Power Series Method For Solving First Order Stiff Systems On Piecewise Uniform Mesh

International Journal of Application or Innovation in Engineering & Management (IJAIEM)
Web Site: www.ijaiem.org Email: [email protected]

Volume 4, Issue 9, September 2015
ISSN 2319 - 4847
Power series Method for solving first order stiff

systems on Piecewise uniform mesh
1
B. Sumithra , 2A. Tamilselvan
Department of Mathematics, Anna University, BIT Campus

Tiruchirappalli - 620 024, Tamilnadu, India.
2
Department of Mathematics, Bharathidasan University

Tiruchirappalli - 620 024, Tamilnadu, India.
ABSTRAC
In this paper, a power series method is suggested on a piecewise uniform mesh to solve a system of initial value problems of
first order stiff Ordinary Differential Equations(ODEs). The numerical results exhibited the performance of this method and
the results revealed that this method is very effective and convenient than the power series method without piecewise uniform
mesh.
Key words: System of stiff differential equations, Initial value problem, Power series method, Piecewise uniform mesh.
1 INTRODUCTION
The study of differential equations in pure and applied mathematics, physics and engineering is extensive. All of these
disciplines are concerned with the properties of differential equations of various types. Pure mathematics focuses on the
existence and uniqueness of solutions, while applied mathematics emphasizes the rigorous justification of the methods
for approximating solutions. Differential equations play an important role in modelling virtually every physical,
technical, or biological process from celestial motion, to bridge design, to interactions between neurons. Differential
equations like those used to solve real - life problems may not essentially be directly solvable, i.e. need not have closed
form solutions instead solutions can be approximated using numerical methods. Many fundamental laws of physics and
chemistry can be formulated as differential equations. In biology and economics, differential equations are used to
perfect the behavior of complex systems. The mathematical theory of differential equations first developed together
with the sciences, where the equations had originated and where the results initiated application. However, diverse
problems, sometimes originating in quite distinct scientific fields, may give rise to identical differential equations.
Whenever this happens, mathematical theory behind the equations can be viewed as a unifying principle behind diverse
phenomena. For example, consider propagation of light and sound in the atmosphere, and of waves on the surface of a
pond.
In the past power series method played a minor role in the numerical solutions of ordinary and partial differential
equations as it is often difficult to operate with power series. However, power series have certain advantages. A
truncated series forms a closed approximation of the solution which can be evaluated at any point in the region where
the series converges. Instability, which causes difficulties for finite difference solutions, does not affect the power series
solutions. The series solution has a great accuracy which permits the study of analytic properties of the solution to an
extent which is unachievable with a finite difference solution, the series solution can be used as an intermediate result
which can be integrated and differentiated easily. If a finite difference solution is only an intermediate step in the
solution of a problem, computer storage problems can be a major concern.
In [22], Tomasz Blenski states that, in a system of equations the time constant characterizing the system differ from
each other by orders of magnitude. Such a system of ODEs is called as stiff system of equations. For this stiff systems,
in initial moment ( close to origin ) the standard perturbation approach breaks down for t. Therefore for small t the
standard expansion be supplemented by an inner solution. This approach of stiff system of ODEs leads to the singular
perturbation method. Singular perturbation problem (spp) works well in piecewise uniform mesh rather than uniform
mesh which is discussed in [14]. So in stiff system also we can implemented the piecewise uniform mesh logic.
For a detailed discussion on ODEs, stiff nature, stiff ODEs and power series method one may refer to [1, 2, 3, 7, 8, 9,
11, 13, 21] and to name a few.
In this paper, the Power series method proposed by Nuran Guzel, Mustafa Bayram [15] for the numerical solution of
initial value problems in ODEs are reviewed. The focus of this paper is to improve the performance of the Power series
method by applying it in a piecewise uniform mesh (like Shishkin mesh). One can refer [17], [18], [19] and [20] for
numerical methods for stiff ODEs on piecewise uniform mesh. Possible approaches to the construction of such
piecewise uniform meshes and also some special schemes are given in [6, 12, 14]. As a result, the proposed method
have improved the accuracy and required less computational time.
Page 11

ISSN 2319 - 4847
Numerical solution of dynamic systems is often leads to stiff systems of first order ODEs with Initial Value Problems
(IVPs) in general form,
yi ' = fi (t, yi )
(1)
with initial condition
yi (t 0 ) = y0, i ,
i = 1, 2,..., n
t [0, T]
As given in [15], the solutions of (1) can be assumed that
y = y 0 et,
(2)
where e is a vector function which is the same size as y 0 . Substitute (2) into (1) and neglect higher order term. We
have the linear equation of e in the form
Ae = B,
(3)
where A and B is a constant matrixes. Solve this equation of (3), the coefficients of t in (2) can be determined.
Repeating above procedure for higher order terms, we can get the arbitrary order power series of the solutions for (1).
2 DESCRIPTION OF THE METHOD
As given in [15], we define another type power series in the form
f (x) = f 0 f1t f 2 t 2 ...... (f n p1e1 ........ p m em )t n ,
(4)
where p1 , p 2 ,..., p m are constant. e1 , e 2 ,...,e m are basis of vector e, m is size of vector e, y is a vector with m
elements in (2). Every element can be represented by the power series in (4).
yi = yi,0 yi,1t yi,2 t 2 .... ei t n ,

where yi is i
th
(5)
element of y. Substitute (5) into (1), we can get the following
fi = (fi,n pi,1e1 .... pi,m e m )t n j Q(t n j1 )

where f i is i
th
'
(6)
'
'
element of f (y, y , t) in (1) and j is 0 if f (y, y , t) have y ,1 if do not. From (6) and (3), we can
determine the linear equation in (3) as follow
Ai, j = Pi, j ,
(7)
Bi = f i,n .
(8)
Solve this linear equation, we have ei (i = 1,..., m) . Substitute ei into (5), we have yi (i = 1,..., m) which are
polynomials of degree n . Repeating this procedure from (5) to (8), we can get the arbitrary order power series of the
solution for differential-algebraic equations in (1).
3 DESCRIPTION OF PIECEWISE UNIFORM MESH

A piecewise uniform mesh is constructed on the interval [0,T] as follows:
Choose a point satisfying 0 <
T
m
and assume that N = 2 with m 2 . The point is called a transition
4
point and it divides the interval [0,T] into the two subintervals:
[0, T] = [0, ] [ , T].

If the solution has fast varying component in the neighbourhood of t = 0 , it is natural to have more number of mesh
points in the neighbourhood of t = 0 . Therefore the corresponding piecewise uniform mesh is constructed by dividing
N
3N
(0, ) into
equal mesh elements and (, T) into
equal mesh elements. This will give a better information
4
4
about the solution near t = 0 .
The piecewise uniform mesh is used with the following location of the transition point
T
= min{ , lnN}
4
(9)
Consider the parameter as
= 10( 6)
(10)
and
Page 12

ISSN 2319 - 4847
4
N
, j = 0(1) ,
N
4
4 (T )
N
where h 2 =
, j = ( 1)(1)N.
3N
4
t j = jh1
where h1 =
(11)
N
t j = ( j )h 2
4
T
If = T / 4, (i.e).,
< lnN then h = N 1 .
4
In such a case the method can be analysed using the standard techniques. We therefore assume that
(12)
= lnN
'
'
Let step size of t to be h and substitute it into the power series of y and y , we have y and y at t . If we repeat
above procedure, we have numerical solution of differential-algebraic equations in (1) [10, 16]. This method has been
used to obtained approximate numerical and theoretical solutions of a large class of differential-algebraic equation [4,
5] references therein.
4 NUMERICAL EXAMPLE
The numerical results of power series method with piecewise uniform mesh is compared with uniform mesh. The
comparison is based in terms of maximum error and average error. These results are tabulated and the numerical
results are recorded in terms of the following quantities:
As the formula given in [17], [18] and [19] for uniform mesh we have,
h=
(b a)
, where b is the end value of t and a is the initial value of t.
N
The calculation of error (for all piecewise uniform mesh and uniform mesh) is given as:
errorj =| u j(exactsolution) u j(approximate) | .
For maximum error(MAXE)(for all piecewise uniform mesh and uniform mesh), we compute using the formula which
is defined as follows:
MAXE N = max(errorj )
The average error for power series method with piecewise uniform mesh is defined as:
N
4
(error )
j
AVE1 =
j=1
where N1 =
N1
4
,
h1
j = 0(1)
N
,
4
(error )
j
j=
AVE2 =
N
1
4
4 (T )
,
(N 2 )
3 h2
AVE = max{AVE1, AVE2}
where N 2 =
j=(
N
1)(1)N.
4
The average error for power series method with uniform mesh is defined as:
Nj=1errorj
AVE =
N
whereN =
(b a)
,
h
b is the end value of t and a is the initial value of t.

Example 4.1 An illustrating example of a stiff system given in [15], be discussed now.
u'(t) = 1002u(t) 1000v 2 (t)

v'(t) = u(t) v(t)(1 v(t)) t [0,1],
u(0) = 1, v(0) = 1.
(13)
(14)
(15)
Exact solution:
u(t) = e 2t
Page 13

ISSN 2319 - 4847
v(t) = e t .
If we apply the power series method [15] to the given equation of system, the solutions of (4.1) and (4.2) can be
supposed as
u(t) = u e1 t u(t) = 1 e1t
(16)
v(t) = v e 2 t v(t) = 1 e 2 t
(17)
Substitude (4.4) and (4.5) into (4.1) and (4.2), and neglect higher order terms, we have
e1 2 Q(t) = 0
(18)
1 e 2 Q(t) = 0.
(19)
These formulas correspond to (2.3). The linear equation that corresponds to (2.4) can be given in the following
Ae = B,
(20)
where
e1
1 0
2
, B= , e= .
A=
0 1
1
e 2
From equation (4.8), we have linear equation
1 0 e1
2
0 1 e = 1
2

Solve this linear equation, we have
2
e=
1
and
u(t) = 1 2t,
v(t) = 1 t
(21)
(22)
From (4.9) and (4.10) the solutions of (4.1) and (4.2) can be supposed as
u(t) = 1 2t e1t 2 ,
v(t) = 1 t e2 t
(23)
(24)
In this manner, substitute (4.11) and (4.12) into (4.1) and (4.2) and neglect higher order terms, we have
(2e1 4)t Q(t 2 ) = 0, (25)

(2e 2 1)t Q(t 2 ) = 0 (26)
where
2 0
4 e1
B= e= .
A=
0 2,
1, e 2
From (4.13) and (4.14) we have linear equation
2 0 e1
4
0 2 e = 1
2

Solve this linear equation,we have
2
e= 1

2
Therefore
u(t) = 1 2t 2t 2 ,
1
v(t) = 1 t t 2 .
2
(27)
(28)
Repeating above procedure we have
Page 14

4
2
u(t) = 1 2t 2t 2 t 3 t 4 ....,
3
3
1
1
1
v(t) = 1 t t 2 t 3 t 4 .....
2
6
24
ISSN 2319 - 4847
(29)
(30)
Further proceed with the equations (29) and (30) by applying the piecewise uniform mesh method. The piecewise
uniform mesh is used with the following location of the transition point
T
= min{ , lnN}
4
(31)
= 10( 6)
(32)
where as
and
t j = jh1
t j = ( j )h 2
4
4
N
, j = 0(1) ,
N
4
4 (T )
N
where h 2 =
, j = ( 1)(1)N
3N
4
m
where N = 2 , m 2.
where h1 =
(33)
And the numerical results obtained by applying the piecewise uniform mesh method to the example 4.1 are given in
table 1.
Example 4.2 Let us consider the following system of differential equation discussed in [15], be done with piecewise
uniform mesh.
u(t) = 20u 0.25v 19.75w, u(0) = 1,

v(t) = 20u 20.25v 0.25w, v(0) = 0,
w (t) = 20u 19.75v 0.25w, w(0) = 1,
The analytic solution of the problem is
u = [exp
v = [exp
1
2t
1
2t
w = [exp
exp 20t (cos(20t) sin(20t))] / 2,

exp 20t (cos(20t) sin(20t))] / 2,
1
2t
exp 20t (cos(20t) sin(20t))] / 2.
If we apply the power series method [15] to the given equation system, following solution is obtained,
1 3199 2 85333 3 3413333 4

33
12799999 6
u = 1 t
t
t
t
t5
t ....,
4
16
32
256
250000
36
79 3199 2 1 3 3386667 4
1262212226 6
v=
t
t t
t 106666t 5
t ....,
4
16
96
256
355
81 3201 2 1 3 3413333 4 213333333 5 21688755 6
w = 1 t
t t
t
t
t ....
4
16
96
256
2000
61
(34)
(35)
(36)
Further proceed the Power series method by applying the piecewise uniform mesh method (9), (10) and (11) to (34),
(35) and (36). And the numerical results obtained by applying the piecewise uniform mesh method to the
example 4.2 are given in table 2.
Page 15

ISSN 2319 - 4847
5 CONCLUSION
The method has been applied to the solution of stiff systems of differential equation with Piecewise uniform mesh
outperformed the power series method with uniform mesh in terms of error, average error and accuracy. Hence, the
power series method with Piecewise uniform mesh is more efficient than the power series method with uniform mesh
for stiff systems.
Acknowledgements The authors are grateful to Dr. N. Ramanujam, Professor of Emeritus, Bharathidasan University,
Tiruchirappalli - 24 for his suggestions on this work.
REFERENCES
[1] P. Amodio, F. Mazzia, Numerical solution of differential-algebraic equations and computation of consistent
initial/boundary conditions, J. Comput. Appl. Math. (1997).
[2] U. M. Ascher, L. R. Petzold, Computer Methods for Ordinary Differential Equations and Differential-Algebraic
Equations, Society for Industrial and Applied Mathematics, Philadelphia, PA, (2003).
[3] K. E. Brenan, S.L. Campbell, L. R. Petzold, Numerical Solution of Initial-Value Problems in DifferentialAlgebraic Equations, North-Holland, Amsterdam, (1989).
[4] E. Celik, M. Bayram, Arbitrary order numerical method for solving differential-algebraic equation by Pade series,
Appl. Math. Comput. (2003).
[5] E. Celik, E. Karaduman, M. Bayram, Numerical method to solve chemical differential algebraic equations, Int. J.
Quant. Chem. (2002).
[6] C. Clavero, J.L. Gracia and F. Lisbona, High order methods on Shishkin meshes for singular perturbation
problems of convection diffusion type, Numerical Algorithms, (1999)
[7] Coddington, Earl A, Levinson, Norman, Theory of Ordinary Differential Equations, McGraw-Hill, New York,
(1955).
[8] E. Hairer, G. Wanner, Solving Ordinary Differential Equations II: Stiff and Differential- Algebreis Problems,
Springer-Verlag, (1991).
[9] Hille, Einar, Ordinary Differential Equations in the Complex Plane, Mineola Dover Publications, (1976)
[10] T. E. Hull, W. H. Enright, B. M. Fellen, A. E. Sedgwick, Comparing numerical methods for ordinary differential
equations, SIAM J. Numer. Anal. (1972).
[11] Jay A. Leavitt, Methods and Applications of Power Series, Math. Comp. (1966).
[12] Kaailash C. Patidar, High order parameter uniform numerical Method for Singular Perturbation problems, Appl.
Math and comp, (2007).
[13] G. Lewis, Two Methods Using Power Series for Solving Analytic Initial Value Problems, A.E.C. Research and
Development Report, NYU-2881, New York University, Courant Inst. of Math. Sci., (1960).
[14] J. J. H. Miller, E. O Riordan, G. I. Shishkin, Fitted numerical methods for singular perturbation problems, World
Scientific Publishing Co.pvt.Ltd, Dublin, (2012).
[15] Nuran Guzel, Mustafa Bayram, On the numerical solution of stiff systems , Applied Mathematics and
Computation, (2005).
[16] W. H. Press, B. P. Flannery, S. A. Teukolsky, W. T. Vetterling, Numerical Recipes, Cambridge University Press,
Cambridge, (1988).
[17] B. Sumithra, A. Tamilselvan, Numerical Solution of Stiff System by Backward Euler Method, Applied
Mathemaical Sciences, Vol. 9, No. 67, (2015).
[18] B. Sumithra, A. Tamilselvan, Numerical Solution of Stiff System by Trapezoidal Method, International Journal of
Mathematical Sciences and Engineering Applications, Vol. 9 No. I (March, 2015).
[19] B. Sumithra, A. Tamilselvan, Trapezoidal Method for solving first order stiff systems on Piecewise uniform mesh,
International Journal of Mathematical Engineering Science, (In Press).
[20] B. Sumithra, A. Tamilselvan, Numerical Solution of Stiff System by Second Order Backward Difference Formula,
Global Journal of Pure and Applied Mathematics, (In Press).
[21] Teschl, Gerald, Ordinary Differential Equations and Dynamical Systems, Providence American Mathematical
Society, (2012).
[22] Tomasz Blenski, Modified singular perturbation method for a stiff system of linear evolution equations, J. Math.
Phys., (1981).
Table 1: Values of MAXE(u),AVE(u),MAXE(v),AVE(v) for the solution components u and v
respectively for the Example 4.1
8
Number of mesh points N

16
32
Page 16

ISSN 2319 - 4847
piecewise uniform mesh
6.2102e-007
1.3328e-006
1.8753e-006
uniform mesh
1.3288e-006
1.8724e-006
2.2038e-006
uniform mesh
7.7628e-008
1.6611e-007
8.3298e-008
1.1703e-007
5.8603e-008
6.8867e-008
1.9648e-008
4.2252e-008
5.9511e-008
uniform mesh
4.2127e-008
5.9419e-008
6.9969e-008

uniform mesh
2.456e-009
5.2658e-009
2.6407e-009
3.7137e-009
1.8597e-009
2.1865e-009
MAXE(u)
AVE(u)
MAXE(v)
AVE(v)
( 6)
Figure 1:
, the solution obtained by the suggested numerical method is displayed in
For the example-1 with n = 32, = 10

Figure 1.
Table 2: Values of MAXE(u),AVE(u),MAXE(v),AVE(v) for the solution components u and v
respectively for the Example 4.2
MAXE(u)

uniform mesh
32
0.22965e-001
0.63190e-001
0.64219e-001 0.10154 e+000

0.10069e+000 0.12560e+000
0.28706e-002
0.78995e-002
0.40137e-002
0.62934e-002
0.31731e-002
0.39249e-002
uniform mesh
0.41502e-003
0.57076e-002
0.58649e-002
0.11910e-001
0.12059e-001
0.16414e-001

uniform mesh
0.51878e-004
0.71345e-003
0.36656e-003
0.74439e-003
0.37684e-003
0.51292e-003
AVE(u)
uniform mesh
MAXE(v)
AVE(v)
Number of mesh points N

16
Page 17

ISSN 2319 - 4847
( 6)
Figure 2:
, the solution obtained by the suggested numerical method is displayed in
Page 18
For the example-2 with n = 32, = 10

Figure 2.

Power Series Method For Solving First Order Stiff Systems On Piecewise Uniform Mesh

Uploaded by

Copyright:

Available Formats

Power Series Method For Solving First Order Stiff Systems On Piecewise Uniform Mesh

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Power Series Method For Solving First Order Stiff Systems On Piecewise Uniform Mesh

Uploaded by

Copyright:

Available Formats

International Journal of Application or Innovation in Engineering & Management (IJAIEM)

Web Site: www.ijaiem.org Email: [email protected]

ISSN 2319 - 4847