Research article Special Issues

A coupled system of Laplacian and bi-Laplacian equations with nonlinear dampings and source terms of variable-exponents nonlinearities: Existence, uniqueness, blow-up and a large-time asymptotic behavior

  • In this paper, we consider a coupled system of Laplacian and bi-Laplacian equations with nonlinear dampings and source terms of variable-exponents nonlinearities. This system is supplemented with initial and mixed boundary conditions. First, we establish the existence and uniqueness results of a weak solution, under suitable assumptions on the variable exponents. Second, we show that the solutions with positive-initial energy blow-up in a finite time. Finally, we establish the global existence as well as the energy decay results of the solutions, using the stable-set and the multiplier methods, under appropriate conditions on the variable exponents and the initial data.

    Citation: Salim A. Messaoudi, Mohammad M. Al-Gharabli, Adel M. Al-Mahdi, Mohammed A. Al-Osta. A coupled system of Laplacian and bi-Laplacian equations with nonlinear dampings and source terms of variable-exponents nonlinearities: Existence, uniqueness, blow-up and a large-time asymptotic behavior[J]. AIMS Mathematics, 2023, 8(4): 7933-7966. doi: 10.3934/math.2023400

    Related Papers:

    [1] Attaullah, Ramzi Drissi, Wajaree Weera . Galerkin time discretization scheme for the transmission dynamics of HIV infection with non-linear supply rate. AIMS Mathematics, 2022, 7(6): 11292-11310. doi: 10.3934/math.2022630
    [2] Attaullah, Muhammad Jawad, Sultan Alyobi, Mansour F. Yassen, Wajaree Weera . A higher order Galerkin time discretization scheme for the novel mathematical model of COVID-19. AIMS Mathematics, 2023, 8(2): 3763-3790. doi: 10.3934/math.2023188
    [3] Dongjie Gao, Peiguo Zhang, Longqin Wang, Zhenlong Dai, Yonglei Fang . A novel high-order symmetric and energy-preserving continuous-stage Runge-Kutta-Nyström Fourier pseudo-spectral scheme for solving the two-dimensional nonlinear wave equation. AIMS Mathematics, 2025, 10(3): 6764-6787. doi: 10.3934/math.2025310
    [4] Mohamed Adel, Mohamed M. Khader, Mohammed M. Babatin, Maged Z. Youssef . Numerical investigation for the fractional model of pollution for a system of lakes using the SCM based on the Appell type Changhee polynomials. AIMS Mathematics, 2023, 8(12): 31104-31117. doi: 10.3934/math.20231592
    [5] Eman A. A. Ziada, Salwa El-Morsy, Osama Moaaz, Sameh S. Askar, Ahmad M. Alshamrani, Monica Botros . Solution of the SIR epidemic model of arbitrary orders containing Caputo-Fabrizio, Atangana-Baleanu and Caputo derivatives. AIMS Mathematics, 2024, 9(7): 18324-18355. doi: 10.3934/math.2024894
    [6] Sara S. Alzaid, Pawan Kumar Shaw, Sunil Kumar . A numerical study of fractional population growth and nuclear decay model. AIMS Mathematics, 2022, 7(6): 11417-11442. doi: 10.3934/math.2022637
    [7] Azadeh Ghanadian, Taher Lotfi . Approximate solution of nonlinear Black–Scholes equation via a fully discretized fourth-order method. AIMS Mathematics, 2020, 5(2): 879-893. doi: 10.3934/math.2020060
    [8] Khalid K. Ali, Mohamed A. Abd El Salam, Mohamed S. Mohamed . Chebyshev fifth-kind series approximation for generalized space fractional partial differential equations. AIMS Mathematics, 2022, 7(5): 7759-7780. doi: 10.3934/math.2022436
    [9] Nurain Zulaikha Husin, Muhammad Zaini Ahmad . Hybridization of the shooting and Runge-Kutta Cash-Karp methods for solving Fuzzy Boundary Value Problems. AIMS Mathematics, 2024, 9(11): 31806-31847. doi: 10.3934/math.20241529
    [10] Haifa Bin Jebreen . On the numerical solution of Fisher's equation by an efficient algorithm based on multiwavelets. AIMS Mathematics, 2021, 6(3): 2369-2384. doi: 10.3934/math.2021144
  • In this paper, we consider a coupled system of Laplacian and bi-Laplacian equations with nonlinear dampings and source terms of variable-exponents nonlinearities. This system is supplemented with initial and mixed boundary conditions. First, we establish the existence and uniqueness results of a weak solution, under suitable assumptions on the variable exponents. Second, we show that the solutions with positive-initial energy blow-up in a finite time. Finally, we establish the global existence as well as the energy decay results of the solutions, using the stable-set and the multiplier methods, under appropriate conditions on the variable exponents and the initial data.



    Epidemiology is the foundation of public health and is defined as the study of the distribution and determinants of diseases or disorders within groups of people, and the development of knowledge on how to prevent and control them. Epidemiological research helps us understand not only who has a disorder or disease but why and how it was brought to this individual or region. Nowadays, epidemiologists use the insights gathered in their research to determine how illness within a population affects our society and systems on a larger scale, and, in turn, they provide recommendations for interventions. The hepatitis B virus (HBV) became widespread, and epidemiologists around the world worked to control the spread. The epidemiology of HBV infection is diverse, with population prevalence, age and mode of acquisition and likelihood of progression to chronic infection mutually interdependent. Hepatitis B is a life-threatening liver disease caused by the HBV [1,2], which has a circular genome made of partly double-stranded DNA and is difficult to eradicate after infection due to the development of cccDNA [1,3]. HBV may induce both acute sickness and chronic liver infection, with sustained blood levels of HBV surface antigen (HBsAg), IgG anti-core antigen (anti-HBc), and HBV DNA [3]. Chronic infection may progress to cirrhosis or liver cancer [4]. According to the World Health Organization, around 2 billion individuals worldwide have been infected with the virus, with approximately 350 million having HBV. Each year, HBV causes approximately 600, 000 fatalities. Hepatitis B is prevalent in China, with an estimated 93 million people afflicted [5]. Due to the high risk of infection and the high number of fatalities associated with HBV, it is critical to improve our knowledge of the dynamics of HBV illness. HBV is spread by contact with the blood or other bodily fluids of the infected individuals [2]. Generally, the average age at which people get affected is influenced by the population's infection incidence [5]. Transmission from mother to baby occurs throughout pregnancy, delivery and the postpartum period. Vaccination, including passive immunoprophylaxis and postnatal proactive vaccination, is effective in protecting infants against contracting HBV during labor and delivery and in the postnatal period, but it cannot halt HBV prenatal infection [6,7,8,9,10,11]. Mathematical models may assist us in gaining insight into disease transmission, evaluating the success of different preventative methods, and ultimately bringing the illness under control. Mathematical modeling of a dynamical system is an interesting study area that has grabbed the interest of a majority of researchers. Dynamical systems have a wide range of applications, including population growth models, biomedical research, biological systems, and engineering. Aniji et al. [12] established a mathematical model of hepatitis B virus infection for antiviral treatment that depicts the probable interaction between HBV and target liver cells. Zada et al. [13] evaluated and analyzed the hepatitis B epidemic model with optimal control. Zhao et al. [14] designed a mathematical model of hepatitis B viral transmission and addressed its potential use in China's immunization plan. Means et al. [15] explored the Hepatitis B Virus's geographical impacts and potential function. Khatun et al. [16] explored viral dynamics, which aids in the understanding of the dynamics of HIV, hepatitis B virus (HCB), hepatitis C virus (HCV), and many other viruses. Additionally, they explored the hepatitis B virus model with Cytotoxic T Lymphocyte (CTL) immunological responses. Zhang et al. [17] developed a mathematical model to explain how hepatitis B spreads. The stability of equilibria and illness persistence are investigated. The study demonstrated that the model's dynamics are entirely determined by the basic reproductive number. Khan et al. [18] reported a detailed analysis of the Hepatitis B model using the Caputo-Fabrizio fractional derivative. The required findings are obtained by establishing the criteria for the existence and uniqueness of the solution to the considered model using fixed point theory. The continuous Galerkin-Petrov (cGP) discretization scheme is a very popular and powerful approach to handling complex dynamical systems. Numerous researchers addressed so many interesting properties and some feasible practical applications of the proposed scheme's implementation. Attaullah and Sohaib [19] employed a continuous Galerkin-Petrov (cGP) discretization scheme with a mathematical model of Human Immunodeficiency Virus (HIV) infection, which contained a system of nonlinear differential equations. They validated the results of the proposed scheme in comparison to other traditional schemes. Hussain et al. [20] utilized the Galerkin method with the Chen model and performed a comparison between the findings of the Galerkin technique and the fourth-order Runge-Kutta (RK4) method for the presented problem. They concluded that the Galerkin scheme achieved more accurate results at larger time step sizes than the RK4 method. Hussain [21] implemented a novel class of higher-order Galerkin temporal discretization techniques for non-stationary incompressible flow problems. Attaullah et al. [22] implemented the Galerkin time discretization scheme for the transmission dynamics of HIV infection with nonlinear supply rate and discussed the impact of different physical parameters on the population dynamics of healthy/infected cells and virus particles. Schieweck [23] described A-stable discontinuous Galerkin-Petrov time discretization of higher-order. For the heat equation, higher-order Galerkin temporal discretizations and fast multigrid solvers were used by Hussain et al. [24]. For the non-stationary Stokes equations, Hussain et al. [25] demonstrated an accurate and efficient higher-order Galerkin time-stepping approach. For nonlinear systems of ordinary differential equations, higher-order variational time discretizations were utilized by Matthies and Schieweck [25]. Attaullah et al. [26] numerically simulated the HIV model using a Galerkin approach. To validate the numerical results, a detailed evaluation of the outcomes achieved by the Galerkin scheme is contrasted with the findings of the RK4 technique. Various simulations were carried out to evaluate the findings of the Galerkin and RK4 approach, demonstrating the credibility of the cGP(2) technique. The comparison analysis and graphical plots illustrate that the Galerkin scheme is more appropriate than the RK4 method for achieving high accuracy at larger step sizes. For the approximate solution of the HIV model with a variable source term and full logistic proliferation, Attaullah et al. [27] implemented the Galerkin-technique and compared the results with the findings of the RK4 method. Khan et al. [28] described a mathematical model for the transmission dynamics of acute and chronic hepatitis B and developed an optimal control method for hepatitis B spread in a community. Numerous models of HBV transmission dynamics have been developed to examine the effect of preventative and control methods such as vaccination, antiviral therapy and the incidence rate [29,30,31,32]. These models have provided valuable insight into the effectiveness of different control strategies. However, the majority of these models have been studied primarily analytically. The mathematical model proposed in the paper is the extension of the model presented by Khan et al. [28] by including a non-linear saturated incidence rate [32] and investigated numerically. We carry out a comprehensive investigation of the various influence of several clinical parameters. The main contributions are as follows:

