
Citation: Nadejda Komendantova, Masoud Yazdanpanah, Roshanak Shafiei. Studying young people’ views on deployment of renewable energy sources in Iran through the lenses of Social Cognitive Theory[J]. AIMS Energy, 2018, 6(2): 216-228. doi: 10.3934/energy.2018.2.216
[1] | Rashmi Murali, Sangeeta Malhotra, Debajit Palit, Krishnapada Sasmal . Socio-technical assessment of solar photovoltaic systems implemented for rural electrification in selected villages of Sundarbans region of India. AIMS Energy, 2015, 3(4): 612-634. doi: 10.3934/energy.2015.4.612 |
[2] | Austin Bushur, Kevin Ward, Tommy Flahaven, Tom Kelly, Jin H. Jo, Matt Aldeman . Techno-economic evaluation of installing EV and PV combined infrastructure on Academic Institution’s Parking Garages in Illinois, USA. AIMS Energy, 2019, 7(1): 31-45. doi: 10.3934/energy.2019.1.31 |
[3] | Edwin Collado, Easton Li Xu, Hang Li, Shuguang Cui . Profit maximization with customer satisfaction control for electric vehicle charging in smart grids. AIMS Energy, 2017, 5(3): 529-556. doi: 10.3934/energy.2017.3.529 |
[4] | Mansoor Soomro, Zeeshan Ali Shaikh, Mazhar Baloch, Abdul Manan Shaikh, Sohaib Tahir Chauhdary . Development of wind and solar systems for power charging: An application of an electric vehicle to grid systems. AIMS Energy, 2024, 12(3): 664-685. doi: 10.3934/energy.2024031 |
[5] | Nikolaos S. Korakianitis, Nikolaou Nikolaos, Georgios A. Vokas, George Ch. Ioannidis, Stavros D. Kaminaris . Simulation-Based evaluation of dynamic wireless charging systems for electric vehicles: Efficiency, limitations, and future directions. AIMS Energy, 2025, 13(3): 590-631. doi: 10.3934/energy.2025023 |
[6] | Mulualem T. Yeshalem, Baseem Khan . Design of an off-grid hybrid PV/wind power system for remote mobile base station: A case study. AIMS Energy, 2017, 5(1): 96-112. doi: 10.3934/energy.2017.1.96 |
[7] | Rasool M. Imran, Kadhim Hamzah Chalok . Innovative mode selective control and parameterization for charging Li-ion batteries in a PV system. AIMS Energy, 2024, 12(4): 822-839. doi: 10.3934/energy.2024039 |
[8] | Imam Abadi, Chairul Imron, Mohammad Musa Bachrowi, Dwi Nur Fitriyanah . Design and implementation of battery charging system on solar tracker based stand alone PV using fuzzy modified particle swarm optimization. AIMS Energy, 2020, 8(1): 142-155. doi: 10.3934/energy.2020.1.142 |
[9] | Anbazhagan Geetha, S. Usha, J. Santhakumar, Surender Reddy Salkuti . Forecasting of energy consumption rate and battery stress under real-world traffic conditions using ANN model with different learning algorithms. AIMS Energy, 2025, 13(1): 125-146. doi: 10.3934/energy.2025005 |
[10] | Nadwan Majeed Ali, Handri Ammari . Design of a hybrid wind-solar street lighting system to power LED lights on highway poles. AIMS Energy, 2022, 10(2): 177-190. doi: 10.3934/energy.2022010 |
The term hantavirus represents several groups of RNA-containing viruses (that are members of the virus family of Bunyaviridae) that are carried by rodents, particularly deer mice. The virus is found in their urine and feces, but it does not make the animal sick. The hantavirus is a rare but potentially very serious disease that affects a small number of people every year. People become infected through contact with hantavirus-infected rodents or their urine and droppings. Infection with hantavirus can progress to Hantavirus Pulmonary Syndrome (HPS), which can be fatal. The Sin Nombre hantavirus, first recognized in 1993, is one of several New World hantaviruses circulating in the US. Old World hantaviruses, including Seoul virus, are found across the world and can cause Hemorrhagic Fever with Renal Syndrome (HFRS). Hantavirus does not seem to spread from human to human. The humans get the virus from the mice but have no feedback effects on the mice in the infection process. Furthermore, the transmission of the infection is horizontal, i.e., no mice are born infected [1], infection may only be contracted in adulthood from other mice mainly through agammaessive encounters, such as fights among them, or through inhalation of the aerosolized virus [2,3,4]. The infection and persistence of hantavirus in its rodent host has little or no effect on survival [5]. In order to have a better understanding of the spread of the disease in humans, it is necessary to understand the transmission dynamics of hantavirus in rodent populations.
Several epidemic models that have been applied to hantavirus infection in rodents are available. For instance, Abramson and Kenkre [6] and Abramson and Kenkre et al. [7] formulated a two-equation reaction diffusion model for susceptible and infected deer mice. Logistic growth is assumed with carrying capacity K. K is chosen as a control parameter of the dynamics. They analyzed the traveling waves of a model of the hantavirus infection. Allen et al. [8] developed an SI epidemic model for a host with two viral infections. The model was applied to a hantavirus and an arenavirus that infect cotton rats. The first virus is transmitted horizontally whereas the second is transmitted vertically. Considering the movement characteristics of the mice that carry the infection, Kenkre et al. [9] considered two types of mice, stationary adult mice and itinerant juvenile mice. Gedeon et al. [10] applied their model to hantavirus infection in deer mice. Their goal was to compare the relative importance of direct (from infected to healthy individuals) and indirect (by the contaminated environment) transmission in sylvan and peridomestic environments. Sauvage et al. [11] formulated a model which was applied to Puumala virus infection in bank voles. Host population was divided into two age classes, juveniles and adults. It was assumed that individuals can live in favourable or unfavourable patches. Their study showed that indirect transmission significantly increased the probability for the virus to persist in the host population. These two transmission modes also have been discussed by Wolf [12] and Wolf et al. [13]. Abramson and Kenkre [1] gave a stage-dependent model with maturation delay, in their model, a virus-free young mice variable is introduced, the adult population was subdivided into susceptibles and infectives. The spatial version of [1] was presented in [14]. Based on the model in [6], Buceta et al. [15] studied the impact of seasonality on hantavirus, they have shown that the alternation of seasons may cause outbreaks of the disease even if neither season by itself satisfies the environmental requirements for propagation of the disease. Allen et al. [16] constructed two gender-structured SEIR model, the first model is a system of ordinary differential equations, while the second model extends the first model to a system of stochastic differential equations. These models are studied mainly from the numerical simulation point of view. A spatio-temporal SEIR compartmental model was proposed in Burger et al. [17].
It has been noted that environmental conditions are directly connected to outbreaks of Hanta. For example, the Four Corners Region, where an important number of cases of HPS have occurred, has a desert climate. The largest climate variations within this region come from periods of rain and of drought [15]. The influence of the environmental conditions plays a fundamental role in the evolution of the population. Resource availability would fluctuate as seed and fruit production vary over a 3-year period [18].
As in most infectious diseases, there is a lag between exposure and infectivity, which is usually called the incubation period. Because the life expectancy of rodents is relatively short, then the incubation period cannot be neglected [16]. Since infected mice that survive the incubation period will remain infectious for the rest of their lives. Thus, the incubation period directly influences the number of infectious mice. However, explicit delay effects related to finite incubation periods have received little attention.
Motivated by the works of [10,11,15,16,17], in this paper we formulate a periodic time-delayed model by taking into account the seasonality. The model contains a time delay accounting for the incubation period of the virus.
The rest of this paper is organized as follows. In the next section, based on the work of Gedeon et al. [10] we present the model, and study its well-posedness, also we introduce the basic reproduction number R0. In section 3, we establish the threshold dynamics in terms of R0. In section 4, we study the autonomous case of the periodic model, and prove the global stability of the virus-free equilibrium and the global attractivity of the endemic equilibrium. In section 5, we perform numerical simulations to illustrate our analytic results. A brief discussion section completes the paper.
Our model was built on the framework of [10]. In [10], S(t), I(t) and G(t) denote the susceptible, infectious adult mice at time t, and the number of contaminated sites in the environment. In our model, we add the exposed class, we let E(t) denote the exposed adult mice at time t. We make the same assumptions as those in [10], that is, we assume that there is a discrete number of sites that are visited by mice. Each site is small enough to be infected by a single mouse. Also we assume that the total number of sites M is large and we represent it by a continuous variable.
We let N(t) be the total population of mice, then N(t)=S(t)+E(t)+I(t). The direct contact rate c(t) of mice is the average number of direct contacts between mice per mouse per unit time at time t. This rate depends on a number of factors, and in particular, climatic ones, but for simplicity in this paper we assume c(t) to be periodic. These contacts may involve biting and scratching and are thought to occur predominantly between sexually active males [2,19]. Suppose the transmission probability that given an active contact between susceptible and infected mice is denoted by β, which is called a direct transmission rate.
For the indirect transmission route, we let ˉc(t) be the number of contacts between a mouse and all the potentially contaminated sites per susceptible mouse per unit time at time t, which is also assumed to be periodic. The probability that given a contact between susceptible mouse and contaminated site is denoted by α, this is called an indirect transmission rate.
Indirect transmission is due to the fact that infected individuals can excrete viruses in their feces, vomit, urine or others. To model the process of site contamination by infected mice, let d(t) be the number of contacts between the mice and the site that can lead to transmission of the infection per uncontaminated site per unit time at time t. The probability of site contamination is called the site contamination rate, which is denoted by γ, given a contact between an uncontaminated site and an infected mouse.
