key: cord-0828773-u1fa64eb
authors: Saha, Sangeeta; Samanta, Guruprasad; Nieto, Juan J.
title: Impact of optimal vaccination and social distancing on COVID-19 pandemic
date: 2022-04-30
journal: Math Comput Simul
DOI: 10.1016/j.matcom.2022.04.025
sha: e650af5cbf8dc8b25125b2859770fc61d391d57d
doc_id: 828773
cord_uid: u1fa64eb

The first COVID–19 case was reported at Wuhan in China at the end of December 2019 but till today the virus has caused millions of deaths worldwide. Governments of each country, observing the severity, took non-pharmaceutical interventions from the very beginning to break the chain of higher transmission. Fortunately, vaccines are available now in most countries and people are asked to take recommended vaccines as precautionary measures. In this work, an epidemiological model on COVID–19 is proposed where people from the susceptible and asymptomatically infected phase move to the vaccinated class after a full two-dose vaccination. The overall analysis says that the disease transmission rate from symptomatically infected people is most sensitive on the disease prevalence. Moreover, better disease control can be achieved by vaccination of the susceptible class. In the later part of the work, a corresponding optimal control problem is considered where maintaining social distancing and vaccination procedure change with time. The result says that even in absence of social distancing, only the vaccination to people can significantly reduce the overall infected population. From the analysis, it is observed that maintaining physical distancing and taking vaccines at an early stage decreases the infection level significantly in the environment by reducing the probability of becoming infected.

The spread of coronavirus first started from Wuhan in China in the middle of December 2019 [8, 27, 9] . The first case was reported near Hunan seafood market where live animals are traded [22] . But within few weeks it spreads all over the Chinese province and in few months, the whole world was affected by this virus. Within three months, WHO declared COVID-19 as a pandemic observing its severity. In order to reduce the higher disease transmission, the Government of almost every country took some non-pharmaceutical interventions (such as maintaining physical distancing, using face-mask and alcohol-based hand sanitizers) from the very beginning. In the case of COVID-19, the respiratory system mainly becomes affected. In many cases, people have mild symptoms and recover on their own. But some may show severe problems (breathing problems, fluctuation of oxygen level, etc.). The most common symptoms of COVID-19 include fever, dry cough, loss of taste, diarrhea, etc. and if a patient suffers from breathing problems, pain or pressure in the chest, etc., they are advised for medical help as soon as possible [8, 17, 43, 7, 40] . The huge spread of the virus led the Governments of each country to call for partial or full lockdown at different times of March 2020. According to the data of Worldometer and dashboard of John Hopkins University, there was 178,488,817 number of COVID-19 cases on 20 th June 2021 worldwide among which the count of active cases and death cases were 11,471,056 and 178,488,817 respectively [11, 42] . Fortunately, vaccines for COVID-19 are available right now. According to the data of 20 th June 2021, a total of 781,873,425 people are vaccinated across the world with two-dose of full vaccination and 1,704,015,929 people have got the first dose. It is true that the vaccination procedure takes few months to implement and also people need some time to develop immunity after inoculation. With this vaccination, the level of infectivity can be decreased in the lockdown period.

In India, the first COVID-19 case was reported on 20 th January 2020 at Kerala. The number of confirmed cases crossed a benchmark of 5 million on 16 th September 2020 and surpassed the 10 million mark on 19th December 2020. The severity of infection crossed with two-crore confirmed cases on May 4, 2021. The continued rise of COVID-19 positivity rate has indicated that more people are carrying the virus with time. The government in India, from the very beginning, has announced some preventive measures to reduce the high transmission. One of the precautionary measures is maintaining social distancing to reduce the number of times people come into close contact with each other. The other preventive measures include adopting a self-quarantine strategy when slight symptoms are shown, using of face-masks and alcohol-based hand sanitizer, etc. When the number of infected cases reached 500, the Government of India announced for a 14hours 'Janta Curfew' on 22 nd March 2020 and called for nationwide lockdown on 25 th March 2020. The lockdown initially was for a fortnight, but as the situation became worse, it was expanded up to May 2020 through different phases. Then the unlockdown period started but with a long list of restrictions. State-wise lockdown was imposed a few times later depending on the severity of the infection. According to the data of PIB, Government of India, the active cases declined to 7,29,243 on 20 th June 2021 and less than 60,000 daily cases were reported (after 81 days) with 58,419 new cases [33] . Till J o u r n a l P r e -p r o o f Journal Pre-proof impact of vaccines is not considered there [12, 13, 14, 28, 29, 32, 35, 36, 37, 38] . Bhopal and Bhopal (2020) in their work have presented the significance of the epidemiological data on COVID-19 which are arranged by sex and age group [4] . However, there are some literature available discussing the effect of underreporting infections and the impact of vaccination on COVID-19 transmission [1, 2] . In this work, we have emphasized how the non-pharmaceutical intervention (social distancing) as well as the pharmaceutical intervention (vaccination) reduce the chance of becoming infected by the virus. It means how the precautionary measures decrease the contact rate of people with infected ones and people become less infected because of the immunity developed by vaccination. A compartmental epidemic model on COVID-19 is proposed with a separated compartment of vaccinated people. It is considered here that people from susceptible and asymptomatic states take vaccines to develop the immunity to protect themselves. Section 2 contains the proposed model on COVID-19 with non-negative initial conditions; Section 3 shows that the system is biologically well-defined. In Section 4, the basic reproduction number (R 0 ) is derived along with the endemic equilibrium point. In Section 5 it is observed how some of the system parameters are sensitive to R 0 and affect the disease transmission. The stability criteria of the equilibrium points are obtained in Section 6, and Section 7 shows the change of stability of disease-free equilibrium point through transcritical bifurcation. The consequent section contains the numerical simulation of the proposed model without any optimal control interventions. In the later part, an optimal control problem is formulated in Section 9 to reduce the overall infected population in the system and the next section contains the corresponding numerical scenarios to support the analytical part. The work ends with a brief conclusion.

