Professional Documents
Culture Documents
Applied Mathematics and Computation: Xiaoyun Wang, Lijuan Wei, Juan Zhang
Applied Mathematics and Computation: Xiaoyun Wang, Lijuan Wei, Juan Zhang
a r t i c l e i n f o a b s t r a c t
Keywords: In this paper we present a kind of technique to obtain the asymptotical solution of an SEIR
SEIR model (where S is the susceptible population, E is the exposed population, I is the infectious pop-
Dynamical analysis ulation and R is the recovered population) epidemic model by employing the method of
Perturbation method perturbation. At first, we investigate the epidemic model by analysing its dynamical behav-
ior. Then, we use the method of perturbation to obtain the analytical solution of the model.
We assign values for parameters and draw figures to observe the magnitude of error of the
perturbation method in a macro view. Finally, we analyse macroscopically the two com-
parison chart on analytical solution and the exact solution, and know that it is feasible
to analyse the solution of the epidemic model by using the perturbation method.
Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction
It is well known that the epidemic models of studying transmission dynamics of infectious diseases in population have
been extensively developed. Hawer firstly proposed that the rate of spread of infection was proportional to the product of the
density of susceptible individuals times the density of infectious individuals and studied a discrete-time model for measles.
Sir Ronald Ross [1–3] translated it into a continuous time framework to study the dynamics of malaria of measles between
mosquitoes and humans, and won Nobel prize in 1902. In 1927, Kermark and Mckendrick extended the ideas of Hawer and
Ross and established the threshold theory [4]. Since then, the concept of compartment modeling has been used up to now,
and after several decades, many scholars established various different differential equation models of disease and ecology to
research corresponding disease and ecology analysis.
Now, establishing reasonable differential equations has become an important content in analyzing the epidemiological
characteristics of infectious diseases, which can also provide some useful control measures. For the established differential
equations are often qualitative analysis of the equations, such as the stability of equilibria, bifurcation theory or the steady
state and so on. And there have been a series of mature methods to apply to study them. The majority of applications refer to
epidemic models such as SIR, SEIR, SEIS, SEIRS models and so on, where the population is divided into several subclass
according to the epidemic characteristic: S is the susceptible class, E is the exposed class, I is the infectious class and R is
the recovered class. So far, there have been a lot of researches [5–12] in this area.
q
Project supported by the National Sciences Foundation of Shanxi (2011011002-1), the Postdoctoral Sciences Foundation of China (20100471857) and
the Special Foundation by the China Postdoctoral Science Foundation (201104653).
⇑ Corresponding author at: College of Mathematics, Taiyuan University of Technology, Taiyuan, Shanxi 030024, China.
E-mail address: xywang0708@126.com (X. Wang).
http://dx.doi.org/10.1016/j.amc.2014.01.090
0096-3003/Ó 2014 Elsevier Inc. All rights reserved.
480 X. Wang et al. / Applied Mathematics and Computation 232 (2014) 479–486
However, in many cases, we need to understand the model in a certain period of time the state, which was unable to
obtain from the equations of the qualitative analysis. So, we need to study the solution of the model. But, considering that
the majority of the exact solution is not to be obtained easily, here we are more interested in getting their approximate
analytic solutions, which can be solved by the perturbation methods [13].
The perturbation methods [14,15] all rely on there being a parameter in the problem that is relatively small. When using
perturbation methods, one is capable of solving problems that nonlinear problem, which can translate a nonlinear problem
into many linear subproblems and then use the solutions of the first several linear subproblems to approximate the solution
of the nonlinear problem using the perturbation parameter [16]. In recent years, many scholars have done a lot of work,
especially in some nonlinear problems [17–25] and other interdisciplinary issues [26–30]. In 1880s or 1890s, Lindstedt
and Poincare et al. had used the perturbation method to study some problems in celestial mechanics and found the ninth
planet in solar system. After 1950s, the perturbation methods have been a common way to deal with the nonlinear problem.
