author declares there is no conflict of interest.



    [1] A. Mody, A. H. Sohn, C. Iwuji, R. K. J. Tan, F. Venter, E. H. Geng, HIV epidemiology, prevention, treatment, and implementation strategies for public health, Lancet, 402 (2024), 471–492. https://doi.org/10.1016/S0140-6736(23)01381-8 doi: 10.1016/S0140-6736(23)01381-8
    [2] UNAids, Fact Sheet 2024 - Global HIV Statistics, 2024. Available from: https://www.unaids.org/sites/default/files/media_asset/UNAIDS_FactSheet_en.pdf.
    [3] P. Essunger, A. S. Perelson, Modeling HIV infection of CD4+ T-cell subpopulations, J. Theor. Biol., 170 (1994), 367–391. https://doi.org/10.1006/jtbi.1994.1199 doi: 10.1006/jtbi.1994.1199
    [4] D. D. Ho, A. U. Neumann, A. S. Perelson, W. Chen, J. M. Leonard, M. Markowitz, Rapid turnover of plasma virions and CD4 lymphocytes in HIV-1 infection, Nature, 373 (1995), 123–126. https://doi.org/10.1038/373123a0 doi: 10.1038/373123a0
    [5] D. Kirschner, Using mathematics to understand HIV immune dynamics, Not. AMS, 43 (1996), 191–202.
    [6] M. A. Nowak, C. R. M. Bangham, Population dynamics of immune responses to persistent viruses, Science, 272 (1996), 74–79. https://doi.org/10.1126/science.272.5258.74 doi: 10.1126/science.272.5258.74
    [7] S. Bonhoeffer, R. M. May, G. M. Shaw, M. A. Nowak, Virus dynamics and drug therapy, Proc. Natl. Acad. Sci., 94 (1997), 6971–6976. https://doi.org/10.1073/pnas.94.13.6971 doi: 10.1073/pnas.94.13.6971
    [8] D. Finzi, R. F. Siliciano, Viral dynamics in HIV-1-infection, Cell, 93 (1997), 665–671. https://doi.org/10.1016/s0092-8674(00)81427-0 doi: 10.1016/s0092-8674(00)81427-0
    [9] R. J. De Boer, A. S. Perelson, Target cell limited and immune control models of HIV infection: A comparison, J. Theor. Biol., 190 (1998), 201–214. https://doi.org/10.1006/jtbi.1997.0548 doi: 10.1006/jtbi.1997.0548
    [10] A. S. Perelson, P. W. Nelson, Mathematical analysis of HIV-1-dynamics in vivo, SIAM Rev., 41 (1999), 3–44. https://doi.org/10.1137/S0036144598335107 doi: 10.1137/S0036144598335107
    [11] M. A. Stafford, L. Corey, Y. Cao, E. S. Daar, D. D. Ho, A. S. Perelson, Modeling plasma virus concentration during primary HIV infection, J. Theor. Biol., 203 (1999), 285–301. https://doi.org/10.1006/jtbi.2000.1076 doi: 10.1006/jtbi.2000.1076
    [12] A. S. Perelson, Modelling viral and immune system dynamics, Nat. Rev. Immunol., 2 (1999), 28–36. https://doi.org/10.1038/nri700 doi: 10.1038/nri700
    [13] A. Korobeinikov, A Lyapunov function and global properties for SIR and SEIR epidemiological Models with Nonlinear Incidence, Math. Biosci. Eng., 1 (2004), 57–60. https://doi.org/10.3934/mbe.2004.1.57 doi: 10.3934/mbe.2004.1.57
    [14] A. Korobeinikov, Global properties of basic virus dynamics models, Bull. Math. Biol., 66 (2004), 879–883. https://doi.org/10.1016/j.bulm.2004.02.001 doi: 10.1016/j.bulm.2004.02.001
    [15] L. Wang, M. Y. Li, Mathematical analysis of the global dynamics of a model for HIV infection of CD4+ T cells, Math. Biosci., 200 (2006), 44–57. https://doi.org/10.1016/j.mbs.2005.12.026 doi: 10.1016/j.mbs.2005.12.026
    [16] R. M. Ribeiro, Dynamics of CD4+ T cells in HIV-1 infection, Immunol. Cell Biol., 85 (2006), 287–294. https://doi.org/10.1038/sj.icb.7100056 doi: 10.1038/sj.icb.7100056