The novel beta coronavirus causes a pandemic situation worldwide from 2020. Different models have been proposed in order to curb the high transmission rate of this virus. In this section, we have proposed a compartmental SIRV (Susceptible-Infected-Recovered-Vaccinated) model with a separate compartment of vaccinated people. In India, the first vaccination started on 16 th January 2021. All the people, if eligible, were requested from the very first day to take two doses of vaccine maintaining a certain time interval. The total population (N ) is divided into the following subpopulations: susceptible population (S), asymptomatically or pre-symptomatically infected population who are exposed to coronavirus without showing any symptoms (I 1 ), symptomatically infected and so quarantined population (I 2 ), hospitalized population (H), recovered population (R) and vaccinated population (V ). In a susceptible environment, people become infected and move to an asymptomatic state when they come in contact with asymptomatically infected class (I 1 ) and symptomatically infected class (I 2 ) with rates β 1 and β 2 respectively. The recruitment rate in the susceptible class (Λ) is assumed to be constant. It denotes the new susceptible people, who are coming by birth or immigration. The parameter d denotes the natural death rate which is incorporated in each class. It is considered that q 1 portion of susceptible and q 2 portion of asymptomatically infected people move to the vaccinated compartment after completing a full two-dose vaccination procedure. A person, whether is COVID-19 positive, is detected mostly by RT-PCR test. The swab from a person's throat or nose is used in this test. Besides, there are several tests like TrueNAT, antigen testing, etc. to detect COVID-19 in the human body. But these tests do not J o u r n a l P r e -p r o o f Journal Pre-proof show appropriate results every time and may result in false-negative tests. So, a person who tests negative through the tests still may have COVID-19. Moreover, in some cases, the symptoms develop after one or two weeks and so, a person turns out COVID positive even after two or three tests. So, people from asymptomatically infected class move to symptomatically infected class with rate κ when symptoms are shown and even when one has Covid-19 positive report. Further, deterioration of health condition indicates hospitalization of an infected person whereas people may recover by natural immunity also. So, people from symptomatically infected class move to either hospitals with rate φ 1 for regular observation or to recovered class with rate φ 2 . Moreover, asymptomatically infected people can also move to the recovered class with a rate of α if they recuperate by natural immunity. Also, hospitalized people move to recovered class (after proper medical treatment) with rate ψ. There are some reports stated that the recovery from the disease does not guarantee permanent recovery and so some of the recovered people move back to susceptible class further with rate constant η [37] . Lastly, the parameters µ 1 and µ 2 denote the disease-related death rates at symptomatically infected class and hospitalized compartment respectively. So, the model is proposed in system (1) as follows: Figure 1 contains a schematic diagram of the model system for clear understanding.

The following two theorems in this section show that the variables in system (1) are positive and bounded with time and so, the proposed system is biologically well-posed.

Theorem 3.1. Solutions of system (1) in R 6 + are positive for t > 0. Proof. As the right side functions of system (1) is continuous and locally Lipschitzian, so, there exists an unique solution of the system on [0, τ ) with 0 < τ ≤ +∞ [18] . Let us show, S(t) > 0, ∀ t ∈ [0, τ ). If the statement is not true, then ∃ t 1 ∈ (0, τ ) such that S(t 1 ) = 0,Ṡ(t 1 ) ≤ 0 and S(t) > 0, ∀ t ∈ [0, t 1 ). Then we have I 1 (t) ≥ 0, ∀ t ∈ [0, t 1 ). If it does not hold, then ∃ t 2 ∈ (0, t 1 ) such that I 1 (t 2 ) = 0,İ 1 (t 2 ) < 0 and (1) which is a contradiction toV (

. Suppose it is not true, then ∃ t 4 ∈ (0, t 2 ) such that I 2 (t 4 ) = 0,İ 2 (t 4 ) < 0 and I 2 (t) ≥ 0, ∀ t ∈ [0, t 4 ). The third equation gives

Similarly, we can show that H(t) ≥ 0 and R(t) ≥ 0, ∀ t ∈ [0, t 2 ). From the second equation of (1), we have

. By the above steps, we have

From the first equation of (1) we get

which contradictsṠ(t 1 ) ≤ 0. Hence we have, S(t) > 0, ∀ t ∈ [0, τ ) with 0 < τ ≤ +∞. Following the previous steps we get I 1 (t) ≥ 0,

J o u r n a l P r e -p r o o f Journal Pre-proof Theorem 3.2. Solutions of system (1) starting from R 6 + are bounded with time.

Proof.

Here N (0) is total population size at initial time.

for any > 0. The solutions of the system remain in the region: Ω ≡ (S, I 1 ,

and an endemic equilibrium point E * (S * , I * 1 , I * 2 , H * , R * , V * ).

Basic reproduction number R 0 is obtained by the process developed by van den Driessche and Watmough [39] . Consider, x ≡ (I 1 , I 2 ). Denote, p 0 = d + q 1 , p 1 = q 2 + κ + α + d, p 2 = φ 1 + φ 2 + d + µ 1 , p 3 = ψ + d + µ 2 and p 4 = d + η. Then we have:

where F(x) and ν(x) contain the compartment containing new infection term and other terms respectively. So, at the disease-free equilibrium E 0 = (S 0 , 0, 0, 0, 0, V 0 ) we have

The spectral radius of the next generation matrix F V −1 is R 0 and is given by:

Endemic equilibrium point E * (S * , I * 1 , I * 2 , H * , R * , V * ) J o u r n a l P r e -p r o o f Journal Pre-proof

Solving these equations, we get S * = S 0 R 0 , I * 1 =

Hence, from the calculation we get the following theorem as 

From the expression of basic reproduction number it is observed that R 0 depends on recruitment rate (Λ), natural death rate (d), disease transmission rates (β 1 , β 2 ), disease related death rate (µ 1 ), vaccination rates (q 1 , q 2 ), moving rate of asymptomatically infected people to symptomatically infected and recovered classes (κ, α) and moving rates of symptomatically infected people into hospitalized and recovered classes (φ 1 , φ 2 ). It is shown below how β 1 , β 2 , φ 1 , q 1 , q 2 affect on the transmission of the disease.

