You are on page 1of 6

1st Reading

November 17, 2010 19:41 WSPC S1793-5245 242-IJB 00105

1 International Journal of Biomathematics


2 Vol. 3, No. 4 (December 2010) 1–6
3 c World Scientific Publishing Company
4 DOI: 10.1142/S1793524510001057

5 APPROXIMATE TRAVELING WAVE SOLUTION OF AVIAN


6 FLU TELEGRAPH REACTION DIFFUSION EQUATION

7 SHAIMAA A. A. AHMED
8 Department of Mathematics, Faculty of Sciences
9 Mansoura University, P. O. Box 35516, Egypt
10 shaimaa.math@yahoo.com

11 Received 4 May 2009


12 Revised 22 November 2009

13 It is known that, the telegraph equation is more suitable than ordinary diffusion equation
14 in modeling reaction diffusion in several branches of sciences. In this paper we gener-
15 alize the governing equation of distributed-infective model which represents the spread
16 of avian flu to the telegraph reaction diffusion equation and presents its approximate
17 traveling wave solution by using linear piecewise approximation.

18 Keywords: Diffusion; telegraph reaction diffusion equation; avian flu; traveling waves;
19 linear piecewise approximation.

20 1. Introduction
21 The problem we are concerned with is one of avian flu which diffuses globally
22 over the world. It spreads by the wild birds which may carry the bird flu viruses
23 while they are migrating without showing symptoms and spread the disease to other
24 countries. Avian viruses generally do not affect humans, but in 1997, an outbreak of
25 bird flu in Hong Kong infected 18 people, six of whom died. This virus was different
26 because it moved directly from chickens to people, rather than having been altered
27 by infecting pigs as an intermediate host. Since then, human cases of bird flu have
28 been reported in Asia, Europe and the Middle East. This paper is divided into
29 four sections. After the introduction, the second section is devoted to explaining
30 some basics of avian flu. Traditionally, contact and dispersal have been modeled
31 with local operators, resulting in reaction–diffusion equations [13]. We present the
32 standard diffusion equation, the modified diffusion equation or (telegraph equation)
33 and the corresponding telegraph reaction diffusion (TRD) equation in Sec. 3. Then
34 we give an application of telegraph reaction diffusion equation to avian flu in Sec. 4.

35 2. Basics of Avian Flu


36 Influenza A (H5N1) virus — also called “H5N1 virus” — is influenza A virus sub-
37 type that occurs mainly in birds, is highly contagious among birds, and can be

1
1st Reading
November 17, 2010 19:41 WSPC S1793-5245 242-IJB 00105

2 S. A. A. Ahmed

1 deadly to them. H5N1 virus does not usually infect people, but infections with these
2 viruses have occurred in humans. Most of these cases have resulted from people hav-
3 ing direct or close contact with H5N1 — infected poultry or H5N1-contaminated
4 surfaces [7].
5 Although more than 300 people have contracted the H5N1 virus, experts point
6 out that cross-infection to humans is still relatively rare, and usually occurs where
7 people have been in close contact with infected birds [8].
8 But as H5N1 continues its journey through Europe and Africa from its original
9 hotspot of south-east Asia, there is still concern that it may combine with a human
10 strain to produce a mutation that is more dangerous and difficult to combat [9].

11 3. Telegraph Reaction Diffusion Equation


12 The classical model for spatial spread is the diffusion equation (the second
13 Fickian law)

14 ut = D̄∆u, (3.1)

15 for a density u = u(t, x), x ∈ Rn , and D̄is the diffusion constant.


16 Consider a particular move on the real line with a correlated random walk with
17 parameters γ and µ/2. Then we have the Goldstien–Kac system [11]
1
u+ +
t + γux = µ(u− + u+ ),
2
18 (3.2)
1
u− − + −
t − γux = µ(u + u ),
2
19 by introducing in this equation the total density u = (u+ + u− ) and the probability
20 flow ν = γ(u+ − u− ), then we have

ut + νx = 0,
21 (3.3)
νt + γ 2 ux = −µν.

22 Using the Kac trick

23 utt + µut = γ 2 uxx . (3.4)

24 Also we can find that ν satisfies this telegraph equation or damped wave equation.
25 The conservation law in several dimensions where a particular density u : Rn → R
26 and a particular flux ν : Rn → R is

27 ut + div ν = 0. (3.5)

28 The first Fickian law

29 ν = −D̄ grad u, (3.6)


1st Reading
November 17, 2010 19:41 WSPC S1793-5245 242-IJB 00105

Approximate Traveling Wave Solution of Avian Flu Telegraph Reaction Diffusion Equation 3

1 where D is the diffusion coefficient. Cattaneo has argued that an instantaneous


2 adaptation of the flux to the gradient as stated in this equation is physically unre-
3 alistic. He assumed that the flux adapts to the gradient by a negative feedback loop
4 with small time constant τ ,

ut + div ν = 0,
5 (3.7)
τ νt + D̄ grad u + ν = 0.

