Research article Special Issues

Almost periodic solutions for a SVIR epidemic model with relapse


  • This paper is devoted to a nonautonomous SVIR epidemic model with relapse, that is, the recurrence rate is considered in the model. The permanent of the system is proved, and the result on the existence and uniqueness of globally attractive almost periodic solution of this system is obtained by constructing a suitable Lyapunov function. Some analysis for the necessity of considering the recurrence rate in the model is also presented. Moreover, some examples and numerical simulations are given to show the feasibility of our main results. Through numerical simulation, we have obtained the influence of vaccination rate and recurrence rate on the spread of the disease. The conclusion is that in order to control the epidemic of infectious diseases, we should increase the vaccination rate while reducing the recurrence rate of the disease.

    Citation: Yifan Xing, Hong-Xu Li. Almost periodic solutions for a SVIR epidemic model with relapse[J]. Mathematical Biosciences and Engineering, 2021, 18(6): 7191-7217. doi: 10.3934/mbe.2021356

    Related Papers:

    [1] Ning Huang, Zhengtao Xi, Yingying Jiao, Yudong Zhang, Zhuqing Jiao, Xiaona Li . Multi-modal feature fusion with multi-head self-attention for epileptic EEG signals. Mathematical Biosciences and Engineering, 2024, 21(8): 6918-6935. doi: 10.3934/mbe.2024304
    [2] Zizhuo Wu, Qingshan She, Zhelong Hou, Zhenyu Li, Kun Tian, Yuliang Ma . Multi-source online transfer algorithm based on source domain selection for EEG classification. Mathematical Biosciences and Engineering, 2023, 20(3): 4560-4573. doi: 10.3934/mbe.2023211
    [3] Jin Zhang, Nan Ma, Zhixuan Wu, Cheng Wang, Yongqiang Yao . Intelligent control of self-driving vehicles based on adaptive sampling supervised actor-critic and human driving experience. Mathematical Biosciences and Engineering, 2024, 21(5): 6077-6096. doi: 10.3934/mbe.2024267
    [4] Ansheng Ye, Xiangbing Zhou, Kai Weng, Yu Gong, Fang Miao, Huimin Zhao . Image classification of hyperspectral remote sensing using semi-supervised learning algorithm. Mathematical Biosciences and Engineering, 2023, 20(6): 11502-11527. doi: 10.3934/mbe.2023510
    [5] Sakorn Mekruksavanich, Wikanda Phaphan, Anuchit Jitpattanakul . Epileptic seizure detection in EEG signals via an enhanced hybrid CNN with an integrated attention mechanism. Mathematical Biosciences and Engineering, 2025, 22(1): 73-105. doi: 10.3934/mbe.2025004
    [6] Xu Yin, Ming Meng, Qingshan She, Yunyuan Gao, Zhizeng Luo . Optimal channel-based sparse time-frequency blocks common spatial pattern feature extraction method for motor imagery classification. Mathematical Biosciences and Engineering, 2021, 18(4): 4247-4263. doi: 10.3934/mbe.2021213
    [7] Xiaotong Ding, Lei Yang, Congsheng Li . Study of MI-BCI classification method based on the Riemannian transform of personalized EEG spatiotemporal features. Mathematical Biosciences and Engineering, 2023, 20(7): 12454-12471. doi: 10.3934/mbe.2023554
    [8] Tianhui Sha, Yikai Zhang, Yong Peng, Wanzeng Kong . Semi-supervised regression with adaptive graph learning for EEG-based emotion recognition. Mathematical Biosciences and Engineering, 2023, 20(6): 11379-11402. doi: 10.3934/mbe.2023505
    [9] Jiacan Xu, Donglin Li, Peng Zhou, Chunsheng Li, Zinan Wang, Shenghao Tong . A multi-band centroid contrastive reconstruction fusion network for motor imagery electroencephalogram signal decoding. Mathematical Biosciences and Engineering, 2023, 20(12): 20624-20647. doi: 10.3934/mbe.2023912
    [10] Yue Zhang, Haitao Gan, Furong Wang, Xinyao Cheng, Xiaoyan Wu, Jiaxuan Yan, Zhi Yang, Ran Zhou . A self-supervised fusion network for carotid plaque ultrasound image classification. Mathematical Biosciences and Engineering, 2024, 21(2): 3110-3128. doi: 10.3934/mbe.2024138
  • This paper is devoted to a nonautonomous SVIR epidemic model with relapse, that is, the recurrence rate is considered in the model. The permanent of the system is proved, and the result on the existence and uniqueness of globally attractive almost periodic solution of this system is obtained by constructing a suitable Lyapunov function. Some analysis for the necessity of considering the recurrence rate in the model is also presented. Moreover, some examples and numerical simulations are given to show the feasibility of our main results. Through numerical simulation, we have obtained the influence of vaccination rate and recurrence rate on the spread of the disease. The conclusion is that in order to control the epidemic of infectious diseases, we should increase the vaccination rate while reducing the recurrence rate of the disease.



    Abbreviations: Aij: area of wall cell (i,j), m2; e: energy carried by sample Monte Carlo bundle, W; Eb(T): spectral blackbody emissive power at temperature T, W/(m2·m); H: total exercise times; L: length of the domain, m; M, N: lattice indices in square mesh corresponding to x, y position, respectively; n: refractive index of medium; Nij: number of sample bundles emitted by cell (i,j); Nijkl: number of sample bundles emitted by cell (i,j) and absorbed by cell (k,l); Nij: number of sample bundles emitted by cell (i,j) in the BDMC method; Nijkl: number of sample bundles emitted by cell (i,j) and absorbed by cell (k,l)in the BDMC method; N0: the number density of sampling bundles for a reference case; Nr: the number of sample bundles chosen for a reference volume cell Vr; P: performance parameter for Monte Carlo simulation; RDλ,ijkl: radiative exchange factor of cell (i,j) to cell (k,l) for non-gray medium in the TMC method; ¯RDλ,ijkl: modified form of RDλ,ijkl; RDijkl: radiative exchange factor of cell (i,j) to cell (k,l) for gray medium in the TMC method; ¯RDijkl: modified form of RDijkl; RDijkl: radiative exchange factor of cell (i,j) to cell (k,l) in the BDMC method; ¯RDijkl: modified form of RDijkl; sn, ˆsn: bundles trajectories and coupled bundles trajectories; t: time, s; T: absolute temperature, K; Vij: volume of medium cell (i,j), m3; Wij: bundle weight of cell (i,j); δijkl: reciprocity error of a couple of radiative exchange factor by the TMC method; δijkl: reciprocity error of a couple of radiative exchange factor by the BDMC method; ΔP: variation of performance parameter; Δr: relative error of reciprocity; ελ: spectral emissivity; Φλ,ij: spectral radiative energy emitted from cell Vij (or Aij); Φλ,ijkl: spectral radiative energy emitted from cell Vij (or Aij), that is absorbed by cell Vkl (or Akl); Φijkl: net radiative energy exchange, W; γ: standard deviation; κaλ: spectral absorption coefficient of medium; κe: extinction coefficient, m-1; σ: Stefan-Boltzmann constant, (σ=5.67×108W/m2K4); τ: optical thickness (τ=κeL); ω: scattering albedo of medium. Subscripts, a: medium absorption; h: hth sampling; i,j: order number of cell; r: reference cell; w: wall

    Radiative heat transfer in participating medium is described by radiative transfer equation (RTE), which is an integro-differential equation. Many methods have been developed to solve the RTE, the common used numerical methods such as the discrete ordinate method (DOM) and the finite volume method (FVM), rely on both spatial discretization and angular discretization, which are difficult to solve the RTE with high accuracy. Monte Carlo (MC) method is a stochastic statistical method based on the physical processes, it has been applied to solve radiative heat transfer in various participating media [1,2,3,4,5,6]. Moreover, the results predicted by MC method can often be treated as benchmark solutions due to its high accuracy of solution [7,8].