J o u r n a l P r e -p r o o f

Computing the normalized forward sensitivity index for the parameters β 1 , β 2 , φ 1 , q 1 and q 2 by the method of Arriola and Hyman, we have [3] :

From the expression of R 0 and also from the calculation it is observed that the virus transmission rates (β 1 , β 2 ) maintain a directly proportional relation with R 0 . It means increasing β i (i = 1, 2) escalates the basic reproduction number resulting in the occurrence of an epidemic situation in the system. It is evident that if people from the susceptible class come in contact with infected people (both asymptomatically and symptomatically) frequently without any precautionary measures, then the disease invades the population easily, and even at a larger rate. On the other hand, the rate at which symptomatically infected people move to hospitals, if increases, the prevalence can be reduced to some certain extent with time, i.e., the hospitalization rate is inversely proportional with R 0 . If more people get admitted to the hospitals for clinical treatment without ignoring the slightest symptoms, then the chance of an epidemic or pandemic outbreak reduces. Moreover, the vaccination rates (q 1 , q 2 ) are inversely proportional with R 0 which means basic reproduction number decreases with increase of q 1 and q 2 . It is biologically relevant because if more people are provided with vaccines at susceptible and asymptomatic stages, then the chances of becoming infected reduce which lessens the higher disease transmission. From the sensitivity index, it is observed that the transmission rate from symptomatically infected is most sensitive among all the parameters to reduce the disease prevalence and vaccination reduces the count of the symptomatically infected population. Hence, lowering the virus transmission through social distancing and vaccination along with other precautionary measures would help to handle this pandemic situation with time.

J o u r n a l P r e -p r o o f

The Jacobian matrix of system (1) is given as:

Theorem 6.1. Disease-free equilibrium (E 0 ) of the proposed system is locally asymptotically stable (LAS) for R 0 < 1 when P i > 0 for i = 1, 2, .., 5.

Proof. Jacobian matrix corresponding to DFE E 0 = Λ p 0 , 0, 0, 0, 0, q 1 S 0 d is given as follows:

The characteristic equation of the corresponding Jacobian matrix is λ 6 + P 1 λ 5 + P 2 λ 4 + P 3 λ 3 + P 4 λ 2 + P 5 λ + P 6 = 0, where,

Theorem 6.2. The endemic equilibrium point E * of system (1) is LAS for R 0 > 1 when the conditions (i) and (ii), as mentioned in the proof, are satisfied.

Proof. The Jacobian matrix at the endemic equilibrium point E * is given as: Let us consider:

,

,

According to Routh-Hurwitz criterion [31] , E * is locally asymptomatically stable (LAS) when Ω i > 0 for i = 1, 2, 3, 4, 5, 6, i.e.,

Global stability of E 0 : Now we show the global stability of the disease-free equilibrium point with the help of Lyapunov function.

Proof. Let us consider the Lyapunov function V 1 = p 2 I 1 + β 2 S 0 I 2 , where S 0 = λ p 0 , p 0 = (d+q 1 ), p 1 = q 2 +κ+α+d ans p 2 = φ 1 +φ 2 +d+µ 1 . Here V 1 is a positive definite function, J o u r n a l P r e -p r o o f Journal Pre-proof the time derivative of V 1 computed along the solutions of system (1) is as follows:

Hence, by LaSallea's invariance principle [26] , E 0 is globally asymptotically stable when S ≤ S 0 with R 0 < 1.

The result of the central manifold theory, discussed by Castillo-Chavéz and Song [6] , is stated in the following theorem:

Theorem 7.1. Consider the following system of ODEs with a parameter Φ:

Let O be taken as an equilibrium point of the mentioned system with f (O, Φ) = O for all Φ. Let us further assume

) be the linearization matrix of the mentioned system at the equilibrium O and Φ evaluated at 0. B has a simple zero eigenvalue and other eigenvalues of the matrix have negative real parts. (II) B contains a right eigenvector w which is non-negative and also a left eigenvector v corresponding to the zero eigenvalue. If f k is considered to be the k th component of f and

then the local dynamics of the system around O is determined by the sign of a and b.

1. a > 0, b > 0: (i) O is locally asymptotically stable and there exists a positive unstable equilibrium for Φ < 0 and |Φ| 1. (ii) Further O is unstable and there exists a negative and locally asymptotically stable equilibrium for 0 < Φ 1.

Further O is locally asymptotically stable, and there exists a positive unstable equilibrium for 0 < Φ 1.

The components of the right eigenvector w may not be non-negative and it depends on the positivity of corresponding component of equilibrium (Remark 1 in [6] ).

Even if some components of w become negative, then also the theorem can be applied, though on that case one needs to compare w with the equilibrium. The comparison is necessary as the general parameterization of the center manifold theory before changing the coordinate is

and V = x 6 , then the system (1) can be rewritten as:

We have considered Φ = β 2 as bifurcation parameter for R 0 = 1.

The characteristic equation of the corresponding Jacobian matrix is λ 6 + P 1 λ 5 + P 2 λ 4 + P 3 λ 3 + P 4 λ 2 + P 5 λ + P 6 = 0, where,