Assume that B(t) is the recruitment rate for mice and μ(t) is the death rate of the mice. Since hantavirus is not lethal to mice, then we assume the death rates are the same for the infected and the susceptible classes [3]. The mice are infected for the rest of their lives [3,20], so there is no recovery. We assume that the environment eliminates viruses with time at a rate δ(t), which is called the disinfection rate. Suppose that τ is the average incubation period. Then we obtain the following system
{dS(t)dt=B(t)−μ(t)S(t)−β1(t)S(t)I(t)N(t)−β2(t)S(t)G(t)M,dE(t)dt=β1(t)S(t)I(t)N(t)+β2(t)S(t)G(t)M−μ(t)E(t)−[β1(t−τ)S(t−τ)I(t−τ)N(t−τ)+β2(t−τ)S(t−τ)G(t−τ)M]e−∫tt−τμ(r)dr,dI(t)dt=[β1(t−τ)S(t−τ)I(t−τ)N(t−τ)+β2(t−τ)S(t−τ)G(t−τ)M]e−∫tt−τμ(r)dr−μ(t)I(t),dG(t)dt=a(t)I(t)N(t)(M−G(t))−δ(t)G(t), | (2.1) |
where β1(t)=βc(t),β2(t)=αˉc(t),a(t)=γd(t). All parameters are positive, continuous, and ω-periodic functions for some ω>0. It is easy to see that the equation for E(t) can be rewritten as one integral equation
E(t)=∫tt−τ[β1(ξ)S(ξ)I(ξ)N(ξ)+β2(ξ)S(ξ)G(ξ)M]e−∫tξμ(r)drdξ. | (2.2) |
The dynamics of the mouse population is governed by the following equation
dN(t)dt=B(t)−μ(t)N(t). | (2.3) |
It is easy to see that (2.3) has a unique positive ω-periodic solution
N∗(t)=[∫t0B(r)e∫r0μ(s)dsdr+∫ω0B(r)e∫r0μ(s)dsdre∫ω0μ(s)ds−1]e−∫t0μ(s)ds, |
which is globally asymptotically stable.
Let C=C([−τ,0],R4), C+=C([−τ,0],R4+). Then (C,C+) is an ordered Banach space equipped with the maximum norm and the partial order induced by the positive cone C+. For any given continuous function x:[−τ,σ)→R4 with σ>0, we can define xt∈C as xt(θ)=x(t+θ), ∀θ∈[−τ,0] for any t∈[0,σ).
For a given continuous ω-periodic function g(t), let
ˆg=maxt∈[0,ω]g(t),¯g=mint∈[0,ω]g(t). |
Let
W:=C([−τ,0],R3+)×C([−τ,0],[0,M]). |
In view of (2.2), we choose the initial data for system (2.1) in Xδ0, which is defined as
Xδ0={ϕ∈W:∑3i=1ϕi(s)≥δ0,∀s∈[−τ,0],ϕ2(0)=∫0−τ[β1(ξ)ϕ1(ξ)ϕ3(ξ)∑3i=1ϕi(ξ)+β2(ξ)ϕ1(ξ)ϕ4(ξ)M]e−∫0ξμ(r)drdξ} |
for small δ0∈(0,¯Bˆμ).
Lemma 2.1. For any ϕ∈Xδ0, system (2.1) has a unique nonnegative solution u(t,ϕ) with u0=ϕ for all t≥0, and solutions are ultimately bounded and uniformly bounded.
Proof. Given ϕ∈Xδ0, define f(t,ϕ)=(f1(t,ϕ),f2(t,ϕ),f3(t,ϕ),f4(t,ϕ)) with
f1(t,ϕ)=B(t)−μ(t)ϕ1(0)−β1(t)ϕ1(0)ϕ3(0)∑3i=1ϕi(0)−β2(t)ϕ1(0)ϕ4(0)M,f2(t,ϕ)=β1(t)ϕ1(0)ϕ3(0)∑3i=1ϕi(0)+β2(t)ϕ1(0)ϕ4(0)M−μ(t)ϕ2(0)−[β1(t−τ)ϕ1(−τ)ϕ3(−τ)∑3i=1ϕi(−τ)+β2(t−τ)ϕ1(−τ)ϕ4(−τ)M]e−∫tt−τμ(r)dr,f3(t,ϕ)=[β1(t−τ)ϕ1(−τ)ϕ3(−τ)∑3i=1ϕi(−τ)+β2(t−τ)ϕ1(−τ)ϕ4(−τ)M]e−∫tt−τμ(r)dr−μ(t)ϕ3(0),f4(t,ϕ)=a(t)ϕ3(0)∑3i=1ϕi(0)(M−ϕ4(0))−δ(t)ϕ4(0). |
Since f(t,ϕ) is continuous in (t,ϕ)∈R+×Xδ0, and f(t,ϕ) is Lipschitz in ϕ on each compact subset of Xδ0, it then follows that system (1) has a unique solution u(t,ϕ) on its maximal interval [0,σϕ) of existence with u0=ϕ (see [21,Theorems 2.2.1 and 2.2.3]).
In view of the second equation of system (2.1), we have
(dE(t)dt+μ(t)E(t))e∫t0μ(s)ds=[β1(t)S(t)I(t)N(t)+β2(t)S(t)G(t)M−(β1(t−τ)S(t−τ)I(t−τ)N(t−τ)+β2(t−τ)S(t−τ)G(t−τ)M)e−∫tt−τμ(r)dr]e∫t0μ(s)ds. |
By integrating on both sides, we obtain
E(t)e∫t0μ(s)ds−E(0)=∫t0[β1(s)S(s)I(s)N(s)+β2(s)S(s)G(s)M]e∫s0μ(ρ)dρds−∫t0[β1(s−τ)S(s−τ)I(s−τ)N(s−τ)+β2(s−τ)S(s−τ)G(s−τ)M]e∫s−τ0μ(ρ)dρds=∫t0[β1(s)S(s)I(s)N(s)+β2(s)S(s)G(s)M]e∫s0μ(ρ)dρds−∫t−τ−τ[β1(s)S(s)I(s)N(s)+β2(s)S(s)G(s)M]e∫s0μ(ρ)dρds=∫tt−τ[β1(s)S(s)I(s)N(s)+β2(s)S(s)G(s)M]e∫s0μ(ρ)dρds−∫0−τ[β1(s)S(s)I(s)N(s)+β2(s)S(s)G(s)M]e∫s0μ(ρ)dρds. |
Hence, if E(0)=∫0−τ[β1(s)S(s)I(s)N(s)+β2(s)S(s)G(s)M]e∫s0μ(ρ)dρds is satisfied, we then have
E(t)=∫tt−τ[β1(s)S(s)I(s)N(s)+β2(s)S(s)G(s)M]e−∫tsμ(ρ)dρds. | (2.4) |
We see that u2(t)≥0, ∀t∈[0,m] whenever ui(t)≥0 for all i≠2 and t∈[0,m]⊆[0,σϕ). Let ϕ=(ϕ1,ϕ2,ϕ3,ϕ4)∈Xδ0 be given. If ϕi(0)=0 for some i∈{1,3,4}, then fi(t,ϕ)≥0. If ϕ4(0)=M, then f4(t,ϕ)≤0. By [22,Theorem 5.2.1] and its proof, it follows that for any ϕ∈Xδ0, ui(t,ϕ)≥0 for i=1,3,4 for all t∈[0,σϕ). By equation (2.4), we have E(t)≥0 for all t∈[0,σϕ). Therefore, it follows that for any ϕ∈Xδ0, system (2.1) has a unique nonnegative solution u(t,ϕ) with u0=ϕ satisfies u(t,ϕ)∈W for all t∈[0,σϕ).
Note that dN(t)dt≥¯B−ˆμN(t). For the system
dydt=¯B−ˆμy(t), |
the equilibrium ¯Bˆμ is globally asymptotically stable. For any 0<δ0<¯Bˆμ, dydt|y=δ0=¯B−ˆμδ0>0. So if y(0)≥δ0, then y(t)≥δ0 for any t≥0. By the comparison principle,
N(t)≥δ0ifN(0)=3∑i=1ϕi(0)≥δ0. |
This implies that u(t,ϕ)∈Xδ0 for all t∈[0,σϕ). Also we have
dG(t)dt≤a(t)(M−G(t))−δ(t)G(t)=a(t)M−(a(t)+δ(t))G(t) |
for all t∈[0,σϕ). Thus, both N(t) and G(t) are bounded on [0,σϕ), it follows that σϕ=∞ (see [21,Theorem 2.3.1]), and that all solutions are ultimately bounded. Moreover, when N(t)>max{ˆB¯μ,ˆaM¯a+¯δ} and G(t)>max{ˆB¯μ,ˆaM¯a+¯δ}, we have
dN(t)dt<0anddG(t)dt<0. |
This implies that all solutions are uniformly bounded.