    However, for the statistical nature of the MC method, the high computational cost is still a considerable disadvantage of MC simulation. Many attempts have been made in order to improve the computational efficiency of the method [9]. For example, several sampling approaches were developed to improve the speed of convergence, such as the importance sampling method [10], the rejection sampling method [11], the differential sampling method [12] and the weight-equivalent sampling method [13]. In addition, parallel computing technique was introduced into the MC simulation to improve computation efficiency evidently [1,9]. Howell [9] analyzed the advantages of various programming strategies of the MC method for radiative heat transfer in absorbing and scattering medium. All of these improvements were mainly mathematical efforts and hardware improvement of computer, which are universal to the MC simulations.

    It has been noticed that the physical feature of thermal radiation transfer can also provide possibility to improve the MC simulation of radiative heat transfer in participating medium. For example, Walters [14] developed a reverse method based on the reciprocity principle for radiative heat transfer in a generalized enclosure containing an absorbing, emitting and scattering medium, the reverse method was proved to be efficient. Cherkaoui et al. [15,16] developed a net exchange Monte Carlo method based on a net-exchange formulation, provided an efficient way of systematically fulfilling the reciprocity principle, the computing time was proved much smaller than the conventional Monte Carlo approach. Lataillade [17] and his cooperators applied the net exchange Monte Carlo approach for radiative heat transfer in optically thick medium with spectral dependent radiative properties. Eymet et al. [18] extended this method to absorbing, emitting, and scattering media. Tessé et al. [19] improved the forward Monte Carlo (FMC) method based on the reciprocity principle, the method was used for radiative transfer in real gases, and proved to be a better choice for optically thick or nearly isothermal media compared with the forward Monte Carlo method. In addition, the reverse Monte Carlo (RMC) methods, based on the reversibility of radiative transfer trajectory, has been developed to solve the radiative transfer in absorbing and scattering media [20,21,22,23,24]. Kovtanyuk et al. [25] presented a recursive algorithm based on modification of Monte Carlo method, the modified method was used to solve the coupled radiative and conductive heat transfer in an absorbing and scattering medium, and was proved to be more accurate. Soucasse et al. [26] proposed a Monte Carlo formulation for radiative transfer in quasi-isothermal media which consists in directly computing the difference between the actual radiative field and the equilibrium radiative field at the minimum temperature in the medium.

    In the present study, a bidirectional Monte Carlo (BDMC) method based on reversibility of bundle trajectory and reciprocity of radiative energy exchange was developed to solve thermal radiation transfer in absorbing and scattering medium. Two types of sampling models for MC simulation were presented, namely the equivalent sampling and the weight sampling. The equivalent sampling was chosen for the uniform mesh while the weight sampling was more suitable to the non-uniform mesh. The bidirectional information of tracing a sampling bundle was utilized by the BDMC method, the solution precision or efficiency can be evidently improved. Radiative heat transfer in a two-dimensional rectangular domain with absorbing and scattering media was solved by the BDMC method and the TMC method, respectively. The radiative exchange factors and the temperature profiles were investigated, in addition, the performance parameter defined by Howell [9] was also calculated to evaluate the two MC methods.

    Radiative heat transfer in a two-dimensional (2-D) rectangular domain with absorbing, emitting and/or scattering medium was investigated. Figure 1 shows the 2-D rectangular geometry as well as the coordinate system. The four walls were assumed to be diffuse and gray. Radiative properties, such as the absorption coefficient, the scattering coefficient were assumed to be constant. The rectangular domain was divided into M × N = MN cells, any of which was depicted by Vij (medium cell) or Aij (wall cell), wherein the subscripts I∈[1, M] and j∈[1, N]. Monte Carlo method has robust adaptability and can be extended to more complex cases.

    Figure 1.  Schematic diagram of radiative transfer in a rectangular domain with participating medium in the BDMC method.

    In the present study, the radiative exchange factor RDλ,ijkl was introduced to decouple the solution of radiative transfer from that of the temperature profile. It was defined as a fraction of the spectral radiative energy emitted from cell Vij (or Aij), that is absorbed by cell Vkl (or Akl) [27,28].

    RDλ,ijkl=Φλ,ijkl/Φλ,ij (1)

    As Eq 1, where Φλ,ij is the spectral radiative energy emitted from cell Vij (or Aij), Φλ,ijkl is the spectral radiative energy emitted from cell Vij (or Aij), and absorbed by cell Vkl (or Akl), taking into account possible wall reflections. It is obvious that the conservation relation M,Nk=1,l=1RDλ,ijkl=1 is tenable according to the definition of radiative exchange factors. Note that the radiative exchange factor depends only on the system geometry and the radiative properties distribution of the medium [29]. The reciprocity relation between a couple of radiative exchange factors was given by [30].

    (κaλ)ijVijRDλ,ijkl=(κaλ)klVklRDλ,klij (2)

    or

    4(κaλ)ijVijRDλ,ijkl=(ελ)klAklRDλ,klij (3)

    As Eqs 2 and 3, where, κaλ is the spectral absorption coefficient of the medium, while ελ is the spectral emissivity of the boundary wall. For convenience, the radiative exchange factors were usually modified as

    ¯RDλ,ijkl=4(κaλ)ijVijRDλ,ijkl (4)

    or

    ¯RDλ,ijkl=(ελ)ijAijRDλ,ijkl (5)

    here, Eqs 4 and 5 are used for volume and wall cells, respectively. The reciprocity relation can be written as

    ¯RDλ,ijkl=¯RDλ,klij (6)

    As Eq 6, after solving the radiative exchange factors, the net radiative energy exchange Φijkl from Vij to Vkl can be calculated from

    Φijkl=0¯RDλ,ijkl[Ebλ(Tij)Ebλ(Tkl)]dλ (7)

    As Eq 7, where, is the Stefan-Boltzmann constant ( = 5.67×10-8 W/m2K4), E(T) is the spectral blackbody emissive power at temperature T.

    In the MC simulation, Φλ,ij is represented by a large number of independent sampling bundles, the propagating process of each sampling bundle can be tracked and counted. For example, the number of total sampling bundles for Φλ,ij is Nij, each bundle with the same energy of eij=Φλ,ij/Nij. If Nijkl bundles among them are finally absorbed by Vkl, then, Φλ,ijkl=eijNijkl, the radiative exchange factor can be calculated from [31].