has a zero eigenvalue at R 0 = 1 as P 6 | R 0 =1 = 0. The right eigenvector corresponding to the zero eigenvalue of Table 1 : Parameter values used for numerical simulation of system (1) Figure 2 shows that a trajectory starting from the mentioned initial point converges to the DFE E 0 (1 × 10 8 , 0, 0, 0, 0, 1.2 × 10 9 ) for β 2 = 1 × 10 −11 along with other parametric values from Table 1 , and we get R 0 as 0.949 < 1 here. So, the basic reproduction number, when lies below unity, we get an infection-free system. Figure 3 depicts that for the parametric values of Table 1 , the system converges to the endemic equilibrium point E * (2.85×10 7 , 9.33×10 6 , 5.77×10 8 , 1.40×10 7 , 6.60×10 6 , 4.54×10 8 ). Here we get R 0 = 3.506 which exceeds unity. So, taking β 2 as the regulating parameter it is observed that the system undergoes a forward (transcritical) bifurcation at β 2 = β 2[T C] = 1.0623 × 10 −11 and E 0 becomes unstable for β 2 > β 2[T C] (see Figure (4.a) ). Also from Figure (4.b) , it is observed that E 0 is stable for R 0 < 1 and becomes unstable for R 0 > 1. Also, a stable branch of endemic equilibrium evolves from R 0 = 1. controlled to some certain extent. It is the reason φ 1 maintains an inversely proportional relation with R 0 . Moreover, if people from susceptible and asymptomatically infected classes start to take vaccines at the early stages with a higher rate, then the chances of becoming infected become lower. It ultimately reduces the fatality of the current situation.

The tornado plot of the sensitivity index of the parameters is shown in Figure 6 . For the parametric values of Table 1 , the calculated sensitivity index are as follows: Γ β 1 = 0.0377, Γ β 2 = 0.9622, Γ φ 1 = −0.0551, Γ q 1 = −0.9231 and Γ q 2 = −0.0079. So, it is observed that β 2 and q 1 are more sensitive than others. Vaccination to susceptible people help to control the disease transmission to a greater extent in fact. In Figure 7 , the impact of vaccination on disease propagation is illustrated. Both Figures  (7.a) the increase of vaccination rates (q 1 , q 2 ). It means the infection level and fatality starts to decrease if more people take the recommended vaccines as a precautionary measure before getting severely infected. In Figure 8 , the asymptomatically infected population is plotted with increasing virus transmission rates (β 1 , β 2 ) and it is observed that the count of infected individuals increases more for higher value of β 2 than β 1 . It means frequent contact with symptomatically infected people than the people who are in the pre-symptomatic state actually increases the infection in the system.

The model system is reintroduced in this section by implementing some control interventions which can reduce the disease burden. Maintaining social distances and proper hygiene is one of the important precautionary measures which is advised to be followed by each and every one. Besides it, there are different vaccines are available now and people are asked to take the proper dosage of available vaccines to avoid further infection. So, these strategies are incorporated into this system to reduce the rapid transmission. Worldwide high disease transmission ensures that there must be Covid cases without showing J o u r n a l P r e -p r o o f Journal Pre-proof any kind of symptoms and this asymptomatic transmission of COVID-19 has made the situation worse in terms of controlling the spread. So, maintaining a safe distancing in population as well as vaccination to all the people are considered to be the control policies. The analysis is performed to observe the impact of the control policies to reduce the incidence of transmission of disease and also to obtain the optimal cost burden. Let us first describe the control strategies one by one.

Increase the awareness of social distancing and maintaining hygiene: People can be aware of a disease and its prevalence when they are provided with the necessary information. It helps them to bring behavioral changes and instigate to take precautionary measures for not becoming infected. Day-to-day updates on different news portals and live tracking sites help to increase the cautiousness among the population. Now people become infected when they come in contact with any of the infected classes (asymptomatic and symptomatic). So, maintaining a proper physical distance is one of the main ways to curb the higher disease transmission. In this work, it is considered that the u 1 portion of the susceptible population maintains that social distancing and takes other precautionary measures (using the face masks, maintaining enough hygiene, etc.). So, only (1 − u 1 )S of susceptible individuals move to the pre-symptomatic or asymptomatic stage after contact with infected people. In system (7), u 1 denotes the intensity of maintaining physical distancing with 0 ≤ u 1 ≤ 1, where u 1 = 0 means not maintaining the distancing at all and u 1 = 1 means full maintenance of distancing. As the awareness depends on the infectivity and disease fatality, so, u 1 (t) is taken as one control intervention.

Increasing the vaccination rates of population: Vaccination is another strategy that reduces the rate of infection only if people go through the procedure as early as possible. Fortunately, there are many vaccines available now for Covid-19. The vaccination programs, firstly, may take some time for implementation, and also some time is required for individuals to develop immunity after inoculation. Governments of almost every country have conducted several awareness programs to make people to understand the importance of the recommended vaccines and requested people to be vaccinated to avoid the infection further. Taking vaccines at an early stage can decrease the disease burden. So, instead of constant values, time-dependent vaccination rate functions q 1 (t) and q 2 (t) are considered here with the restrictions 0 ≤ q 1 (t) ≤ 1 and 0 ≤ q 2 (t) ≤ 1. By implementing these control policies, the overall chances of becoming infected with coronavirus would be lessened. Here, 1 denotes when all people take vaccines as an important precautionary measure and 0 denotes the case when no person becomes vaccinated. The main work is to determine optimal control strategies with minimum implemented cost. So, the region for the control interventions u 1 (t), q 1 (t) and q 2 (t) is given as:

where T f is the final time up to which the control policies are executed, and also u 1 (t), q i (t) for i = 1, 2 are measurable and bounded functions.

Journal Pre-proof 9 .1 Deduction of Total Cost which needs to be minimized (i) Cost incurred in maintaining social distancing and proper hygiene: The total cost incurred maintaining social distancing and other precautionary measures is given by:

The integrand term w 2 u 2 1 (t) represents the cost of spreading awareness regarding social distancing and maintaining hygiene. This cost is comparatively higher because it considers the associated efforts for convincing people. There is some literature revealing the cost incurred for some mitigation strategies like self-protective measures etc. with second-order nonlinearity term [23, 5] . This work analyzes how the optimal control strategy representing social distancing reduces the overall count of the infective population in the system. (ii) Cost incurred in vaccination: Total cost associated with the vaccination of susceptible and asymptomatically infected individuals is:

Here w 1 I 2 (t) denotes the cost associated with symptomatically infected population for losing manpower [16, 23, 21] . The terms w 3 q 2 1 (t) and w 4 q 2 2 (t) denote the expenditure of vaccination procedure provided to susceptible and asymptomatically infected people respectively. These two terms also include the opportunity losses in terms of productivity loss due to the overall vaccination procedure. The control policies q 1 (t) and q 2 (t) are considered up to second-order non-linearity terms [16, 23, 21] . The following control problem is considered based on previous discussions along with the cost functional J to be minimized:

subject to the model system:

with initial conditions S(0) > 0, I 1 (0) ≥ 0, I 2 (0) ≥ 0, H(0) ≥ 0, R(0) ≥ 0 and V (0) > 0. We have already considered p 2 = (φ 1 + φ 2 + d + µ 1 ), p 3 = (ψ + d + µ 2 ) and p 4 = (d + η).