It is easy to see that system (2.1) has a unique virus-free periodic solution E0(t)=(N∗(t),0,0,0), where N∗(t) is the positive periodic solution of (2.3). Linearizing system (2.1) at its virus-free periodic solution E0(t)=(N∗(t),0,0,0), we then obtain the following system of periodic linear equations for the infective variables E,I, and G
{dE(t)dt=β1(t)I(t)+β2(t)N∗(t)MG(t)−μ(t)E(t)−[β1(t−τ)I(t−τ)+β2(t−τ)N∗(t−τ)MG(t−τ)]e−∫tt−τμ(r)dr,dI(t)dt=[β1(t−τ)I(t−τ)+β2(t−τ)N∗(t−τ)MG(t−τ)]e−∫tt−τμ(r)dr−μ(t)I(t),dG(t)dt=a(t)MN∗(t)I(t)−δ(t)G(t). | (2.5) |
Since the first equation of system (2.5) is decoupled from the second and third equations of system (2.5), it suffices to use the following system to define the basic reproduction number
{dI(t)dt=[β1(t−τ)I(t−τ)+β2(t−τ)N∗(t−τ)MG(t−τ)]e−∫tt−τμ(r)dr−μ(t)I(t),dG(t)dt=a(t)MN∗(t)I(t)−δ(t)G(t). | (2.6) |
Let F:R→L(C,R2) be a map and V(t) be a continuous 2×2 matrix function on R defined as follows
F(t)ϕ=([β1(t−τ)ϕ1(−τ)+β2(t−τ)N∗(t−τ)Mϕ2(−τ)]e−∫tt−τμ(r)dra(t)MN∗(t)ϕ1(0)), |
and
V(t)=(μ(t)00δ(t)). |
Let Φ(t,s) be the evolution matrices of the linear ordinary differential system
dydt=−V(t)y, |
that is
∂Φ(t,s)∂t=−V(t)Φ(t,s),∀t⩾sandΦ(s,s)=I,∀s∈R, |
where I is the 2×2 identity matrix. It then easily follows that
Φ(t,s)=(e−∫tsμ(r)dr00e−∫tsδ(r)dr). |
Let Cω be the ordered Banach space of all continuous and ω-periodic functions from R to R2, which is equipped with the maximum norm and the positive cone C+ω={v∈Cω:v(t)⩾0,∀t∈R}. Then we can define one linear operator on Cω by
[Lv](t)=∫∞0Φ(t,t−s)F(t−s)v(t−s+⋅)ds,∀t∈R,v∈Cω. |
According to [23], the basic reproduction number is defined as R0=r(L), the spectral radius of L.
Let
Y=C([−τ,0],R2),Y+=C([−τ,0],R2+). |
Then (Y,Y+) is an ordered Banach space.
Let P(t) be the solution maps of (2.6), that is, P(t)ϕ=ut(ϕ), where u(t,ϕ) is the unique solution of (2.6) with u0=ϕ∈Y. Then P:=P(ω) is the Poincareˊ map associated with (2.6). Let r(P) be the spectral radius of P . By [23,Theorem 2.1], we have the following result:
Lemma 2.2. R_0-1 has the same sign as r(P)-1 .
The following lemma indicates that the periodic semiflow P(t) is eventually strongly monotone.
Lemma 2.3. For any \varphi and \psi in \mathcal{Y}_{+} with \varphi > \psi (that is, \varphi\geq\psi but \varphi\neq\psi ), the solutions v(t, \varphi) and v(t, \psi) of system (2.6) with v_0 = \varphi and v_0 = \psi , respectively, satisfy v_i(t, \varphi) > v_i(t, \psi) for all t > \tau , i = 1, 2 , and hence, P(t)\varphi \gg P(t)\psi in \mathcal{Y}_{+} for all t > 2\tau .
Proof. For any \varphi, \psi\in \mathcal{Y}_{+} with \varphi\geq \psi , let v(t, \varphi) and v(t, \psi) be the unique solutions of system (3.1) satisfying v_0 = \varphi and v_0 = \psi , respectively. By [22,Theorem 5.1.1], we have v(t, \varphi)\geq v(t, \psi) for all t\geq0 ; that is, P(t):\mathcal{Y}_{+}\rightarrow\mathcal{Y}_{+} is monotone. Next we prove that P(t):\mathcal{Y}_{+}\rightarrow\mathcal{Y}_{+} is eventually strongly monotone. Let \varphi, \psi\in \mathcal{Y_{+}} satisfy \varphi > \psi . Denote v(t, \varphi) = (\overline{y}_1(t), \overline{y}_2(t)) and v(t, \psi) = (y_1(t), y_2(t)) .
Claim\, 1. \, \, \, There \, \, exists\, \, t_0\in[0, \tau]\, \, such\, \, that\, \, \overline{y}_1(t) > y_1(t)\, \, for\, \, all\, \, t\geq t_0.
We first prove that \overline{y}_1(t_0) > y_1(t_0) for some t_0\in[0, \tau] . Otherwise, we have \overline{y}_1(t) = y_1(t) , \forall t\in [0, \tau] , and hence \frac{d\overline{y}_1(t)}{dt} = \frac{dy_1(t)}{dt} for all t\in (0, \tau) . Thus, we have
\beta_1(t-\tau)[\overline{y}_1(t-\tau)-y_1(t-\tau)]+\frac{\beta_2(t-\tau)N^*(t-\tau)}{M}[\overline{y}_2(t-\tau)-y_2(t-\tau)] = 0, \forall t\in[0, \tau]. |
Since P(t) is monotone, it follows that \overline{y}_1(t-\tau) = y_1(t-\tau) and \overline{y}_2(t-\tau) = y_2(t-\tau) for all t\in [0, \tau] , that is, \varphi_1(\theta) = \psi_1(\theta) and \varphi_2(\theta) = \psi_2(\theta) for all \theta\in [-\tau, 0] , a contradiction to the assumption that \varphi > \psi .
Let g_1(t, y) = \left[\beta_1(t-\tau)y_1(t-\tau)+\frac{\beta_2(t-\tau)N^*(t-\tau)}{M}y_2(t-\tau)\right]e^{-\int^t_{t-\tau}\mu(r)dr}-\mu(t)y . Since
\begin{equation*} {\begin{array}{l} \frac{d\overline{y}_1(t)}{dt} = \left[\beta_1(t-\tau)\overline{y}_1(t-\tau)+\frac{\beta_2(t-\tau)N^*(t-\tau)}{M}\overline{y}_2(t-\tau)\right] e^{-\int^t_{t-\tau}\mu(r)dr}-\mu(t)\overline{y}_1(t) \\[8pt] \quad\quad\, \, \, \geq \left[\beta_1(t-\tau)y_1(t-\tau)+\frac{\beta_2(t-\tau)N^*(t-\tau)}{M}y_2(t-\tau)\right] e^{-\int^t_{t-\tau}\mu(r)dr}-\mu(t)\overline{y}_1(t) = g_1(t, \overline{y}_1(t)), \end{array}} \end{equation*} |
we have \frac{d\overline{y}_1(t)}{dt}-g_1(t, \overline{y}_1(t))\geq0 = \frac{dy_1(t)}{dt}-g_1(t, y_1(t)), \forall t\geq t_0 . Since \overline{y}_1(t_0) > y_1(t_0) , the comparison theorem for ordinary differential equations [24,Theorem 4] implies that \overline{y}_1(t) > y_1(t) for all t\geq t_0 .
Claim\, 2. \, \, \, \overline{y}_2(t) > y_2(t)\, \, for\, \, all\, \, t > t_0.
Let g_2(t, y) = \frac{a(t)M}{N^*(t)}y_1(t)-\delta(t)y . Then we have
\begin{equation*} {\begin{array}{l} \frac{d\overline{y}_2(t)}{dt} = \frac{a(t)M}{N^*(t)}\overline{y}_1(t)-\delta(t)\overline{y}_2(t) \\[8pt] \quad\quad\, \, \, \gt \frac{a(t)M}{N^*(t)}y_1(t)-\delta(t)\overline{y}_2(t) = g_2(t, \overline{y}_2(t))\, \, \forall t \gt t_0, \end{array}} \end{equation*} |
and hence, \frac{d\overline{y}_2(t)}{dt}-g_2(t, \overline{y}_2(t)) > 0 = \frac{dy_2(t)}{dt}-g_2(t, y_2(t)) \forall t > t_0 . Since \overline{y}_2(t_0)\geq y_2(t_0) , it follows from [24,Theorem 4] that \overline{y}_2(t) > y_2(t) for all t > t_0 .
In view of the above two claims, we obtain
(\overline{y}_1(t), \overline{y}_2(t))\gg (y_1(t), y_2(t)), \forall t \gt t_0. |
Since t_0\in [0, \tau] , it follows that (\overline{y}_{1t}, \overline{y}_{2t})\gg (y_{1t}, y_{2t}) , \forall t > 2\tau , that is, v_t(\varphi)\gg v_t(\psi) for all t > 2\tau . This shows P(t): \mathcal{Y}_{+}\rightarrow \mathcal{Y}_{+} is strongly monotone for any t > 2\tau .
First we show that the virus will be endemic if R_0 > 1 . Let
\begin{equation*} {\begin{array}{l} X_0 = \{\phi = (\phi_1, \phi_2, \phi_3, \phi_4)\in \mathcal{X}_{\delta_0}: \phi_3(0) \gt 0, \phi_4(0) \gt 0\}, \\[8pt] \partial X_0 = \mathcal{X}_{\delta_0}\setminus X_0 = \{\phi\in \mathcal{X}_{\delta_0}: \phi_3(0) = 0\, \, \mbox{or}\, \, \phi_4(0) = 0\}. \end{array}} \end{equation*} |
Theorem 3.1. Assume that R_0 > 1 , then there exists a positive constant \eta > 0 such that any solution (S(t, \phi), E(t, \phi), I(t, \phi), G(t, \phi)) of system (2.1) with \phi \in X_0 satisfies
\liminf\limits_{t\rightarrow\infty}(I(t, \phi), G(t, \phi))\geq(\eta, \eta). |
Proof. Let Q(t) be the solution maps of (2.1) on \mathcal{X}_{\delta_0} , that is, Q(t)\phi = u_t(\phi) , t\geq0 , where u(t, \phi) is the unique solution of (2.1) satisfying u_0 = \phi\in \mathcal{X}_{\delta_0} . Then Q: = Q(\omega) is the Poincar \acute{e} map associated with (2.1).
