Stability and Hopf Bifurcation Analysis of an Epidemic Model by Using the Method of Multiple Scales

A delayed epidemic model with nonlinear incidence rate which depends on the ratio of the numbers of susceptible and infectious individuals is considered. By analyzing the corresponding characteristic equations, the effects of time delay on the stability of the equilibria are studied. By choosing time delay as bifurcation parameter, the critical value of time delay at which a Hopf bifurcation occurs is obtained. In order to derive the normal form of the Hopf bifurcation, an extended method of multiple scales is developed and used. Then, the amplitude of bifurcating periodic solution and the conditions which determine the stability of the bifurcating periodic solution are obtained. The validity of analytical results is shown by their consistency with numerical simulations.


Introduction
Mathematical models describing the transmission of infectious diseases have played an important role in understanding the mechanism of disease transmission and controlling the spread of infectious diseases.In the literatures, many classical epidemic models have been proposed and studied, and many authors also attempt to develop more realistic mathematical models.In 1927, by using "compartment modeling," Kermack and Mckendrick [1] described an epidemic model and computed the theoretical number of infectious individuals.From then on, "compartment modeling" is used until now.Compartmental models are also called  models or  models, where , , and  denote the number of susceptible, infectious, and recovered individuals.Some  and  models have been analyzed in detail [2][3][4][5][6][7].However, many diseases incubate inside the hosts for a period of time before the hosts become infectious.Therefore, the  epidemic models, where  denotes the number of individuals who are infected but not yet infectious, are developed to investigate the role of the incubation period in disease transmission.Some authors have studied the  epidemic models [8][9][10][11][12].For some diseases, it is necessary that some infectious individuals are quarantined.Then, the  and  models, where  denotes the number of infectious individuals that have been quarantined and controlled, are studied [13][14][15][16].
Incidence rate plays an important role in the transmission of disease.In earlier models [17], based on the law of mass action, the bilinear incidence rate , where  is the average number of contacts per infectious individual per day is used.Capasso and Serio [18] introduced a saturated incidence rate () into epidemic models, where () = /(1 + ), where  measures the infection force of the disease and 1/(1 + ) describes the psychological effect or inhibition effect from the behavioral change of the susceptible individuals with the increase of the infectious individuals.
Based on the work of Capasso and Serio [18], in this paper, we investigate an  epidemic model with a saturated incidence rate (⋅), where (⋅) depends on the ratio of  and .Therefore, we have Then, the saturated incidence rate is /( + ).Considering the effects of time delay, we obtain the following  epidemic model: where (), (), (), and () denote the number of sus- In this model, we assume that the susceptible individuals were infected before time delay  which is the latent period.Then, the infected individuals become infectious individuals and some of infectious individuals are quarantined.In classes  and , some individuals are cured and removed.The outline of this paper is as follows.In Section 2, the equilibria and their stability are analyzed.The critical value of time delay at which Hopf bifurcation occurs is obtained.In Section 3, by using an extended method of multiple scales, we obtained the normal form of the Hopf bifurcation.In Section 4, The above theoretical results are validated by numerical simulations with the help of dynamical software WinPP.In Section 5, conclusions are given.
For  = 0, the characteristic equation becomes Remark 1.For  = 0, if  +  > 0 and  +  > 0, then the roots of ( 8) are real and negative, and then the equilibrium  is asymptotically stable.
For  ̸ = 0, if  =  ( > 0) is a root of ( 7), then we have Separating the real and imaginary parts, we have which leads to the following fourth-degree polynomial equation: Then, by discussing the distribution of the solutions of (11), we analyze the stability of the equilibrium and the existence of Hopf bifurcation.The distribution of the solutions of ( 11) can be divided into the following three cases.
then (7) has no positive root.
then (7) has only one positive root  + .
then (7) has two positive roots: and no such solutions otherwise.
Noticing that  + in Case 2 is the same as  + in Case 3, we suppose that (7) has two positive roots  ± .Then, from (10), we can determine that at which (7) has a pair of purely imaginary roots ± ± .Supposing that then Substituting () into ( 7) and taking the derivative with respect to , we have which, together with ( 10) and ( 16), leads to Therefore, we can obtain the following theorem.