    1.Implemented Galerkin scheme and the RK4 scheme correctly to the HBV model and presented the solutions numerically and graphically.

    2.The significant effects of the various physical parameters are visualized accurately.

    3.To authenticate the validity and reliability of the suggested scheme, employed the proposed schemes to other mathematical models (Chen System and the HIV infection model) existing in the literature.

    4.The comparative study is performed to validate the accuracy (adopting various step sizes) and computational cost (in terms of time) of Galerkin and RK4-scheme.

    5.Finally, invoked the methods to the problem having an exact solution and confirm the accuracy and time consuming with different step sizes.

    The mathematical model addressed here is as follows:

    d˜Sdt=eη˜S(t)˜I2(t)1+ξ1˜I2(ϕ0+λ)˜S(t), (2.1)
    d˜I1dt=η˜S(t)˜I2(t)1+ξ1˜I2(ϕ0+ξ+ψ1)˜I1(t), (2.2)
    d˜I2dt=ξ˜I1(t)(ϕ0+ϕ1+ψ2)˜I2(t), (2.3)
    d˜Rdt=ψ1˜I1(t)+ψ2˜I2(t)+λ˜S(t)ϕ0˜R(t). (2.4)

    In the this model the population ˜T(t) is divided into four compartments: susceptible individuals ˜S(t), infected individuals ˜I1(t), infected individuals with chronic hepatitis B (CHB), ˜I2(t) and recovered individuals ˜R(t). The initial conditions and the detailed information of all parameters included in the model are presented in Table 1.

    Table 1.  The description of variables, parameters, with their values involved in model (unit = day1mm3) [28].
    Variable Explanation Value
    ˜S(0) Susceptible individuals 100
    ˜I1(0) Infected individuals with acute HBV 40
    ˜I2(0) Infected individuals with chronic HBV 20
    ˜R(0) Recovered individuals 5
    e Birth rate 0.4
    η The rate of moving from susceptible to infected with acute hepatitis B 0.005
    ξ The rate of moving from the acute stage to infected with chronic hepatitis B. 0.01
    ψ1 The rate of rehabilitation from the acute stage to the recovered stage 0.05
    ψ2 The rate of rehabilitation from the chronic stage to the recovered stage 0.06
    ϕ0 Natural death rate 0.003
    ϕ1 Death rate happening due to hepatitis B 0.002
    λ Vaccination rate of Hepatitis B 0.02
    ξ1 Saturation constant 0.004

     | Show Table
    DownLoad: CSV

    The Galerkin technique has been widely employed in recent years to handle a wide range of nonlinear problems in science and engineering, such as [23,24,25,33,34,35,37,43,44]. In this paper, we are using this approach for the models discussed in [19,20,27,28,38]. The system of ODEs for the considered model can be written as follows:

    Find ˜u:[0,tmax]V=Rd such that

    dt˜u(t)=F(t,˜u(t))tI,˜u(0)=˜u0, (3.1)

    where dt shows the derivative of ˜u(t) w.r.t. time, I = [0, T] is the time interval, ˜u(t)=(˜u1(t),˜u2(t),˜u3(t))V˜u(0)=(˜u1(0),˜u2(0),˜u3(0))V are the initial values of ˜u(t) at time t=0. We also assume that (˜u1(t),˜u2(t),˜u3(t))=(T(t),I(t),V(t)), which implies that (˜u1(0),˜u2(0),˜u3(0))=(T(0),I(0),V(0)). The function F=(f1,f2,f3) is nonlinear and is defined as F : I × K K.

    The weak formulation of Problem (3.1) is: Find ˜u X such that ˜u(0)=˜u0 and

    Idt˜u(t),v(t)dt=IF(t,˜u(t),v(t))dtforallvY, (3.2)

    where , represents the usual inner product in L2(I,K), and X, Y represent the solution space and test space, respectively. To explain the time discretization of variational type Problem(3.1).

    Characterize the function t ˜u(t), and we define the space C(I,K)=C0(I,K) as the space of continuous functions ˜u : I K equipped with the norm

    ˜uC(I,K)=suptI˜uK.

    We will use the space L2(I,K) as the space of discontinuous functions, which is given by

    L2(I,K)={˜u:IK:˜uL2(I,K)=(I˜u2Kdt<)1/2}.

    In time discretization, we split the intervals I into N sub-intervals In=[tn1,tn], where n=1,,N and 0=t0<t1<t2<<tN1<tN=T. The symbol τ will indicate the time discretization parameter, as well as the maximum time step size τ=max1nNτn, where τn=tntn1, the length of nth time interval In. The following set of time intervals Mτ={ˉI1,,ˉIN} will be called the time-mesh. We find out the solution ˜u : I K on each time interval In by a function ˜uτ : I K which is a piecewise polynomial of some order l w.r.t time. The time-discrete solution space for ˜uτ is XlτX and is defined by

    Xlτ={˜uC(I,K):˜u|InPl(In,K)for allInMτ}, (3.3)

    where

    Pl(In,K)={˜u:InK:˜u(t)=ls=0Usts,foralltIn,UsK,s}.