6 Using w = −div ν and eliminating wthen we have a telegraph equation or damped


7 wave equation

8 τ utt + ut = D̄∆u. (3.8)

9 The corresponding telegraph reaction diffusion (TRD) equation or nonlinear


10 damped wave equation given by [6] is

∂2u ∂u ∂2u
11 τ 2
+ [1 − τ f  (u)] = D̄ 2 + f (u), (3.9)
∂t ∂t ∂x

12 where f (u) represents the interacting term.


13 The ODE and PDE neglect correlations between movements of different species,
14 and they assume instantaneous results of interactions.
15 The telegraph reaction-diffusion equation is known to be [2–4] more suitable for
16 modeling complex systems than usual diffusion equation because

17 • The TRD equation (3.9) is a form of a PDE of the joint system modeling reaction
18 and diffusion which include the delay and correlations.
19 • TRD results from correlated random walk.
20 • A third motivation is that starting from discrete space time one does not obtain
21 the standard diffusion equation but the telegraph equation.

22 4. Application of Telegraph Reaction Diffusion Equation


23 to Avian Flu
24 Here we use the method presented in [12] for solving the telegraph equation of DI
25 model given by:

∂2I  ∂I ∂2I
26 τ + [1 − τ f (I)] = D̄ + f (I), (4.1)
∂t2 ∂t ∂x2

27 where f (I) = βI(N − I) + D R k(x, y)I(y, t)dy, I(x, t) ≥ 0 is the densities of
28 infectives at point xat time t, k(x, y) is the density function for the proportion
29 of infectives at y that contact susceptibles  at x. The kernel k(x, y) is an integrable
30 function that satisfies k(x, y) ≥ 0 and R k(x, y)dx = 1 and D(x) ≥ 0 is the rate
1st Reading
November 17, 2010 19:41 WSPC S1793-5245 242-IJB 00105

4 S. A. A. Ahmed

1 at which infective individuals move from x to some new location. We will consider
2 that it is constant. This gives

∂2I ∂I ∂2I
3 τ 2 + [1 − τ β(N − 2I)] = D̄ 2 + βI(N − I) + D k(x, y)I(y, t)dy, (4.2)
∂t ∂t ∂x R

by transforming this equation using traveling wave coordinate into the ordinary
differential equation

(τ c2 − D̄)I  (z) − c[1 − τ β(N − 2I)]I  (z) − βI(N − I)



+ DI − D k(z − y)I(y)dy = 0, (4.3)
R

4 where z = x − ct, I(x, t) = I(z) and  is the derivative with respect to z.


5 Now, we will approximate the wave shapes by linearizing about both I = 0
6 and I = N , as Jones and Sleeman [10] have done for Fisher’s equation (see also
7 [1, 5, 14]).
8 Linearizing about I = 0

9 (τ c2 − D̄)I0 − c[1 − τ βN ]I0 − βN I0 + DI0 − D k(z − y)I0 (y)dy = 0, (4.4)
R

10 solving this linear equation gives characteristic equation of the form M (θ) = f (θ)
11 with f (λ) is square function since the derivatives in Eq. (4.4) is of the second order,
12 also we have M  (θ) > 0. Thus, f (θ) has at most two intersections with M (θ).
13 Using the ansatz I0 (z) = Ae−λz produces the characteristic equation

14 DM (λ) = αλ2 + γλ − (βN − D), (4.5)



15 where α = τ c2 − D̄, γ = (1 − βN τ )c and M (λ) = R k(z − y)eλ(z−y) dy this charac-
16 teristic equation has no real roots for c < c∗ , a real double root, λ, for c = c∗ , and
17 two distinct real roots, λ1 and λ2 , for c > c∗ . Both roots are positive and so the
18 corresponding solutions satisfy condition I0 (z) → 0 as z → ∞, I0 (0) = N/2 is

A1 e−λ1 z + A2 e−λ2 z , if c > c∗ ,
19 I0 (z) = (4.6)
(A1 + A2 z)e−λz , if c = c∗ ,

linearizing about I = N

(τ c2 − D̄)IN
 
− c[1 + τ βN ]IN − (2βN τ + βN )(IN − N )