From the third, fourth equations of system (2.1) and equation (2.2), it is easy to see that Q(t)X_0\subseteq X_0 for all t\geq0 . Lemma 2.1 implies that \{Q^n: \mathcal{X}_{\delta_0}\rightarrow \mathcal{X}_{\delta_0}\}_{n\geq0} is point dissipative and Q^n is compact for sufficiently large n . It then follows from [25,Theorem 2.9] that Q admits a global attractor in \mathcal{X}_{\delta_0} . Now we prove that Q is uniformly persistent with respect to (X_0, \partial X_0) .
Let M_1 = (N^*_0, 0, 0, 0) , where N^*_0(\theta) = N^*(\theta) for all \theta\in[-\tau, 0]. Since \lim_{\phi\rightarrow M_1}\|Q(t)\phi-Q(t)M_1\| = 0 uniformly for t\in[0, \omega] , for any given \varepsilon > 0 , there exists a \eta_1 > 0 such that for any \phi\in X_0 with \|\phi-M_1\| < \eta_1 , we have \|Q(t)\phi-Q(t)M_1\| < \varepsilon for all t\in[0, \omega] .
Claim \, 3. \quad\limsup_{n\to\infty}\|Q(n\omega)\phi-M_1\|\geq\eta_1\, \, for\, \, all\, \, \phi\in X_0.
Suppose, by contradiction, then \limsup_{n\to\infty}\|Q(n\omega)\psi-M_1\| < \eta_1 for some \psi\in X_0 . Then there is an integer N_1\geq1 such that \|Q(n\omega)\psi-M_1\| < \eta_1 for all n\geq N_1 . For t\geq N_1\omega , we have t = n\omega+t_1 , with n\geq N_1 , t_1\in[0, \omega) and \|Q(t)\psi-Q(t)M_1\| = \|Q(t_1)Q(n\omega)\psi-Q(t_1)Q(n\omega)M_1\| = \|Q(t_1)Q(n\omega)\psi-Q(t_1)M_1\| < \varepsilon for all t\geq N_1\omega . Therefore, N^*(t)-\varepsilon < S(t) < N^*(t)+\varepsilon , 0\leq E(t) < \varepsilon , 0 < I(t) < \varepsilon , 0 < G(t) < \varepsilon for all t\geq N_1\omega .
let P_\varepsilon(t) be the solution maps of the following system on \mathcal{Y}_{+} :
\begin{equation} \left\{\begin{array}{rl} \frac{d\breve{I}(t)}{dt}& = \left[\frac{\beta_1(t-\tau)(N^*(t-\tau)-\varepsilon)}{N^*(t-\tau)+3\varepsilon}\breve{I}(t-\tau) +\frac{\beta_2(t-\tau)(N^*(t-\tau)-\varepsilon)}{M}\breve{G}(t-\tau)\right]e^{-\int^t_{t-\tau}\mu(r)dr}\\&\quad-\mu(t)\breve{I}(t), \\[8pt] \frac{d\breve{G}(t)}{dt}& = \frac{a(t)(M-\varepsilon)}{N^*(t)+3\varepsilon}\breve{I}(t)-\delta(t)\breve{G}(t). \end{array}\right. \end{equation} | (3.1) |
and P_\varepsilon: = P_\varepsilon(\omega) . Since R_0 > 1 , then \lim_{\varepsilon\rightarrow0^+}r(P_\varepsilon) = r(P) > 1 . Then we can fix a sufficiently small \varepsilon > 0 , such that r(P_\varepsilon) > 1 , and N^*(t)-\varepsilon > 0 , M-\varepsilon > 0 for all t\geq 0 . By the arguments similar to system (2.6), it is easy to verify that P_\varepsilon(t) is strongly monotone on \mathcal{Y}_{+} for t > 2\tau . It follows from [21,Theorem 3.6.1] that the linear operator P_\varepsilon(t) is compact on \mathcal{Y}_{+} . Choose an integer n_0 > 0 such that n_0\omega > 2\tau . Since P^{n_0}_\varepsilon = P_\varepsilon(n_0\omega) , [26,Lemma 3.1] implies that r(P_\varepsilon) is a simple eigenvalue of P_\varepsilon having a strongly positive eigenvector. It then follows from [27,Lemma 1] that there is a positive \omega -periodic function v^*(t) = (v^*_1(t), v^*_2(t)) such that u^*_\varepsilon(t) = e^{\frac{\ln r(P_\varepsilon)}{\omega}t}v^*(t) is a positive solution of system (3.1).
For all t\geq N_1\omega+\tau , by system (2.1), we have
\begin{equation*} \left\{\begin{array}{rl} \frac{dI(t)}{dt}&\geq\left[\frac{\beta_1(t-\tau)(N^*(t-\tau)-\varepsilon)}{N^*(t-\tau)+3\varepsilon}I(t-\tau) +\frac{\beta_2(t-\tau)(N^*(t-\tau)-\varepsilon)}{M}G(t-\tau)\right]e^{-\int^t_{t-\tau}\mu(r)dr}\\&\quad-\mu(t)I(t), \\[8pt] \frac{dG(t)}{dt}&\geq\frac{a(t)(M-\varepsilon)}{N^*(t)+3\varepsilon}I(t)-\delta(t)G(t). \end{array}\right. \end{equation*} |
Since I(t, \psi) > 0 , G(t, \psi) > 0 for all t\geq0 , we can choose a sufficiently small k > 0 such that
I(t, \psi), G(t, \psi)\geq ku^*_\varepsilon(t), \forall t\in [N_1\omega+\tau, N_1\omega+2\tau]. |
By [22,Theorem 5.1.1] it follows that I(t, \psi), G(t, \psi)\geq ku^*_\varepsilon(t), \forall t\geq N_1\omega+2\tau. Thus we have I(t, \psi)\rightarrow \infty , G(t, \psi)\rightarrow \infty , a contradiction.
Define
M_\partial: = \{\phi\in \partial X_0: Q^n(\phi)\in \partial X_0, \forall n\geq0\}. |
For any given \psi\in \partial X_0 , we have \psi_3(0) = 0 or \psi_4(0) = 0 .
If \psi_3(0) = 0 , we have the following two cases.
Case 1. I(t, \psi) = 0 for all t\geq0 .
From the third equation of (2.1), we have G(t-\tau, \psi) = 0 for all t\geq \tau . Then from equation (2.2), we get E(t, \psi) = 0 for all t\geq \tau . In this case, Q^n(\psi)\rightarrow M_1 as n\rightarrow\infty .
Case 2. I(t_2, \psi) > 0 for some t_2 > 0 .
From the third and fourth equations of (2.1), we have I(t, \psi) > 0 and G(t, \psi) > 0 for all t\geq t_2 . Thus Q^n(\psi)\in X_0 for n\omega > t_2 .
For the case \psi_4(0) = 0 , we can do similar analysis. Finally, we see that for any \psi\in M_\partial , Q^n(\psi)\rightarrow M_1 as n\rightarrow\infty . Thus \omega(\psi) = M_1 for any \psi\in M_\partial , and M_1 can not form a cycle in \partial X_0 .
By Claim 3, we see that M_1 is an isolated invariant set for Q in X , and W^s(M_1)\cap X_0 = \emptyset , where W^s(M_1) is the stable set of M_1 for Q . By the acyclicity theorem on uniform persistence for maps (see [28,Theorem 1.3.1 and Remark 1.3.1], it follows that Q : X\rightarrow X is uniformly persistent with respect to X_0 . Thus, [28,Theorem 3.1.1] implies that the periodic semiflow Q(t) : X\rightarrow X is also uniformly persistent with respect to X_0 .
It remains to prove the practical uniform persistence. By [25,Theorem 4.5], with \rho(x) = d(x, \partial X_0) , it then follows that Q: X_0\rightarrow X_0 has a compact global attractor A_0 . For any \phi\in A_0 , we have \phi_i(0) > 0 for all i = 3, 4 . Let B_0 = \cup_{t\in[0, \omega]}Q(t)A_0 . Then \psi_i(0) > 0 , i = 3, 4 , for all \psi\in B_0 . It is easy to see that \lim_{t\rightarrow\infty}d(Q(t)\phi, B_0) = 0 for all \phi\in X_0 . Let X_{+} = C([-\tau, 0], \mathbb{R}^{4}_{+}) , and define a continuous function p: X_{+}\rightarrow \mathbb{R}_{+} by
p(\phi) = \min\{\phi_3(0), \phi_4(0)\}, \forall \phi \in X_{+}. |
Clearly, p(\phi) > 0 for all \phi\in B_0 . Since B_0 is a compact subset of X_+ , we have \inf_{\phi\in B_0}p(\phi) = \min_{\phi\in B_0}p(\phi) > 0 . By the attractiveness of B_0 , it then follows that there exists \eta > 0 such that
\liminf\limits_{t\rightarrow\infty}\min(I(t, \phi), G(t, \phi)) = \liminf\limits_{t\rightarrow\infty}p(Q(t)\phi)\geq \eta, \forall\phi\in X_0. |
This completes the proof.
The following theorem shows that the virus will be cleared from the population if R_0 < 1 .
Theorem 3.2. If R_0 < 1 , then the virus-free periodic solution E_0(t) = (N^*(t), 0, 0, 0) is globally attractive for system (2.1) in \mathcal{X}_{\delta_0} .
Proof. By the aforementioned analysis for (2.3), we know that (2.3) has a positive \omega -periodic solution N^*(t) which is globally asymptotically stable. It then follows that for any \epsilon > 0 , we can choose a sufficiently large integer n_1 > 0 such that n_1\omega > \tau and N^*(t)-\epsilon < N(t) = S(t)+E(t)+I(t) < N^*(t)+\epsilon for t > n_1\omega-\tau . Thus, for t > n_1\omega , we have
\begin{equation*} \left\{\begin{array}{rl} \frac{dI(t)}{dt}&\leq\left[\beta_1(t-\tau)I(t-\tau)+\frac{\beta_2(t-\tau)(N^*(t-\tau)+\epsilon)}{M}G(t-\tau)\right]e^{-\int^t_{t-\tau}\mu(r)dr}-\mu(t)I(t), \\[8pt] \frac{dG(t)}{dt}&\leq \frac{a(t)M}{N^*(t)-\epsilon}I(t)-\delta(t)G(t). \end{array}\right. \end{equation*} |
It then suffices to show that positive solutions of the auxiliary system
\begin{equation} \left\{\begin{array}{rl} \frac{d\widetilde{I}(t)}{dt}& = \left[\beta_1(t-\tau)\widetilde{I}(t-\tau)+\frac{\beta_2(t-\tau)(N^*(t-\tau)+\epsilon)}{M}\widetilde{G}(t-\tau)\right]e^{-\int^t_{t-\tau}\mu(r)dr} \\&\quad\, -\mu(t)\widetilde{I}(t), \\[8pt] \frac{d\widetilde{G}(t)}{dt}& = \frac{a(t)M}{N^*(t)-\epsilon}\widetilde{I}(t)-\delta(t)\widetilde{G}(t) \end{array}\right. \end{equation} | (3.2) |
tend to zero when t tends to infinity. Let P_\epsilon(t) be the solution maps of system (3.2) on \mathcal{Y} , and P_\epsilon: = P_\epsilon(\omega) . Since R_0 < 1 , then by Lemma 2.2, we have \lim_{\epsilon\rightarrow 0^+} r(P_\epsilon) = r(P) < 1 . Thus we can fix \epsilon > 0 small enough such that r(P_\epsilon) < 1 . According to [27,Lemma 1], there is a positive \omega -periodic function \varsigma(t) = (\varsigma_1(t), \varsigma_2(t)) , such that u_\epsilon(t) = e^{\frac{\ln r(P_\epsilon)}{\omega}t}\varsigma(t) is a positive solution of system (3.2).
Let l > 0 large enough, such that (I(t), G(t))\leq lu_\epsilon(t) for t\in[n_1\omega, n_1\omega+\tau] . Then [22,Theorem 5.1.1] implies that (I(t), G(t))\leq lu_\epsilon(t) for all t\geq n_1\omega+\tau . Hence, I(t)\rightarrow0 , G(t)\rightarrow0 as t\rightarrow\infty . It then follows from the theory of asymptotically periodic semiflow (see [28,Theorem 3.2.1]) that
\lim\limits_{t\rightarrow\infty}(S(t)-N^*(t)) = 0, \, \lim\limits_{t\rightarrow\infty}E(t) = 0. |
This completes the proof.
In this section, we study the corresponding autonomous system of system (2.1), that is, we suppose that all coefficients in system (2.1) are time independent and positive. In this case, system (2.1) becomes
\begin{equation} \left\{\begin{array}{rl} \dfrac{dS(t)}{dt}& = B-\mu S(t)-\frac{\beta_1S(t)I(t)}{N(t)}-\frac{\beta_2S(t)G(t)}{M}, \\[10pt] \dfrac{dE(t)}{dt} & = \frac{\beta_1S(t)I(t)}{N(t)}+\frac{\beta_2S(t)G(t)}{M}-\mu E(t)\\[10pt]& \quad-\left[\dfrac{\beta_1S(t-\tau)I(t-\tau)}{N(t-\tau)}+\dfrac{\beta_2S(t-\tau)G(t-\tau)}{M}\right]e^{-\mu\tau}, \\[8pt] \dfrac{dI(t)}{dt} & = \left[\frac{\beta_1 S(t-\tau)I(t-\tau)}{N(t-\tau)}+\frac{\beta_2S(t-\tau)G(t-\tau)}{M}\right]e^{-\mu\tau} -\mu I(t), \\[10pt] \dfrac{dG(t)}{dt} & = \frac{aI(t)}{N(t)}(M-G(t))-\delta G(t). \end{array}\right. \end{equation} | (4.1) |
For system (4.1), there is always the virus-free equilibrium P_0 = (N^*, 0, 0, 0) with N^* = \frac{B}{\mu} .
Let
\begin{equation*} \begin{array}{ccc} F_1 = \left( \begin{array}{ccc} \beta_1e^{-\mu\tau}& \frac{\beta_2 N^*e^{-\mu\tau}}{M}\\ 0&0\\ \end{array} \right), &F_2 = \left( \begin{array}{ccc} 0 &0 \\ \frac{aM}{N^*} &0 \end{array} \right)\, \, \, \mbox{and} &V = \left( \begin{array}{ccc} \mu & 0 \\ 0 & \delta \end{array} \right). \end{array} \end{equation*} |
It then follows that F(\phi) = F_1\phi(-\tau)+F_2\phi(0) . By [23,Corollary 2.1], we can define the basic reproduction number for system (4.1) as \mathbf{R_0} = r(\hat{F}V^{-1}) , where \hat{F} = F_1+F_2 . It then easily follows that
\begin{equation*} \begin{array}{ccc}\hat{F}V^{-1} & = \left( \begin{array}{ccc} \frac{\beta_1e^{-\mu\tau}}{\mu}& \frac{\beta_2 N^*e^{-\mu\tau}}{\delta M}\\ \frac{aM}{\mu N^*}& 0\\ \end{array} \right)\end{array} \end{equation*} |
and
\mathbf{R_0} = \frac{1}{2}\left(C_1+\sqrt{C_1^2+4C_0}\right), |
where C_0 = \frac{a\beta_2e^{-\mu\tau}}{\mu\delta} , C_1 = \frac{\beta_1e^{-\mu\tau}}{\mu} . We define \Theta = \frac{\beta_1\delta+\beta_2a}{\mu\delta}e^{-\mu\tau}, it is easy to verify that \mathbf{R_0}-1 has the same sign as \Theta-1 .
Now consider endemic equilibria with I = I^* > 0 . Let the right-hand sides be zero, then system (4.1) admits another equilibrium: P^* = (S^*, E^*, I^*, G^*) , where
\begin{equation*} {\begin{array}{l} S^* = N^*-e^{\mu\tau}I^*, \\[8pt] E^* = (e^{\mu\tau}-1)I^*, \\[8pt] G^* = \dfrac{aM I^*}{aI^*+\delta N^*}, \\[8pt] \end{array}} \end{equation*} |
and I^* is a positive real root of equation g_1(I) = g_2(I) , where
\begin{equation*} g_1(I) = \frac{\beta_1}{N^*}+\frac{\beta_2a }{aI+\delta N^*}, \quad g_2(I) = \frac{\mu e^{\mu\tau}}{N^*-e^{\mu\tau}I}. \end{equation*} |
Note that g_1(0) = \frac{\beta_1}{N^*}+\frac{\beta_2a }{\delta N^*} , g_1(I) is decreasing with respect to I > 0 and \lim_{I\rightarrow+\infty}g_1(I) = \frac{\beta_1}{N^*} . Also g_2(0) = \frac{\mu e^{\mu\tau}}{N^*} , g'_2(I) > 0 for I\in(0, N^* e^{-\mu\tau}) , and \lim_{x\rightarrow (N^* e^{-\mu\tau})^-} g_2(x) = +\infty . Then g_1(I) = g_2(I) has a unique positive root in (0, N^* e^{-\mu\tau}) if and only if g_1(0) > g_2(0) , which is equivalent to \mathbf{R_0} > 1 . Therefore, if \mathbf{R_0} > 1 , system (4.1) has a unique endemic equilibrium P^* .
Define
\begin{array}{rl} \Omega_{\delta_0} = &\{\phi\in W: \sum^3_{i = 1}\phi_i(s)\geq\delta_0, \forall s\in[-\tau, 0], \\[8pt] &\phi_2(0) = \int^0_{-\tau}\left[\frac{\beta_1\phi_1(\xi)\phi_3(\xi)}{\sum^3_{i = 1}\phi_i(\xi)} +\frac{\beta_2\phi_1(\xi)\phi_4(\xi)}{M}\right]e^{\mu \xi} d\xi\}.\end{array} |
The following theorem provides the global stability of the virus-free equilibrium.
Theorem 4.1. If \mathbf{R_0} < 1 , then the virus-free equilibrium P_0 is globally asymptotically stable for system (4.1) in \Omega_{\delta_0} , and is unstable if \mathbf{R_0} > 1 .
Proof. Linearizing system (4.1) at the virus-free equilibrium P_0 , we obtain the characteristic equation
\begin{equation*} (\lambda+\mu)^2g(\lambda, \tau) = 0, \label{4.2} \end{equation*} |
with
g(\lambda, \tau) = \lambda^2+(\mu+\delta)\lambda-e^{-\mu\tau}e^{-\lambda\tau}(\beta_1\lambda+\beta_1\delta+\beta_2a)+\mu\delta. |
It is clear that the local stability of P_0 is determined by the roots of g(\lambda, \tau) = 0. If \tau = 0, then g(\lambda, 0) = \lambda^2+(\mu+\delta-\beta_1)\lambda+\mu\delta-(\beta_1\delta+\beta_2a). If \mathbf{R_0} < 1 , then \Theta = \frac{\beta_1\delta+\beta_2a}{\mu\delta} < 1, which implies that \mu+\delta-\beta_1 > 0 and \mu\delta-(\beta_1\delta+\beta_2a) > 0 . Then by the Routh-Hurwitz criterion, all the roots of g(\lambda, 0) = 0 have negative real parts, while if \mathbf{R_0} > 1 , then \mu\delta-(\beta_1\delta+\beta_2a) < 0 , and g(\lambda, 0) = 0 has one positive real root. That is, for \tau = 0 , P_0 is locally asymptotically stable for \mathbf{R_0} < 1 , and is unstable if \mathbf{R_0} > 1 .