    The discrete test space for ˜uτ is YlτY and is defined by

    Ykτ={vL2(I,K):v|InPk1(In,K)InMτ}, (3.4)

    which is composed of l1 piecewise polynomials and is discontinuous at the time step end nodes. We multiply Eq (3.1) by the test function vτYkτ and integrate it over the whole interval I. We get the discrete-time problem: Find ˜uτXkτ such that ˜uτ(0)=0 and

    I˜uτ(t),vτ(t)dt=IF(t,˜uτ(t),vτ(t))dtvτYlτ, (3.5)

    where , represents the usual inner product in L2(I,K). This discretization is known as the exact continuous Galerkin-Petrov method or briefly the "exact cGP(l)-method'' of order k. The Galerkin-Petrov name is due to the fact that the solution space Xlτ is different from the test space Ylτ. The term "exact" denotes that the time integral on the right-hand side of the Eq (3.5) is determined exactly. Because the discrete test space Ylτ is discontinuous, Eq (3.5) can be solved in a time marching procedure in which local problems on the time interval are handled successively. Therefore, we choose the test function vτ(t)=vψ(t) with arbitrary time-independent vK and a scalar function ψ:IR which is zero on I|In and a polynomial of the order less than or equal to l1 on the time interval In=[tn1,tn]. Then, we get from (3.5) the In-problem of the exact continuous Galerkin-Petrov method of order l: Find ˜uτ|InPl(In,K) such that

    Indt˜uτ(t),vψ(t)dt=InF(t,˜uτ(t)),vψ(t)dtvKψPl1(In), (3.6)

    with the initial condition ˜uτ|In(tn1)=˜uτ|In1(tn1) for n2 and ˜uτ|In(tn1)=˜u0 for n=1.

    In the usual case of a nonlinear function F,, (where , denotes the usual inner product in L2 norm) we need to calculate the integrals numerically on the right hand side of the Eq (3.6). The (l+1)-point Gauss-Lobatto formula is exact if the function is to be integrated as a polynomial of degree less than or equal to 2l1. As a result, this formula is applied to the integral on the left hand side of (3.6) will give the exact value. Then, the "In-problem of the numerically integrated cGP(l) method'' is: Find ˜uτ|InPl(In,K) such that ˜uτ(tn1)=˜un1,

    ls=0ˆwsdt˜uτ(tn,s)ψ(tn,s)=ks=0ˆwsF(tn,s,˜uτ(tn,s))ψ(tn,s)ψPl1(In), (3.7)

    where ˆws are the weights and ˆt[1,1],s=0,1,2,3,,l represent the nodes on the reference interval. To find ˜uτ on each time interval In, we represent it by a polynomial ansatz

    ˜uτ(t)=ks=0Usnϕn,s(t)tIn, (3.8)

    where the coefficients Usn are the components of K and the real-valued functions ϕn,sPl(In) are the Lagrange basis functions with respect to l+1 suitable nodal points tn,sIn satisfying conditions mentioned below:

    ϕn,s(tn,r)=δr,s,r,s=0,1,2,,l, (3.9)

    where δr,s is the Kronecker delta that is defined as:

    δr,s={1ifr=s0ifrs

    Like in [36], the tn,s have been defined as the quadrature points of (l+1)-point Gauss-Lobatto formula on the time interval In. For the selection of initial conditions we can set tn,0=tn1 which implies the initial condition for (3.6)

    U0n=˜uτ|In1ifn2,forn=1U0n=˜u0. (3.10)

    We define the basis functions ϕn,sPl(In) via the affine reference transformation ˉT:ˆIIn where ˆI = [-1, 1] and

    t=ˉT(ˆt)=tn+tn12+τn2ˆtInˆtIn,n=1,2,3,N. (3.11)

    Let ^ϕsPl(ˆI), s = 0, 1, ,l, demonstrate the basis functions that meet the requirements

    ^ϕs(^tr)=δr,s,r,s=0,1,,l,

    where ^t0=1 and ^tr, r=1,2,,l, are the quadrature points for the reference interval ˆI. Then, we define the basis functions on the original time interval In by the mapping

    ϕn,s(t)=^ϕs(ˆt)withˆt=ˉT1n(t)=2τn(t+tn1tn2)ˆI. (3.12)

    Likewise, the test basis functions ψn,r are defined by the suitable reference basis functions ˆψPl1(ˆI), i.e.,

    ψn,r(t)=^ψr(ˉT1n(t))tIn,r=1,2,,l.

    From the representation (3.8), we get for dt˜uτ

    dt˜uτ(t)=ks=0Usnϕn,s(t)tIn. (3.13)

    By putting (3.13) in (3.6), we get

    Indt˜uτ(t),vψ(t)dt=Inls=0Usn,vϕs(t)ψ(t). (3.14)

    The integral is now transformed into the reference interval hatI and computed using the (l+1)-point Gauss-Lobatto quadrature formula, which leads for each test basis function ψPl1 and for all vK,

    ˆInls=0Usn,vˆϕs(ˆt)ˆψ(ˆt)dˆt=ˆInF(ωn(ˆt),ls=0Usn(ˆt)),vˆψ(ˆt)dˆtvK,lμ=0ˆwμls=0Usn,vˆϕs(ˆtμ)ˆψ(ˆtμ)=lμ=0ˆwμF(ωn(ˆtμ),ls=0Usnˆϕs(ˆtμ)),vˆψ(ˆtμ), (3.15)

    where ˆwμ are the weights and ˆtμ[1,1] are the points of integration with ˆt0=1 and ˆtl=1. If we choose the test functions ψn,iPl1(In) such that

    ˆψ(^tμ)=(ˆw)1δr,μr,μ=1,2,,l. (3.16)

    Now find the coefficients that are unknown UsnK for s=1,,l,

    ls=0αr,sUsn=τn2{F(tn,r,Usn)+βrF(tn,0,Usn)}i=1,2,,l, (3.17)

    where Usn=Usn1 for n>1, U01=˜u0 for n=1 and

    αr,s=^ϕs(^tr)+βr^ϕs(t0),βr=ˆw0ˆψr(ˆt0).

    We will discuss the cGP(k) method for the cases l=1andl=2 in the following.

    We used the two point Gauss-Lobatto formula with tn,0=tn1,tn,1=tn and weights ˆw0=ˆw1=1, which gives the well-known Trapezoidal rule. We obtain α1,0=1,α1,1=1 and β1=1. For the single coefficient U1n=˜uτ(tn)K, the problem leads to the following block equation:

    α1,1U1nαn,0U0n=τn2{F(tn,U1n)+F(tn1,U0n)}. (3.18)

    The three-point Gauss-Lobatto formula (Simpson rule) is used to define the quadratic basis functions with weights ˆw0=ˆw2=1/3,ˆw1=4/3 and ˆt0=1, ˆt1=0, ˆt2=1. Then, we get

    αr,s=(54114242),βr=(121),r=1,2,s=0,1,2.

    Thus, the system to be solved for U1n,U2nK from the known U0n=U2n1 becomes:

    α1,1U1n+α1,2U2n=α1,0U0n+τn2{F(tn,1,U1n)+β1F(tn,0,U0n)}, (3.19)
    α2,1U1n+α2,2U2n=α2,0U0n+τn2{F(tn,2,U2n)+β2F(tn,0,U0n)}, (3.20)

    where U0n indicate the initial conditions at the current time interval. For the cGP(k)-method, the discrete solution space consists of continuous piecewise polynomial functions in time of degree k>1 and the discrete test space of discontinuous Galerkin (dG) having polynomial functions of degree k1 (see [33,34,35,36] for more details). In the dG(k)-method, both the solution and test space are constructed by means of discontinuous polynomial functions of degree k. With respect to the computational costs, which mainly depend on the size of the resulting block system that has to be solved for each time interval, the cGP(k)-method is comparable to the dG(k1)-method. However, concerning the discretization error in time, the accuracy of the cGP(k)-method is one order higher than that of the dG(k1)-method. Furthermore, it is known that all cGP(k)-methods are A-stable, while all dG(k)-methods are even strongly A-stable (or L-stable), i.e., the dG-methods have better damping properties with respect to high frequency error components (see [33,34,35,36] for more details).