    [17] S. Alizon, C. Magnus, Modelling the course of an HIV infection: Insights from ecology and evolution, Viruses, 4 (2012), 1984–2013. https://doi.org/10.3390/v4101984 doi: 10.3390/v4101984
    [18] A. S. Perelson, R. M. Ribeiro, Modeling the within-host dynamics of HIV infection, BMC Biol., 11 (2013), 96. https://doi.org/10.1186/1741-7007-11-96 doi: 10.1186/1741-7007-11-96
    [19] D. Kirschner, G. Webb, Immunotherapy of HIV-1 infection, J. Biol. Syst., 6 (1998), 71–83. https://doi.org/10.1142/S0218339098000091 doi: 10.1142/S0218339098000091
    [20] A. Mojaver, H. Kheiri, Mathematical analysis of a class of HIV infection models of CD4+ T-cells with combined antiretroviral therapy, Appl. Math. Comput., 259 (2015), 258–270. https://doi.org/10.1016/j.amc.2015.02.064 doi: 10.1016/j.amc.2015.02.064
    [21] H. F. Huo, R. Chen, X. Y. Wang, Modelling and stability of HIV/AIDS epidemic model with treatment, Appl. Math. Modell., 40 (2016), 6550–6559. https://doi.org/10.1016/j.apm.2016.01.054 doi: 10.1016/j.apm.2016.01.054
    [22] A. N. Timsina, Y. R. Liyanage, M. Martcheva, N. Tuncer, A novel within-host modelof HIV and nutrition, Math. Biosci. Eng., 21 (2024), 5577–5603. https://doi.org/10.3934/mbe.2024246 doi: 10.3934/mbe.2024246
    [23] S. Xu, Saturated lysing efficiency of CD8+ cells induced monostable, bistable and oscillatory HIV kinetics, Math. Biosci. Eng., 21 (2024), 7373–7393. https://doi.org/10.3934/mbe.2024324 doi: 10.3934/mbe.2024324
    [24] A. L. Jenner, M. Smalley, D. Goldman, W. F. Goins, C. S. Gobbs, R. B. Puchaslki, et al., Agent-based computational modeling of glioblastoma predicts that stromal density is central to oncolytic virus efficacy, iScience, 25 (2022), 104395. https://doi.org/10.1016/j.isci.2022.104395 doi: 10.1016/j.isci.2022.104395
    [25] A. Surendran, A. L. Jenner, E. Karimi, B. Fiset, D. F. Quail, L. A. Walsh, et al., Agent-based modelling reveals the role of the tumor Microenvironment on the short-term success of combination temozolomide/immune checkpoint blockade to treat glioblastoma, J. Pharmacol. Exp. Ther., 387 (2023), 66–77. https://doi.org/10.1124/jpet.122.001571 doi: 10.1124/jpet.122.001571
    [26] B. Wacker, J. C. Schlüter, An age- and sex-structured SIR model: Theory and an explicit-implicit numerical solution algorithm, Math. Biosci. Eng., 17 (2020), 5752–5801. https://doi.org/10.3934/mbe.2020309 doi: 10.3934/mbe.2020309
    [27] B. Wacker, J. C. Schlüter, Time-continuous and time-discrete SIR models revisited: theory and applications, Adv. Differ. Equations, 2020 (2020), 556. 10.1186/s13662-020-02995-1 doi: 10.1186/s13662-020-02995-1
    [28] B. Wacker, J. C. Schlüter, Qualitative analysis of two systems of nonlinear first-order ordinary differential equations for biological systems, Math. Methods Appl. Sci., 45 (2022), 4597–4624. https://doi.org/10.1002/mma.8056 doi: 10.1002/mma.8056
    [29] B. Wacker, J. C. Schlüter, A non-standard finite-difference-method for a non-autonomous epidemiological model: analysis, parameter identification and applications, Math. Biosci. Eng., 20 (2023), 12923–12954. https://doi.org/10.3934/mbe.2023577 doi: 10.3934/mbe.2023577