Now, with the development of technological, it is playing an ever more important role in the astronomy, physics, biology,
chemistry, control theory and so on [31]. And with the application of perturbation methods, it contains more methods, such
as the methods of multiple scales, asymptotic matching, the method of Lindstedt–Poincare, variation of parameter, correc-
tional d approach, homotopy perturbation technique, iterative perturbation method, homotopy analysis and so on. In a
word, the perturbation methods have attracted more and more attention of domestic and abroad academic circles.
However, for epidemic models, there have been a few papers to apply perturbation method to solve and discuss the cor-
responding asymptotic solution directly. In this paper, we propose an SEIR epidemic model, which is obtained by modifying
the model [10]. Li [10] proposed a dynamical model whose birth rate (in proportion to the total population) was the same as
the death rate. In this paper, we change the birth rate into the constant import to obtain SEIR epidemic model. Moreover, we
use qualitative and perturbation methods to discuss behavior of solution, respectively and discuss the veracity of perturba-
tion method in a macro view.
2. Qualitative method
In this section, we propose the dynamical mathematical model. Because the solution of the model cannot be obtained
directly, we can depend on analysing its dynamical behavior: the equilibria, the basic reproduction number and the stability
of the equilibria to see the trend of the solution, especially the trend of the epidemic.
In the paper we consider an SEIR model. Letting SðtÞ be the number of susceptible individuals, EðtÞ be the number of
exposed individuals, IðtÞ be the number of infective individuals and RðtÞ be the number of removed individuals at time t.
Using Fig. 1 we formulate the following SEIR model:
This model is a four differential equations system as following:
8 dS
>
> dt
¼ A eSI bS þ aR;
>
>
< dE ¼ eSI ðb þ dÞE;
dt
ð1Þ
>
> dI
¼ dE ðb þ l þ cÞI;
>
> dt
: dR
dt
¼ cI ðb þ aÞR;
where 0 < e 1. Let N ¼ S þ E þ I þ R. From the model (1), we can know that
dN
¼ A bN lI:
dt
Then,
A
lim sup NðtÞ ¼ :
t!þ1 b
So, we can obtain the positive invariant set of the model (1):
A
C ¼ ðS; E; I; RÞj0 < S; E; I; R; S þ E þ I þ R 6 :
b
Equilibria are obtained by setting the right side of each of the four differential equations equal to zero. If I ¼ 0, it is easy to
deduce the disease-free equilibrium P0 ¼ fS0 ; 0; 0; 0g, where S0 ¼ Ab. The basic reproduction number, defined as the expected
number of secondary cases produced, in a completely susceptible population, by a typical infected individual during its
entire period of infectiousness, can be also obtained easily as following:
eS0 d eAd
Rr ¼ ¼ :
ðb þ dÞðb þ l þ cÞ bðb þ dÞðb þ l þ cÞ
lþcÞ 0 1ÞðaþbÞbðbþdÞðbþlþcÞ
If Rr > 1, we can derive the unique endemic equilibrium: P ¼ ðS ; E ; I ; R Þ, where S ¼ ðbþdÞðbþ
de
, I ¼ ðReðbþdÞðbþ lþcÞðaþbÞacde ,
cI
E ¼ ðbþldþcÞI and R ¼ aþb .
Assume: (H) the parameters are positive, in the model, which are shown in Table 1.
According to (H), Rr < 1 in model (1). Therefore, we get the following theorem.
Proof.
(a) The Jacobian matrix in the disease-free equilibrium P 0 can obtained as following:
0 1
b 0 eS0 0
B 0 ðb þ dÞ eS0 0 C
B C
JjP0 ¼ B C:
@ 0 d ðb þ l þ cÞ 0 A
0 0 c ðb þ aÞ
Consequently, we can obtain characteristic equations
Table 1
Description of parameters in model (1).
However, the stability of the endemic equilibrium is hard to prove. Because of the high dimensionality of model (1), it is
highly nontrivial to prove the global stability of the unique endemic equilibrium and establish rigorously the threshold
theory. The previous work only deals with models that can be reduced to a 2 or 3 dimensional system.
If Rr < 1, we can also use the perturbed solution to solve the model. Next we will introduce the perturbed solution.
3. Perturbed solution
In this section, we are interested in using what are known as perturbation methods to find approximate solution of the
model (1).