The functional J denotes the total incurred cost as stated and the integrand

J o u r n a l P r e -p r o o f Journal Pre-proof denotes the cost at time t. Positive parameters w 1 , w 2 , w 3 and w 4 are weight constants balancing the units of the integrand [16, 23] . The optimal control interventions u * 1 , q * 1 and q * 2 , exist in Ψ, mainly minimize the cost functional J.

Theorem 9.1. The optimal control interventions u * 1 , q * 1 and q * 2 in Ψ of the control system (6)-(7) exist such that J(u * 1 , q * 1 , q * 2 ) = min[J(u 1 , q 1 , q 2 )].

Proof. Proof is given in Appendix.

Theorem 9.2. If the optimal controls u * 1 , q * i for i = 1, 2 and corresponding optimal states (S * , I * 1 , I * 2 , H * , R * , V * ) exist for the control system, then we have adjoint variables λ = (λ 1 , λ 2 , ..., λ 6 ) ∈ R 6 satisfying the canonical equations:

with transversality conditions λ i (T f ) = 0 for i = 1, 2, ..., 6. The corresponding optimal controls u * 1 , q * 1 and q * 2 are given as:

Proof. Proof is given in Appendix.

In the proposed model, mainly two types of control strategies are implemented in order to reduce the disease burden and minimize the cost incurred for the implementation of the control intervention. Maintaining social distancing, proper hygiene, and vaccination at the early stage-these are the main precautionary measures one should maintain to avoid being infected with coronavirus. We consider u 1 and q i for i = 1, 2 as the control variables where u 1 fraction of people in the susceptible environment maintain proper Table 2 : Parametric values for numerical simulation of model (7) distancing and other precautionary measures, whereas q 1 , q 2 are the vaccination rates of people in susceptible and pre-symptomatic stages respectively. All the parametric values and positive weight constants, which are used to perform the numerical simulation here, are listed in Table 2 . The initial population size is considered as follows: S(0) = 1.38 × 10 6 , I 1 (0) = 3 × 10 5 , I 2 (0) = 180, H(0) = 2 × 10 2 , R(0) = 500 and V (0) = 10 2 to solve the control system in equations (6)- (7) . The numerical simulations are performed in MATLAB using forward-backward sweep method for control interventions [24] . It is assumed that the control interventions are implemented for two month, i.e., T f = 60 days. Figure 9 depicts the dynamics of model (7) when no time-dependent control policies are implemented, i.e., when u 1 = 0 and q i = 0.0006 for i = 1, 2. At T f = 60, the population becomes (5160981.3975, 9928.9060, 273778.6960, 406.8855, 3452.8266, 120750.1001) . The count of the asymptomatically infected population decreases at a slower rate whereas the symptomatically infected population increases throughout the time. It is observed that the number of overall infected (both asymptomatically and symptomatically) individuals remains significantly higher in this case.

Next, we consider the cases when people maintain physical distancing to avoid further infection and the vaccines are provided at constant rates. People in hospitals are already under strict restrictions and so, we are not considering any extra control policy for them. Figure 10 depicts the population profiles when u 1 = u * 1 and q 1 = q 2 = 0.0006. At T f = 60, the population becomes (5181165.1330, 3307.8807, 260453.4143, 394.8829, 3078.1103, 120908.0058). When only u 1 is implemented, the susceptible population increases as only a fraction of susceptible maintain physical distancing and rest move to asymptomatically infected class (I 1 ). The overall count of infected people decreases by implementing social distancing as a control strategy. The count of recovered people also decreases as a lesser number of the population becomes infected. The corresponding graph of optimal control intervention is depicted in Figure 11 . The control variable works with the highest intensity almost throughout the period.

Next, let us consider the case when vaccination provided to susceptible class depends on the severity of virus transmission, and hence, is considered to be time-dependent. Figure 12 shows the dynamics of model (7) Figure 12 : Diagrams of the population in presence of optimal control q * 1 and u 1 = 0 and q 2 = 0.0006. vaccinated people increases at a higher rate. Moreover, a decrease in the susceptible population leads to a declination of the infected population as well as the recovered population. Figure 13 depicts corresponding graph of optimal control intervention of q * 1 when u 1 = 0, q 2 = 0.0006. From this figure, it is observed that this control strategy works with the highest intensity immediately after implementation and remains at its highest throughout the whole time period. Now, we consider that situation when susceptible people are given the vaccines at a constant rate, but people who are in the asymptomatic or pre-symptomatic state are given vaccines depending on the severity (so, time-dependent). Figure 14 shows the dynamics of model system (7) for q 2 = q * 2 and u 1 = 0, q 1 = 0.0006. At T f = 60, the population becomes (5153294.3107, 94.9403, 18546.0253, 185.9842, 198.6641, 397705.2411) . Here the count of asymptomatically infected people significantly decreases than the case when the vaccination rate is constant. As these people, after full vaccination, move to the vaccinated compartment, so, the count of vaccinated people automatically increases. The Figure 14 : Diagrams of the population in presence of control policy q * 2 and u 1 = 0 and q 1 = 0.0006. graphs reveal that the overall infected population falls off at a significant level. Figure 15 depicts corresponding graph of optimal control policy q * 2 when u 1 = 0, q 1 = 0.0006. From this figure, it is observed that q 2 works with the highest intensity almost for the whole time before decreasing in the last week. Now, we consider the case when overall vaccination strategy changes with time depending on the infection level in absence of social distancing (u 1 = 0, q i = q * i for i = 1, 2). Figure 16 depict the dynamics of model system (7) for these control policies and at T f = 60, population becomes (65015.9027, 3.0362, 18446.8118, 185.8953, 196.0166, 5486177.5950) . The declination of the susceptible population and asymptomatically infected population in fact reduces the level of overall infection in the system as well as increases the count of the vaccinated population to a higher extent. As a larger number of people take vaccines at the pre-symptomatic state or even before getting infected, so, the count of infected population as well as recovered population decrease. Figure 17 depicts the optimal graphs of control policies q * 1 , q * 2 when u 1 = 0. The control strategy denoting q 1 works with the Figure 16 : Diagrams of the population in presence of optimal controls q * 1 and q * 2 when u 1 = 0. highest intensity after one or days of implementation and remains at its highest value throughout the time period. On the other hand, q 2 works with the highest intensity for almost two weeks and then decreases with time.