    RDλ,ijkl=Nijkl/Nij (8)

    As Eq 8, the number Nijkl can be counted from the MC simulation, however, only an approximate value of Nijkl can be obtained because of the pseudo randomicity [32] in the sampling process. In fact, the reciprocity correlation shown in Eq 6 cannot be strictly satisfied, and random errors always exist. In order to obtain more accurate results, one need to increase the sampling bundles Nij, which results in the increasing computing cost.

    In the TMC method, the propagation trajectories of Nij sampling bundles for cell Vij are tracked and counted to get the value of Nijkl for any cell Vkl. Similarly, the number Nklij is obtained after tracing another Nkl propagation trajectories of the sampling bundles for cell Vkl. Wherein, only the forward propagation information of a trajectory is used, the number of sampling bundles for cell Vij is Nij.

    According to the reversibility of light propagation, for a tracing trajectory from Vkl to Vij, a bundle from Vij can also propagate to Vkl along the reverse direction of the tracing trajectory. The information of the reverse direction can be used in the MC simulation, the bidirectional information can be obtained from the forward tracing, which results in the development of the present BDMC method.

    In order to simplify the description of the BDMC method, the formula in the following sections were given for gray medium with gray walls. The basic idea of BDMC is to calculate the radiative exchange factor by the reversibility of light beams. According to the reversibility of light beams, the transmission path of the bundles received by cells can also be used as the transmission path of the bundles transmitted by cells. In this way, the effective information of sampling bundles if fully utilized without increasing the number of sampling bundles and calculation amount. Theoretically, the statistical sampling bundles in calculation can be doubled. And it can better satisfy the relationship of the reciprocal of radiative exchange factor. Figure 1 shows the schematic diagram of radiative transfer in a rectangular domain with participating medium in the BDMC method, it can be noticed that there are Nij tracing trajectories departing from cell Vij, and M,Nk=1,l=1Nklij tracing trajectories getting into cell Vij at the same time. According to the reversibility of bundle trajectories, there must be M,Nk=1,l=1Nklij tracing trajectories depart from cell Vij in the reverse directions. It is believed that the M,Nk=1,l=1Nklij trajectories are independent of the emitted Nij trajectories because of the randomness of the sampling process. Thus, the number of total sampling bundles for Vij can be counted anew as

    Nij=Nij+M,Nk=1,l=1Nklij (9)

    while the number of bundles emitted by Vij and absorbed by Vkl is counted anew as

    Nijkl=Nijkl+Nklij (10)

    then, the radiative exchange factor RDijkl in the BDMC method can be calculated by

    RDijkl=NijklNij=Nijkl+NklijNij+M,Nk=1,l=1Nklij (11)

    where, RDijkl is used to denote the radiative exchange factor in the BDMC method. The information of the bidirectional trajectories has been used, as is shown in Eq 11, it indicates that the effective sampling bundles are increased, while the computing cost remains unchanged. Therefore, more accurate results can be predicted from Eq 11 than that from Eq 8 under the same computing cost.

    It should be pointed out that Eqs 9 and 11 are valid only if the sampling bundles for cell Vij and any other cells are equivalent. According to the reversibility of light beams, the transmission path of the bundles received by cells can also be used as the transmission path of the bundles transmitted by cells. In this way, the effective information of sampling bundles if fully utilized without increasing the number of sampling bundles and calculation amount. In other words, the forward and reverse bundles in the BDMC method should have the same contribution for radiative exchange factors calculation. To use Eq 11, the equivalent sampling was introduced, in which the number of sampling bundles for cell (i, j) was determined by

    Nij=4(κa)ijVijN0orNij=εijAijN0 (12)

    where N0 is the sampling density for a reference case defined as

    N0=Nr4(κa)rVr (13)

    where, Nr is the number of sample bundles of the reference cell Vr, the subscript r refers to the reference cell.

    The bidirectional counting of sampling bundles employed in the BDMC method is equivalent to double the sampling number if the random error does not exist. In fact, any of the Monte Carlo methods always accompanied by random error. For the BDMC method, that leads to Nij2Nij and the energy of a bundle is not strictly equal to other cells. For any cell, it can be easily demonstrated by Eq 11, and the radiative exchange factors calculated by the BDMC method satisfied the conservation relation strictly. But the reciprocity relation between a couple of radiative exchange factors cannot be satisfied strictly. According to Eq 11, the following equation can be obtained

    RDijkl(Nij+M,Nk=1,l=1Nklij)=RDklij(Nkl+M,Ni=1,j=1Nijkl) (14)

    According to Eq 8, for gray medium, Nijkl=NijRDijkl, Nklij=NklRDklij. Therefore,

    RDijkl(Nij+M,Nk=1,l=1NklRDklij)=RDklij(Nkl+M,Ni=1,j=1NijRDijkl) (15)

    replace Nij and Nkl according to Eq 12, in addition, considering Eqs 4 and 5, then Eq 15 can be transformed into

    ¯RDijkl+RDijklM,Nk=1,l=1¯RDklij=¯RDklij+RDklijM,Ni=1,j=1¯RDijkl (16)

    where, ¯RDijkl and ¯RDklij are the modified forms for RDijkl and RDklij, respectively.

    Supposing the reciprocity errors for a couple of radiative exchange factors ¯RDijkl and ¯RDklij predicted in the TMC method are δijkl and δklij, respectively, one can write

    ¯RDklij=¯RDijkl+δijkland¯RDijkl=¯RDklij+δklij (17)

    where, it is obvious that δklij=δijkl.

    For radiative exchange factors predicted in the BDMC method, similar equations can be written as follows

    ¯RDklij=¯RDijkl+δijkland¯RDijkl=¯RDklij+δklij (18)

    where, δijkl and δklij are the reciprocity errors for a couple of radiative exchange factors ¯RDijkl and ¯RDklij in the BDMC method, and δklij=δijkl.Combine Eqs 16–18, then

    δklij=RDklijM,Ni=1,j=1(¯RDklij+δklij)RDijklM,Nk=1,l=1(¯RDijkl+δijkl) (19)

    Considering the conservation relation M,Ni=1,j=1RDλ,klij=1, M,Nk=1,l=1RDλ,ijkl=1, δklij=δijkl, δklij=δijkl and Eq 4, then, M,Ni=1,j=1¯RDklij=4(κa)klVkl, and M,Nk=1,l=1¯RDijkl=4(κa)ijVij, moreover, considering Eq 18, the following error relation can be derived from Eq 19

    δijkl=12(RDijklM,Nk=1,l=1δijkl+RDklijM,Ni=1,j=1δijkl) (20)

    therefore,

    |δijkl|12(RDijkl|M,Nk=1,l=1δijkl|+RDklij|M,Ni=1,j=1δijkl|) (21)

    Because of the randomicity of the reciprocity error δijkl, the value of |M,Nk=1,l=1δijkl| would be very close to 0 if M and N tend to infinity. The formulae |M,Nk=1,l=1δijkl|=O(|δijkl|) and |M,Nk=1,l=1δijkl|=O(|δijkl|) would be valid for most of the discrete elements as M×N was large in the present study, where O(|δijkl|) is less than |δijkl|. Note that the values of RDijkl and RDklij do not greater than unity, therefore

    |δijkl|O(|δijkl|) (22)

    The reciprocity error for a couple of radiative exchange factors predicted by the BDMC method is always smaller than those predicted by the TMC method.