The Normal Form of Hopf Bifurcation
In this section, we suppose that system (3) undergoes a Hopf bifurcation from equilibrium  * ( * ,  * ).Time delay  is chosen as the bifurcation parameter and its critical value is   .By using the method of multiple scales [19], we will derive the normal form of the Hopf bifurcation.By translating the equilibrium to the origin, system (3) is rewritten as the following equation: where x = ((), ())  ∈  2 is the state variable vector and x  = (( − ), ( − ))  .According to the MMS, a monoparametric family of solution of the type is as follows: where   =    ( = 0,2) and  is a nondimensional parameter.The solution does not depend on slow scale  1 =  because secular terms first appear at ( 3 ).Therefore, we assume a two-scale expansion of the solution of (25).
First, for -order terms, we have where Equation ( 29) has the following general solution: where  is complex constant and u is the right eigenvector given by Next, for  2 -order terms, we have where F 0 xx , F 0 xx  , F 0 x  x  are the second-order partial derivatives of F(x, x  ) with respect to x and x  .
Substituting (31) into (34), we obtain Solving (35), it yields where vectors z 11 and z 11 ∈  2 are obtained by solving the following equations: that is, where Finally, for  3 -order terms, we get where x  are the third-order partial derivatives of F(x, x  ) with respect to x and x  .
Substituting (31) and (36) into (40) and eliminating the secular terms, we can obtain the equations including  2 .Eliminating the coefficients of  2  by using the left eigenvectors and reabsorbing parameter  [20], the normal form is determined by where the expressions of coefficients     and  111 are given by where  = k H (  F 0 x   −  + E)u and k is the left eigenvector obtained by solving the following equations: where H denotes the transpose conjugate and T denotes the transpose.
To express the normal form in real form, a polar form representation for the complex amplitude is introduced: Substituting (44) into (41) and separating the real and imaginary parts in (41), the generalized amplitude and phase modulation equations are drawn: where  1 = Re(    ),  1 = Im(    ),  111 = Re( 111 ), and  111 = Im( 111 ).From (45), we can get that the amplitude of the periodic solution is The stability of the periodic solution is determined by the sign of the eigenvalue of the Jacobian matrix of (45 1 ).The eigenvalue is Then, we can obtain the following theorem.Theorem 3. The amplitude of the bifurcating periodic solution of system (3) is 2√− 1 / 111 ; the stability of the bifurcating periodic solution is determined by : the bifurcating periodic solution is stable (unstable), if  < 0 ( > 0).
By using the dynamical software WinPP, some numerical simulations are given.First, by Theorem 2, endemic equilibria  * are asymptotically stable when  ∈ [0,   ) as shown in Figure 1 and the time histories of the state variables  and  are also given in Figure 1.
Endemic equilibria  * are unstable when  >   .When  crosses critical value   , a stable periodic solution yields as shown in Figure 2.
By using the method developed in this paper, the amplitude of the periodic solution can be obtained.The relationship between the amplitude and bifurcation parameter  is given in (45).Then, the analysis results are compared with that of numerical simulations as shown in Figure 3.

Conclusions
In this paper, the stability and Hopf bifurcation of an epidemic model with time delay are studied.The stability of disease-free equilibrium and endemic equilibrium is studied and the conditions of the stability and unstability of the two equilibria are obtained.Then, the critical value of  at which a Hopf bifurcation occurs is obtained.By using an extended method of multiple scales, the normal form of Hopf bifurcation is obtained.In order to verify the theoretical results, the Runge-Kutta scheme is adopted to produce the numerical results.The comparison between analytical predictions and   numerical results of the amplitude of Hopf bifurcation reveals a qualitatively and quantificationally excellent agreement.
In order to obtain the normal form of Hopf bifurcation occurring in delayed differential equations, we develop an extended method of multiple scales.Choosing time delay as a bifurcation parameter and using the method, one can get the normal form of Hopf bifurcation without the tedious computation encountered in the center manifold reduction.Moreover, when we develop the method, we choose dimensional delayed differential equation.Therefore, the method can be used for studying the Hopf bifurcation occurring in a general -dimensional delayed differential equation.
Through observing the results, there are two contributions in this paper.First, we improve an epidemic model and introduce a nonlinear incidence rate which depends on the ratio of the numbers of susceptible and infectious individuals.Second, we extend the method of multiple scales.By using this method, the normal form of Hopf bifurcation occurring in -dimensional delayed differential equations can be obtained.

Figure 3 :
Figure 3: Comparison between the theoretical solution (solid) and numerical simulation (dot) for the amplitude of periodic solution.
ceptible, infectious, quarantined, and recovered individuals at time .Λ is the recruitment rate of the population. is the natural death rate of the population. 1 and  2 are the extra disease-related death rate in classes  and , respectively. is the rate constant for individuals leaving infectious compartment  for quarantine compartment . 1 and  2 are the removal rate constants from classes  and , respectively.