Substituting (4) into (1), expanding the nonlinear terms in e and equating the coefficients of the same powers of e, we
have
8
dS0
>
> dt ¼ A bS0 þ aR0 ;
>
>
>
< dE0 ¼ ðb þ dÞE ;
dt 0
ð5Þ
>
> dI0
¼ dE ðb þ l þ cÞI0 ;
>
> dt 0
>
: dR0
dt
¼ cI0 ða þ bÞR0 :
8
dS1
>
> ¼ A bS1 þ aR1 S0 I0 ;
>
> dt
>
< dE1 ¼ ðb þ dÞE þ S I ;
dt 1 0 0
ð8Þ
> dI1 ¼ dE1 ðb þ l þ cÞI1 ;
>
>
> dt
>
: dR1
dt
¼ cI1 ða þ bÞR1 :
D1 ¼ lþcc13bd
d
; D2 ¼ lþccabd
14 d
; D3 ¼ lþcCb2d
5d C6 d
; D4 ¼ bþd ; D5 ¼ Cb7 d ; D6 ¼ aþb
C8 d
;
C9 d C1 d
D7 ¼ bþd ; D8 ¼ lCþ10cþb
d
; D9 ¼ C 11 d; D10 ¼ lþ2cd ; D11 ¼ lAþ3 Bc3dd ; D12 ¼ ðlAþ3cBdÞ
3d
2 ;
D13 ¼ C 13 .
We use perturbation methods to obtain the asymptotic solution (10) of system (1). Next, we show that the solution is
uniformly valid about e in t 2 ½0; T 0 . Firstly, we construct auxiliary functions:
8
> X
m
>
> SðtÞ ¼ Si ðtÞei þ r 1 emþ1 ;
>
<
i¼0
ð11Þ
>
> X
m
>
> Si ðtÞei r 1 emþ1 ;
: SðtÞ ¼
i¼0
8
> X
m
>
> EðtÞ ¼ Ei ðtÞei þ r 2 emþ1 ;
>
<
i¼0
ð12Þ
>
> X
m
>
> Ei ðtÞei r 2 emþ1 ;
: EðtÞ ¼
i¼0
8
> X
m
>
> IðtÞ ¼ Ii ðtÞei þ r 3 emþ1 ;
>
<
i¼0
ð13Þ
>
> X
m
>
> Ii ðtÞei r 3 emþ1 ;
: IðtÞ ¼
i¼0
484 X. Wang et al. / Applied Mathematics and Computation 232 (2014) 479–486
8
> X
m
>
> RðtÞ ¼ Ri ðtÞei þ r 4 emþ1 ;
>
<
i¼0
ð14Þ
>
> X
m
>
> Ri ðtÞei r 4 emþ1 ;
: RðtÞ ¼
i¼0
where ðSi ; Ei ; Ii ; Ri Þði ¼ 1; 2; . . .Þ is the solution by (7); m is an arbitrary the nonnegative integer; ri ði ¼ 1; 2; 3; 4Þ are appropri-
ate large constants. It is not difficult to prove by using the theory of differential inequality, selecting sufficiently constant
ri ði ¼ 1; 2; 3; 4Þ in e; ðS; E; I; RÞ and ðS; E; I; RÞ are the supersolution and subsolution of (4) in t 2 ½0; T 0 , respectively. So, the
system exists a set of solution ðSðtÞ; EðtÞ; IðtÞ; RðtÞÞ and satisfies:
8
>
> SðtÞ 6 SðtÞ 6 SðtÞ;
>
>
>
< EðtÞ 6 EðtÞ 6 EðtÞ;
ð15Þ
>
> IðtÞ 6 IðtÞ 6 IðtÞ;
>
>
>
:
RðtÞ 6 RðtÞ 6 RðtÞ:
Thus, the solution (10) by using perturbation methods that is the solution of (1) in t 2 ½0; T 0 , which is a set of valid asymp-
totic expansion on e.
4. Application
To intuitively observe the veracity of perturbation method, we can assign values for parameters and draw the figures
of differential model and the approximate solutions to observe the magnitude of error of the perturbation method in a
macro view. The values of parameters are shown in Table 1. The results are shown in Figs. 1 and 2. From that, we can
see that though the approximate solutions have certain error, the trend of the results are coincident, which is satisfac-
tory (see Fig. 3).