Next we will show that g(\lambda, \tau) = 0 has no pure imaginary roots if \mathbf{R_0} < 1 . Obviously, i\rho ( \rho\in \mathbb{R} ) is a root of g(\lambda, \tau) = 0 if and only if \rho satisfies
-\rho^{2}+(\mu+\delta)\rho i-e^{-\mu\tau}(\cos\rho\tau-i\sin\rho\tau)(\beta_1\rho i+\beta_1\delta+\beta_2a)+\mu\delta = 0. |
Separating the real and the imaginary parts, we have
\begin{equation*} \left\{\begin{array}{rl} \mu\delta-\rho^2& = \beta_1e^{-\mu\tau}\rho\sin\rho\tau+(\beta_1\delta+\beta_2a)e^{-\mu\tau}\cos\rho\tau, \\[10pt] (\mu+\delta)\rho & = \beta_1e^{-\mu\tau}\rho\cos\rho\tau-(\beta_1\delta+\beta_2a)e^{-\mu\tau}\sin\rho\tau. \end{array}\right. \end{equation*} |
Squaring the two equations and adding them gives
\begin{equation} z^2+(\mu^2+\delta^2-\beta^2_1e^{-2\mu\tau})z+\mu^2\delta^2(1-\Theta^2) = 0, \end{equation} | (4.2) |
where z = \rho^2 . If \mathbf{R_0} < 1 , we have \beta_1e^{-\mu\tau} < \mu , hence, (4.2) has no nonnegative root, which implies that g(\lambda, \tau) = 0 has no zero root and pure imaginary roots. Therefore, P_0 is locally asymptotically stable for all \tau\geq0 if \mathbf{R_0} < 1 . Together with Theorem 3.2, we can show that P_0 is globally asymptotically stable in \Omega_{\delta_0} when \mathbf{R_0} < 1 .
We let
S^\infty = \limsup\limits_{t\rightarrow\infty}S(t), \quad S_\infty = \liminf\limits_{t\rightarrow\infty} S(t). |
We can define I^\infty, I_\infty, G^\infty, G_\infty in a similar way. Now we will use the method of fluctuations (see, e.g., [29]) to show the global attractivity of the endemic equilibrium P^* . To do this, we need the following assumption:
(H) (e^{-\mu\tau}a-\delta)\beta_2a > \beta_1(a+\delta)^2.
The condition (H) is technically needed in the arguments supporting the global attractivity of the endemic equilibrium P^* for system (4.1).
Theorem 4.2. Let \mathrm{(H)} hold. If \mathbf{R_0} > 1 , then (4.1) has a unique endemic equilibrium P^* = (S^*, E^*, I^*, G^*) such that \lim_{t\rightarrow\infty}u(t, \phi) = P^* for any \phi\in \Omega_{\delta_0} with \phi_3(0) > 0 and \phi_4(0) > 0 .
Proof. The whole mouse population satisfies the following equation:
\frac{dN(t)}{dt} = B-\mu N(t). |
Then N^* is globally asymptotically stable. Hence, we have the following limiting system:
\begin{equation} \left\{\begin{array}{rl} \dfrac{dS(t)}{dt}& = B-\mu S(t)-\widetilde{\beta}_1S(t)I(t)-\widetilde{\beta_2}S(t)G(t), \\[10pt] \dfrac{dE(t)}{dt} & = \widetilde{\beta}_1S(t)I(t)+\widetilde{\beta_2}S(t)G(t)-\mu E(t)\\[10pt]& \quad-[\widetilde{\beta}_1S(t-\tau)I(t-\tau)+\widetilde{\beta}_2S(t-\tau)G(t-\tau)]e^{-\mu\tau}, \\[8pt] \dfrac{dI(t)}{dt} & = [\widetilde{\beta}_1S(t-\tau)I(t-\tau)+\widetilde{\beta}_2S(t-\tau)G(t-\tau)]e^{-\mu\tau} -\mu I(t), \\[10pt] \dfrac{dG(t)}{dt} & = \widetilde{a}I(t)(M-G(t))-\delta G(t), \end{array}\right. \end{equation} | (4.3) |
where \widetilde{\beta}_1 = \frac{\beta_1}{N^*} , \widetilde{\beta}_2 = \frac{\beta_2}{M} , \widetilde{a} = \frac{a}{N^*} . Since the second equation of system (4.3) is decoupled from the other three equations of system (4.3), we then consider the following system:
\begin{equation} \left\{\begin{array}{rl} \dfrac{dS(t)}{dt}& = B-\mu S(t)-\widetilde{\beta}_1S(t)I(t)-\widetilde{\beta_2}S(t)G(t), \\[10pt] \dfrac{dI(t)}{dt} & = [\widetilde{\beta}_1S(t-\tau)I(t-\tau)+\widetilde{\beta}_2S(t-\tau)G(t-\tau)]e^{-\mu\tau} -\mu I(t), \\[10pt] \dfrac{dG(t)}{dt} & = \widetilde{a}I(t)(M-G(t))-\delta G(t). \end{array}\right. \end{equation} | (4.4) |
From the proof of Lemma 2.1, we can show that
\Gamma = C\left([-\tau, 0], \left[0, \frac{B}{\mu}\right]^2\times[0, M]\right) |
is positively invariant for system (4.4). By the arguments similar to those in Theorem 3.1, it is easy to verify that system (4.4) is uniformly persistent in the sense that there exists a \eta_2 > 0 such that for any given \psi = (\psi_1, \psi_2, \psi_3)\in \Gamma with \psi_2(0) > 0 , \psi_3(0) > 0 , the solution (S(t, \psi), I(t, \psi), G(t, \psi)) of (4.4) satisfies \liminf_{t\rightarrow\infty}(I(t, \psi), G(t, \psi))\geq (\eta_2, \eta_2).
For any given \psi\in \Gamma with \psi_2(0) > 0 , \psi_3(0) > 0 , let (S(t), I(t), G(t)) = (S(t, \psi), I(t, \psi), G(t, \psi)) . It is clear that S^\infty\geq S_\infty , \frac{B}{\mu}\geq I^\infty\geq I_\infty\geq \eta_2 > 0 , G^\infty\geq G_\infty\geq\eta_2 > 0 . Moreover, there exist sequences t^i_n and \sigma^i_n , i = 1, 2, 3 , such that
\lim\limits_{n\rightarrow \infty} S(t^1_n) = S^\infty, \, S'(t^1_n) = 0, \, \lim\limits_{n\rightarrow \infty}S(\sigma^1_n) = S_\infty, \, S'(\sigma^1_n) = 0, \forall n\geq1; |
\lim\limits_{n\rightarrow \infty} I(t^2_n) = I^\infty, \, I'(t^2_n) = 0, \, \lim\limits_{n\rightarrow \infty}I(\sigma^2_n) = I_\infty, \, I'(\sigma^2_n) = 0, \forall n\geq1; |
\lim\limits_{n\rightarrow \infty} G(t^3_n) = G^\infty, \, G'(t^3_n) = 0, \, \lim\limits_{n\rightarrow \infty}G(\sigma^3_n) = G_\infty, \, G'(\sigma^3_n) = 0, \forall n\geq1. |
It then follows from the first equation of (4.4), we have
B-\mu S^\infty-\widetilde{\beta}_1S^\infty I_\infty-\widetilde{\beta_2}S^\infty G_\infty\geq0 \geq B-\mu S^\infty-\widetilde{\beta}_1S^\infty I^\infty-\widetilde{\beta_2}S^\infty G^\infty, |
and
B-\mu S_\infty-\widetilde{\beta}_1S_\infty I_\infty-\widetilde{\beta_2}S_\infty G_\infty\geq0 \geq B-\mu S_\infty-\widetilde{\beta}_1S_\infty I^\infty-\widetilde{\beta_2}S_\infty G^\infty, |
we see that
\begin{equation} \frac{B}{\mu+\widetilde{\beta}_1I_\infty+\widetilde{\beta}_2G_\infty}\geq S^\infty\geq S_\infty\geq \frac{B}{\mu+\widetilde{\beta}_1I^\infty+\widetilde{\beta}_2G^\infty}. \end{equation} | (4.5) |
By the second and third equations of (4.4), by a similar argument, it then follows that
\begin{equation} \frac{(\widetilde{\beta}_1S^\infty I^\infty+\widetilde{\beta_2}S^\infty G^\infty)e^{-\mu\tau}}{\mu}\geq I^\infty\geq I_\infty\geq \frac{(\widetilde{\beta}_1S_\infty I_\infty+\widetilde{\beta_2}S_\infty G_\infty)e^{-\mu\tau}}{\mu}, \end{equation} | (4.6) |
and
\begin{equation} \frac{\widetilde{a}MI^\infty}{\widetilde{a}I^\infty+\delta}\geq G^\infty\geq G_\infty\geq \frac{\widetilde{a}MI_\infty}{\widetilde{a}I_\infty+\delta}. \end{equation} | (4.7) |
Using (4.5) and (4.6), we get
\begin{equation} \frac{(\widetilde{\beta}_1I^\infty+\widetilde{\beta_2}G^\infty)Be^{-\mu\tau}}{\mu(\mu+\widetilde{\beta}_1I_\infty+\widetilde{\beta}_2G_\infty)}\geq I^\infty\geq I_\infty\geq \frac{(\widetilde{\beta}_1I_\infty+\widetilde{\beta_2}G_\infty)Be^{-\mu\tau}}{\mu(\mu+\widetilde{\beta}_1I^\infty+\widetilde{\beta}_2G^\infty)}. \end{equation} | (4.8) |
Then (4.7) and (4.8) imply that
\frac{Be^{-\mu\tau}}{\mu}(\widetilde{\beta}_1+\frac{\widetilde{\beta}_2\widetilde{a}M}{\widetilde{a}I^\infty+\delta}) \geq\mu+\widetilde{\beta}_1 I_\infty+\frac{\widetilde{\beta}_2\widetilde{a}MI_\infty}{\widetilde{a}I_\infty+\delta}, |
\frac{Be^{-\mu\tau}}{\mu}(\widetilde{\beta}_1+\frac{\widetilde{\beta}_2\widetilde{a}M}{\widetilde{a}I_\infty+\delta}) \leq\mu+\widetilde{\beta}_1 I^\infty+\frac{\widetilde{\beta}_2\widetilde{a}MI^\infty}{\widetilde{a}I^\infty+\delta}. |
Simplifying the above two inequalities, we get
\widetilde{\beta}_1(\widetilde{a}I^\infty+\delta)(\widetilde{a}I_\infty+\delta)(I^\infty-I_\infty)\geq \widetilde{\beta}_2\widetilde{a}M\left(\frac{Be^{-\mu\tau}}{\mu}\widetilde{a}-\delta\right)(I^\infty-I_\infty). |
Since \widetilde{\beta}_1(\widetilde{a}I^\infty+\delta)(\widetilde{a}I_\infty+\delta)\leq \widetilde{\beta}_1(\widetilde{a}\frac{B}{\mu}+\delta)^2 , then if condition (H) holds, we have I^\infty = I_\infty . By (4.5) and (4.7), we get S^\infty = S_\infty and G^\infty = G_\infty . Then we have \lim_{t\rightarrow\infty}(S(t), I(t), G(t)) = (S^*, I^*, G^*) for any \psi\in \Gamma with \psi_2(0) > 0 and \psi_3(0) > 0 . By the theory of chain transitive sets [28], we can lift the global attractivity for system (4.4) to system (4.1). It follows that \lim_{t\rightarrow\infty} u(t, \phi) = P^* , for any \phi\in \Omega_{\delta_0} with \phi_3(0) > 0 and \phi_4(0) > 0 .