    This method is very well-known of order four established by Kutta [39] and extensively utilized for initial value problems (see [40] for more details). The Runge-Kutta method of order four is used to solve numerically the first order initial value problems. Let

    ˙y=g(t,y),atb, (3.21)

    be the initial value problem with the initial condition y(a)=α, and let N>0 be an integer and set h=baN is the step size. Partition the whole interval into N sub-intervals with mesh points ti=a+ih, for i=0,1,2,,N1. Then, the Runge-Kutta method of order four is described as

    yi+1=yi+16(k1+2(k2+k3)+k4),fori=0,1,2,,N1, (3.22)

    where

    k1=hg(ti,yi),k2=hg(ti+h2,yi+k12),k3=hg(ti+h2,yi+k22),k4=hg(ti+h,yi+k3). (3.23)

    The Runge Kutta method of order four (RK-4) agrees with the Taylor series method up to terms of O(h4). This method can be extended to solve a system of n first-order differential equations. The generalization of the method is as follows.

    Let

    dy1dt=g1(t,y1,y2,,yn),dy2dt=g2(t,y1,y2,,yn),dyndt=gn(t,y1,y2,,yn), (3.24)

    be the nth-order system of first-order initial value problems with the initial conditions

    y1(a)=α1,y2(a)=α2,,yn(a)=αn.

    Use the notation yji, for each i=0,1,2,,N and j=1,2,,n, to denote an approximation to yj(ti). That is, yji approximates the jth solution y(t) of (3.24) at the ith mesh points ti. For the initial condition, set

    y10=α1,y20=α2,,yn0=αn.

    Suppose that the values y1i,y2i,,yni have been computed. We obtain y1i+1,y2i+1,,yni+1 by first calculating

    kj1=hgj(ti,y1i,y2i,,yni), (3.25)
    kj2=hgj(ti+h2,y1i+k112,y2i+k212,,yni+kn12), (3.26)
    kj3=hgj(ti+h2,y1i+k122,y2i+k222,,yni+kn22), (3.27)
    kj4=hgj(ti+h,y1i+k13,y2i+k23,,yni+kn3), (3.28)

    for each j=1,2,,n; and then

    yji+1=yji+16(kj1+2(kj2+kj3)+kj4), (3.29)

    for each j=1,2,,n. The values k11,k21,,kn1, must be computed before any of the terms of the form kj2 can be determined.

    This section relates to the implementation of the cGP(2)-scheme and RK4 scheme for the model presented and figure out the correct results. All the parameters, values are presented in Table 1. The findings and absolute errors achieved utilizing cGP(2) scheme (at τ = 0.1) and RK4 method (at τ = 0.1) were compared and presented in Tables 25 and in Figure 2. The findings achieved utilizing the presented approach are considerably comparable to those obtained using the RK4-method with differing computational time. From comparison one could see that the results are not precise but similar up to nine digits approximately. The results demonstrated that the proposed methodologies yielded reliable and precise solutions and that both results overlapped significantly during the stipulated time period. We also solved the model using cGP(4) scheme (at τ = 0.2) and RK4 (at τ = 0.1) and found the absolute errors between the findings. Comparing the absolute values of the errors presented in Tables 69, we see that the higher order method cGP(4) achieves with a quite large time step size higher accuracy than the lower order method cGP(2) with smaller time step displayed in Tables 25. Moreover, the analysis of the numerical costs for different step sizes in terms of time show that the cGP(2)-method is more time consuming as in comparison to RK4-method as presented in Table 10 and Figure 2e. Furthermore, Figure 1 depicts the influence of varying the parameters ξ1 and ψ2 on the dynamical behavior of the state variables. By boosting the value of ξ1, we observed that the population of susceptible individuals grows, and the population of infected individuals with acute HBV, infected individuals with chronic HBV and recovered individuals declines. By increasing ψ2, the concentration of susceptible and recovered people increases, whereas the population of acutely infected and chronically infected individuals declines. Also, graphical visualizations of error solutions and errors phase portraits using cGP(2)-scheme (at τ = 0.0005 where τ is the step size) and RK4-scheme (at τ = 0.0005) are shown in Figure 3 for t[0,10]. The findings demonstrate that the cGP(2) scheme outperforms the RK4 scheme in terms of time step accuracy while incurring higher computational costs.

    Table 2.  The comparative analysis of the outcomes of Galerkin and RK4 methods for ˜S(t).
    |cGP(2)τ=0.1RK4τ=0.1|
    ti cGP(2)-scheme RK4 -scheme |cGP(2)-RK4|
    0.0 100.000000000000 100.000000000000 0.00000000000000E-009
    0.1 98.8923978777600 98.8923978779475 0.18746959540294E-009
    0.2 97.8012267618895 97.8012267622551 0.36556002669386E-009
    0.3 96.7261679090720 96.7261679096068 0.53472604122362E-009
    0.4 95.6669105844358 95.6669105851311 0.69528027779597E-009
    0.5 94.6231518165269 94.6231518173745 0.84763485119765E-009
    0.6 93.5945961609257 93.5945961619179 0.99213082194183E-009
    0.7 92.5809554721666 92.5809554732958 1.12910925054166E-009
    0.8 91.5819486836314 91.5819486848903 1.25892540836503E-009
    0.9 90.5973015951057 90.5973015964875 1.38184930165153E-009
    1.0 89.6267466676998 89.6267466691980 1.49820778005960E-009

     | Show Table
    DownLoad: CSV
    Table 3.  The comparative analysis of the results of Galerkin and RK4 methods for ˜I1(t).
    |cGP(2)τ=0.1RK4τ=0.1|
    ti cGP(2)-scheme RK4 -scheme | cGP(2)- RK4 |
    0.0 40.0000000000000 40.0000000000000 0.00000000000000E-009
    0.1 40.6647754297249 40.6647754293823 0.34259528547409E-009
    0.2 41.3115173101931 41.3115173095231 0.66994942926613E-009
    0.3 41.9406193275814 41.9406193265987 0.98267349812886E-009
    0.4 42.5524654340669 42.5524654327855 1.28134303167826E-009
    0.5 43.1474301351317 43.1474301335651 1.56653356953029E-009
    0.6 43.7258787669986 43.7258787651598 1.83877801873678E-009
    0.7 44.2881677645815 44.2881677624830 2.09858086464010E-009
    0.8 44.8346449203167 44.8346449179703 2.34645369801001E-009
    0.9 45.3656496342260 45.3656496316431 2.58284416077004E-009
    1.0 45.8815131555476 45.8815131527394 2.80822831655314E-009

     | Show Table
    DownLoad: CSV
    Table 4.  The comparative analysis of the results of Galerkin and RK4 methods for ˜I2(t)..
    |cGP(2)τ=0.1RK4τ=0.1|
    ti cGP(2)-scheme RK4 -scheme | cGP(2)- RK4 |
    0.0 20.0000000000000 20.0000000000000 0.0000000000000E-10
    0.1 19.9106250505375 19.9106250505917 0.5414690917860E-10
    0.2 19.8224827434847 19.8224827435910 1.0622258628246E-10
    0.3 19.7355473148203 19.7355473149766 1.5628387473043E-10
    0.4 19.6497935549745 19.6497935551789 2.0440538150979E-10
    0.5 19.5651967956791 19.5651967959298 2.5065105546673E-10
    0.6 19.4817328971831 19.4817328974781 2.9508484544749E-10
    0.7 19.3993782358234 19.3993782361612 3.3777070029828E-10
    0.8 19.3181096919390 19.3181096923178 3.7876546343796E-10
    0.9 19.2379046381163 19.2379046385345 4.1812242557171E-10
    1.0 19.1587409277559 19.1587409282118 4.5589843011840E-10

     | Show Table
    DownLoad: CSV
    Table 5.  The comparative analysis of the results of Galerkin and RK4 methods for ˜R(t)..
    |cGP(2)τ=0.1RK4τ=0.1|
    ti cGP(2)-scheme RK4 -scheme | cGP(2)- RK4 |
    0.0 5.0000000000000 5.0000000000000 0.0000000000000E-10
    0.1 5.5187126241351 5.5187126242367 1.0162182206841E-10
    0.2 6.0378169434692 6.0378169436686 1.9942092421843E-10
    0.3 6.5572635287196 6.5572635290131 2.9353230956985E-10
    0.4 7.0770041268788 7.0770041272629 3.8408920488564E-10
    0.5 7.5969916319757 7.5969916324469 4.7121240243087E-10
    0.6 8.1171800566980 8.1171800572530 5.5502624718429E-10
    0.7 8.6375245048468 8.6375245054825 6.3564797869731E-10
    0.8 9.1579811445943 9.1579811453075 7.1318595473713E-10
    0.9 9.6785071825168 9.6785071833045 7.8775030942779E-10
    1.0 10.1990608383783 10.1990608392378 8.5944229510915E-10