7
x 10
2.5
1.5
E
0.5
0
0 2 4 6 8 10
t (month)
Fig. 2. E with time. The full line is the solution of differential equations by the instruct ‘‘ode45’’ in matlab. The broken line is the solution obtained by the
perturbation method.
X. Wang et al. / Applied Mathematics and Computation 232 (2014) 479–486 485
6
x 10
8
I
3
0
0 2 4 6 8 10
t (month)
Fig. 3. I with time. The full line is the solution of differential equations by the instruct ‘‘ode45’’ in matlab. The broken line is the solution obtained by the
perturbation method.
5. Discussion
For infectious disease models, the conventional dynamic analysis only deal with the limit situation or steady state. In
actual applications, some complex infectious disease models cannot be analysed by qualitative theory or we need to know
solution dynamic of the models in some certain time. Under the circumstances we can use perturbation method to deal with
them by obtaining the expression of the solution, then conduct quantitative and qualitative research, make a prediction for
the trend of the development of some diseases and take the corresponding measures. So, in this paper, we introduce an SEIR
model and study it from two aspects. On the one hand, we analyze dynamical behavior of the model by the conventional
qualitative and stability theory. On the other hand, we use perturbation method to obtain the analytical solution of the mod-
el and give the contrast figure about the analytical solution and the exact solutions. Through observing the results, we can
draw the following conclusions:
1. By macroscopically analyzing the two comparison charts, it can be seen that analytical solution and the exact solution are
consistent in the trend and wave by and large. So, we can know that it is feasible to analyse the solution of the model
infectious diseases using the perturbation method and the analytic solutions are effective.
2. From the above two graphics, we can see that the error of the solution is a little bigger, which can be interpreted in the
following reasons: the model dimension is higher and the initial value can be taken different and the order of solution
analytical formula we take is only the first order. So, there must exist errors between exact solution and analytical solu-
tion. In a word, we can know that the perturbation method is feasible, but it will need further improvement and research.
So, our research direction in future is to improve the perturbation method and to deal with the higher order number of
asymptotic expansion to get more accurate analytical solution.
References
[1] R. Ross, The Prevention of Malaria in Mauritius, Waterlow and Sons, London, 1908. pp. 29–40.
[2] R. Ross, H.P. Hudson, An application of the theory of to the study of a priori pathometry. Part II, Proc. R. Soc. London A 92 (1916) 204–230.
[3] R. Ross, H.P. Hudson, An application of the theory of probabilities to the study of a priori pathometry. Part II, Proc. R. Soc. London A 93 (1917) 212–225.
[4] M. Kermack, A. Mckendrick, Contributions to the mathematical theory of epidemics, Proc. R. Soc. A 115 (1927) 700–721.
[5] J. Arino, C.C. McCluskey, P. van den Driessche, Global results for an epidemic model with vaccination that exhibits backward bifurcation, SIAM J. Appl.
Math. 64 (1) (2003) 260–276.
[6] B. Buonomo, A. d’Onofrio, D. Lacitignola, Global stability of an SIR epidemic model with information dependent vaccination, Math. Biosci. 216 (2008) 9–
16.
[7] B. Buonomo, D. Lacitignola, On the dynamics of an SEIR epidemic model with a convex incidence rate, Ric. Mat. 57 (2) (2008) 261–281.
[8] M.Y. Li, J.S. Muldowney, Global stability for the SEIR model in epidemiology, Math. Biosci. 125 (1995) 155–164.
[9] M.Y. Li, H.L. Smith, L. Wang, Global dynamics of an SEIR epidemic model with vertical transmission, SIAM J. Math. Anal. 62 (1) (2001) 58–69.
[10] M.Y. Li, J.R. Graef, L.C. Wang, J. Karsai, Global dynamics of a SEIR model with varying total population size, Math. Biosci. 160 (1999) 191–213.
[11] X.Z. Meng, L.S. Chen, The dynamics of a new SIR epidemic model concerning pulse vaccination strategy, Appl. Math. Comput. 197 (2008) 582–597.