    For some cases, the equivalent sampling may be inconvenient or inaccurate. For example, if the differences of apparent radiation characteristics and/or cell volume are very large among different cells, the equivalent sampling will result in a very small sampling number for a cell, and may be a very large sampling number for another. The former is not expected for the statistical analysis, while the latter increase computational cost.

    The weight sampling was introduced into the BDMC method to avoid the above shortages of the equivalent sampling. For weight sampling, the number of sampling bundles for a cell was determined only based on the statistical request, but the weight of a sampling bundle was taken into account in the final statistical calculation. If the number of sampling bundles for cell Vij was Nij, then, its bundle weight Wij was defined as

    Wij=(κa)ijVijNr(κa)rVrNij,orWij=εijAijNr4(κa)rVrNij (23)

    where Nr is the number of sample bundles for reference cell Vr. Then, the radiative exchange factor RDijkl in the BDMC method should be calculated by

    RDijkl=Nijkl+NklijWkl/WijNij+M,Nk=1,l=1(NklijWkl/Wij) (24)

    Eq 11 should be substituted by Eq 24 if the weight sampling is employed.

    Radiative transfer in a rectangular domain with absorbing, emitting and isotropic scattering gray medium was solved separately by the BDMC method and the TMC method. The radiative exchange factors, the radiative equilibrium temperature profiles, and the performance parameter defined by Farmer and Howell [1,9], predicted by the BDMC method were compared with those predicted by the TMC method. In addition, the performance of the equivalent sampling and the weight sampling in the MC simulation were also examined.

    Table 1.  Discrete parameters of the uniform and non-uniform mesh systems.
    (i[2,M1], j[2,N1])
    mesh number mesh size
    uniform mesh M×N=21×21 Δxi,j=Δyi,j=const
    non-uniform mesh M×N=21×21 Δxi,j=ΔxcK|iM+12|, Δyi,j=ΔycK|jN+12|

     | Show Table
    DownLoad: CSV

    The radiative properties of the medium considered were supposed to be uniform and depicted by constant extinction coefficient κe, scattering albedo ω, and refractive index n = 1.0. The length and width of the rectangular domain were Lx=Ly=L. Optical thickness was defined as τ=κeL. The walls were assumed to be diffuse and gray, with constant emissivity of ε. The surfaces were imposed to constant temperature of TWE, TWW, TWN, and TWS, respectively, see Figure 1. The dimensionless coordinates were defined as X=x/Lx, and Y=y/Ly. The simulations were conducted using both uniform and non-uniform grids. The discrete parameters of the two types of mesh systems were listed in Table 1, where, xc = yc were the width and length of the largest volume cell in the non-uniform mesh system, and xc = yc, the parameter K = 1.04, M = N = 21, the mesh sizes in the non-uniform mesh system were expressed as Δxij=ΔxcK|i11|, and Δyij=ΔycK|j11|. For medium cells, i.e., i[2,M1], and j[2,N1], the maximum width of the cells was Δxi,j=Δxc for i = 11, and the minimum width of the cells was Δxi,j=ΔxcK9 for i = 2 or 20. Similarly, the maximum length of the cells was Δyi,j=Δyc for j = 11, while the minimum length of the cells was Δyi,j=ΔycK9 for j = 2 or 20. The maximum/minimum ratio of the mesh size is (ΔxcΔyc)/(ΔxcK9ΔycK9) = 2.026.

    The sampling bundles needed in the MC simulation should be estimated as the results were calculated based on statistical computation. Figures 2 and 3 show the radiative equilibrium temperature profiles at dimensionless locations of Y = 0.3 and Y = 0.5 predicted by the TMC method with different sampling bundles of Nb = 104, 105, and 106, respectively. Figure 2 shows the results for absorbing medium, while the results shown in Figure 3 were for absorbing and isotropic scattering medium with scattering albedo of = 0.5. For both the two cases, the wall temperatures were imposed separately as TWE = TWW = 1000 K and TWN = TWS = 1500 K, the emissivity of the walls were assumed to be constant and equal to 0.5, the optical thickness = keL = 15. For both the absorbing and absorbing-scattering medium, the predicted temperature profiles were found to be close to each other for different sampling bundles of Nb = 104, 105, and 106, moreover, the temperature difference for Nb = 105 and Nb = 106 were less than 0.5%, therefore, the results were considered to reach convergence for sampling bundles Nb = 105 in the present study, and the sampling bundles of Nb = 105 was also chosen as the reference sampling bundles Nr in the BDMC method.

    Figure 2.  Radiative equilibrium temperature profiles predicted by the TMC method in absorbing, emitting, non-scattering medium at locations Y = 0.3 and 0.5.
    Figure 3.  Radiative equilibrium temperature profiles predicted by the TMC method in absorbing, emitting and isotropic scattering medium (scattering albedo = 0.5) at Y = 0.3 and 0.5.

    In this section, the radiative exchange factors predicted by the BDMC method were compared with those predicted by the TMC method, the numerical simulations were implemented using both uniform and non-uniform grids, meanwhile, the equivalent sampling and the weight sampling were also employed separately. The predicted radiative exchange factors always satisfy the conservation relation strictly, therefore, only the reciprocity for radiative exchange factors was examined in the present study. The relative error of the reciprocity for a couple of radiative exchange factors was defined as

    Δr=2|¯RDijkl¯RDklij|¯RDijkl+¯RDklij×100% (25)

    where, ¯RDijkl refers not only to the modified form of the radiative exchange factors given by Eq 8 for the TMC method, but also to those given by Eq 11 (BDMC method, equivalent sampling) and by Eq 24 (BDMC method, weight sampling).

    Table 2 shows the modified radiative exchange factors predicted by the BDMC method and the TMC method employing the equivalent sampling and the uniform grids as well as the corresponding relative error Dr defined by Eq 25. The emissivity of the walls were taken as = 0.5, the optical thicknesses of the rectangle medium along the length and width direction were τ=1.0, the reference sampling bundles was taken as Nr=105. It indicates that the reciprocity error for a couple of radiative exchange factors predicted by the BDMC method was at least an order of magnitude less than those predicted by the TMC method with the same sampling bundles. The maximum relative error for all the elements considered did not exceed 0.58% in the BDMC method, while it reached 30.1% in the TMC method.

    Table 2.  Comparison of radiative exchange factors predicted by the BDMC method and the TMC method using the equivalent sampling and the uniform grids.
    (ε=0.5, τ=1.0, ω=0.0, Nr=105, Ni,j=7.22×107)
    Cell 1 Cell 2 TMC method BDMC method
    (i, j) (k, l ) ¯RDi,jk,l ¯RDk,li,j Δr(%) ¯RDi,jk,l ¯RDk,li,j Δr(%)
    (11, 11) (1, 11) 1.115 × 10-4 1.065 × 10-4 4.59 1.091 × 10-4 1.091 × 10-4 0.00
    (11, 11) (11, 1) 1.044 × 10-4 1.098 × 10-4 5.04 1.071 × 10-4 1.066 × 10-4 0.47
    (11, 11) (11, 21) 1.014 × 10-4 1.065 × 10-4 4.90 1.040 × 10-4 1.041 × 10-4 0.10
    (11, 11) (21, 11) 1.021 × 10-4 1.098 × 10-4 7.27 1.060 × 10-4 1.061 × 10-4 0.09
    (11, 11) (2, 2) 1.410 × 10-5 1.747 × 10-5 21.3 1.579 × 10-5 1.586 × 10-5 0.44
    (11, 11) (2, 20) 1.936 × 10-5 1.684 × 10-5 13.9 1.811 × 10-5 1.813 × 10-5 0.11
    (11, 11) (20, 2) 1.305 × 10-5 1.768 × 10-5 30.1 1.537 × 10-5 1.546 × 10-5 0.58
    (11, 11) (20, 20) 1.452 × 10-5 1.873 × 10-5 25.3 1.663 × 10-5 1.670 × 10-5 0.42

     | Show Table
    DownLoad: CSV

    Table 3 shows the radiative exchange factors predicted by the BDMC method using the equivalent sampling and the weight sampling in a non-uniform mesh system. The computing parameters were ε=0.5, τ=1.0, ω=0.0, and Nr=105. The reciprocity error for a couple of radiative exchange factors predicted by the BDMC method with the equivalent sampling was unacceptable for the non-uniform mesh, it indicates that the equivalent sampling was not recommended to solve radiative transfer in the BDMC method with non-uniform grids. If there is obvious difference of heat capacity between cells, the equivalent sampling will lead to big difference of sampling number for the same cells. The calculation accuracy will come under influence if the equivalent sampling is applied. However, the BDMC method with the weight sampling can predict quite satisfactory results for all the cells considered.

    Table 3.  Comparison of radiative exchange factors predicted by the BDMC method with the equivalent sampling and the weight sampling using the non-uniform grids.
    (ε=0.5, τ=1.0, ω=0.0, Nr=105, Ni,j=8.12×107)
    Cell 1 Cell 2 BDMC (equivalent sampling) BDMC (weight sampling)
    (i, j) (k, l ) ¯RDi,jk,l ¯RDk,li,j Δr(%) ¯RDi,jk,l ¯RDk,li,j Δr(%)
    (11, 11) (1, 11) 6.730 × 10-4 5.000 × 10-4 29.5 9.413 × 10-4 9.414 × 10-4 0.01
    (11, 11) (11, 1) 6.697 × 10-4 4.950 × 10-4 30.0 9.364 × 10-4 9.364 × 10-4 0.00
    (11, 11) (11, 21) 6.541 × 10-4 4.855 × 10-4 29.6 9.257 × 10-4 9.239 × 10-4 0.19
    (11, 11) (21, 11) 6.654 × 10-4 4.940 × 10-4 29.6 9.335 × 10-4 9.332 × 10-4 0.03
    (11, 11) (2, 2) 6.993 × 10-5 2.837 × 10-5 84.6 4.374 × 10-5 4.384 × 10-5 0.23
    (11, 11) (2, 20) 7.047 × 10-5 2.855 × 10-5 84.7 5.052 × 10-5 5.049 × 10-5 0.06
    (11, 11) (20, 2) 6.849 × 10-5 2.783 × 10-5 84.4 4.442 × 10-5 4.452 × 10-5 0.22
    (11, 11) (20, 20) 7.317 × 10-5 2.971 × 10-5 84.5 4.378 × 10-5 4.386 × 10-5 0.18

     | Show Table
    DownLoad: CSV

    Table 4 shows the radiative exchange factors predicted by the BDMC method and the TMC method employing the weight sampling and the non-uniform grids. It can be seen that the reciprocity error in the TMC method was larger than those in the BDMC method for all the cells considered. The BDMC method showed greatly superior to the TMC method when the weight sampling and the non-uniform grids was used.

    Table 4.  Comparison of radiative exchange factors by the BDMC method and the TMC method using the weight sampling and the non-uniform grids.
    (ε=0.5, τ=1.0, ω=0.0, Nr=105, Ni,j=8.12×107)
    Cell 1 Cell 2 TMC method BDMC method
    (i, j) (k, l) ¯RDi,jk,l ¯RDk,li,j (i, j) (k, l) ¯RDi,jk,l ¯RDk,li,j
    (11, 11) (1, 11) 9.492 × 10-4 9.344 × 10-4 (11, 11) (1, 11) 9.492 × 10-4 9.344 × 10-4
    (11, 11) (11, 1) 9.485 × 10-4 9.252 × 10-4 (11, 11) (11, 1) 9.485 × 10-4 9.252 × 10-4
    (11, 11) (11, 21) 9.180 × 10-4 9.344 × 10-4 (11, 11) (11, 21) 9.180 × 10-4 9.344 × 10-4
    (11, 11) (21, 11) 9.428 × 10-4 9.252 × 10-4 (11, 11) (21, 11) 9.428 × 10-4 9.252 × 10-4
    (11, 11) (2, 2) 4.156 × 10-5 4.596 × 10-5 (11, 11) (2, 2) 4.156 × 10-5 4.596 × 10-5
    (11, 11) (2, 20) 5.664 × 10-5 4.446 × 10-5 (11, 11) (2, 20) 5.664 × 10-5 4.446 × 10-5
    (11, 11) (20, 2) 4.223 × 10-5 4.667 × 10-5 (11, 11) (20, 2) 4.223 × 10-5 4.667 × 10-5
    (11, 11) (20, 20) 3.988 × 10-5 4.772 × 10-5 (11, 11) (20, 20) 3.988 × 10-5 4.772 × 10-5

     | Show Table
    DownLoad: CSV

    The advantage of the BDMC method was further verified by comparing the predicted temperature profiles. For gray medium, the radiative equilibrium temperature Tij of cell (i, j) satisfies the energy equation

    4(κa)ijVijT4ij=M,Nk=1,l=1¯RDklijT4kl (26)

    or

    εijAijT4ij=M,Nk=1,l=1¯RDklijT4kl (27)

    Utilizing the predictions of radiative exchange factors, the radiative equilibrium temperature profile can be solved iteratively from Eq 26 or Eq 27. In the iteration, the iteration is stopped by setting the residuals. The number of iterations is not certain, which is related to the residuals.

    Figure 4 shows the radiative equilibrium temperature profiles for an absorbing, emitting, and non-scattering medium predicted by the TMC method and the BDMC method employing the equivalent sampling and the uniform grids. The wall temperatures were imposed separately as TWE = TWW =1000 K and TWN = TWS =1500 K, the emissivity of the walls were assumed to be constant and equal to 0.5, the optical thickness = keLx = keLy = 15. First, the BDMC predictions were very close to the TMC predictions, it indicates that the development of the present BDMC for predicting temperature profiles is correct, in addition, the BDMC predictions were shown smoother than those predicted by the TMC method, the BDMC method converged faster than the TMC method. Similar superior of the BDMC method to the TMC method can be seen from Figure 5, where the weight sampling and the non-uniform grids were employed and the medium was absorbing and isotropic scattering with scattering albedo of = 0.5.