    [30] B. Wacker, Framework for solving dynamics of Ca2+ ion concentrations in liver cells numerically: Analysis of a non-negativity-preserving non-standard finite-difference-method, Math. Methods Appl. Sci., 46 (2023), 16625–16643. https://doi.org/10.1002/mma.9464 doi: 10.1002/mma.9464
    [31] O. Diekmann, J. A. P. Heesterbeek, J. A. J. Metz, On the definition and the computation of the basic reproduction number R0 in models for infectious diseases in heterogeneous populations, J. Math. Biol., 28 (1990), 365–382. https://doi.org/10.1007/BF00178324 doi: 10.1007/BF00178324
    [32] P. van den Driessche, J. Watmough, Reproduction numbers and sub-threshold endemic equilibria for compartmental models of disease transmission, Math. Biosci., 180 (2002), 29–48. https://doi.org/10.1016/S0025-5564(02)00108-6 doi: 10.1016/S0025-5564(02)00108-6
    [33] I. M. Elbaz, H. El-Metwally, M. A. Sohaly, Viral kinetics, stability and sensitivity analysis of the within-host COVID-19 model, Sci. Rep., 13 (2023), 11675. https://doi.org/10.1038/s41598-023-38705-6 doi: 10.1038/s41598-023-38705-6
    [34] H. K. Khalil, Nonlinear Systems, Prentice-Hall, Upper Saddle River, 2001.
    [35] D. G. Schaeffer, J. W. Cain, Ordinary Differential Equations: Basics and Beyond, Springer-Verlag, New York, 2016. https://doi.org/10.1007/978-1-4939-6389-8
    [36] B. Wacker, J. C. Schlüter, A cubic nonlinear population growth modelfor single species: theory, an explicit-implicit solution algorithm and applications, Adv. Differ. Equations, 2021 (2021), 236. https://doi.org/10.1186/s13662-021-03399-5 doi: 10.1186/s13662-021-03399-5
    [37] M. Reed, B. Simon, Functional Analysis, Academic Press, San Diego, 1980.
    [38] A. Hurwitz, Über die Bedingungen, unter welchen eine Gleichung nur Wurzeln mit negativen reellen Theilen besitzt, Math. Ann., 46 (1895), 273–284. https://doi.org/10.1007/BF01446812 doi: 10.1007/BF01446812
    [39] J. W. Eaton, D. Bateman, S. Hauberg, R. Wehbring, GNU Octave Version 6.1.0 Manual: A High-level Interactive Language for Numerical Computations, 2020. Available from: https://www.gnu.org/software/octave/doc/v6.1.0/.
    [40] J. R. Dormand, P. J. Prince, A family of embedded Runge-Kutta formulae, J. Comput. Appl. Math., 6 (1980), 19–26. https://doi.org/10.1016/0771-050X(80)90013-3 doi: 10.1016/0771-050X(80)90013-3
    [41] L. F. Shampine, M. W. Reichert, The MATLAB ODE suite, SIAM J. Sci. Comput., 18 (1997), 1–22. https://doi.org/10.1137/S1064827594276424 doi: 10.1137/S1064827594276424
    [42] T. M. Hoang, High-order nonstandard finite difference methods preserving dynamical properties of one-dimensional dynamical systems, Numerical Algorithms, 2024 (2024), 1–31. https://doi.org/10.1007/s11075-024-01792-1 doi: 10.1007/s11075-024-01792-1
    [43] R. E. Mickens, Nonstandard Finite Difference Models of Differential Equations, World Scientific, 1993. https://doi.org/10.1142/2081
  • mbe-21-12-343-Supplementary.pdf
  • This article has been cited by:

    1. Benjamin Wacker, Qualitative Study of a Dynamical System for Computer Virus Propagation—A Nonstandard Finite‐Difference‐Methodological View, 2025, 0170-4214, 10.1002/mma.10798
  • 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(868) PDF downloads(52) Cited by(1)

Figures and Tables

Figures(2)  /  Tables(3)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog