Hindawi Publishing Corporation

Mathematical Problems in Engineering

Volume 2013, Article ID 989381, 11 pages
http://dx.doi.org/10.1155/2013/989381

Research Article
A Numerical Algorithm for Solving Stiff
Ordinary Differential Equations

S. A. M. Yatim,1 Z. B. Ibrahim,2 K. I. Othman,3 and M. B. Suleiman2

1
School of Distance Education, Universiti Sains Malaysia, 11800 Penang, Malaysia
2
Department of Mathematics, Faculty of Science, Universiti Putra Malaysia, Selangor, 43400 Serdang, Malaysia
3
Department of Mathematics, Faculty of Computer and Mathematical Sciences, Universiti Teknologi MARA, Selangor,
40450 Shah Alam, Malaysia

Correspondence should be addressed to Z. B. Ibrahim; [email protected]

Received 6 August 2012; Revised 24 October 2012; Accepted 7 November 2012

Academic Editor: J. Rodellar

Copyright © 2013 S. A. M. Yatim et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

An advanced method using block backward differentiation formula (BBDF) is introduced with efficient strategy in choosing the step
size and order of the method. Variable step and variable order block backward differentiation formula (VSVO-BBDF) approach is
applied throughout the numerical computation. The stability regions of the VSVO-BBDF method are investigated and presented in
distinct graphs. The improved performances in terms of accuracy and computation time are presented in the numerical results with
different sets of test problems. Comparisons are made between the proposed method and MATLAB’s suite of ordinary differential
equations (ODEs) solvers, namely, ode15s and ode23s.

1. Introduction preferred solvers to be used for comparison purposes [7, 8].

Stiff ODE solvers that are available in MATLAB ODE suite
Many studies on solving the equations of stiff ordinary are ode15s and ode23s which are based on the numerical
differential equations (ODEs) have been done by researchers differentiation formulas [8, 9] and the modified Rosenbrock
or mathematicians specifically. With the numbers of numer- formula of order 2, respectively [8].
ical methods that currently exist in the literature, extensive An improved method has been identified in [10], which
research has been done to unveil the comparison between was expanded from the method incorporated with BDF
their rate of convergence, number of computations, accuracy, proposed by Gear. Since then, the study on producing block
and capability to solve certain type of test problems [1]. The approximations 𝑦𝑛+1 , 𝑦𝑛+2 , 𝑦𝑛+3 , . . . , 𝑦𝑛+𝑘 also known as Block
well-known numerical methods that are used widely are from Backward Differentiation Formulae (BBDF) [11] has attracted
the class of BDFs or commonly understood as Gear’s Method much attention. BBDF method by using variable step size in
[2]. However, many other methods that have evolved to this [12] demonstrates the competency of computing concurrent
date are for solving stiff ODEs which arise in many fields solution values at different points. Consequently, study in
of the applied sciences [3–6]. The problems considered in [13, 14] is an extension of a previous study in a way that
this paper are for the numerical solution of the initial value the accuracy is improved by increasing the order of the
problem, method up to order 5. Thus, these studies lead to enhancing
𝑦󸀠 = 𝑓 (𝑥, 𝑦) (1) the existing method to become Variable Order Variable Step
Block Backward Differentiation Formula of order 3 until 5
with given initial values 𝑦(𝑎) = 𝑦0 in the given interval 𝑥 ∈ (VSVO-BBDF).
[𝑎, 𝑏]. In Section 2, we introduce the general formulation of
The existing numerical methods have often been com- VSVO-BBDF of order 3 to 5. Order conditions are listed in
pared to one another to find the best approximation and Section 3 while in Section 4, we consider the implementation
the best method. MATLAB ODE suite is one of the most of VSVO-BBDF. The analysis of stability regions is illustrated
2 Mathematical Problems in Engineering

qh rh rh h h For VSVO-BBDF of order 𝑃5 (𝑃 = 5)

𝑃 (𝑥) = 𝑃 (𝑥𝑛+1 + 𝑠ℎ)

P3 (𝑞 + 2𝑟 + 1 + 𝑠) (2𝑟 + 1 + 𝑠) (𝑟 + 1 + 𝑠)
P4
= 𝑦𝑛
(2 (𝑞 + 2𝑟)) 𝑟2
P5
Xn−3 Xn−2 Xn−1 Xn Xn+1 Xn+2 (𝑞 + 2𝑟 + 1 + 𝑠) (2𝑟 + 1 + 𝑠) (1 + 𝑠)
+ 𝑦𝑛−1
Figure 1: VSVO-BBDF method of order (𝑃3–𝑃5). − (𝑞 + 𝑟) 𝑟2 (6)

(𝑞 + 2𝑟 + 1 + 𝑠) (𝑟 + 1 + 𝑠) (1 + 𝑠)
in Section 5. This is followed by the strategy in choosing the + 𝑦𝑛−2
(2𝑞𝑟2 )
step size and order of VSVO-BBDF in Section 6. Numerical
results are presented in Section 7. The appendix describes the (2𝑟 + 1 + 𝑠) (𝑟 + 1 + 𝑠) (1 + 𝑠)
algorithm applied for VSVO-BBDF method. + 𝑦𝑛−3 .
−𝑞 (−𝑞 − 𝑟) (−𝑞 − 2𝑟)

2. VSVO-BBDF Method Formulation

Substituting 𝑠 = 0 and 𝑠 = 1 gives the predictor for the
Two values of 𝑦𝑛+1 and 𝑦𝑛+2 were computed simultaneously first and second points, respectively. Therefore by letting 𝑟 =
in block by using earlier blocks with each block containing a 1, 𝑞 = 1, 𝑟 = 2, 𝑞 = 2 and 𝑟 = 1, 𝑞 = 10/19. This produced
maximum of two points (Figure 1). The orders of the method the following coefficients (Tables 1, 2, and 3) for the first
(𝑃3, 𝑃4, and 𝑃5) are distinguished by the number of back and second points of predictor formulae for VSVO-BBDF
values contained in total blocks. The ratio distance between method.
current (𝑥𝑛 ) and previous step (𝑥𝑛−1 ) is represented as 𝑟 and The interpolating polynomial of the function 𝑦(𝑥) using
𝑞 in Figure 1. In this paper, the step size is given selection Lagrange polynomial in (2) gives the following corrector for
𝑝 𝑝
to decrease to half of the previous steps or increase up to a the first point 𝑦𝑛+1 and second point 𝑦𝑛+2 . The resulting
factor of 1.9. For simplicity, 𝑞 is assigned as 1, 2, and 10/19 Lagrange polynomial for each order was given as follows.
for the case of constant, halving and increasing the step size, For VSVO-BBDF of order 𝑃3 (𝑃 = 3)
respectively. The zero stability is achieved for each of these
cases and explained in the next section. 𝑃 (𝑥) = 𝑃 (𝑥𝑛+1 + 𝑠ℎ)
We find approximating polynomials 𝑃𝑘 (𝑥), by means of
a 𝑘-degree polynomial interpolating the values of 𝑦 at given (𝑟 + 1 + 𝑠) (𝑠 + 1) (𝑠)
points, that are (𝑥𝑛−3 , 𝑦𝑛−3 ), (𝑥𝑛−2 , 𝑦𝑛−2 ), (𝑥𝑛−1 , 𝑦𝑛−1 ), . . . , = 𝑦𝑛+2
2𝑟 + 4
(𝑥𝑛+2 , 𝑦𝑛+2 ):
(𝑟 + 1 + 𝑠) (𝑠 + 1) (𝑠 − 1)
𝑘 + 𝑦𝑛+1
𝑃𝑘 = ∑ 𝑦 (𝑥𝑛+1−𝑗 ) ⋅ 𝐿 𝑘,𝑗 (𝑥) , (2) −1 − 𝑟 (7)
𝑗=0 (𝑟 + 1 + 𝑠) (𝑠 − 1) (𝑠)
+ 𝑦𝑛
where 2𝑟
𝑘
(𝑥 − 𝑥𝑛+1−𝑖 )
𝐿 𝑘,𝑗 (𝑥) = ∏ for each 𝑗 = 0, 1, . . . , 𝑘. (1 + 𝑠) (𝑠 − 1) (𝑠)
+ 𝑦 .
𝑖=0 (𝑥𝑛+1−𝑗 − 𝑥𝑛+1−𝑖 ) −𝑟 (−1 − 𝑟) (−𝑟 − 2) 𝑛−1
𝑖 ≠ 𝑗
(3)
𝑝 𝑝 For VSVO-BBDF of order 𝑃4 (𝑃 = 4)
Predictors for the first point and second point 𝑦𝑛+2
𝑦𝑛+1
were computed using back values as the interpolating points.
𝑃 (𝑥) = 𝑃 (𝑥𝑛+1 + 𝑠ℎ)
The resulting Lagrange polynomial for each order was given
as follows. (2𝑟 + 1 + 𝑠) (𝑟 + 1 + 𝑠) (1 + 𝑠) (𝑠)
For VSVO-BBDF of order 𝑃3 (𝑃 = 3) = 𝑦𝑛+2
2 (2𝑟 + 2) (𝑟 + 2)
(𝑟 + 1 + 𝑠) (1 + 𝑠)
𝑃 (𝑥) = 𝑃 (𝑥𝑛+1 + 𝑠ℎ) = 𝑦𝑛 + 𝑦 . (4) (2𝑟 + 1 + 𝑠) (𝑟 + 1 + 𝑠) (1 + 𝑠) (𝑠 − 1)
𝑟 −𝑟 𝑛−1 + 𝑦𝑛+1
− (2𝑟 + 1) (𝑟 + 1)
For VSVO-BBDF of order 𝑃4 (𝑃 = 4)
(2𝑟 + 1 + 𝑠) (𝑟 + 1 + 𝑠) (𝑠) (𝑠 − 1) (8)
𝑃 (𝑥) = 𝑃 (𝑥𝑛+1 + 𝑠ℎ) + 𝑦𝑛
4𝑟2
(2𝑟 + 1 + 𝑠) (𝑟 + 1 + 𝑠) (2𝑟 + 1 + 𝑠) (1 + 𝑠)
= 2
𝑦𝑛 + 𝑦𝑛−1 (2𝑟 + 1 + 𝑠) (1 + 𝑠) (𝑠) (𝑠 − 1)
2𝑟 −𝑟2 + 𝑦𝑛−1
−𝑟2 (−𝑟 − 1) (−𝑟 − 2)
(𝑟 + 1 + 𝑠) (1 + 𝑠)
+ 𝑦𝑛−2 . (𝑟 + 1 + 𝑠) (1 + 𝑠) (𝑠) (𝑠 − 1)
2𝑟2 + 𝑦𝑛−2 .
(5) 2𝑟2 (−2𝑟 − 1) (−2𝑟 − 2)
Mathematical Problems in Engineering 3

Table 1: Coefficients for the first and second point of predictor formulae for VSVO-BBDF when 𝑟 = 1, and 𝑟 = 1, 𝑞 = 1.

Order 𝑦𝑛 𝑦𝑛−1 𝑦𝑛−2 𝑦𝑛−3

ℎ𝑓𝑛+1 2 −1
𝑃=3
ℎ𝑓𝑛+2 3 −2
ℎ𝑓𝑛+1 3 −3 1
𝑃=4
ℎ𝑓𝑛+2 6 −8 3
ℎ𝑓𝑛+1 4 −6 4 −1
𝑃=5
ℎ𝑓𝑛+2 10 −20 15 −4

Table 2: Coefficients for the first and second point of predictor formulae for VSVO-BBDF when 𝑟 = 2, and 𝑟 = 2, 𝑞 = 2.

Order 𝑦𝑛 𝑦𝑛−1 𝑦𝑛−2 𝑦𝑛−3

3 1
ℎ𝑓𝑛+1 −
𝑃=3 2 2
ℎ𝑓𝑛+2 2 −1
15 5 3
ℎ𝑓𝑛+1 −
𝑃=4 8 4 8
ℎ𝑓𝑛+2 3 −3 1
35 35 21 5
ℎ𝑓𝑛+1 − −
𝑃=5 16 16 16 16
ℎ𝑓𝑛+2 −1 4 −6 4

For VSVO-BBDF of order 𝑃5 (𝑃 = 5) where 𝑦𝑛+𝑗 ≈ 𝑦(𝑥𝑛+𝑗 ) and 𝑓𝑛+𝑗 ≡ 𝑓(𝑥𝑛+𝑗 , 𝑦𝑛+𝑗 ), coefficients
𝑃 (𝑥) = 𝑃 (𝑥𝑛+1 + 𝑠ℎ) 𝛼𝑗 , 𝛽𝑗 are suitably chosen constants subject to conditions
𝛼𝑘 = 1, |𝛼0 | + |𝛽0 | ≠ 0, and 𝑘 is defined as the order of
= (((𝑞 + 2𝑟 + 1 + 𝑠) (2𝑟 + 1 + 𝑠) (𝑟 + 1 + 𝑠) (1 + 𝑠) 𝑠) the particular method employed. This method is said to be
explicit if 𝛽𝑘 = 0 and implicit otherwise.
−1
×(2 (𝑞 + 2𝑟 + 2) (2𝑟 + 2) (𝑟 + 2)) ) 𝑦𝑛+2 Substituting 𝑠 = 0 and 𝑠 = 1 gives the corrector for the
first and second points, respectively. Therefore by letting 𝑟 =
1, 𝑞 = 1, 𝑟 = 2, 𝑞 = 2 and 𝑟 = 1, 𝑞 = 10/19. This produced
+ (((𝑞 + 2𝑟 + 1 + 𝑠) (2𝑟 + 1 + 𝑠) (𝑟 + 1 + 𝑠)
the following coefficients (Tables 4, 5, and 6), as
× (1 + 𝑠) (𝑠 − 1)) in Definition 1 for the first and second points of VSVO-
BBDF method.
−1
× (− (𝑞 + 2𝑟 + 1) (2𝑟 + 1) (𝑟 + 1)) ) 𝑦𝑛+1
3. Order Conditions for General VSVO-BBDF
(𝑞 + 2𝑟 + 1 + 𝑠) (2𝑟 + 1 + 𝑠) (𝑟 + 1 + 𝑠) 𝑠 (𝑠 − 1)
+ 𝑦𝑛 As similar to analysis for order of Linear Multistep Method
4 (𝑞 + 2𝑟) 𝑟2 (LMM) given in [15], we use the following to determine the
(𝑞 + 2𝑟 + 1 + 𝑠) (2𝑟 + 1 + 𝑠) (1 + 𝑠) 𝑠 (𝑠 − 1) order of VSVO-BBDF method.
+ 𝑦𝑛−1
−𝑟2 (𝑞 + 𝑟) (−𝑟 − 1) (−𝑟 − 2) Definition 2. The LMM [15] and the associated difference
operator 𝐿 defined by
(𝑞 + 2𝑟 + 1 + 𝑠) (𝑟 + 1 + 𝑠) (1 + 𝑠) 𝑠 (𝑠 − 1)
+ 𝑦𝑛−2 𝑗
2𝑞𝑟2 (−2𝑟 − 1) (−2𝑟 − 2)
𝐿 [𝑧 (𝑥) ; ℎ] = ∑ [𝛼𝑘 𝑧 (𝑥 + 𝑘ℎ) − ℎ𝛽𝑘 𝑧󸀠 (𝑥 + 𝑘ℎ)] (10)
𝑘=0
+ (((2𝑟 + 1 + 𝑠) (𝑟 + 1 + 𝑠) (1 + 𝑠) 𝑠 (𝑠 − 1))
are said to be of order 𝑝 if 𝑐𝑜 = 𝑐1 = ⋅ ⋅ ⋅ = 𝑐𝑝 = 0, 𝐶𝑝+1 ≠ 0.
× (−𝑞 (−𝑞 − 𝑟) (−𝑞 − 2𝑟)
The general form for the constant 𝐶𝑞 is defined as
−1
× (−𝑞 − 2𝑟 − 1) (−𝑞 − 2𝑟 − 2)) ) 𝑦𝑛−3 . 𝑗
1
(9) 𝐶𝑞 = ∑ [𝑘𝑞 𝛼𝑘 − 𝑘𝑞−1 𝛽𝑘 ] , 𝑞 = 2, 3, . . . 𝑝 + 1.
𝑘=0
(𝑞 − 1)!
Linear Multistep Method (LMM) given in [15] is given in (11)
the definition below.
Consequently, BBDF method can be represented in standard
Definition 1. The linear 𝑘-step method can be represented in form by an equation ∑𝑘𝑗=0 𝐴 𝑗 𝑦𝑛+𝑗 = ℎ ∑𝑘𝑗=0 𝐵𝑗 𝑓𝑛+𝑗 , where 𝐴 𝑗
standard form by an equation ∑𝑘𝑗=0 𝛼𝑗 𝑦𝑛+𝑗 = ℎ ∑𝑘𝑗=0 𝛽𝑗 𝑓𝑛+𝑗 , and 𝐵𝑗 are 𝑟 by 𝑟 matrices with elements 𝑎𝑙,𝑚 and 𝑏𝑙,𝑚 for
4 Mathematical Problems in Engineering

Table 3: Coefficients for the first and second point of predictor formulae for VSVO-BBDF when 𝑟 = 10/19, and 𝑟 = 1, 𝑞 = 10/19.

Order 𝑦𝑛 𝑦𝑛−1 𝑦𝑛−2 𝑦𝑛−3

29 19
ℎ𝑓𝑛+1 −
𝑃=3 10 10
24 19
ℎ𝑓𝑛+2 −
5 5
1131 741 551
ℎ𝑓𝑛+1 − −
𝑃=4 200 100 200
348 551 228
ℎ𝑓𝑛+2 25
25 25
67 201 67 67
ℎ𝑓𝑛+1 − −
𝑃=5 16 29 10 32
43 688 129 6859
ℎ𝑓𝑛+2 − −
4 29 5 580

Table 4: Coefficients for the first and second point of VSVO-BBDF when 𝑟 = 1, and 𝑟 = 1, 𝑞 = 1.

Order 𝛼𝑘,0 𝛼𝑘,1 𝛼𝑘,2 𝛼𝑘,3 𝛼𝑘,4 𝛼𝑘,5

1 1 1
ℎ𝑓𝑛+1 −1
𝑃=3 6 2 3
1 3 11
ℎ𝑓𝑛+2 − −3
3 2 6
1 1 3 5 1
ℎ𝑓𝑛+1 − −
𝑃=4 12 2 2 6 4
1 4 25
ℎ𝑓𝑛+2 − 3 −4
4 3 12
1 1 7 1 5 5
ℎ𝑓𝑛+1 − − −
𝑃=5 12 2 6 3 4 6
5 61 107 77 15
ℎ𝑓𝑛+2 − −13 −
6 12 6 6 4

𝑙, 𝑚 = 1, 2, . . . , 𝑟. Since VSVO-BBDF for variable order (𝑃) The general form of VSVO-BBDF method is
is a block method, we extend the Definition 2 in the form of
𝑗 𝑦𝑛+1 = 𝛼1 ℎ𝑓𝑛+1 + 𝜃1 𝑦𝑛+2 + 𝜓1 ,
󸀠 (15)
𝐿 [𝑧 (𝑥) ; ℎ] = ∑ [𝐴 𝑘 𝑧 (𝑥 + 𝑘ℎ) − ℎ𝐵𝑘 𝑧 (𝑥 + 𝑘ℎ)] . (12) 𝑦𝑛+2 = 𝛼1 ℎ𝑓𝑛+2 + 𝜃1 𝑦𝑛+1 + 𝜓2 ,
𝑘=0
with 𝜓1 and 𝜓2 are the back values. By setting,
And the general form for the constant 𝐶𝑞 is defined as
1 0 𝑦 𝛼1 0
𝑗
1 𝐼=[ ], 𝑦𝑛+1,𝑛+2 = [ 𝑛+1 ] , 𝐵=[ ],
𝐶𝑞 = ∑ [𝑘𝑞 𝐴 𝑘 − 𝑘𝑞−1 𝐵𝑘 ] , 𝑞 = 2, 3, . . . 𝑝 + 1. 0 1 𝑦𝑛+2 0 𝛼2
𝑘=0
(𝑞 − 1)!
𝑓𝑛+1 𝜓1
(13) 𝐹𝑛+1,𝑛+2 = [ ], 𝜉𝑛+1,𝑛+2 = [ ].
𝑓𝑛+2 𝜓2
𝐴 𝑘 is equal to the coefficients of 𝑦𝑘 , where 𝑘 = 𝑛 = (𝑃 − (16)
2), . . . , 𝑛 + 1, 𝑛 + 2 and 𝑃 = 3, 4, 5.
Equation (15) in matrix-vector form is equivalent to
4. Implementation of VSVO-BBDF Method
(𝐼 − 𝐴) 𝑦𝑛+1,𝑛+2 = ℎ𝐵𝐹𝑛+1,𝑛+2 + 𝜉𝑛+1,𝑛+2 . (17)
Throughout this section, we illustrate the effect of Newton-
type scheme which in a general form of Equation (17) is simplified as
−1
(𝑖+1) (𝑖) 𝜕𝐹 (𝑖) 𝑓̂𝑛+1,𝑛+2 = (𝐼 − 𝐴) 𝑦𝑛+1,𝑛+2 − ℎ𝐵𝐹𝑛+1,𝑛+2 − 𝜉𝑛+1,𝑛+2 = 0. (18)
𝑦𝑛+1,𝑛+2 − 𝑦𝑛+1,𝑛+2 = −[(𝐼 − 𝐴) − ℎ𝐵 (𝑦𝑛+1,𝑛+2 )]
𝜕𝑦
(𝑖) (𝑖)
Newton iteration is performed to the system 𝑓̂𝑛+1,𝑛+2 = 0;
×[(𝐼 − 𝐴) 𝑦𝑛+1,𝑛+2 −ℎ𝐵𝐹 (𝑦𝑛+1,𝑛+2 )−𝜉] . by taking the analogous form of (14) where 𝐽𝑛+1,𝑛+2 =
(14) (𝑖)
(𝜕𝐹/𝜕𝑌)(𝑌𝑛+1,𝑛+2 ) is the Jacobian matrix of 𝐹 with respect
Mathematical Problems in Engineering 5

Table 5: Coefficients for the first and second point of VSVO-BBDF when 𝑟 = 2, and 𝑟 = 2, 𝑞 = 2.

Order 𝛼̂𝑘,0 𝛼̂𝑘,1 𝛼̂𝑘,2 𝛼̂𝑘,3 𝛼̂𝑘,4 𝛼̂𝑘,5

1 3 1 3
ℎ𝑓𝑛+1 −
𝑃=3 24 4 3 8
1 8 7
ℎ𝑓𝑛+2 − 1 −
12 3 4
1 5 15 8 5
ℎ𝑓𝑛+1 − −
𝑃=4 80 48 16 15 16
1 1 3 16 23
ℎ𝑓𝑛+2 − −
30 4 2 5 12
1 1 1 17 12 11
ℎ𝑓𝑛+1 − −
𝑃=5 168 24 8 24 7 12
23 1 43 37 160 45
ℎ𝑓𝑛+2 − − −
336 2 24 6 21 16

Table 6: Coefficients for the first and second point of VSVO-BBDF when 𝑟 = 10/19, and 𝑟 = 1, 𝑞 = 10/19.

Order 𝛼𝑘,0 𝛼𝑘,1 𝛼𝑘,2 𝛼𝑘,3 𝛼𝑘,4 𝛼𝑘,5

6859 29 19 29
ℎ𝑓𝑛+1 −
𝑃=3 13920 20 29 96
6859 12 96 91
ℎ𝑓𝑛+2 − −
6960 5 29 48
6854 89167 1131 1292 13
ℎ𝑓𝑛+1 − −
𝑃=4 15600 46400 400 1131 64
13718 6859 174 64 3095
ℎ𝑓𝑛+2 − −
9425 1200 25 13 1392
2476099 7 124 47 80 851
ℎ𝑓𝑛+1 − − −
𝑃=5 8020704 8 87 96 67 1032
12380495 53 452 931 2692 165
ℎ𝑓𝑛+2 − − −
4010352 6 29 48 201 43

(𝑖+1)
to 𝑌. Equation (14) is separated to three different matrices Step 2. Calculate the corrected value for 𝑦𝑛+1,𝑛+2 with the
denoted as (𝑖+1)
value 𝑒𝑛+1,𝑛+2 from (1).
(𝑖+1) (𝑖+1) (𝑖)
𝐸1,2 = 𝑦𝑛+1,𝑛+2 − 𝑦𝑛+1,𝑛+2 , (𝑖+1) ̂−1 𝐵̂ for the second iteration.
Step 3. Solve 𝑒𝑛+1,𝑛+2 =𝐴
̂ = (𝐼 − 𝐴) − ℎ𝐵 𝜕𝐹 (𝑦(𝑖)
𝐴 ), (19) (𝑖+1)
𝜕𝑌 𝑛+1,𝑛+2 Step 4. Let 𝑦𝑛+1,𝑛+2 be equal to the second iteration of
(𝑖+1)
𝑒𝑛+1,𝑛+2 =𝐴 ̂
̂−1 𝐵.
𝐵̂ = (𝐼 − 𝐴) 𝑦𝑛+1,𝑛+2
(𝑖) (𝑖)
− ℎ𝐵𝐹 (𝑦𝑛+1,𝑛+2 ) − 𝜉𝑛+1,𝑛+2 . The error is defined as
󵄨 󵄨
Two-stage Newton iteration works to find the approxi- error = 󵄨󵄨󵄨󵄨𝑦approximate − 𝑦actual 󵄨󵄨󵄨󵄨 , (21)
mating solution to (1) with two simplified strategies based on
evaluating the Jacobian (𝐽𝑛+1,𝑛+2 ) and LU factorization of 𝐴̂ and the maximum error is defined as
[16]. Two-stage Newton iteration is carried out as follows.
MAXE = ( max (max (error))) , (22)
1<𝑖<TS 1<𝑖<𝑛
(𝑖+1)
Step 1. Compute the values for 𝑒𝑛+1,𝑛+2 =𝐴 ̂ where
̂−1 𝐵,
where TS is the total steps and 𝑛 is the number of equations.
𝜕𝐹𝑛+1 Consequently, the rest of the method is carried out as the
[1 − 𝛼1 ℎ 𝜕𝑦 𝜃1 ] appendix.
̂= [
𝐴 ]
𝜕𝐹𝑛+2 ] ,
𝑛+1
[
𝜃2 1 − 𝛼2 ℎ
[ 𝜕𝑦𝑛+2 ] (20) 5. Stability Conditions for
(𝑖) (𝑖) (𝑖) VSVO-BBDF Method
𝑦𝑛+1 + 𝛼1 ℎ𝐹𝑛+1 + 𝜉1 𝜃1 𝑦𝑛+2
𝐵̂ = [ ].
To begin this section, we provide some definitions for the
(𝑖) (𝑖) (𝑖)
[ 𝜃2 𝑦𝑛+1 𝑦𝑛+2 + 𝛼2 ℎ𝐹𝑛+2 + 𝜉2 ] stability conditions of VSVO-BBDF method as in (15).
6 Mathematical Problems in Engineering

Table 7: Roots for different step sizes and orders.

Roots for different step size 𝑃=3 𝑃=4 𝑃=5

𝑡= 𝑡 = −0.24414201370, 𝑡 = −0.9671746307𝑒 − 1,
𝑟 = 1, and
−0.4347826087𝑒 − 1, 𝑡 = 0.02079175991, 𝑡 = 0.01195305716,
𝑟 = 1, 𝑞 = 1
and 𝑡 = 1
and 𝑡 = 1 and 𝑡 = 1
𝑡 = −0.052708171410, 𝑡 = −0.06178960900,
𝑟 = 2, and 𝑡 = −0.01315789474,
𝑟 = 2, 𝑞 = 2 and 𝑡 = 1 𝑡 = 0.003257621961, 𝑡 = −0.001737052270,
and 𝑡 = 1 and 𝑡 = 1
10 𝑡 = −0.93455113330, 𝑡 = −0.4375589404,
𝑟= , and 𝑡 = −0.1098846524,
19
10 and 𝑡 = −1 𝑡 = 0.08581158625, 𝑡 = −0.007529578675,
𝑟 = 1, 𝑞 =
19 and 𝑡 = 1 and 𝑡 = 1

Definition 3. A method is said to be zero stable if the roots of Meanwhile changing the order of the method is designed for
the polynomial 𝜌(𝑧) = 𝜋(𝑧, 0) satisfy the condition 𝑧1 = 1 ≤ finding the best approximation. The essential components of
|𝑧2 |, 𝑧2 ≠ 1. VSVO-BBDF algorithm are the local truncation error (LTE),
strategies for deciding when to change step size and order,
Definition 4. A method is said to be absolutely stable in a and techniques for changing the step size and order. Strategies
region 𝑅 for a given ℎ𝜆 if for that ℎ𝜆 all the roots 𝑟𝑠 of the proposed in [17] are applied in this study for choosing the
stability polynomial 𝜋(𝑟, ℎ𝜆) = 𝜌(𝑟) − ℎ𝜆𝜎(𝑟) = 0 satisfy step size and order. The strategy is to estimate the maximum
|𝑟𝑠 | < 1, 𝑠 = 1, 2, . . . , 𝑘. step size for the following step. Methods of order 𝑃 − 1, 𝑃,
and 𝑃 + 1 are selected depending on the occurrence of
Definition 5. A method is said to be stiffly stable if R1 and R2 every successful step. Consequently, the new step size ℎnew is
are contained in the absolute stability region, and it is accurate obtained from which order produces the maximum step size.
for all ℎ ∈ R2 when applied to the scalar test equation 𝑦󸀠 = The user initially will have to provide an error tolerance
𝜆𝑦; 𝜆 is a complex constant with 𝑅𝜆 < 0, where R1 = {ℎ𝜆 | limit, TOL, on any given step and obtain the LTE for each
𝑅ℎ𝜆 < −𝑡}, R2 = {ℎ𝜆 | −𝑡 ≤ 𝑅ℎ𝜆 ≤ 𝑢, −𝑣 ≤ ∐ ℎ𝜆 < 𝑣}, and iteration. The LTE is obtained from
𝑡, 𝑢, and 𝑣 are positive constants.
(𝑃+1) (𝑃)
LTE𝑘 = 𝑦𝑛+2 − 𝑦𝑛+2 , 𝑃 = 3, 4, 5, (23)
The stability polynomial, 𝑅(𝑡, ̂ℎ) associated with the
method of (15), is given by det(𝐴𝑡2 −𝐵𝑡−𝐶), while the absolute (𝑃+1)
where 𝑦𝑛+2 (𝑃)
is the (𝑃 + 1)th order method and 𝑦𝑛+2 is the 𝑘th
stability region of this method in the ℎ𝜆 plane is determined order method. By finding the LTEs, the maximum step size is
by solving det(𝐴𝑡2 − 𝐵𝑡 − 𝐶 = 0). Consequently, to determine defined as
zero stable, we substitute ̂ℎ = ℎ𝜆 = 0 to 𝑅(𝑡, ̂ℎ) for each order
of VSVO-BBDF method. TOL 1/𝑃
Hence, the roots for the three different step size and order ℎ𝑃−1 = ℎold × ( ) ,
LTE𝑃−1
(𝑃) selections obtained by using Maple are listed in Table 7.
The stability region was given by the set of points deter- TOL 1/(𝑃+1)
mined by the boundary 𝑡 = 𝑒𝑖𝜃 , 0 ≤ 𝜃 ≤ 2𝜋. We obtained the ℎ𝑃 = ℎold × ( ) , (24)
LTE𝑃
stability region by finding the region for which |𝑡| < 1. Since
all of the roots in Table 7 have modulus less than or equal to TOL 1/(𝑃+2)
1, the method (15) is said to be zero stable. Figure 2 shows the ℎ𝑃+1 = ℎold × ( ) ,
LTE𝑃+1
stability for orders 𝑃3, 𝑃4, and 𝑃5 of VSVO-BBDF method,
respectively. The stability regions lie outside the closed region where ℎold is the step size from previous block and ℎmax is
for each case. obtained from the maximum step size given in the above
Based on Figure 2, VSVO-BBDF method possesses the equations.
region absolute stability, which contains almost the whole of There are 3 cases of the possibilities in choosing the step
the half-plane Re(ℎ𝜆) < 0. size.

6. Choosing Order and Step Size Case 1. From order 3 to order 4 and from order 4 to order 3,
the step size is allowed to increase, decrease, or be maintained,
To increase the efficiency in BBDF algorithm, VSVO-BBDF that is, (𝑟 = 1), (𝑟 = 2) or (𝑟 = 10/19).
algorithm is designed to have the capacity to vary not only
the step size, but also the order of the method employed. Case 2. From order 3 to order 5 and from order 4 to order 5,
The importance of choosing the step size is to achieve the step size is allowed to decrease, or be maintained, that is,
reduction in computation time and number of iterations. (𝑟 = 1) (𝑞 = 1), (𝑟 = 2) (𝑞 = 2), or (𝑟 = 1) (𝑞 = 10/19).
Mathematical Problems in Engineering 7

3
8
r = 10/19
2 6
r=1 r = 10/9
4
1
2 r=1
r=2
r=2

Im
0 0
Im

−2
−1
−4

−2 −6

−8
−3

0 1 2 3 4 −2 0 2 4 6 8 10 12
Re Re
(a) (b)
8

4 r = 2, q = 2

r = 1 q = 10/19
2
r = 1, q = 1
0
Im

−2

−4

−6

−8
0 2 4 6 8 10 12
Re
(c)
Figure 2: Stability regions for order (a) 𝑃3, (b) 𝑃4, and (c) 𝑃5.

Case 3. From order 5 to order 3 and from order 5 to order 4, 7. Numerical Results
the step size is allowed to decrease, or be maintained, that is,
(𝑟 = 1) (𝑟 = 2) or (𝑟 = 10/19). We carry out numerical experiments arising from problems
in physics to compare the performance of VSVO-BBDF
The successful step is dependent on the condition LTE < method with stiff ODE solvers in MATLAB version 7 in
TOL. If this condition fails, the values of 𝑦𝑛+1 , 𝑦𝑛+2 are Windows XP or later. The syntax for using ode15s and ode23s
rejected, and the current step is reiterated with step size is similar for every test problem. For example,
selection (𝑞 = 2). On the contrary, the step size increment [𝑋, 𝑌] = ode15s (“odefun”, int, init, options) , (25)
for each successful step is defined as
where odefun is a function stored that evaluates the right
side of the differential equations, int is a vector of the
ℎnew = 𝑐 × ℎmax and if integration interval, [𝑥𝑜 , 𝑥end ], init is a column vector of
initial conditions, [𝑦0 ], and options is an adjustable format
ℎnew > 1.9 × ℎold then ℎnew = 1.9 × ℎold , of optional parameters to change the default integration
properties.
We created the above option structure to vary the relative
where 𝑐 is the safety factor, 𝑝 is the order of the method while tolerance (RelTol) and absolute tolerance (AbsTol) according
ℎold and ℎnew is the step size from previous and current block, to specific error tolerance. These test problems are performed
respectively. In this paper, 𝑐 is set to be 0.8 so as to make sure under different conditions of error tolerances—(a) 10−2 , (b)
the rejected step is being reduced. 10−4 , and (c) 10−6 .
8 Mathematical Problems in Engineering

Table 8: Numerical results for Problem 1.

TOL MTD TS AVEE MAXE TIME
VSVO-BBDF 21 2.9370𝑒 − 005 2.8298𝑒 − 004 0.0103
10−2 ode15s 28 1.3000𝑒 − 003 8.4000𝑒 − 003 0.0313
ode23s 19 1.0000𝑒 − 003 4.5000𝑒 − 003 0.1406
VSVO-BBDF 48 1.0716𝑒 − 006 3.2212𝑒 − 006 0.0105
10−4 ode15s 60 3.0358𝑒 − 005 1.6621𝑒 − 004 0.0156
ode32s 42 1.1285𝑒 − 004 2.5683𝑒 − 004 0.0313
VSVO-BBDF 164 1.6733𝑒 − 008 3.1232𝑒 − 008 0.0115
10−6 ode15s 100 7.2564𝑒 − 007 2.7506𝑒 − 006 0.0313
ode23s 143 7.3558𝑒 − 006 1.2514𝑒 − 005 0.0469

Table 9: Numerical results for Problem 2.

TOL MTD STs AVEE MAXE TIME
VSVO-BBDF 22 7.1459𝑒 − 005 2.5736𝑒 − 004 0.0106
10−2 ode15s 29 9.0287𝑒 − 004 5.2000𝑒 − 003 0.0781
ode23s 25 3.3626𝑒 − 004 1.1000𝑒 − 003 0.0781
VSVO-BBDF 54 7.4173𝑒 − 006 3.7659𝑒 − 004 0.0111
10−4 ode15s 55 1.7139𝑒 − 005 8.5506𝑒 − 005 0.1250
ode32s 118 1.3783𝑒 − 005 6.9774𝑒 − 005 0.2031
VSVO-BBDF 194 6.3429𝑒 − 009 3.2882𝑒 − 008 0.0143
10−6 ode15s 197 2.1320𝑒 − 007 1.0790𝑒 − 006 0.2344
ode23s 773 3.3163𝑒 − 007 2.8081𝑒 − 006 0.4531

The test problems and solutions are listed below. This paper considers the comparison of four different fac-
tors, namely, number of steps taken, average error, maximum
Problem 1. Consider, error, and computation time. From Table 8, among the three
𝑦󸀠 = −100 (𝑦 − 𝑥) + 1, 𝑦 (0) = 1, 0 ≤ 𝑥 ≤ 10, (26) methods tested, our method, VSVO-BBDF method, requires
the shortest execution time, smallest maximum error, and
with solution average error for each given tolerance level. From Figure 3,
𝑦 (𝑥) = 𝑒−100𝑥 + 𝑥. (27) we can see more clearly that VSVO-BBDF gives the lowest
maximum error for every tolerance level.
Problem 2. Consider, Again by comparing the four factors mentioned earlier,
𝑦1󸀠 = −1002𝑦1 + 1000𝑦2 2 , 𝑦1 (0) = 1, 0 ≤ 𝑥 ≤ 10, we can see VSVO-BBDF in Table 9 gives the minimum
(28) maximum error for every tolerance level except for TOL 10𝑒−
𝑦2󸀠 = 𝑦1 − 𝑦2 (1 + 𝑦2 ) , 𝑦2 (0) = 1, 4. However, our method prevails in terms of average error
for each given tolerance level. VSVO-BBDF once again
with solution
requires the shortest execution time for each given tolerance
𝑦1 = 𝑒−2𝑥 , level. Figure 4 illustrates the efficiency of VSVO-BBDF as
(29) compared to ode15s and ode23s.
𝑦2 = 𝑒−𝑥 . From Table 10, VSVO-BBDF method gives the shortest
See Kaps [18]. execution time for every given tolerance level. While in
terms of average error and maximum error, VSVO-BBDF
Problem 3. Consider, method once again gives the best result as compared to ode15s
and ode23s. Figure 5 clarifies the efficiency of the proposed
𝑦1󸀠 = −2𝑦1 + 𝑦2 + 2 sin (𝑥) , 𝑦1 (0) = 2, 0 ≤ 𝑥 ≤ 10,
method based on its total steps and tolerance level.
𝑦2󸀠 = 998𝑦1 − 999𝑦2 + 999 (cos (𝑥) − sin (𝑥)) , 𝑦2 (0) = 3,
(30) 8. Conclusion
with solution
The objective is met when VSVO-BBDF method outper-
𝑦1 (𝑥) = 2𝑒−𝑥 + sin (𝑥) , formed ode15s and ode23s in terms of average error as well
(31) as maximum error. In most of the cases, VSVO-BBDF has
𝑦2 (𝑥) = 2𝑒−𝑥 + cos (𝑥) .
successfully managed to reduce the number of total steps
See Lambert [15]. taken. As for the computation time wise, it gave lesser values
Mathematical Problems in Engineering 9

Table 10: Numerical results for Problem 3.

TOL MTD TSs AVEE MAXE TIME
VSVO-BBDF 35 4.6584𝑒 − 005 3.0045𝑒 − 004 0.0111
10−2 ode15s 45 6.9000𝑒 − 003 1.4620𝑒 − 002 0.0156
ode23s 137 4.0745𝑒 − 004 3.0000𝑒 − 002 0.2031
VSVO-BBDF 84 2.5775𝑒 − 006 1.1002𝑒 − 005 0.3281
10−4 ode15s 93 7.6548𝑒 − 005 2.7591𝑒 − 004 0.0469
ode32s 1211 2.4915𝑒 − 005 6.3075𝑒 − 005 0.0156
VSVO-BBDF 380 2.4244𝑒 − 008 8.9627𝑒 − 008 0.0199
10−6 ode15s 186 2.1681𝑒 − 006 6.1936𝑒 − 006 0.0781
ode23s 2829 5.3155𝑒 − 007 1.5667𝑒 − 006 2.3281

−2 −2

−3 −3

−4 −4
log |max error|
log |max error|

−5 −5

−6 −6

−7 −7

−8 −8
0 20 40 60 80 100 120 140 160 180 10e−2 10e−4 10e−6
TOL

VSVO-BBDF VSVO-BBDF
ode15s ode15s
ode23s ode23s
(a) (b)

Figure 3: Efficiency curves for Problem 1.

for all cases. Therefore, we can conclude that VSVO-BBDF The algorithm for VSVO-BBDF code is given as follows.
can serve as an alternative solver for solving stiff ordinary
differential equations which arise in engineering and applied Step 1. Let order = 3.
sciences.
Step 2. Calculate initial step size (ℎ) and 𝜀.
Appendix Step 3. Calculate initial array 𝑦1 , . . . 𝑦𝑖 , 𝑖 = order − 1.
The following notation is used in the algorithm of VSVO-
BBDF: Step 4. Calculate the predictor values for 𝑦𝑖+1 , 𝑦𝑖+2 , 𝑖 =
ℎ: step size, order − 1.
𝜀: tolerance, Step 5. Calculate 𝐹𝑖 for 𝑖 = 1, . . . , order + 1.
TS: total steps,
Step 6. Calculate the corrector values for 𝑦𝑖+1 , 𝑦𝑖+2 , 𝑖 =
Order: order of the method,
order − 1.
JCBN: jacobian,
E: increment. Step 7. Calculate 𝐹𝑖 for 𝑖 = 1, . . . , order + 1.
10 Mathematical Problems in Engineering

−2 −2

−3 −3

−4 −4

log |max error|

log |max error|

−5 −5

−6 −6

−7 −7

−8 −8
0 100 200 300 400 500 600 700 800 10e−2 10e−4 10e−6
Total steps TOL

VSVO-BBDF VSVO-BBDF
ode15s ode15s
ode23s ode23s
(a) (b)

Figure 4: Efficiency curves for Problem 2.

−1 −1

−2 −2

−3 −3

−4 −4
log |max error|
log |max error|

−5 −5

−6 −6

−7 −7

−8 −8
0 500 1000 1500 2000 2500 3000 10e−2 10e−4 10e−6
Total steps TOL

VSVO-BBDF VSVO-BBDF
ode15s ode15s
ode23s ode23s
(a) (b)

Figure 5: Efficiency curves for Problem 3.

Mathematical Problems in Engineering 11

Implementation of VSVO-BBDF. References

Step 8. Calculate Jacobian (JACBN). [1] W. H. Enright, T. E. Hull, and B. Lindberg, “Comparing numer-
ical methods for stiff systems of O.D.E:s,” BIT, vol. 15, no. 1, pp.
Step 9. Perform LU-Decomposition to calculate the incre- 10–48, 1975.
ment 𝐸. [2] G. D. Byrne, A. C. Hindmarsh, K. R. Jackson, and H. G. Brown,
Consider, “A comparison of two ode codes: gear and episode,” Computers
and Chemical Engineering, vol. 1, no. 2, pp. 125–131, 1977.
(𝑖+1) (𝑖+1) (𝑖+1)
𝐸1,2 = 𝑦𝑛+1,𝑛+2 − 𝑦𝑛+1,𝑛+2 , 𝑛 = order − 1. (A.1) [3] R. Sacks-Davis, “Fixed leading coefficient implementation of
sd-formulas for stiff ODES,” ACM Transactions on Mathemat-
Test for convergence. ical Software, vol. 6, no. 4, pp. 540–562, 1980.
[4] A. S. Mahmood, L. Casasús, and W. Al-Hayani, “The decom-
Step 10. If 𝐸 > 𝜀, convergence is false. Else go to Step 11. position method for stiff systems of ordinary differential equa-
tions,” Applied Mathematics and Computation, vol. 167, no. 2, pp.
Step 11. Calculate LTE𝑖 , 𝑖 = order. 964–975, 2005.
[5] J. Ibáñez, V. Hernández, E. Arias, and P. A. Ruiz, “Solving
Change order. initial value problems for ordinary differential equations by two
approaches: BDF and piecewise-linearized methods,” Computer
Step 12. Let order equal 4 and order equal 5. Physics Communications, vol. 180, no. 5, pp. 712–723, 2009.
[6] H. Aminikhah and J. Biazar, “A new analytical method for
Step 13. Repeat from Step 2 until Step 11 for each order system of ODEs,” Numerical Methods for Partial Differential
correspondingly. Equations, vol. 26, no. 5, pp. 1115–1124, 2010.
[7] S. Abelman and K. C. Patidar, “Comparison of some recent
Step size and order control. numerical methods for initial-value problems for stiff ordinary
differential equations,” Computers and Mathematics with Appli-
Step 14. Calculate ℎmax for choosing the new step size. cations, vol. 55, no. 4, pp. 733–744, 2008.
Consider, [8] L. F. Shampine and M. W. Reichelt, “The MATLAB ode suite,”
SIAM Journal on Scientific Computing, vol. 18, no. 1, pp. 1–22,
1/(𝑘+1) 1997.
TOL
ℎ𝑘 = ℎold × ( ) . (A.2) [9] L. F. Shampine, M. W. Reichelt, and J. A. Kierzenka, “Solving
LTE2,𝑘
index-I DAEs in MATLAB and simulink,” SIAM Review, vol.
41, no. 3, pp. 538–552, 1999.
The final step size after a successful step (LTE < 𝜀) is given
[10] T. Kato and K. Ikeuchi, “Variable order and variable step-
by
size integration method for transient analysis programs,” IEEE
Transactions on Power Systems, vol. 6, no. 2, pp. 206–213, 1991.
TOL 1/𝑝
ℎnew = 𝑐 × ℎold × ( ) . (A.3) [11] Z. B. Ibrahim, M. Suleiman, and K. I. Othman, “Fixed coeffi-
LTE cients block backward differentiation formulas for the numer-
ical solution of stiff ordinary differential equations,” European
The final step size after a failure step (LTE > 𝜀) is given by Journal of Scientific Research, vol. 21, no. 3, pp. 508–520, 2008.
[12] Z. B. Ibrahim, K. I. Othman, and M. B. Suleiman, “Variable
1
ℎnew ≥ × ℎold . (A.4) step size block backward differentiation formula for solving stiff
2 odes,” in Proceedings of the World Congress on Engineering, vol.
2, pp. 785–789, London, UK, 2007.
Step 15. If ℎnew ≥ 1.9 × ℎold then ℎnew = 1.9 × ℎold .
[13] S. A. M. Yatim, Z. B. Ibrahim, K. I. Othman, and F. Ismail, “Fifth
Else ℎnew = ℎold . order variable step block backward differentiation formulae for
solving stiff ODEs,” Proceedings of World Academy of Science,
Step 16. Update values: Engineering and Technology, vol. 62, pp. 998–1000, 2010.
[14] S. A. M. Yatim, Z. B. Ibrahim, K. I. Othman, and M. B. Suleiman,
𝑦𝑖 = 𝑦𝑖+2 , 𝑥𝑖 = 𝑥𝑖+2 for 𝑖 = 1, . . . , order − 1. (A.5) “Quantitative comparison of numerical method for solving
stiff ordinary differential equations,” Mathematical Problems in
Step 17. Repeat Step 1 to Step 16 until ℎ = end point. Engineering, vol. 2011, Article ID 193691, 12 pages, 2011.
[15] J. D. Lambert, Computational Methods in Ordinary Differential
The Abbreviations Used in Figures 3–5 Equation, John Wiley & Sons, New York, NY, USA, 1991.
and Tables 8–10 [16] K. R. Jackson, “The numerical solution of large systems of stiff
IVPs for ODEs,” Applied Numerical Mathematics, vol. 20, no. 1-
TSs: The total number of steps taken 2, pp. 5–20, 1996.
TOL: The initial value for the local error estimate [17] G. Hall and J. M. Watt, Modern Numerical Methods for Ordinary
MAXE: The maximum error Differential Equations, Clarendon Press, Oxford, UK, 1976.
AVEE: The average error [18] P. Kaps and G. Wanner, “A study of Rosenbrock-type methods
MTD: The method used of high order,” Numerische Mathematik, vol. 38, no. 2, pp. 279–
TIME: The total execution time (seconds). 298, 1981.
Advances in Advances in Journal of Journal of
Operations Research
Hindawi Publishing Corporation
Decision Sciences
Hindawi Publishing Corporation
Applied Mathematics
Hindawi Publishing Corporation
Algebra
Hindawi Publishing Corporation
Probability and Statistics
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

The Scientific International Journal of

World Journal
Hindawi Publishing Corporation
Differential Equations
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Submit your manuscripts at

http://www.hindawi.com

International Journal of Advances in

Combinatorics
Hindawi Publishing Corporation
Mathematical Physics
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Journal of Journal of Mathematical Problems Abstract and Discrete Dynamics in

Complex Analysis
Hindawi Publishing Corporation
Mathematics
Hindawi Publishing Corporation
in Engineering
Hindawi Publishing Corporation
Applied Analysis
Hindawi Publishing Corporation
Nature and Society
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

International
Journal of Journal of
Mathematics and
Mathematical
Discrete Mathematics
Sciences

Journal of International Journal of Journal of

Hindawi Publishing Corporation Hindawi Publishing Corporation Volume 2014

Function Spaces
Hindawi Publishing Corporation
Stochastic Analysis
Hindawi Publishing Corporation
Optimization
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014