     | Show Table
    DownLoad: CSV
    Table 6.  The comparative analysis of the outcomes of cGP(4) and RK4 methods for ˜S(t).
    |cGP(4)τ=0.2RK4τ=0.1|
    ti cGP(4)-scheme RK4 -scheme |cGP(4)-RK4|
    0.0 100.000000000000 100.000000000000 0.0000E-011
    0.2 97.8012267622895 97.8012267622551 3.4404E-011
    0.4 95.6669105851458 95.6669105851311 1.4694E-011
    0.6 93.5945961618257 93.5945961619179 9.2200E-011
    0.8 91.5819486848314 91.5819486848903 5.8904E-011
    1.0 89.6267466691998 89.6267466691980 1.8048E-011

     | Show Table
    DownLoad: CSV
    Table 7.  The comparative analysis of the results of cGP(4) and RK4 methods for ˜I1(t).
    |cGP(4)τ=0.2RK4τ=0.1|
    ti cGP(4)-scheme RK4 -scheme | cGP(4)- RK4 |
    0.0 40.0000000000000 40.0000000000000 0.0000E-011
    0.2 41.3115173095931 41.3115173095231 7.0003E-011
    0.4 42.5524654327669 42.5524654327855 1.8602E-011
    0.6 43.7258787651986 43.7258787651598 3.8803E-011
    0.8 44.8346449179167 44.8346449179703 5.3603E-011
    1.0 45.8815131527576 45.8815131527394 1.8197E-011

     | Show Table
    DownLoad: CSV
    Table 8.  The comparative analysis of the results of cGP(4) and RK4 methods for ˜I2(t).
    |cGP(4)τ=0.2RK4τ=0.1|
    ti cGP(4)-scheme RK4 -scheme | cGP(4)- RK4 |
    0.0 20.0000000000000 20.0000000000000 0.0000E-12
    0.2 19.8224827435947 19.8224827435910 3.7019E-12
    0.4 19.6497935551745 19.6497935551789 4.3983E-12
    0.6 19.4817328974731 19.4817328974781 4.9987E-12
    0.8 19.3181096923190 19.3181096923178 1.1973E-12
    1.0 19.1587409282159 19.1587409282118 4.0998E-12

     | Show Table
    DownLoad: CSV
    Table 9.  The comparative analysis of the results of cGP(4) and RK4 methods for ˜R(t).
    |cGP(4)τ=0.2RK4τ=0.1|
    ti cGP(4)-scheme RK4 -scheme | cGP(4)- RK4 |
    0.0 5.0000000000000 5.0000000000000 0.0000E-13
    0.2 6.0378169436682 6.0378169436686 3.9968E-13
    0.4 7.0770041272628 7.0770041272629 1.0036E-13
    0.6 8.1171800572580 8.1171800572530 5.0004E-13
    0.8 9.1579811453043 9.1579811453075 3.2010E-13
    1.0 10.199060839238 10.199060839237 4.9916E-13

     | Show Table
    DownLoad: CSV
    Table 10.  The computational costs in terms of the CPU time (in seconds) for the Galerkin-scheme and RK4 scheme t [0, 1].
    CPU time (in seconds)
    1/τ Galerkin-scheme RK4 -scheme
    10 0.254317 0.054033
    100 0.377437 0.042027
    200 0.732050 0.065334
    500 1.393989 0.117353
    1000 2.039698 0.144833
    1500 2.835332 0.151022
    5000 5.925991 0.173021
    10000 10.267904 0.243410
    20000 20.300213 0.466144
    100000 283.102165 5.214400

     | Show Table
    DownLoad: CSV
    Figure 1.  The influence of changing ξ (left) and ψ (right) on the dynamics of ˜S(t), ˜I1(t), ˜I2(t) and ˜R(t) for t[0, 200].
    Figure 2.  Graphical comparison using Galerkin-scheme (τ=0.1) and RK4-scheme (τ=0.1) for ˜S(t), ˜I1(t), ˜I2(t) and ˜R(t) for t[0, 1] and computational costs in terms of time for both schemes.
    Figure 3.  The geometrical representation of errors solutions and error phase portraits using Galerkin-scheme (τ=0.0005) and RK4-scheme (τ=0.0005) for t [0, 10].

    ● Windows: 8.1 single language

    ● Processor: Intel(R)Core(TM)i5-5200CPU@2.20GHz 2.20GHz

    ● Installed memory(RAM): 4.00GB

    ● System type: 64 bit Operating System, x64-based processor

    The Chen system was initially established in 1999 by Chen and Ueta [41] and described as follows:

    dX(t)dt=α(Y(t)X(t)) (4.1)
    dY(t)dt=(γα)X(t)X(t)Y(t)+γY(t) (4.2)
    dZ(t)dt=X(t)Y(t)βZ(t), (4.3)

    where X(t), Y(t) and Z(t) are the dependent variables, and α, β and γ are positive constant parameters. According to bifurcation analysis assuming the parameters α=35 and γ=28, system of Equations 4.1–4.3 shows non-chaotic and chaotic behaviour for β=12 and β=3, respectively [20].

    For the Chen system solutions, Hussain et al. [20] employed and investigated the Galerkin-method for the model of both non-chaotic behavior and chaotic behavior. They numerically compared the Galerkin-method solutions to the classic RK4-method solutions and determined that the Galerkin solutions with larger time steps achieved comparable accuracy to the RK4 solutions with significantly smaller time steps. This opposed the claim that the findings of the Galerkin-method and RK4-method are same.

    Herein, we consider the Chen system with non-chaotic behavior and implement the proposed methods for same and different step sizes to strengthen and confirm the claim that (1) the results of Galerkin and RK4 are not equal, as could be seen from Tables 1116. The computations are conducted with several time step sizes, and the accuracy is assessed at various time intervals. It could be observed that the results are not similar at τ=0.01 while identical up to eight digit, when τ=0.0001. (2) Tables 17 and Figure 4(g) show the elapsed time of both schemes, and it may be worthy to note that the RK4-scheme required less time as compared to the cGP(2)-scheme. In addition, Figure 4 shows the differences between the outcomes obtained through the Galerkin and RK4 methods for different values of τ in the non-chaotic case. This confirms that the RK4 technique requires less computing cost than the Galerkin scheme.

    Table 11.  Differences in Galerkin and RK4 outcomes for X(t)..
    |Galerkinτ=0.01RK4τ=0.01|
    ti Galerkin-scheme RK4 -scheme |Galerkin-RK4|
    1 -27.048832665674844 -27.054660707952376 0.005828042277532
    2 -24.121382804250171 -24.220203102041907 0.098820297791736
    3 -17.595913326656611 -17.836012791172081 0.240099464515470
    4 -9.479112397807544 -9.831210400585679 0.352098002778135
    5 -1.519967659034344 -1.930913832633865 0.410946173599521
    6 5.629377958266195 5.185500567394190 0.443877390872006
    7 12.096029332377837 11.616145579345067 0.479883753032770
    8 18.108332396521128 17.598089053219176 0.510243343301951
    9 23.372140795726757 22.908872091375947 0.463268704350810
    10 26.704118501387580 26.483143514708306 0.220974986679273

     | Show Table
    DownLoad: CSV
    Table 12.  Differences in Galerkin and RK4 solutions for Y(t).
    |Galerkinτ=0.01RK4τ=0.01|
    ti Galerkin-scheme RK4 -scheme | Galerkin- RK4 |
    1 -25.757727035512879 -25.829785688541214 0.072058653028336
    2 -15.733229367223030 -15.960756261129955 0.227526893906925
    3 -4.687271172508644 -5.014096669129265 0.326825496620621
    4 4.359806983134549 4.021272883035516 0.338534100099033
    5 11.202102936238061 10.866833989119010 0.335268947119051
    6 16.973500381147989 16.603394521003853 0.370105860144136
    7 22.551140692377597 22.122303983970319 0.428836708407278
    8 27.775451248198650 27.358265955699231 0.417185292499418
    9 31.042892281454858 30.879307980173685 0.163584301281173
    10 29.717909515486923 30.160333148606682 0.442423633119759

     | Show Table
    DownLoad: CSV
    Table 13.  Differences in cGP(2) and RK4 solutions for Z(t).
    |Galerkinτ=0.01RK4τ=0.01|
    ti Galerkin-scheme  RK4 -scheme | Galerkin- RK4 |
    1 37.235989947410992 37.195681864725273 0.040308082685719
    2 38.702841501348885 38.727949536128541 0.025108034779656
    3 34.855832136470781 35.030555096436252 0.174722959965472
    4 28.471059015282727 28.746019629584858 0.274960614302131
    5 22.625844253119464 22.890177162419192 0.264332909299728
    6 18.991435889835248 19.140513211975794 0.149077322140545
    7 18.286069422857828 18.221904191712490 0.064165231145338
    8 20.890272803905422 20.521203756415815 0.369069047489607
    9 26.631030747685497 25.950053980017987 0.680976767667509
    10 33.728856634945707 32.981125844075343 0.747730790870364

     | Show Table
    DownLoad: CSV
    Table 14.  Differences in cGP(2) and RK4 solutions for X(t).
    |Galerkinτ=0.0001RK4τ=0.0001|
    ti Galerkin-scheme  RK4 -scheme | Galerkin- RK4 |
    1 -27.047610303301788 -27.047610303357622 0.005583444817603E-008
    2 -24.094604479340664 -24.094604480173757 0.083309359411032E-008
    3 -17.530298570737838 -17.530298572724817 0.198697946984794E-008
    4 -9.383231382528516 -9.383231385407889 0.287937318432796E-008
    5 -1.408518132110425 -1.408518135446550 0.333612537595229E-008
    6 5.749603796814950 5.749603793216482 0.359846730191293E-008
    7 12.226070443083062 12.226070439194331 0.388873111489829E-008
    8 18.246170817362785 18.246170813251055 0.411172962344608E-008
    9 23.495090238172907 23.495090234528103 0.364480357006869E-008
    10 26.756926094741043 26.756926093227204 0.151383972024632E-008

     | Show Table
    DownLoad: CSV
    Table 15.  Differences in Galerkin and RK4 outcomes for Y(t).
    |Galerkinτ=0.0001RK4τ=0.0001|
    ti Galerkin-scheme  RK4 -scheme | Galerkin- RK4 |
    1 -25.739166712238003 -25.739166712853841 0.061583804722432E-008
    2 -15.671765498592148 -15.671765500480387 0.188823889857304E-008
    3 -4.598724349917238 -4.598724352588283 0.267104471873836E-008
    4 4.451217413219246 4.451217410479821 0.273942468709265E-008
    5 11.292738780804578 11.292738778094927 0.270965117010746E-008
    6 17.074033073890451 17.074033070883853 0.300659763752265E-008
    7 22.667547308391388 22.667547304916646 0.347474227169187E-008
    8 27.886618576432806 27.886618573137842 0.329496430140352E-008
    9 31.079875020723971 31.079875019688370 0.103560182651563E-008
    10 29.584671502608845 29.584671506692231 0.408338607371661E-008

     | Show Table
    DownLoad: CSV
    Table 16.  Differences in Galerkin and RK4 outcomes for Z(t).
    |Galerkinτ=0.0001RK4τ=0.0001|
    ti Galerkin-scheme  RK4 -scheme | Galerkin- RK4 |
    1 37.246938185955891 37.246938185624877 0.033101343888120E-008
    2 38.695920712480053 38.695920712709558 0.022950530365051E-008
    3 34.807827360519170 34.807827361976500 0.145733025647132E-008
    4 28.396215621242757 28.396215623489454 0.224669705062297E-008
    5 22.554831144255033 22.554831146374244 0.211921147297289E-008
    6 18.952840363341220 18.952840364482952 0.114173204224244E-008
    7 18.306723791470940 18.306723790831505 0.063943517147891E-008
    8 20.994876585309012 20.994876582156568 0.315244363946476E-008
    9 26.818990233189414 26.818990227565898 0.562351587518606E-008
    10 33.927750982739248 33.927750976850653 0.588859450090240E-008

     | Show Table
    DownLoad: CSV
    Table 17.  The computational costs in terms of the CPU-time (in seconds) for the Galerkin and RK4 schemes for t [0, 10].
    CPU time (in seconds)
    1/τ Galerkin-scheme  RK4 -scheme
    10 0.351107 0.049666
    100 1.395488 0.060902
    200 1.805245 0.064809
    500 2.922045 0.098383
    1000 4.807289 0.133293
    1500 6.021356 0.169597
    5000 13.376732 0.411494
    10000 26.135305 1.037192
    20000 49.576749 1.942809
    100000 386.666995 6.432528

     | Show Table
    DownLoad: CSV
    Figure 4.  Comparison of Galerkin-scheme (τ=0.01) and RK4-scheme (τ=0.01) (Left) versus Galerkin-scheme (τ=0.0001) and RK4-scheme (τ=0.0001) (Right) for X(t), Y(t) and Z(t) where t[0,10] along with computational cost in terms of time.

    In the field of HIV infection of healthy T-cells, mathematical models have become indispensable tools for better understanding the dynamical behaviour, transmission, disease progression, and immunological interactions of HIV. In humans, the HIV virus mainly targets healthy T-cells. Whenever HIV enters the body, it infects a significant number of healthy T-cells, causing their depletion gradually. As a consequence, the body's immune system is destroyed, weakening the host immunological response to viral diseases and eventually leading to acquired immunodeficiency syndrome (AIDS). The reduction in the number of these cells is used as a major indication to monitor the development of HIV infection and the symptoms of AIDS. These cells proliferate at a steady rate from bone marrow and thymus precursors. To characterize the dynamical behavior of HIV infection, several mathematical models have been introduced. The model proposed by Attaullah et al. [19] is as follows:

    d˜T(t)dt=s0μ˜T˜T(t)+α˜T(t)(1˜T(t)+˜I(t)˜Tmax)β˜V(t)˜T(t), (5.1)
    d˜I(t)dt=β˜V(t)˜T(t)˜I(t)μ˜I, (5.2)
    d˜V(t)dt=γμ˜I˜I(t)μ˜V˜V(t). (5.3)

    Herein, ˜T, ˜I and ˜V describe the densities of healthy/infected T-cells and virus respectively. The explanation of parameters, along their values and initial condition for the state variables are illustrated in Table 18.

    Table 18.  The description of variables and parameters with their values (unit = day1mm3) [19].
    Variable Explanation Value
    ˜T(0) Density of healthy T-cells 0.1
    ˜I(0) Density of infected T-cells 0
    ˜V(0) Density of free HIV viruses 0.1
    Parameters and constants
    s0 The generation rate of healthy T-cells 0.1
    α Proliferating of healthy T-cells 3
    μ˜T Death rate healthy T-cells 0.02
    μ˜I Death rate infected T-cells 0.3
    μ˜V Death rate of free virus 2.4
    β The infection rate 0.0027
    ˜Tmax Maximum number of healthy T-cells 1500
    γ Production rate of virus by infected T-cells 10

     | Show Table
    DownLoad: CSV

    The aforementioned model is considered as a model problem, and we implement the suggested methods. Table 19 and Figure 5 show the numerical cost in terms of time, and for all other results the readers are advised to see [19] for details. It is worth noting that the RK4 approach required very less time than the cGP(2) approach (see Table 19 and Figure 5). This is further supported by the fact that RK4 takes less time than cGP(2).

    Table 19.  In terms of CPU time (in seconds), the computational costs for Galerkin and RK4 schemes t [0, 10].
    CPU time (in seconds)
    1/τ cGP (2)-Method  RK4 -Method
    10 0.324928 0.061477
    100 0.537519 0.062665
    200 0.795505 0.073890
    500 1.355163 0.105935
    1000 2.381097 0.140199
    1500 2.967904 0.154493
    5000 6.112178 0.425491
    10000 10.263549 0.852300
    20000 19.683633 1.675098
    100000 231.860883 6.786395

     | Show Table
    DownLoad: CSV
    Figure 5.  Graphical comparison for time elapsed by cGP(2)-scheme and RK4-scheme for different time step sizes.

    In order to demonstrate that the provided temporal discretization technique is accurate and efficient in terms of time, as a test problem with the prescribed exact solution, analyze the implementations of the cGP(2) and RK4 approaches. The detail of the considered problem is given on page 332 of [42]. The problem consists of two differential equation as follows:

    dU(t)dt=4U+3V+6, (6.1)
    dV(t)dt=2.4U+1.6V+3.6, (6.2)

    where the initial values are U(0)=0 and V(0)=0. The exact solution to the aforementioned model is

    U(t)=3.375exp(2t)+1.875exp(0.4t)+1.5, (6.3)
    V(t)=2.25exp(2t)+2.25exp(0.4t). (6.4)

    Performed different numerical tests to assess the accuracy of the suggested techniques utilized an equidistant and unequal time step size τ=T/N for all tests, with T=0.5. First, find out the solutions using the cGP(2) and RK4 at τ=0.1 and then using at τ=0.00125 and τ=0.000625, respectively. To determine the numerical solution's accuracy, the numerical solutions were compared with the prescribed analytical exact solution presented in Tables 20 and 21. The computational results clearly demonstrate that the Galerkin-scheme gives highly accurate results at relatively large time step sizes as compared to RK4-scheme. For the same step size (τ=0.1), the Galerkin-scheme gained the accuracy of 105 while the RK4-method achieved the accuracy of 104. Further, RK4 achieved the accuracy of 105 at τ=0.000625, while the Galerkin-method can already reach the same precision and accuracy with τ=0.00125. This is a clear negation to the claim of [28] that Galerkin and RK4 have the same results. On the other hand, we find the order of convergence and compared the experimental orders of convergence with the theoretical order of convergence. The cGP(2)-method is of order 3 in the whole time interval and super-convergent of order 4 at the discrete time points. One can see from Table 24 that the experimental orders of convergence for the cGP(2) method are much more visible. From Table 24, we observe that the EOC coincides with the theoretical orders of convergence for corresponding time discretization schemes.

    Table 20.  Differences between the solutions of cGP(2)-scheme and exact solution for U(t) and V(t)..
    cGP(2)-method cGP(2)-method Errors Errors
    ti U(τ=0.1) V(τ=0.1) |UcGP(2)Uexact| |VcGP(2)Vexact|
    0.0 0.000000000000000 0.000000000000000 0.000000000000000E-005 0.000000000000000E-005
    0.1 0.538262676008231 0.319631223294469 0.123076418656609E-005 0.082037279897085E-005
    0.2 0.968510978848033 0.568790332547925 0.201525662668622 E-005 0.134324181744194E-005
    0.3 1.310734072202835 0.760743151897503 0.247482449533543E-005 0.164950454151214E-005
    0.4 1.581281648929932 0.906331555404320 0.270148609193832E-005 0.180050590703473E-005
    0.5 1.793524283491225 1.014413609321062 0.276457637293781E-005 0.184246865142512E-005

     | Show Table
    DownLoad: CSV
    Table 21.  Differences between the solutions of RK4-scheme and exact solution for U(t) and V(t).
    RK4-method RK4-method Errors Errors
    ti U(τ=0.1) V(τ=0.1) |URK4Uexact| |VRK4Vexact|
    0.0 0.0000000 0.0000000 0.00000E-004 0.00000E-004
    0.1 0.5382550 0.3196263 0.08285E-004 0.05803E-004
    0.2 0.9684983 0.5687817 0.15140E-004 0.09596E-004
    0.3 1.3107170 0.7607328 0.19070E-004 0.12160E-004
    0.4 1.5812630 0.9063208 0.20980E-004 0.13110E-004
    0.5 1.7935050 1.0144020 0.21930E-004 0.12400E-004

     | Show Table
    DownLoad: CSV
    Table 22.  Differences between the solutions of Galerkin-scheme and exact solution for U(t) and V(t).
    Galerkin-scheme Galerkin-scheme Errors Errors
    ti U(τ=0.00125) V(τ=0.00125) |UGalerkinUexact| |VGalerkinVexact|
    0.0 0.000000000000000 0.000000000000000 0.000000000000000E-013 0.000000000000000E-013
    0.1 0.538263906772387 0.319632043667248 0.303090885722668E-013 0.218631863105119E-013
    0.2 0.968512994104611 0.568791675789710 0.487387907810444E-013 0.399721318314869E-013
    0.3 1.310736547027270 0.760744801402005 0.608402217494586E-013 0.549991745625225E-013
    0.4 1.581284350415960 0.906333355910185 0.630606677987089E-013 0.674950994505733E-013
    0.5 1.793527048067532 1.014415451789671 0.659472476627343E-013 0.779111596277818E-013

     | Show Table
    DownLoad: CSV
    Table 23.  Differences between the solutions of RK4-scheme and exact solution for U(t) and V(t).
    RK4-method RK4-method Errors Errors
    ti U(τ=0.000625) V(τ=0.000625) |URK4Uexact| |VRK4Vexact|
    0.0 0.000000000000000 0.000000000000000 0.000000000000000E-013 0.000000000000000E-013
    0.1 0.538263906772406 0.319632043667260 0.113242748511766E-013 0.319632043667268E-013
    0.2 0.968512994104641 0.568791675789730 0.185407245112401E-013 0.568791675789742E-013
    0.3 1.310736547027307 0.760744801402030 0.233146835171283E-013 0.760744801402045E-013
    0.4 1.581284350415998 0.906333355910211 0.255351295663786E-013 0.906333355910227E-013
    0.5 1.793527048067573 1.014415451789698 0.255351295663786E-013 1.014415451789714E-013

     | Show Table
    DownLoad: CSV
    Table 24.  Comparative analysis of orders of convergence for U and V.
    cGP(2)-Scheme
    1τ UUτ EOC VVτ EOC
    20 1.38E-04 2.29 1.35E-03 0.84
    40 1.03E-05 3.75 8.86E-05 3.93
    80 6.88E-07 3.90 5.60E-06 3.98
    160 4.39E-08 3.97 3.51E-07 4.00
    320 2.75E-09 4.00 2.19E-08 4.00
    640 1.71E-10 4.01 1.37E-09 4.01

     | Show Table
    DownLoad: CSV

    Moreover, we computed the required CPU-time in seconds at different time step sizes for the mentioned above system, as given in Table 25 and plotted in Figure 6 for both methods. It can be seen from Table 25 and Figure 6 that the RK4-scheme is much faster than the Galerkin-scheme. Hence, these numerical tests show that the Galerkin approach is less efficient and takes more time as in comparison to the RK4-method.

    Table 25.  The computational costs in terms of CPU time (in seconds) for the Galerkin-scheme and RK4 scheme t [0, 0.5].
    CPU time (in seconds)
    1/τ Galerkin  RK4 -Method
    10 0.277157 0.044306
    100 0.479595 0.045998
    200 0.577860 0.051233
    500 0.925523 0.076149
    1000 1.758500 0.090703
    1500 2.478847 0.150593
    5000 5.433901 0.341417
    10000 8.990064 0.569608
    20000 14.029849 1.242083
    100000 144.766793 5.873934

     | Show Table
    DownLoad: CSV
    Figure 6.  Graphical comparison for the time elapsed by Galerkin-scheme and RK4-scheme for different time step sizes.

    In this work, a higher-order Galerkin and a fourth order classical Runge Kutta scheme were implemented. Various numerical tests were performed to determine the reliability and effectiveness of the described schemes. In addition, the impacts of increasing the parameters ξ1 and ψ2 on the dynamical behavior of the state variables were appropriately discussed. We noticed that increasing the value of ξ1 increases the population of susceptible individuals while decreasing the population of infected individuals with acute HBV, infected individuals with chronic HBV and recovered individuals. By elevating ψ2, the concentration of susceptible and recovered individuals grows, while the number of acutely and chronically infected people decreases. We executed many computational tests and compared the results quantitatively. According to computational simulations, the Galerkin-scheme provides much more accurate solutions at relatively high time step sizes than the RK4-method. Afterwards, the accuracy of the cGP(2) technique confirm that it produces more accurate results at relatively longer step sizes than the RK-4 scheme at a comparable numerical cost. The results of numerical studies revealed that the explicit RK4-scheme used less CPU time in comparison to the accuracy, but the implicit Galerkin-scheme required more CPU time. We proved that cGP(2) works better on time steps than the RK4 method. The proposed scheme could be applicable for solving complex real-world problems.