+ DIN − D k(z − y)IN (y)dy = 0. (4.7)
R

20 Using the ansatz IN (z) = N + Be−µz , gives the characteristic equation

21 DM (µ) = αµ2 + (1 + βN τ )cµ + βN (2τ + 1) + D. (4.8)


1st Reading
November 17, 2010 19:41 WSPC S1793-5245 242-IJB 00105

Approximate Traveling Wave Solution of Avian Flu Telegraph Reaction Diffusion Equation 5

1 The 2nd order curve αµ2 + (1 + βN τ )cµ + βN (2τ + 1) + Dlies above the curve
2 αλ2 + (1 − βN τ )cλ − (βN − D). Since the second one has at least one real root for
3 c ≥ c∗ , then the first one has two real roots for c ≥ c∗ . One root is positive, µ+ ,
4 and one is negative µ− .
5 Using the conditions IN (z) → N as z → −∞, IN (0) = N/2, I0 (0) = IN 
(0) But,
6 the solution corresponding to µ+ does not satisfy the first condition, and so we
7 have
 
1 −µz
8 IN (z) = N 1 − e , (4.9)
2
9 with µ = µ− .
10 After using matching conditions we have for c > c∗
  

 1 −µz

 N 1 − e , for z ≤ 0,
2
11 I(z) = (4.10)

 N

 [−(λ2 + µ)e−λ1 z + (λ1 + µ)e−λ2 z ], for z ≥ 0,
2(λ1 − λ2 )
12 and for c = c∗



 1 −µz
N 1 − e , for z ≤ 0,
2
13 I(z) = (4.11)


 N [1 + (λ + µ)z]e−λz , for z ≥ 0.
2

Fig. 1. Traveling-wave shape. The traveling-wave, I(z),with Laplace kernel. The computation of
the solid graph were for D = 1, D̄ = 1, β = 1, N = 1, α = 0.5, c = 1.268
√ and τ = 0.05. And the
computation of the dot graph were for D = 1, β = 1, N = 1, α = 4 3/9, c = 2 as shown in [12]
which represent the case of τ = 0.
1st Reading
November 17, 2010 19:41 WSPC S1793-5245 242-IJB 00105

6 S. A. A. Ahmed

1 To compute the wave speeds, we linearized around I = 0 as in [12] then we have


2 cmin = 1.268.

3 Acknowledgment
4 The author wishes to thank Prof. E. Ahmed for his support and help.

5 References
6 [1] H. A. Abdusalam, Analytic and approximate solutions for Nagumo telegraph reaction
7 diffusion equation, Appl. Math. Comput. 157 (2004) 515–522.
8 [2] E. Ahmed and H. A. Abdusalam, On modified Black–Scholes equation, Chaos, Soli-
9 tons Fractals 22 (2004) 583–587.
10 [3] E. Ahmed, A. S. Elgazzar and A. S. Hegazi, An overview of complex adaptive systems,
11 Mans. J. Math. 32 (2005).
12 [4] E. Ahmed, A. S. Elgazzar and A. S. Hegazi, Modeling the avian flu, lessons form
13 complex adaptive systems in biology, Appl. Math. Comp. 195 (2008) 351.
14 [5] P. K. Brazhnik and J. J. Tyson, ON Traveling wave solutions of Fisher’s equation in
15 two spatial dimensions, Appl. Math. 60 (1999) 371–391.
16 [6] O. Diekman, V. Capasso and K. P. Hadeler, Mathematics Inspired by Biology,
17 (Springer Berlin, Heidelberg, 1998).
18 [7] http://en.wikipedia.org/wiki/H5N1.
19 [8] http://www.who.int/csr/disease/avian influenza/ en/index.html
20 [9] http://news.bbc.co.uk/1/shared/spl/hi/world/05/bird flu map/html/1.stm
21 [10] D. S. Jones, B. D. Sleeman, Differential Equations and Mathematical Biology (Allen
22 and Unwin, Boston, 1983).
23 [11] M. Kac, A stochastic model related to telegrapher’s equation, Rocky Mountain
24 J. Mathematics (1974) 497–509.
25 [12] J. Medlock and M. Kot, Spreading disease: integro-differential equations old and new,
26 Math. Biosci. 184 (2003) 201–222.
27 [13] J. D. Murray, Mathematical Biology I: An Introduction (Springer-Verlag, Berlin,
28 2002).
29 [14] M.-H. Wang and M. Kot, Speeds of invasion in a model with strong or weak Allee
30 effects, Math. Biosci. 171(1) (2001) 83.

You might also like