In this section, we give numerical simulations that support the theory presented in the previous sections. To compute the basic reproduction number R_0 numerically, we first write the operator L into the integral form in [30]. For v\in C_\omega , we get
\begin{equation*} \begin{array}{rl} [Lv](t)& = \int_0^\infty\Phi(t, t-s)F(t-s)v(t-s+\cdot)ds\\[8pt] & = \int_0^\infty\Phi(t, t-s)\\[8pt]&\quad\times \begin{array}{ccc} \left( \begin{array}{ccc} [\beta_1(t-s-\tau)v_1(t-s-\tau)+\frac{\beta_2(t-s-\tau)N^*(t-s-\tau)}{M}v_2(t-s-\tau)]e^{-\int^{t-s}_{t-s-\tau}\mu(r)dr} \\[8pt] \frac{a(t-s)M}{N^*(t-s)}v_1(t-s)\\ \end{array} \right)\end{array} ds\\[15pt] & = \int_0^\infty\begin{array}{ccc} \left( \begin{array}{ccc} [\beta_1(t-s-\tau)v_1(t-s-\tau)+\frac{\beta_2(t-s-\tau)N^*(t-s-\tau)}{M}v_2(t-s-\tau)]e^{-\int^{t}_{t-s-\tau}\mu(r)dr} \\[8pt] \frac{a(t-s)M}{N^*(t-s)}e^{-\int^t_{t-s}\delta(r)dr} v_1(t-s)\\ \end{array} \right)\end{array}ds\\[15pt] & = \begin{array}{ccc} \left( \begin{array}{ccc} \int_\tau^\infty[\beta_1(t-s)v_1(t-s)+\frac{\beta_2(t-s)N^*(t-s)}{M}v_2(t-s)]e^{-\int^{t}_{t-s}\mu(r)dr}ds \\[8pt] \int_0^\infty\frac{a(t-s)M}{N^*(t-s)}e^{-\int^t_{t-s}\delta(r)dr} v_1(t-s)ds\\ \end{array} \right)\end{array}\\[15pt] & = \int_0^\infty K(t, s)v(t-s)ds, \end{array} \end{equation*} |
where
\begin{equation*} \begin{array}{ccc} K(t, s) = \left( \begin{array}{ccc} \beta_1(t-s)e^{-\int^t_{t-s}\mu(r)dr}& \frac{\beta_2(t-s)N^*(t-s)}{M}e^{-\int^{t}_{t-s}\mu(r)dr} \\ \frac{a(t-s)M}{N^*(t-s)}e^{-\int^t_{t-s}\delta(r)dr} & 0\\ \end{array} \right)\, \, \mbox{if} \, \, s\geq\tau, \end{array} \end{equation*} |
and
\begin{equation*} \begin{array}{ccc} K(t, s) = \left( \begin{array}{ccc} 0& 0 \\ \frac{a(t-s)M}{N^*(t-s)}e^{-\int^t_{t-s}\delta(r)dr} & 0\\ \end{array} \right)\, \, \mbox{if} \, \, s \lt \tau.\end{array} \end{equation*} |
By the \omega -periodicity of v , we obtain
\begin{equation*} [Lv](t) = \int_0^\omega G(t, s)v(t-s)ds, \end{equation*} |
where G(t, s) = \sum^\infty_{n = 0} K(t, s+n\omega) . Consequently, we can use the numerical method in [30] to compute R_0 .
The time unit is taken as month. Baseline parameters are \mu = 0.246 , \beta = 0.0024 , \alpha = 0.0016 , \gamma = 0.5 , d = 60 , \delta = 9.21 , M = 50 . These parameters values are taken from [10]. To reflect the seasonality, we suppose that the birth rate of mice is B(t) = 5.85(1+0.8\sin(\pi t/6)) , the direct contact rate and the indirect contact rate are equal, that is, c(t) = \bar{c}(t) = 60(1+0.8\sin(\pi t/6)) . Since the incubation period is about one week, then we take \tau = \frac{1}{3} .
With this set of parameters, we have R_0 = 1.3586 > 1 , the virus will be endemic and the infection is persistent in the mouse population (see Figure 1(a)). If we can increase the incubation period to \tau = 2 , then R_0 = 0.9876 < 1 . In this case, the long-term behavior of the infectious mice is shown in Figure 1(b), which implies that the virus will be eradicated from the mouse population. These results are coincident with Theorems 3.1 and 3.2.
First, we discuss the influence of the incubation period. Let \tau varies and keep other parameters as above. Our numerical computations demonstrate that R_0 is a decreasing function (see Figure 2). Hence, we may try to prolong the incubation period via medical drugs or control measures to control hantavirus infection. For example, to eradicate the virus, we should keep \tau > 1.923 .
The dependence of \mathbf{R_0} on \tau is also illustrated in Figure 2. One can observe that for each fixed \tau , \mathbf{R_0} is greater than or equal to R_0 . Thus, the autonomous model may overestimate the value of R_0 .
Since contact rate plays a very important role in the spread of hantavirus. To explore the influence of direct contact rate and indirect contact rate on hantavirus transmission, we replace c(t) with \hat{c}(t) = (1-k)c(t) and \bar{c}(t) with \hat{\bar{c}}(t) = (1-l)\bar{c}(t) in our model. Then, Figure 3 shows R_0 is a decreasing function with respect to k and l . Moreover, Figure 3(a) shows that the virus will be endemic even if k = 1 (that is, \hat{c}(t) = 0 ) for l = 0 , Figure 3(b) shows that if we keep l > 0.63 , then the virus can be eradicated for k = 0 . For the parameters we have chosen, R_0 is more sensitive to the indirect transmission than the direct transmission. Thus, indirect transmission can not be neglected.
In this paper, we formulate and analyze a compartmental model for hantavirus infection that incorporating the seasonality, incubation period of the mice, direct and indirect transmission. Using the theory developed in [23], we first introduce the basic reproduction number R_0 for the model. Then we show that R_0 is a threshold parameter for the persistence and extinction of the virus. More precisely, the virus will be endemic when R_0 > 1 , and the virus will be cleared if R_0 < 1 . For the corresponding autonomous system, we obtain an explicit expression of \mathbf{R_0} and establish a threshold result on the global stability in terms of \mathbf{R_0} .
Numerical simulations are performed to illustrate our analytic results. Figure 2 shows that an increase of the incubation period could reduce R_0 , which implies that the virus infection can be relieved by prolonging the length of the incubation period. Moreover, our numerical results show that completely abolishing direct transmission would not eradicate virus (see Figure 3(a)), indirect transmission may play an important role in the virus transmission. On the other hand, Figure 3(b) shows that lowering the indirect transmission to 37\% of the original value would lead to eradication of the virus. Hence, outdoor disinfection is very important to control the transmission of hantavirus.
In this paper, we ignore the age-structure of the mouse population, since juvenile mice must leave to find and establish their own home ranges, while the adults do not move, so it should be more reasonable to consider two types of mice: the adult and juvenile mice as noted in [9]. Also the diffusion of the juvenile mice plays an important role in hantavirus infection. We leave these problems for further investigation.