    In the future, we plan to investigate the numerical and qualitative aspects of the present concept using different fractional order derivatives. The concept of fractional analysis is well known, and it has recently become a prominent study area. The utility of fractional calculus in modeling a wide range of real-world phenomena has been demonstrated. Using fractional order derivatives and integrals, researchers have researched infectious illnesses such as HIV, AIDS, and others.

    This research received funding support from the NSRF via the Program Management Unit for Human Resources & Institutional Development, Research and Innovation, (grant number B05F650018).

    The authors declare that they have no conflicts of interest.



    [1] S. Timoshenko, S. Woinowsky-Krieger, Theory of plates and shells, New York: McGraw-hill, 1959.
    [2] P. Shankar, The eddy structure in stokes flow in a cavity, J. Fluid mech., 250 (1993), 371–383. https://doi.org/10.1017/S0022112093001491 doi: 10.1017/S0022112093001491
    [3] R. Srinivasan, Accurate solutions for steady plane flow in the driven cavity. i. stokes flow, Z. angew. Math. Phys., 46 (1995), 524–545. https://doi.org/10.1007/BF00917442 doi: 10.1007/BF00917442
    [4] V. Meleshko, Steady stokes flow in a rectangular cavity, P. Roy. Soc. A-Math. Phy., 452 (1996), 1999–2022. https://doi.org/10.1098/rspa.1996.0106 doi: 10.1098/rspa.1996.0106
    [5] P. Shankar, M. Deshpande, Fluid mechanics in the driven cavity, Annu. Rev. Fluid Mech., 32 (2000), 93–136. https://doi.org/10.1146/annurev.fluid.32.1.93 doi: 10.1146/annurev.fluid.32.1.93
    [6] N. E. Sevant, M. I. Bloor, M. J. Wilson, Aerodynamic design of a flying wing using response surface methodology, J. Aircraft, 37 (2000), 562–569. https://doi.org/10.2514/2.2665 doi: 10.2514/2.2665
    [7] M. I. Bloor, M. J. Wilson, Method for efficient shape parametrization of fluid membranes and vesicles, Phys. Rev. E, 61 (2000), 4218–4229. https://doi.org/10.1103/PhysRevE.61.4218 doi: 10.1103/PhysRevE.61.4218
    [8] V. Meleshko, Biharmonic problem in a rectangle, In: In fascination of fluid dynamics, Springer, 1998,217–249. https://doi.org/10.1007/978-94-011-4986-0_14
    [9] V. Meleshko, Selected topics in the history of the two-dimensional biharmonic problem, Appl. Mech. Rev., 56 (2003), 33–85. https://doi.org/10.1115/1.1521166 doi: 10.1115/1.1521166
    [10] V. Meleshko, Bending of an elastic rectangular clamped plate: Exact versus 'engineering'solutions, J. Elasticity, 48 (1997), 1–50. https://doi.org/10.1023/A:1007472709175 doi: 10.1023/A:1007472709175
    [11] V. Meleshko, A. Gomilko, Infinite systems for a biharmonic problem in a rectangle, P. Roy. Soc. A Math. Phy., 453 (1997), 2139–2160. https://doi.org/10.1098/rspa.1997.0115 doi: 10.1098/rspa.1997.0115
    [12] S. Antontsev, S. Shmarev, Blow-up of solutions to parabolic equations with nonstandard growth conditions, J. Comput. Appl. Math., 234 (2010), 2633–2645. https://doi.org/10.1016/j.cam.2010.01.026 doi: 10.1016/j.cam.2010.01.026
    [13] S. Antontsev, S. Shmarev, Evolution pdes with nonstandard growth conditions, In: Atlantis studies in differential equations, 2015. https://doi.org/10.2991/978-94-6239-112-3
    [14] B. Guo, W. Gao, Blow-up of solutions to quasilinear hyperbolic equations with p(x, t)-Laplacian and positive initial energy, C. R. Mecanique, 342 (2014), 513–519. https://doi.org/10.1016/j.crme.2014.06.001 doi: 10.1016/j.crme.2014.06.001
    [15] S. A. Messaoudi, O. Bouhoufani, I. Hamchi, M. Alahyane, Existence and blow up in a system of wave equations with nonstandard nonlinearities, Electron. J. Differ. Eq., 2021 (2021), 1–33.
    [16] S. A. Messaoudi, A. A. Talahmeh, M. M. Al-Gharabli, M. Alahyane, On the existence and stability of a nonlinear wave system with variable exponents, Asymptotic Anal., 128 (2022), 211–238. https://doi.org/10.3233/ASY-211704 doi: 10.3233/ASY-211704
    [17] S. H. Park, J. R. Kang, Blow-up of solutions for a viscoelastic wave equation with variable exponents, Math. Method. Appl. Sci., 42 (2019), 2083–2097. https://doi.org/10.1002/mma.5501 doi: 10.1002/mma.5501
    [18] O. Bouhoufani, I. Hamchi, Coupled system of nonlinear hyperbolic equations with variable-exponents: Global existence and stability, Mediterr. J. Math., 17 (2020), 166. https://doi.org/10.1007/s00009-020-01589-1 doi: 10.1007/s00009-020-01589-1
    [19] L. Diening, P. Harjulehto, P. Hästö, M. Ruzicka, Lebesgue and Sobolev spaces with variable exponents, Springer, 2011.
    [20] D. V. Cruz-Uribe, A. Fiorenza, Variable Lebesgue spaces: Foundations and harmonic analysis, Springer Science & Business Media, 2013.
    [21] K. Agre, M. A. Rammaha, Systems of nonlinear wave equations with damping and source terms, Differ. Integral Equ., 19 (2006), 1235–1270. https://doi.org/10.57262/die/1356050301 doi: 10.57262/die/1356050301
    [22] C. O. Alves, M. M. Cavalcanti, V. N. D. Cavalcanti, M. A. Rammaha, D. Toundykov, On existence, uniform decay rates and blow up for solutions of systems of nonlinear wave equations with damping and source terms, Discrete Cont. Dyn. S, 2 (2009), 583–608. https://doi.org/10.3934/dcdss.2009.2.583 doi: 10.3934/dcdss.2009.2.583
    [23] V. Komornik, Decay estimates for the wave equation with internal damping, In: Control and estimation of distributed parameter systems: Nonlinear phenomena, 1994. https://doi.org/10.1007/978-3-0348-8530-0_14
  • This article has been cited by:

    1. Kamil Zeb, Ilyas Khan, Riaz Ahmad, Sayed M. Eldin, Transmission dynamics of a novel HIV/AIDS model through a higher-order Galerkin time discretization scheme, 2023, 13, 2045-2322, 10.1038/s41598-023-34696-6
    2. Rahila Naz, Aasim Ullah Jan, Salah Boulaaras, Rafik Guefaifia, Computational analysis of the Covid-19 model using the continuous Galerkin–Petrov scheme, 2024, 13, 2192-8029, 10.1515/nleng-2024-0028
    3. Khaled Mahdi, Mehmet Yavuz, Salah Boulaaras, Mohamed Haiour, Viet-Thanh Pham, Computational approaches on integrating vaccination and treatment strategies in the SIR model using Galerkin time discretization scheme, 2024, 30, 1387-3954, 758, 10.1080/13873954.2024.2405504
    4. Salah Boulaaras, Aasim Ullah Jan, Tahir Hassan, Taha Radwan, Mathematical modeling and computational analysis of hepatitis B virus transmission using the higher-order Galerkin scheme, 2024, 13, 2192-8029, 10.1515/nleng-2024-0048
  • Reader Comments
  • © 2023 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(1761) PDF downloads(112) Cited by(7)

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog