The phenomenon of extinction is an important property of solutions for many evolutionary equa-tions. In this paper, a numerical simulation for computing the extinction time of nonnegative solu-tions for some nonlinear parabolic equations on general domains is presented. The solution algo-rithm utilizes the Donor-cell scheme in space and Euler’s method in time. Finally, we will give some numerical experiments to illustrate our algorithm.
Nonlinear Parabolic Equations Donor-Cell Scheme Numerical Extinction Time General Domains1. Introduction
There is a large number of nonlinear partial differential equations of parabolic type whose solutions for given initial data become identically nulle in finite time T. Such a phenomenon is called extinction and T is called the extinction time. For certain problems, the extinction time can be computed explicitly, but in many cases one can only know the existence of extinction time.
Since the appearance of the pioneering work of Kalashnikov [1] , extinction phenomenon in nonlinear parabolic equations has been studied extensively by many authors [2] [3] . Particular emphasis has been placed on the question as to the existence of extinction time [4] - [7] .
Generally speaking, it is difficult to simulate extinction phenomenon accurately on general domains. Indeed, it is not at all clear if features of such a phenomenon as extinction can be well reflected in the discretized equation which approximates the original equation. In [8] - [11] , some numerical schemes have been used to study the extinction phenomenon of solutions for some nonlinear parabolic equations.
In this work, we propose a numerical algorithm for computing the extinction time for nonnegative solutions of some nonlinear parabolic equations. Our motivation is to reproduce the extinction phenomenon of some non- linear parabolic equations on general domains.
This paper is organized as follows. In the next section, we present the problem model and some theoretical results. A discretization of this problem is derived in Section 3, while numerical experiments are reported in Section 4 and Section 5 is devoted to concluding remarks.
2. The Model Problem
In this work, we are concerned with the following initial-boundary value problem:
where is a bounded domain with boundary, , and, being given functions.
Furthermore, for any and for all defined in, we will set,.
Nonlinear parabolic equations of type (1) appear in various applications. In particular they are used to des- cribe a phenomenon of thermal propagation in an absorptive medium where stands for temperature [5] . In other applications, is a concentration and the process is described as diffusion with absorption.
The problem of determining necessary and sufficient conditions on the functions and which ensure the existence of an extinction time for solutions of (1)-(3) has been considered by several authors [1] [6] [7] [12] .
In this section, we state the following result:
Theorem 1 Assume that is a nonnegative solution of the problem (1)-(3) where f and F are nondecreasing, nonnegative derivatives functions and if, then
Proof: First, let us set. We have
Multiplying Equation (1) by and integrating over it follows
On one hand thanks to regularity of functions, and, we can write
where is the unit outward to, ds denotes an element of surface area, since vanishes on and
, we have, hence
From (ii) and (iii), we deduce
Since and, and according to (i), we obtain
On the other hand, multiplying Equation (1) by yields
,
which we rewrite as
,
then the application is decreased in. It then follows
On the other hand, the increase of implies that
,
Thus
and according to (vi) we obtain
,
This last inequality implies
.
Then considering (iv), we deduced
Setting, we obtain
.
This gives after integrating
.
Knowing that. it follows. The passage to the limit allows us to write
.
Finally,
.
In addition to the assumption of increase of F in Theorem 2.1, if we assume that
is a positive constant.
then the following result is easily shown.
Corollary 1 Suppose that the assumptions of the Theorem 2.1 are satisfied, and if (4) holds for, then
Indeed, for all solution of (1)-(3), it comes from the assumption (4) that
,
that gives
.
As, then
.
So
,
and as a consequence of Theorem 2.1
. □
In summary, under some assumptions we know that all nonnegative solutions of (1)-(3) have extinction time as. We want to determine whether extinction occurs in finite time for any given and.
It is well known that, in general, there is no classical solution to this nonlinear parabolic equation for arbitrary choices of and. However, there are some works dealing with approximation of extinction time for solutions of (1). For example, in [13] a numerical method to approximate the solutions of (1) has been developed in the case and in [14] an algorithm based on splitting technique was derived to compute the extinction time for solutions on a rectangular domain.
In order to determine the extinction time for some and, we will derive in the next section a numerical scheme based on Donor-cell scheme. Given a sufficiently small parameter, we would like to determine the positive real such that a solution of the problem (1)-(3) has to satisfy the above relation
We shall call satisfying (6) as the -extinction time.