    Figure 4.  Radiative equilibrium temperature profiles predicted by the TMC method and the BDMC method with equivalent sampling using the uniform grid. (ε=0.5, τ=1.0, ω=0, Nr=105, Ni,j=7.22×107).
    Figure 5.  Radiative equilibrium temperature profiles predicted by the TMC method and the BDMC method with weight sampling using the non-uniform grid. (ε=0.5, τ=1.0, ω=0, Nr=105, Ni,j=8.12×107).

    Figure 6 shows the comparison of temperature profiles predicted by the BDMC method with non-uniform grids employing different sampling models of the equivalent sampling and the weight sampling. The temperature profiles predicted by using the weight sampling was smooth, while the predictions using equivalent sampling model contains some small rectangles, this may due to the fact that the number of the sampling bundles for different cells employing equivalent sampling were significantly different, which lead to large random error and affected the temperature profiles.

    Figure 6.  Radiative equilibrium temperature profiles predicted by the BDMC method with two different sampling models using the non-uniform grid. (ε=0.5, τ=1.0, ω=0.5, Nr=105, Ni,j=8.12×107).

    Figure 7 shows the radiative equilibrium temperature profiles predicted by the TMC method and the BDMC method with a relative small number of weight sampling bundles using the non-uniform grids. The temperature profiles predicted by the TMC method became unacceptable as the reference sampling bundles decreased to Nr = 103, while the BDMC method can still predict better results, the BDMC method converged faster than the TMC method. Therefore, the BDMC method would be a better choice if the sampling bundles for radiation computation is limited or efficient computation is required.

    Figure 7.  Radiative equilibrium temperature profiles predicted by the TMC method and the BDMC method with a smaller amount of weight sampling using the non-uniform grid. (ε=0.5, τ=1.0, ω=0, Nr=103, Ni,j=5.86×105).

    Farmer and Howell introduced the performance parameter P to evaluate various MC methods or strategies [1,9], the performance parameter was defined as

    P=γ2t (28)

    where, t is the CPU time spent by the concerned MC simulation and γ2 is the variance of the results. A good method or strategy for the MC simulation tends to minimize the performance parameter P. The variance γ2 is given by

    γ2=1HHh=1γ2h (29)

    where, γ2h is the variance of the hth exercise, h[1,H], and H is the total exercise times, γ2h can be calculated from

    γ2h=1M2N2M,Ni=1,j=1M,Nk=1,l=1[(¯RDijkl)h(¯RDijkl)a]2 (30)

    where, (¯RDijkl)h is the predicted value of ¯RDijkl in the hth exercise, and (¯RDijkl)a is the "exact value" of ¯RDijkl.

    It has been discussed that the acceptable numerical results can be obtained if the sampling bundles Nb in the MC simulation reached 105. In order to compute the performance parameters for the TMC method and the BDMC method, the "exact solution" of the radiative exchange factors should be firstly obtained. The approximate "exact solution" can be obtained by increasing the sampling bundles of the MC simulation due to the high precision of the MC method. To do so, the reference sampling bundles were taken as Nr = 106 and Nr = 108, respectively, the predicted results for the two cases were found to be nearly unchanged, and the maximum relative error for any of the radiative exchange factors correspond to different cells was less than 0.1% as the reference sampling bundles Nr increased from 106 to 108. Therefore, the results employing Nr = 108 were treated as the "exact solution" in the present study.

    For each exercise, only the initial random number and sampling order were changed, therefore, the CPU time spent by each exercise should be nearly unchanged. The total exercise times was set as H = 10. The performance parameter of the BDMC method and the TMC method were written as PBDM and PTM, respectively, while the performance increment ΔP was defined as

    ΔP=PTMPBDMPTM×100% (31)

    Table 5 reports the performance parameter for the TMC method and the BDMC method using the equivalent sampling and the uniform grids. It indicates that the performance parameter for the TMC method was always larger than that for the BDMC method, in addition, the performance increment decrease with the increasing sampling bundles. Table 6 shows the performance parameter for the TMC method and the BDMC method employing the weight sampling and the non-uniform grids. Similar conclusions can be drawn as those employing the equivalent sampling and the uniform grids.

    Table 5.  Performance comparison of the TMC method and the BDMC method using the equivalent sampling and the uniform grids.
    (ε=0.5, ω=0.5, Nr=103)
    PBDM PTM ΔP
    τ=1.0
    Ni,j=7.22×105
    8.24 × 10-10 1.38 × 10-9 40.2 %
    τ=0.01
    Ni,j=3.64×107
    6.13 × 10-10 7.91 × 10-10 22.4%

     | Show Table
    DownLoad: CSV
    Table 6.  Performance comparison of the TMC method and the BDMC method using the weight sampling and the non-uniform grids.
    (ε=0.5, ω=0.5, Nr=103)
    PBDM PTM ΔP
    τ=1.0
    Ni,j=8.12×105
    9.98 × 10-10 1.57 × 10-9 36.56 %
    τ=0.01
    Ni,j=4.54×107
    1.78 × 10-10 2.11 × 10-10 15.68%

     | Show Table
    DownLoad: CSV

    Table 7 shows the performance parameter for the BDMC method with equivalent sampling and the weight sampling using the non-uniform grids. The performance parameter for BDMC with the equivalent sampling was found to be larger than that with the weight sampling, this indicate that the weight sampling was more suitable for non-uniform grids in the BDMC simulations.

    Table 7.  Performance comparison of the equivalent sampling and the weight sampling in the BDMC method using the non-uniform grids.
    (ε=0.5, ω=0.5, Nr=103)
    weight sampling equivalent sampling ΔP
    τ=1.0
    Ni,j=8.12×105
    9.98 × 10-10 2.55 × 10-9 60.84%
    τ=0.01
    Ni,j=4.54×107
    1.78 × 10-10 6.91 × 10-10 74.24%

     | Show Table
    DownLoad: CSV

    The BDMC method along with the equivalent sampling and the weight sampling were developed to solve radiative transfer in absorbing and scattering medium. The BDMC method approximately doubles the information from the bundle tracing to possess a high efficiency by making the best use of the reversibility of the bundle trajectory. The formula for the BDMC method and the corresponding error analysis were derived and presented. Radiative heat transfer in a two-dimensional rectangular domain of absorbing and/or scattering medium were solved by the TMC method and the BDMC method, respectively. The reciprocity of the radiative exchange factors, the radiative equilibrium temperature profiles, and the performance parameter predicted by the two MC methods were examined and compared. The results showed that (1) the BDMC method can greatly improve the reciprocity satisfaction of radiative exchange factors, which is helpful for temperature profile solution, (2) the BDMC method was more accurate or efficient than the TMC method, (3) in the BDMC simulations, the weight sampling was found to be more flexible than the equivalent sampling.

    The authors acknowledge the financial support by National Science Foundation of China (No. 51776052) and Aeronautical Science Foundation of China (No.201927077002).

    The authors declare no conflict of interest.

    Xiaofeng Zhang: Software, Formal analysis Data Curation, Writing - Original Draft, Visualization. Qing Ai: Conceptualization, Methodology, Validation, Resources, Writing - Review & Editing, Supervision, Funding acquisition. Kuilong Song: Validation, Formal analysis, Writing - Review & Editing, Heping Tan: Writing - Review & Editing, Supervision, Project administration.



    [1] W. O. Kermack, A. G. McKendrick, A contribution to the mathematical theory of epidemics, Proc. R. Soc. Lond. Ser. A., 115 (1927), 700-721. doi: 10.1098/rspa.1927.0118
    [2] K. Fushimi, Y. Enatsu, E. Ishiwata, Global stability of an SIS epidemic model with delays, Math. Appl. Sci., 41 (2018), 5345-5354.
    [3] H. A. A. El-Saka, The fractional-order SIS epidemic model with variable population size, Egyptian Math. Soc., 22 (2014), 50-54. doi: 10.1016/j.joems.2013.06.006
    [4] D. Bichara, Y. Kang, C. Castillo-Chavez, R. Horan, C. Perrings, SIS and SIR epidemic models under virtual dispersal, Bull. Math. Biol., 77 (2015), 2004-2034. doi: 10.1007/s11538-015-0113-5
    [5] T. Kuniya, J. Wang, Global dynamics of an SIR epidemic model with nonlocal diffusion, Nonlinear Anal. Real World Appl., 43 (2018), 262-282. doi: 10.1016/j.nonrwa.2018.03.001
    [6] P. A. Naik, J. Zu, M. Ghoreishi, Stablity anaysis and approximate solution of sir epidemic model with crowley-martin type functional response and holling type-II treatment rate by using homotopy analysis method, J. Appl. Anal. Comput., 10 (2020), 1482-1515.
    [7] P. A. Naik, Global dynamics of a fractional-order SIR epidemic model with memory, Int. J. Biomath., 13 (2020), 1-23.
    [8] C. N. Angstmann, B. I. Henry, A. V. Mcgann, A fractional order recovery SIR model from a stochastic process, Bull. Math. Biol., 78 (2016), 468-499. doi: 10.1007/s11538-016-0151-7
    [9] P. A. Naik, J. Zu, K. M. Owolabi, Global dynamics of a fractional order model for the transmission of HIV epidemic with optimal control, Chaos Solitons Fractals, 138 (2020), 1-30.
    [10] P. A. Naik, K. M. Owolabi, M. Yavuz, J. Zu, Chaotic dynamics of a fractional order HIV-1 model involving aids-related cancer cells, Chaos Solitons Fractals, 140 (2020), 1-13.
    [11] P. A. Naik, J. Zu, K. M. Owolabi, Modeling the mechanics of viral kinetics under immune control during primary infection of HIV-1 with treatment in fractional order, Phys. A, 545 (2020), 1-19.
    [12] P. A. Naik, M. Yavuz, S. Qureshi, J. Zu, S. Townley, Modeling and analysis of COVID-19 epidemics with treatment in fractional derivatives using real data from pakistan, Eur. Phys. J. Plus, 135 (2020), 795. doi: 10.1140/epjp/s13360-020-00819-5
    [13] S. X. Zhang, H. Guo, Global analysis of age-structured multi-stage epidemic models for infectious diseases, Appl. Math. Comput., 337 (2018), 214-233.
    [14] P. Liu, H. X. Li, Global behavior of a multi-group SEIR epidemic model with age structure and spatial diffusion, Math. Biosci. Eng., 17 (2020), 7248-7273. doi: 10.3934/mbe.2020372
    [15] A. Ricardo, Analysis of a fractional SEIR model with treatment, Appl. Math. Lett., 84 (2016), 56-62.
    [16] J. Li, Y. Yang, Y. Zhou, Global stability of an epidemic model with latent stage and vaccination, Nonlinear Anal. Real World Appl., 12 (2011), 2163-2173. doi: 10.1016/j.nonrwa.2010.12.030
    [17] H. F Huo, L. X. Feng, Global stability of an epidemic model with incomplete treatment and vaccination, Discrete Dyn. Nat. Soc., 2012 (2012), 87-88.
    [18] M. Baniyaghoub, R. Gautam, Z. Shuai, P. van den Driessche, R. Ivanek, Reproduction numbers for infections with free-living pathogens growing in the environment, Discrete Dyn. Nat. Soc., 6 (2012), 923-940.
    [19] T. Caraco, I. N. Wang, Free-living pathogens: life-history constraints and strain competition, J. Theoret. Biol., 250 (2008), 569-579. doi: 10.1016/j.jtbi.2007.10.029
    [20] S. A. Boone, C. P. Gerba, Significance of fomites in the spread of respiratory and enteric viral disease, Appl. Environ. Microb., 73 (2007), 1687-1696. doi: 10.1128/AEM.02051-06
    [21] A. Gabbuti, L. Romanˊo, P. Blanc, F. Meacci, A. Amendola, A. Mele, et al., Long-term immunogenicity of hepatitis B vaccination in a cohort of italian healthy adolescents, Vaccine, 25 (2007), 3129-3132. doi: 10.1016/j.vaccine.2007.01.045
    [22] F. Xu, R. Cressman, Disease control through voluntary vaccination decisions based on the smoothed best response, Comput. Math. Methods Med., 2014 (2014), 825734.
    [23] F. Xu, R. Cressman, Voluntary vaccination strategy and the spread of sexually transmitted diseases, Math. Biosci., 274 (2016), 94-107. doi: 10.1016/j.mbs.2016.02.004
    [24] P. C. Jentsch, M. Anand, C. T. Bauch, Prioritising COVID-19 vaccination in changing social and epidemiological landscapes: a mathematical modelling study, Lancet Infect. Dis., 21 (2021), 1097-1106. doi: 10.1016/S1473-3099(21)00057-8
    [25] A. Ciccone, B. Zanotti, Covid-vaccination working group on cerebral venous thrombosis after, The importance of recognizing cerebral venous thrombosis following anti-COVID-19 vaccination, Eur. J. Intern. Med., 89 (2021), 115-117. doi: 10.1016/j.ejim.2021.05.006
    [26] J. Q. Li, Y. L. Yang, Y. C. Zhou, Global stability of an epidemic model with latent stage and vaccination, Nonlinear Anal. Real World Appl., 12 (2011), 2163-2173. doi: 10.1016/j.nonrwa.2010.12.030
    [27] H. F. Huo, L. X. Feng, Global stability of an epidemic model with incomplete treatment and vaccination, Discrete Dyn. Nat. Soc., 2012 (2012), 87-88.
    [28] X. N. Liu, Y. Takeuchi, S. Iwami, SVIR epidemic models with vaccination strategies, J. Theor. Biol., 253 (2008), 1-11. doi: 10.1016/j.jtbi.2007.10.014
    [29] H. W. Hethcote, S. A. Levin, Periodicity in Epidemiological Models, Springer Berlin Heidelberg, Berlin, Heidelberg, 1989.
    [30] J. P. Tripathi, S. Abbas, Global dynamics of autonomous and nonautonomous SI epidemic models with nonlinear incidence rate and feedback controls, Nonlinear Dyn., 86 (2016), 337-3351. doi: 10.1007/s11071-016-2892-0
    [31] D. Cheban, Z. X. Liu, Periodic, quasi-periodic, almost periodic, almost automorphic, birkhoff recurrent and poisson stable solutions for stochastic differential equations, J. Differ. Equ., 269 (2020), 3652-3685. doi: 10.1016/j.jde.2020.03.014
    [32] Z. Du, Y. Lv, Permanence and almost periodic solution of a lotka-volterra model with mutual interference and time delays, Appl. Math. Model., 37 (2013), 1054-1068. doi: 10.1016/j.apm.2012.03.022
    [33] H. H. Wu, S. B. Yu, Permanence, extinction, and almost periodic solution of a nicholson's blowflies model with feedback control and time delay, Discrete Dyn. Nat. Soc., 2013 (2013), 1-9.
    [34] N. H. Du, N. N. Nhu, Permanence and extinction for the stochastic SIR epidemic model, J. Differ. Equ., 269 (2020), 9619-9652. doi: 10.1016/j.jde.2020.06.049
    [35] F. Chen, On a nonlinear nonautonomous predator-prey model with diffusion and distributed delay, J. Comput. Appl. Math., 180 (2005), 33-49. doi: 10.1016/j.cam.2004.10.001
    [36] Y. Q. Li, T. W. Zhang, Existence and multiplicity of positive almost periodic solutions for a non-autonomous SIR epidemic model, Bull. Malays. Math. Sci. Soc., 39 (2016), 359-379. doi: 10.1007/s40840-015-0176-3
    [37] P. Y. Liu, H. X. Li, Global stability of autonomous and nonautonomous hepatitis b virus models in patchy environment, J. Appl. Math. Comput., 10 (2020), 1771-1799.
    [38] A. M. Fink, Almost periodic differential equations, Springer-Verlag, New York, 1974.
    [39] C. Y. He, Almost Periodic Differential Equations, Higher Education Publishing House, Beijing, 1992. (in Chinese).
    [40] F. Chen, Z. Li, Y. Huang, Note on the permanence of a competitive system with infinite delay and feedback controls, Nonlinear Anal. Real World Appl., 8 (2007), 680-687. doi: 10.1016/j.nonrwa.2006.02.006
    [41] K. Gopalasamy, Stability and Oscillation in Delay Equation of Population Dynamics, Kluwer Academic Publishers, Dordrecht, 1992.
    [42] C. Zhang, N. J. Huang, D. O'Regan, Almost periodic solutions for a volterra model with mutual interference and holling type III functional response, Appl. Math. Comput., 225 (2013), 503-511.
    [43] 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. doi: 10.1016/S0025-5564(02)00108-6
    [44] N. Chitnis, J. M. Hyman, J. M. Cushing, Determining important parameters in the spread of malaria through the sensitivity analysis of a mathematical model, Bull. Math. Biol., 70 (2008), 1272-1296. doi: 10.1007/s11538-008-9299-0
  • This article has been cited by:

    1. Yixuan Qiu, Feng Lin, Weitong Chen, Miao Xu, Pre-training in Medical Data: A Survey, 2023, 2731-538X, 10.1007/s11633-022-1382-8
    2. Shuo Zhang, Yonghao Ren, Jing Wang, Bo Song, Runzhi Li, Yuming Xu, GSTCNet: Gated spatio-temporal correlation network for stroke mortality prediction, 2022, 19, 1551-0018, 9966, 10.3934/mbe.2022465
    3. Hu Cunlin, Ye Ye, Xie Nenggang, Self-supervised motor imagery EEG recognition model based on 1-D MTCNN-LSTM network, 2024, 21, 1741-2560, 036014, 10.1088/1741-2552/ad48bc
    4. In-Nea Wang, Choel-Hui Lee, Hakseung Kim, Dong-Joo Kim, Negative-Sample-Free Contrastive Self-Supervised Learning for Electroencephalogram-Based Motor Imagery Classification, 2024, 12, 2169-3536, 132714, 10.1109/ACCESS.2024.3459866
    5. Aurora Saibene, Hafez Ghaemi, Eda Dagdevir, Deep learning in motor imagery EEG signal decoding: A Systematic Review, 2024, 610, 09252312, 128577, 10.1016/j.neucom.2024.128577
    6. Tianhui Sha, Yikai Zhang, Yong Peng, Wanzeng Kong, Semi-supervised regression with adaptive graph learning for EEG-based emotion recognition, 2023, 20, 1551-0018, 11379, 10.3934/mbe.2023505
    7. Federico Del Pup, Manfredo Atzori, Applications of Self-Supervised Learning to Biomedical Signals: A Survey, 2023, 11, 2169-3536, 144180, 10.1109/ACCESS.2023.3344531
    8. Sergio Kazatzidis, Siamak Mehrkanoon, 2024, A novel dual-stream time-frequency contrastive pretext tasks framework for sleep stage classification, 979-8-3503-5931-2, 1, 10.1109/IJCNN60899.2024.10650151
    9. Pierre Guetschel, Sara Ahmadi, Michael Tangermann, Review of deep representation learning techniques for brain–computer interfaces, 2024, 21, 1741-2560, 061002, 10.1088/1741-2552/ad8962
    10. Huanyu Wu, Siyang Li, Dongrui Wu, Motor Imagery Classification for Asynchronous EEG-Based Brain–Computer Interfaces, 2024, 32, 1534-4320, 527, 10.1109/TNSRE.2024.3356916
    11. Ilya Mikheev, Helen Steiner, Olga Martynova, Detecting cognitive traits and occupational proficiency using EEG and statistical inference, 2024, 14, 2045-2322, 10.1038/s41598-024-55163-w
    12. Wonjun Ko, Seungwoo Jeong, Sa-Kwang Song, Heung-Il Suk, EEG-Oriented Self-Supervised Learning With Triple Information Pathways Network, 2024, 54, 2168-2267, 6495, 10.1109/TCYB.2024.3410844
    13. Zara Cook, Chengzong Zhao, Livia Murray, Jivan Kesan, Nabil Belacel, Sam M. Doesburg, George Medvedev, Vasily A. Vakorin, Pengcheng Xi, 2024, Decoding Brain Age: A Self-Supervised Graph Neural Network Framework for EEG Analysis, 979-8-3503-6351-7, 1, 10.1109/SENSORS60989.2024.10784646
    14. Heng Li, Zhongwei Lu, Yun Mo, Bao Feng, Tianyou Yu, A two-stage feature redundancy minimization methodology framework for motor imagery EEG classification, 2025, 1573-7721, 10.1007/s11042-025-20605-8
    15. Wonjun Ko, Jeongwon Choe, Jonggu Kang, SLEEP-SAFE: Self-Supervised Learning for Estimating Electroencephalogram Patterns With Structural Analysis of Fatigue Evidence, 2025, 13, 2169-3536, 35805, 10.1109/ACCESS.2025.3545094
    16. Weining Weng, Yang Gu, Shuai Guo, Yuan Ma, Zhaohua Yang, Yuchen Liu, Yiqiang Chen, Self-supervised Learning for Electroencephalogram: A Systematic Survey, 2025, 0360-0300, 10.1145/3736574
  • Reader Comments
  • © 2021 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(3559) PDF downloads(232) Cited by(10)

Figures and Tables

Figures(12)  /  Tables(1)

Other Articles By Authors

/

DownLoad:  Full-Size Img  PowerPoint
Return
Return

Catalog