This work was supported by the National Natural Science Foundation of China (No. 11801431), the Natural Science Basic Research Plan in Shaanxi Province of China (No. 2018JM1011), and China Scholarship Council.
The author declares there is no conflict of interest.
[1] | IEA (2002) World Energy Outlook, chapter 13, 365–389. |
[2] |
Yazdanpanah M, Komendantova N, Ardestani RS (2015) Governance of energy transition in Iran: Investigating public acceptance and willingness to use renewable energy sources through socio-psychological model. Renew Sust Energ Rev 45: 565–573. doi: 10.1016/j.rser.2015.02.002
![]() |
[3] | Komendantova N, Yazdanpanah M (2016) Impacts of Human Factors on Willingness to Use Renewable Energy Sources in the Middle East and North African Region. Environ Energy Econ Int Res 1: 173–184. |
[4] | British Petroleum (2013) BP Statistical Review of World Energy June 2013. Available from: http://www.bp.com/content/dam/bp/pdf/statisticalreview/statistical_review_of_world_energy_2013. |
[5] | SUNA (2014) About renewable energies. Renewable energy organization of Iran. |
[6] |
Alamdari P, Nematollahi O, Alemrajabi AA (2013) Solar energy potentials in Iran: A review. Renew Sust Energ Rev 21: 778–788. doi: 10.1016/j.rser.2012.12.052
![]() |
[7] |
Fadai D, Esfandabadi ZS, Abbasi A (2011) Analyzing the causes of non-development of renewable energy-related industries in Iran. Renew Sust Energ Rev 15: 2690–2695. doi: 10.1016/j.rser.2011.03.001
![]() |
[8] |
Hosseini SE, Andwari AM, Wahid MA, et al. (2013) A review on green energy potentials in Iran. Renew Sust Energ Rev 27: 533–545. doi: 10.1016/j.rser.2013.07.015
![]() |
[9] |
Ghobadian B, Najafi G, Rahimi H, et al. (2009) Future of renewable energies in Iran. Rene Sust Energ Rev 13: 689–695. doi: 10.1016/j.rser.2007.11.010
![]() |
[10] |
Mostafaeipour A, Jadidi M, Mohammadi K, et al. (2014) An analysis of wind energy potential and economic evaluation in Zahedan, Iran. Renew Sust Energ Rev 30: 641–650. doi: 10.1016/j.rser.2013.11.016
![]() |
[11] |
Wieser R, Pickle S, Goldman C (1998) Renewable energy policy and electricity restructuring: A California case study. Energy Policy 26: 465–475. doi: 10.1016/S0301-4215(98)00002-0
![]() |
[12] | Kunreuther H, Gupta S, Bosetti V, et al. (2014) Integrated Risk and Uncertainty Assessment of Climate Change Response Policies, In: Sokona Y, Farahani E, Kadner S, et al. Author, Climate Change 2014: Mitigation of Climate Change. Contribution of Working Group III to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change, New York: Cambridge University Press. |
[13] | Sovacool BK (2014) Energy Security, Ed., London: Sage, Six Volumes,. |
[14] | Wheeler SA (2008) The barriers to further adoption of organic farming and genetic engineering in Australia: Views of agricultural professionals and their information sources. Renew Agric Food Syst 23: 161–170. |
[15] |
Komendantova N, Voccciante M, Battaglini A (2015) Can the BestGrid Process Improve Stakeholder Involvement in Electricity Transmission Projects? Energies 8: 9407–9433. doi: 10.3390/en8099407
![]() |
[16] | Bandura A (1986) Social foundations of thought and action. J Appl Psychol 12: 169. |
[17] |
Lorraine SWPD, Janet-Buckworth F, Timothy EKPD, et al. (2000) Characteristics of exercise behavior among college students: Application of social cognitive theory to predicting stage of change. Prev Med 31: 494–505. doi: 10.1006/pmed.2000.0736
![]() |
[18] |
Winters ER, Petosa RL, Charlton TE (2003) Using social cognitive theory to explain discretionary, "leisure-time" physical exercise among high school students. J Adolesc Health 32: 436–442. doi: 10.1016/S1054-139X(03)00046-6
![]() |
[19] | Plotnikoff RC, Costigan SA, Karunamuni N, et al. (2013) Social cognitive theories used to explain physical activity behavior in adolescents: A systematic review and meta-analysis. Prev Med 56: 245–253. |
[20] |
Zundert RMPV, Nijhof LM, Engels RCME (2009) Testing social cognitive theory as a theoretical framework to predict smoking relapse among daily smoking adolescents. Addict Behav 34: 281–286. doi: 10.1016/j.addbeh.2008.11.004
![]() |
[21] |
Young HN, Lipowski EE, Cline RJ (2005) Using social cognitive theory to explain consumers' behavioral intentions in response to direct-to-consumer prescription drug advertising. Res Soc Adm Pharm 1: 270–288. doi: 10.1016/j.sapharm.2005.03.011
![]() |
[22] |
Stallones L, Acosta MS, Sample P, et al. (2009) Perspectives on safety and health among migrant and seasonal farmworkers in the United States and Mexico: A qualitative field study. J Rural Health 25: 219–225. doi: 10.1111/j.1748-0361.2009.00221.x
![]() |
[23] |
Rinderknecht K, Smith C (2004) Social cognitive theory in an after-school nutrition intervention for urban Native American youth. J Nutr Educ Behav 36: 298–304. doi: 10.1016/S1499-4046(06)60398-9
![]() |
[24] |
Larose R, Mastro D, Eastin MS (2001) Understanding internet usage: A social-cognitive approach to uses and gratifications. Soc Sci Comput Rev 19: 395–413. doi: 10.1177/089443930101900401
![]() |
[25] |
Wang SL, Wu PY (2008) The role of feedback and self-efficacy on web-based learning: The social cognitive perspective. Comput Educ 51: 1589–1598. doi: 10.1016/j.compedu.2008.03.004
![]() |
[26] |
Phipps M, Ozanne LK, Luchs MG, et al. (2013) Understanding the inherent complexity of sustainable consumption: A social cognitive framework. J Bus Res 66: 1227–1234. doi: 10.1016/j.jbusres.2012.08.016
![]() |
[27] |
Hawley L, Harker D, Harker M (2010) A social cognitive approach to tackle inactivity and obesity in young Australians. J Bus Res 63: 116–120. doi: 10.1016/j.jbusres.2009.02.004
![]() |
[28] | Yazdanpanah M, Komendantova N, Shirazi ZN, et al. (2015) Green or in between? Examining youth perceptions of renewable energy in Iran. Energy Res Soc Sci 8: 78–85. |
[29] |
Mousavi B, Lopez N, Biona J, et al. (2017) Driving forces of Iran's CO2 emissions from energy consumption: An LMDI decomposition approach. Appl Energy 206: 804–814. doi: 10.1016/j.apenergy.2017.08.199
![]() |
[30] |
Armitage CJ, Conner M (2001) Efficacy of the Theory of Planned Behaviour: A Meta-Analytic Review. Br J Soc Psychol 40: 471–499. doi: 10.1348/014466601164939
![]() |
[31] | Komendantova N, Battaglini A (2016) Beyond Decide-Announce-Defend (DAD) and Not-in-My-Backyard (NIMBY) models? Addressing the social and public acceptance of electric transmission lines in Germany. Energy Res Soc Sci 22: 224–231. |
[32] |
Komendantova N, Riegler M, Neumueller S (2018) Of transitions and models: Community engagement, democracy, and empowerment in the Austrian energy transition. Energy Res Soc Sci 39: 141–151. doi: 10.1016/j.erss.2017.10.031
![]() |
[33] |
Bakhtiyari Z, Yazdanpanah M, Forouzani M, et al. (2017) Intention of agricultural professionals toward biofuels in Iran: Implications for energy security, society, and policy. Renew Sust Energ Rev 69: 341–349. doi: 10.1016/j.rser.2016.11.165
![]() |
1. | Dickson Kindole, Yasutaka Ando, Application of atmospheric solution precursor plasma spray to photocatalytic devices for small and medium industries in developing countries, 2017, 56, 0021-4922, 01AB10, 10.7567/JJAP.56.01AB10 | |
2. | Energy and Water Needs Analysis: Towards Solar Photovoltaic Water Pumping in Rural Areas of Malawi, 2020, 2, 10.20900/jsr20200013 | |
3. | Saadullah Khan, Aqueel Ahmad, Furkan Ahmad, Mahdi Shafaati Shemami, Mohammad Saad Alam, Siddiq Khateeb, A Comprehensive Review on Solar Powered Electric Vehicle Charging System, 2018, 6, 2308-0477, 54, 10.1080/23080477.2017.1419054 | |
4. | Julia Bello-Bravo, Ian Brooks, Anne Namatsi Lutomia, Jeremy Bohonos, John Medendorp, Barry Pittendrigh, Breaking out: the turning point in learning using mobile technology, 2021, 7, 24058440, e06595, 10.1016/j.heliyon.2021.e06595 | |
5. | David M Sackey, De-Graft Owusu-Manu, Amevi Acakpovi, Frank K Appiah, David J Edwards, Reza M Hosseini, Exploring the state of solar photovoltaic decentralization in Ghana: trends and success factors, 2022, 1042, 1755-1307, 012013, 10.1088/1755-1315/1042/1/012013 | |
6. | Iurii Demidov, Antti Pinomaa, Samuli Honkapuro, Marko Nieminen, 2025, Energy Management System for Inter-Connected Self-Sufficient Microgrids with Blockchain-Based Peer-to-Peer Energy Market, 979-8-3315-1278-1, 1, 10.1109/EEM64765.2025.11050162 |