Implementation of all control strategies works better to control the disease burden than the case when a single control policy or only two control policies are applied. Now, we consider the case where a part of susceptible people maintains physical distancing and takes precautionary measures, and both susceptible and asymptomatically infected people are given vaccines according to the severity to curb the high transmission. Figure  18 depict the population trajectories in presence of all the control policies and at T f = 60, population becomes (65015.8899, 2.6375, 18438.7653, 185.8883, 195.8845, 5486186.2138 ). As only a part of the susceptible moves to the asymptomatic or pre-symptomatic phase, it reduces the count of the infected population. Moreover, the available vaccination procedure also decreases the overall infection in the system as people are advised to take vaccines as early as possible for self-protection. Figure 19 depicts the optimal graphs of Figure 18 : Profiles of populations with optimal control policies u * 1 , q * 1 and q * 2 .

all the control policies. The intensity of u 1 remains at the highest almost all the time and then decreases in the last week of the duration of control implementation. The control q 1 also works with the highest intensity throughout the time period but q 2 works for two weeks after implementation and then decreases with time. Figure 20 describes the cost design analysis (J) and count of symptomatically infected population (I 2 ) of the control system in absence and presence of time-dependent control interventions u * 1 , q * 1 and q * 2 . In Figure ( 20.a) optimal cost profiles are shown which reveals that in absence of control strategies the cost occurred due to productivity loss and it is quite higher as the number of the infected population is higher in this case.

In Figure 21 , the count of symptomatically infected population (I 2 ) and vaccinated population (V ) have been plotted. We have not considered the situation with social distancing and compare the graphs when vaccination rates are constant with time-depending vaccination rates. It is observed that in presence of these optimal control strategies, the count of the infected population decreases significantly even in absence of social distancing. Also, the number of people in the vaccinated compartment increases at a higher rate Figure 22 : Graph of optimal vaccination strategy given to asymptomatically infected people (q * 2 ) in presence and absence of other control interventions.

when the overall vaccination depends on the severity of disease transmission and changes with time. From the figures, it is observed that the vaccination at the early stages itself decreases the level of infectivity significantly which reduces the disease burden. So, the proposed control strategies are useful to reduce the number of the overall infected population in the system. From Figure 22 it is observed that q 2 (t) works with the highest intensity for the longest time when it is applied alone. When all the control interventions are applied, the highest intensity of q 2 remains for almost 15 days and then decreases with a steepness. It means maintaining social distancing (u 1 (t)) and vaccination to susceptible (q 1 (t)) actually decrease the higher need of vaccination of asymptomatically infected people. On the other hand, when a vaccine is provided to both susceptible and asymptomatically infected people, even in absence of social distancing, the intensity of q 2 decreases slowly with time after working with the highest intensity for a fortnight. 

Coronavirus was first reported in Wuhan, China, and then spreads worldwide within few months. Observing the severity, WHO declared the disease as a pandemic that affects not only the health of people but social and economical balance have also been disturbed. The Governments in each country called for a full or partial lockdown statewise or in the whole country observing the fatality. [41] . KWOK et al. in their work have suggested increased vaccination to gain immunity that leads to herd immunity in a country and may stop the spread of Covid-19 eventually [25] . But there are still some uncertainties regarding the effectiveness of the available vaccines and so, achieving herd immunities is not assured for a whole population. Hence, the importance of non-pharmaceutical interventions is stated there. Though we have not dealt with herd immunity in this work, a few remarks can be made here. The novel betacoronavirus is a mutant RNA virus. So, if the reinfection occurs, the severity of the disease will be less if a population gains herd immunity. It is true that achieving herd immunity for Covid-19 is a bit time taking because of vaccine hesitancy, uneven vaccine roll-out, etc. Also, this herd immunity may not prevent infection as such, especially when a mutated strain arises, but it evidently helps to reduce the higher possibility of spread of disease and thus the chance of infection. For example, the mass vaccination strategy for polio vaccine and rotavirus vaccine reduce the growth of these viruses and consequently the spread of diseases even if these are RNA viruses.

In this work, we have proposed a compartmental epidemic model to analyze the transmission of coronavirus. A separate compartment for vaccinated people is considered here, where people from susceptible and asymptomatic states move after a complete two-dose vaccine. The proposed system is biologically well-posed and an endemic equilibrium point exists when the basic reproduction number exceeds unity. It is observed that vaccination of susceptible population is more effective to reduce the overall infected people in the system. Also, the probability of becoming infected increases when a person frequently comes in contact with symptomatically infected people. The second part of the work contains a corresponding optimal control problem. It is considered that the chance of disease transmission is reduced when people adopt some behavioral changes in terms of maintaining social distancing and proper hygiene, and in system (7), the disease transmits Table 1 .

into (1 − u 1 )S amount of susceptible people only. As these changes depend on the severity and vary with time, so it is considered as a control intervention. Moreover, the vaccination procedure also changes with time and is considered to be another control strategy to reduce the disease burden in the system. When only one control policy is implemented in the system, it is observed that the policy works with the highest intensity for a larger time. On the other hand, when all the control strategies work simultaneously, the vaccination to susceptible works with the highest intensity throughout the period. The control which denotes social distancing also works with the highest intensity for quite a long time but decreases in the last few days. And, vaccination to asymptomatically infected people works with the highest intensity for almost two weeks and then decreases with time. It is also observed these control policies reduce the count of the overall infected population significantly, when implemented altogether. The number of people in vaccinated compartment increases at a higher rate here, resulting in reducing the chance of becoming infected. Thus, the vaccination have a significant impact on mitigating COVID-19 outbreaks and the non-pharmaceutical interventions are also equally essential to decease the transmission.