3. Discretization3.1. Discretization of the Studied Domain
Let be a considered domain that we assume to be of irregular shape, we approximate by a domain whose boundary is specified by the set of boundary edges lying on gridlines. We imbed in a rectan- gular domain, of smallest possible size. Given two nonzero integers and
, we set and we introduce on a grid of step and in x
and direction respectively. The set of points such that of, defines the discretization of into cells (rectangular subdomains). For all, cell
occupies the spatial region and has center the point noted.
The cells of are then divided into inner cell (which lie completely in), external cell (which lie in) and boundary cells (which lie in a part of). The problem model is then solved only in the inner cells.
A matrix of size gives a description of the discretized domain. For example, consider three sets of indices, et corresponding to the inner, boundary and external cells, we then admit to define the following matrix
the matrix to identify cell types.
The idea of this numerical treatment of general domains has been suggested by Griebel et al. in [15] . An example of this numerical treatment is illustrated in Figure 1 and its matrix representative is given by the following (8).
3.2. Spatial Discretization
First of all, let us give an approximation of the diffusion operator at the point which we rewrite as
An example of the discretization of a non rectangular domain into cells
Let be the approximation of at the cell center. In the following we do apply a discretization that is similar to the one of Donor-cell scheme where the expression is approached by a progressive finite differences scheme and by a central finite differences scheme.
Furthermore, we set where and we note
If denotes the vector of components then, one can write
where the diagonal matrix whose diagonal is the vector and denotes forward differentiation matrix the -direction.
On the other hand, denoting by the approached value of at the cell center and
noting by
the vector of the value of at point, it follows through the central difference scheme, the relation
which is written by the mean of the Equation (12) as
where denotes central differentiation matrix in the -direction.
Similarly, given, the vector of the approached values of at point, we obtain
where denotes forward differentiation matrix in the -direction.
From Equations (13) and (14), we deduce the approximation of the operator at cell center, in matrix form:
where is the vector of value of at points. Thus, we have defined, an approximation operator to approach the operator.
However, it should be noted that is a vector dependent of.
Considering lexicographic numerotation, we note by the vector of the values of u in points at time t. Knowing that at points, and the differentiation matrices to approach the derivative on the set of the points we can replace respectively by the matrices which are obtained by deleting the rows and columns corresponding to the indices of the points of.
Given the Equation (15), the discrete system approaching the problem (1)-(3) is rewritten by
where we have set
and where is the matrix obtained of by deleting the rows and columns corresponding to
the indices of the points of and is the vector of values of,.
Furthermore, the initial condition is written
where is the vector obtained from by deleting the elements corresponding to the indices of the points of
3.3. Temporal Discretization
For a time step fixed, we consider the sequence defined by et. Then, we
denote the approximation at time of vector solution of (16)-(17). Using the explicit Euler method, the semi-discret scheme is written
where we set.
It should be noticed that if the time step is chosen to be little enough, and F satisfied the growth con- dition (4),
Thus the extinction time is obtained using simple itrations process until the stopping criterion
is satisfied. Here is the given tolerance number. The sequence of computations to be performed is sum- marized as follows
Algorithm 3.1
1. Read and.
2. Compute and.
3. Define (rectangular) and (non rectangular) such.
4. Compute the matrix and.
5. Set.
6. Assign initial value to.
7. Set.
8. While and, do.
9. Compute according to (18) .
10. Set.
11..
End while
4. Numerical Experiments
Let be a bounded domain in. Consider the initial value problem
where is the continuous nonnegative function in, vanishing on, and.
Equation (21) models heat propagation in medium where the solution stands for temperature.
For our numerical experiments we have consider Figure 2 to be our studied domain and we have use discretization parameters, and.
We would like to numerically estimate the extinction time for solutions of problem (21)-(23) with the initial condition given by
First, for fixed accurate value, we estimate the -euclidian norm of the sequence solution of the numerical scheme (18) for various values of parameters. Table 1 and Figure 3 clearly show that the approximation extinction time can be given by
We can see in Table 1 that this value is approximated by
Also, the extinction process is illustrated by Figure 4 where we can appreciate the numerical solution extinct in a finite time.
5. Concluding Remarks
In this paper, a numerical algorithm based on Donor-cell scheme was proposed in order to compute the extinc- tion time for nonnegative solutions of some nonlinear parabolic equations on general domains. We have verified
Discretization of studied domain <inline-formula><inline-graphic xlink:href="http://html.scirp.org/file/3-7402687x240.png" xlink:type="simple"/></inline-formula> into cells
Numerical extinction time relatively to time iteration parameter n
n
13,000
16,000
21,000
31,000
61,000
70,000
100,000
>500,000
T
0.13
0.16
0.21
0.31
0.61
0.6123
0.6123
0.6123
Variation norm of the numerical solution
Extinction phenomenon of the numerical solution
experimentally for a class of nonlinear parabolic equations that the numerical algorithm is efficient for comput- ing the extinction time of solutions.
In the works to come, it will be better to apply the numerical algorithm to study, for example, moving boun- dary problems and extinction problems in environment.
Acknowledgements
We thank the Editor and the referee for their comments.
ReferencesKalashnikov, A.S. (1974) The Propagation of Disturbances in Problems of Non-Linear Heat Conduction with Absorption. USSR Computational Mathematics and Mathematical Physics, 14, 70-85.
http://dx.doi.org/10.1016/0041-5553(74)90073-1Galaktionov, V.A. and Vasquez, J.L. (2002) The Problem of Blow-Up in Nonlinear Parabolic Equations. Discrete and Continuous Dynamics Systems, 8, 399-433.
http://dx.doi.org/10.3934/dcds.2002.8.399Levine, H.A. (1985) The Phenomenon of Quenching: A Servey. North-Holland Mathematics Studied, 110, 275-286.
http://dx.doi.org/10.1016/S0304-0208(08)72720-8Diaz J.I. ,et al. (2001)Qualitative Study of Nonlinear Parabolic Equations: An Introduction Extracta Mathematicae 16, 303-341.Friedman, A. and Herrero, M.A. (1987) Extinction Properties of Semilinear HEAT Equations with Strong Absorption. Journal of Mathematical Analysis and Applications, 124, 530-546.
http://dx.doi.org/10.1016/0022-247X(87)90013-8Gu Y.G. ,et al. (1994)Necessary and Sufficient Conditions for Extinction of Solutions to Parabolic Equations Acta Mathematica Sinica 37, 73-79.Lair, A.V. (1993) Finite Extinction Time for Solutions of Nonlinear Parabolic Equations. Nonlinear Analysis, Theory, Methods and Applications, 21, 1-8.Boni, T.K. (2001) Extinction for Dicretizations of Some Semilinear Parabolic Equations. Comptes Rendus de l’Aca- dmie des Sciences de Paris, Serie I, Mathmatique, 333, 795-800.Mikula, K.B. (1995) Numerical Solution of Nonlinear Diffusion with Finite Extinction Phenomenom. Acta Mathematica Universitatis Comenianae, LXIV, 173-184.Nabongo, D. and Boni, T.K. (2008) Numerical Quenching for a Semilinear Parabolic Equation. Mathematical Modelling and Analysis, 13, 521-538.
http://dx.doi.org/10.3846/1392-6292.2008.13.521-538Nabongo, D. and Boni, T.K. (2008) Quenching for Semidiscretization of a Semilinear Heat Equation with Dirichlet and Neumann Boundary Condition. Commentationes Mathematicae Universitatis Carolinae, 49, 463-475.Lair, A.V. and Oxley, M.K. (1996) Anisotropic Nonlinear Diffusion with Absorption: Existence and Extinction. International Journal of Mathematics and Mathematical Sciences, 19, 427-434.
http://dx.doi.org/10.1155/S0161171296000610Dumitrache, A. (2007) A Numerical Method to Approximate the Solutions of Nonlinear Absorption Diffusion Equation. Proceeding in Applied Mathematics and Mechanics, 7, 4070041-4070042.
http://dx.doi.org/10.1002/pamm.200701050Kim, D. and Proskurowski, W. (2004) An Efficient Approach for Solving a Class of Nonlinear 2D Parabolic PDEs. International Journal of Mathematics and Mathematical Sciences, 2004, 881-899.Griebel, M., Dornseifer, T. and Neunhoeffer, T. (1998) Numerical Simulation in Fluid Dynamics. A Practical Guide, SIAM, Philadephia.