[12] X. Wang, Y.D. Tao, X.Y. Song, Studies of a class of SEIV models, Math. Appl. 23 (2010) 774–780.
[13] A.H. Nayfeh, Perturbation Methods, John Wiley&Sons, New York, 1973.
[14] A.H. Nayfeh, Problems in Perturbation, John Wiley&Sons, New York, 1985.
[15] Y.Z. Wang, Perturbation Method Basis, Science Publishing Company, Beijing, 1994 (in Chinese).
[16] J.Q. Mo, J. Zhu, H. Wang, Asymptotic behavior of the shock solution for a class of nonlinear equation, Prog. Nat. Sci. 13 (2003) 768–777.
[17] W.L. Zhao, Singular perturbations for nonlinear Robin problems, J. Comput. Appl. Math. 81 (1997) 59–74.
[18] Z.J. Du, W.G. Ge, M.R. Zhou, Singular perturbations for third-order nonlinear multi-point boundary value problems, J. Differ. Equ. 218 (2005) 69–90.
486 X. Wang et al. / Applied Mathematics and Computation 232 (2014) 479–486
[19] T.C. Lin, H.S. David, W.Q. Zhang, Numerical solutions of linear and nonlinear singular perturbations problems, Comput. Math. Appl. 55 (2008) 2574–
2592.
[20] R.K. Mohanty, S. Singh, A new fourth order discretization for singular perturbed two dimensional non-linear elliptic boundary value problems, Appl.
Math. Comput. 175 (2006) 1400–1414.
[21] W.M. Wang, An algorithm for solving nonlinear singular perturbation problems with mechanization, Appl. Math. Comput. 169 (2005) 995–1009.
[22] Y.L. Jiang, R.M.M. Chen, O. Wing, Convergence analysis of waveform relaxation for nonlinear differential-algebraic equations of index one, IEEE Trans.
Circuits Syst. I – Fundam. Theory Appl. 47 (2000) 1639–1645.
[23] Y.L. Jiang, A general approach to waveform relaxation solutions of nonlinear differential-algebraic equations: the continuous-time and discrete-time
cases, IEEE Trans. Circuits Syst. I–Reg. Pap. 51 (2004) 1770–1780.
[24] H. Hassani, Z.Y. Xu, A. Zhigljavsky, Singular spectrum analysis based on the perturbation theory, Nonlinear Anal.: Real World Appl. 12 (2011) 2752–
2766.
[25] L.L. Du, X.H. Wu, S.Q. Chen, A novel mathematical modeling of multiple scales for a class of two dimensional singular perturbed problems, Appl. Math.
Model. 35 (2011) 4589–4602.
[26] J.Q. Mo, H. Wang, F. Xie, Perturbation solution of epidemic contagion dynamic ecological model, J. Lanzhou Univ. (Nat. Sci.) 43 (2007) 94–97 (in
Chinese).
[27] X.Y. Wang, Y.L. Jiang, A general method for solving singular perturbed impulsive differential equations with two-point boundary conditions, Appl.
Math. Comput. 171 (2005) 775–806.
[28] X.Y. Wang, Y.L. Jiang, Boundary value problems of singular perturbed partial differential equations with impulsive conditions, Int. J. Nonlinear Sci.
Numer. Simul. 7 (2006) 443–446.
[29] J.G. Jia, M.S. Wang, M.L. Li, Periodic solutions for impulsive delay differential equations in the control model of plankton allelopathy, Chaos Solitons
Fract. 32 (2007) 962–968.
[30] S. Koskie, C. Coumarbatch, Z. Gajic, Exact slow-fast decomposition of the singularly perturbed matrix differential Riccati equation, Appl. Math. Comput.
216 (2010) 1401–1411.
[31] W.C. Qian, Singular Perturbation Theory and its Application in Mechanics, Science Publishing Company, Beijing, 1981 (in Chinese).
[32] N.T.J. Bailey, The Mathematical Theory of Infectious Disease and its Applications, Griffin, London, 1975.
[33] J.L. Salle, S. Lefschetz, Stability by Liapunov’s Direct Method, Academic Press, New York, 1961.