There are many factors present in the environment. But, we try to avoid some of the factors while formulating a mathematical model to reduce the complexity. For example, in this work, we have considered that the people who undergo a full vaccination process can not be infected in near future. But the effect of the vaccine may start to fade after a certain time, and a portion of vaccinated people move to the susceptible phase again.

Here, we try to analyze the dynamics of the system when the vaccinated people become susceptible at a rate ξ due to the waning effect of vaccination. Then the susceptible and vaccinated classes of the proposed system become

while the other compartments remain unchanged. For this model, we get the DFE as E 00 (S 00 , 0, 0, 0, 0, V 00 ), where S 00 = Λ(d+ξ) d(d+q 1 +ξ) and V 00 = Λq 1 d(d+q 1 +ξ) . Also, the basic reproduction number becomes R 00 = S 00 (β 1 p 2 +κβ 2 ) p 1 p 2 , and we get R 00 > R 0 as S 00 > S 0 . It means the basic reproduction number takes a higher value if we incorporate the waning effect of vaccines as we get a larger number of susceptibles in this case. It leads to a situation where the chances of getting infected increases significantly. Also, the scenario in Figure  J Table 1 .

23 reveals that the waning effect increases the count of overall infected people in the system. From figure (24.a), it is observed that the infection level rises sharply for a smaller waning rate, whereas the steepness reduces for an increasing value of ξ. In figure (24.b) , the recovered population also first increases with a higher rate, but with the increase of waning rate, the count tends to a saturation level. So, it can be concluded that the waning effect of vaccination though increases the infected population, but ultimately it leads to higher recovery. There remain some limitations that need to be stated while forming an epidemic model. In a population, it is assumed that each person is moving and has an equal chance of contact with each other. But the mixing of people in a large population is not homogeneous. Also, it is considered that the virus transmission rate maintains a constant value throughout the period of a disease outbreak (pandemic). But, immigration and emigration of people in a population increase the chance of infection and reinfection. Moreover, the proposed model does not account for age structures in the population. As for Covid-19, the older generation was severely affected when the second wave started. Moreover, most of the system parameters are taken as an average basis, i.e., immunity, susceptibility, recovery, etc. are taken to be the same for all people of the population. It may happen that the contact rate becomes higher for a portion of people only. Henceforth, the model system is suitable to describe the pandemic for a large period of time but it is not that fruitful for details in a very small period of time.

• An epidemic model on COVID-19 transmissions is introduced in this work.

• A portion of susceptible and asymptomatically infected class move to the vaccinated class after a full vaccination process.

• The numerical figure suggests that a better disease control can be achieved by vaccination of the people when they are at susceptible phase.

• Impact of social distancing as well as vaccination is observed as these are taken as the control policies to reduce the disease burden.

J o u r n a l P r e -p r o o f Journal Pre-proof

The optimal control functions in the optimal system is obtained with the help of Pontryagin's Maximum Principle [15, 34] . Consider the following Hamiltonian function:

H (S, I 1 , I 2 , H, R, V, u 1 , q 1 , q 2 , λ) = L(S, I 1 , I 2 , H, R, V, u 1 , q 1 , q 2 ) + λ 1 dS dt + λ 2 dI 1 dt + λ 3 dI 2 dt + λ 4 dH dt + λ 5 dR dt + λ 6 dV dt

So, H = w 1 I + w 2 u 2 1 + w 3 q 2 1 + w 4 q 2 2 + λ 1 [Λ − (1 − u 1 (t))(β 1 I 1 + β 2 I 2 )S − dS − q 1 (t)S + ηR] + λ 2 [(1 − u 1 (t))(β 1 I 1 + β 2 I 2 )S − q 2 (t)I 1 − (κ + α + d)

Here the adjoint variables are denoted by λ = (λ 1 , λ 2 , λ 3 , λ 4 , λ 5 , λ 6 ). In order to minimize the cost function, the Hamiltonian function needs to be minimized by Pontryagin's Maximum Principle.

Proof. For the control system (7), let us consider u * 1 , q * i (for i = 1, 2) are the applied optimal control interventions along with optimal state variables S * , I * 1 , I * 2 , H * , R * , V * . Then there exist adjoint variables λ i , for i = 1, 2, .., 6, satisfying the canonical equations:

So, we have dλ 1 dt = λ 1 [(1 − u 1 )(β 1 I 1 + β 2 I 2 ) + d + q 1 ] − λ 2 [(1 − u 1 )(β 1 I 1 + β 2 I 2 )] − λ 6 (q 1 )

with the transversality conditions λ i (T f ) = 0, for i = 1, 2, 3, 4, 5, 6. (β 1 I * 1 + β 2 I * 2 )S * 2w 2 (λ 2 − λ 1 ) , q * 1 = S * 2w 3 (λ 1 − λ 6 ) and q * 2 = I * 1 2w 4 (λ 2 − λ 6 ).

So, in Ψ, we have

which is equivalent as (9).

The optimal system involving optimal control variables u * 1 , q * 1 and q * 2 along with minimized Hamiltonian H * at (S * , I * 1 , I * 2 , H * , R * , V * , λ 1 , λ 2 , λ 3 , λ 4 , λ 5 , λ 6 ) is given as following:

J o u r n a l P r e -p r o o f Journal Pre-proof with non-negative initial conditions and corresponding adjoint system is:

with transversality conditions λ i (T f ) = 0, for i = 1, 2, ..., 6, and the control strategies u * 1 , q * i (for i = 1, 2) are same as in (9) .

COVID-19 underreporting and its impact on vaccination strategies

The impact of COVID-19 vaccination delay: A data-driven modeling analysis for Chicago

Lecture notes, forward and adjoint sensitivity analysis: with applications in Dynamical Systems, Linear Algebra and Optimisation Mathematical and Theoretical Biology Institute

Sex differential in COVID-19 mortality varies markedly by age

Optimal control of an epidemic through educational campaigns

Dynamical models of tuberculosis and their applications

Centers for disease control and prevention: 2019 novel coronavirus

Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study

2020) A novel coronavirus outbreak of global health concern

Theory of ordinary differential equations, Tata McGraw-Hill Education

COVID-19 Dashboard by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University (JHU). Available online

A Fractional Order COVID-19 Epidemic Transmission Model: Stability Analysis and Optimal Control

Stability analysis of a fractional ordered COVID-19 model

Modeling Vaccine Efficacy for COVID-19 Outbreak

Optimal control applied to vaccination and treatment strategies for various epidemiological models

Symptom progression of COVID-19

Theory of functional Differential Equations

Optimal control methods applied to disease models

Wuhan wet market closes amid pneumonia outbreak

The impact of self-protective measures in the optimal interventions for controlling infectious diseases of human population

Optimal control theory: an introduction

Will achieving herd immunity be a road to success to end the COVID-19 pandemic

The stability of dynamical systems

Substantial undocumented infection facilitates the rapid dissemination of novel coronavirus (SARS-CoV2)

Understanding Unreported Cases in the COVID-19 Epidemic Outbreak in Wuhan, China, and the Importance of Major Public Health Interventions

A COVID-19 epidemic model with latency period

Mathematical Biology I. An Introduction

Mathematical modeling of COVID-19 transmission dynamics with a case study of Wuhan

Mathematical theory of optimal processes

Epidemic model of COVID-19 outbreak by inducing behavioural response in population

Modelling the role of optimal social distancing on disease prevalence of COVID-19 epidemic

Modelling the epidemic trend of the 2019 novel coronavirus outbreak in China

An updated estimation of the risk of transmission of the novel coronavirus (2019-ncov)

Reproduction numbers and subthreshold endemic equilibria for compartmental models of disease transmission

Symptoms of Coronavirus

WHO Coronavirus (COVID-19

Worldometers Live Tracking

Clinical course and outcomes of critically ill patients with SARS-CoV-2 pneumonia in Wuhan, China: a single-centered, retrospective, observational study. The Lancet Respiratory Medicine

Acknowledgements: The authors are grateful to the learned editor, reviewers and Prof. Laura Gardini (Editor-in-Chief) for their careful reading, valuable comments and helpful suggestions, which have helped them to improve the presentation of this work significantly. The first author (Sangeeta Saha) is thankful to the University Grants Commission, India for providing SRF. The research of J.J. Nieto has been partially supported by the Agencia Estatal de Investigacion (AEI) of Spain, cofinanced by the European Fund for Regional Development (FEDER) corresponding to the 2014-2020 multiyear financial framework, project PID2020-113275GB-I00; and by Xunta de Galicia under grant ED431C 2019/02, and Instituto de Salud Carlos III, grant COV20/00617.

Now we derive the conditions under which optimal control policies exist with minimized cost function in a finite time span. To establish the existence, we shall take the help of results proved in [15, 16] . Proof of Theorem 9.1Proof. For existence of optimal controls, the following conditions have to be satisfied: (i) Set of solutions of the system (7) with control variables in Ψ = φ.(ii) Ψ is closed, convex and state system can be expressed as a linear function of control variables where the coefficients will depend on time and state variables. (iii) The integrand L of equation- (7) is convex on Ψ and L(S, I 1 ,In optimal system (7), let consider N = S + I 1 +where N (0) is the total population size at the initial stage.Now the solution of (7) is bounded and also the right hand side of the system are locally Lipschitzianin functions in presence of the control variables in Ψ. Therefore, by P icard − Lindelöf theorem condition, the solution of the optimal system along with implemented optimal control strategies exist in Ψ and is non-zero [10] . Also, the set Ψ, in which control variables are defined, is closed and convex. Again each of the equations of the stated system can be written as a linear equation in terms of u 1 , q 1 and q 2 along with coefficients depending on state variables and so, condition (ii) is also satisfied. The integrand function L(S, I 1 , I 2 , H, R, V, u 1 , q 1 , q 2 ) is a convex function on Ψ as the control variables are of order two. Now, L(S, I 1 , I 2 , H, R, V, u 1 , q 1 , q 2 ) = w 1 I 2 + w 2 u 2 1 + w 3 q 2 1 + w 4 q 2 2 ≥ w 2 u 2 1 + w 3 q 2 1 + w 4 q 2 2Let, w = min(w 2 , w 3 , w 4 ) > 0 and k(u 1 , q 1 , q 2 ) = w(u 2 1 + q 2 1 + q 2 2 ) which is a continuous function. Then L(S, I 1 , I 2 , H, R, V, u 1 , q 1 , q 2 ) ≥ k(u 1 , q 1 , q 2 ). Here k is continuous and ||(u 1 , q 1 , q 2 )|| −1 k(u 1 , q 1 , q 2 ) → ∞ whenever ||(u 1 , q 1 , q 2 )|| → ∞. So, condition (iii) is also fulfilled which implies the existence of control variables u * 1 , q *Author contribution: All the authors have accepted responsibility for the entire content of this submitted manuscript and approved submission.

The authors declare that they have no conflict of interest regarding this article.Data Availability: Some data used in this work have been obtained from the official sources as stated in the references [11, 19, 30, 42] . The data that support the findings of this study are available from the corresponding author upon reasonable request.Ethics Statement: This research did not require ethical approval.