A NEW THREE STEP SECOND DERIVATIVE LINEAR MULTISTEP Supported by

METHODS FOR THE SOLUTIONS OF STIFF INITIAL VALUE PROBLEMS

Samson Omagwu1, Tiwalade Modupe Usman2 and Joshua Kyaharnan Victor3

1Mathematics & Statistics Department, Kaduna Polytechnic, Kaduna State, Nigeria
2Computer Science Department, Kaduna Polytechnic, Kaduna State, Nigeria
3Mathematics Department, University of Jos, Plateau State, Nigeria

Received: January 18, 2019 Accepted: May 29, 2019

Abstract: This paper is concerned with the accuracy and efficiency of a three step second derivative linear multistep method
for the approximate solution of stiff initial value problems. The main methods were derived by blending of two
linear multistep methods using continuous collocation approach. These methods are of uniform order four. The
stability analysis of the block methods indicates that the methods are A–stable, consistent and zero stable hence
convergent. Numerical results obtained using the proposed new block methods were compared with those obtained
by the well known ODE solver ODE15S to illustrate its accuracy and effectiveness. The proposed block method is
found to be efficient and accurate hence recommended for the solution of stiff initial value problems.
Keywords: Blended block, linear multistep methods, stiff ODEs, continuous collocation

Introduction Method was used (Kumleng et al., 2013) and where ten step
In this research paper, the construction and application of a block generalized Adams method was used. Many has
three step order four blended block linear multistep method discussed the solution of linear and non linear ODEs from
for the numerical solutions of stiff initial value problems (1) different basis functions, among them are Onumanyi et al
was considered. A potentially good numerical method for the (1994), Sirisena et al. (2004), Kumleng (2012) and so on.
solution of stiff system of ordinary differential equations The three step blended linear multistep method
(ODEs) must have good accuracy and some wide region of The three step blended linear multistep method is constructed
absolute stability. One of the first and most important stability based on the continuous finite difference approximation
requirements for linear multistep methods is A-stability as approach using the interpolation and collocation criteria
proposed by Enright (1974). The three step blended block described by Lie and Norsett (1981) called multistep
linear multistep methods is A-stable hence the application of collocation (MC) and block multistep methods by Onumanyi
the method here which makes it suitable for the solution of et al. (1994, 1999). We define based on the interpolation and
linear and non linear ODEs. collocation methods the continuous form of the k- step second
The solution of stiff system of ODEs has been considered by derivative new method as;
Chollom et al. (2011) where a block hybrid Adams Moulton

(6)
are the continuous coefficients of the method, m is the number of distinct collocation points, h is the step size and from
Onumanyi et al. (1994), we obtain our matrices D and C = D–1 by the imposed conditions expressed as DC = I; where:

FUW Trends in Science & Technology Journal, www.ftstjournal.com

e-ISSN: 24085162; p-ISSN: 20485170; August, 2019: Vol. 4 No. 2 pp. 458 – 461 458
 Three Step Second Derivative Linear Multistep Method for the Approximate Solution……

,
respectively.
In this case, k=3,t=1 and m=5 and it continuous form expressed in the form of (6) is
m- 1
y ( x) = a 2 ( x ) yn + 2 + h å b j ( x ) f n + j 2
+ h l ( x ) y ''
3 n+ 3 (9)
j= 0

y( xn+ 9 ) = a 2 ( x) yn+ 2 + h éëb 0 ( x) f n + b1 ( x) f n+ 1 + b 2 ( x) f n+ 2 + b 3 ( x) f n+ 3 ùû+ h 2l 3 ( x) y ' n+ 3

% (10)
Using the approach of Onumanyi et al. (1999); the matrix form of
1 (xn + 2h) (xn + 2h)2 (xn + 2h)3 (xn + 2h)4 (xn + 2h)5
2
0 1 2xn 3xn 4xn 3 5xn 4
2
0 1 2(xn + h) 3(xn + h) 4(xn + h)3 5(xn + h)4
D= 2 (11)
0 1 2(xn + 2h) 3(xn + 2h) 4(xn + 2h)3 5(xn + 2h)4
2
0 1 2(xn + 3h) 3(xn + 3h) 4(xn + 3h)3 5(xn + 3h)4
[0 0 2 6(xn + 3h) 12(xn + 3h)2 20(xn + 3h)3 ]

Using the Maple software, the inverse of the matrix in (11) is Evaluating the continuous scheme (12) at t = 0, h, 3h gives
obtained and its elements are used in obtaining the continuous the three discrete methods which constitute the three step
coefficients and substituting these continuous coefficients into blended block linear multistep method.
(9) yields the continuous form of our new method. The 1
continuous form as: 𝑦𝑛 = 𝑦𝑛+2 − {43ℎ𝑓𝑛 + 189ℎ𝑓𝑛+1 + 27ℎ𝑓𝑛+2 11ℎ𝑓𝑛+3
135
13𝜏 2 29𝜏 3 𝜏4 𝜏5 ′′
43ℎ − 3ℎ2 𝑦𝑛+3 }
𝑦̃(𝜏 + 𝑥𝑛 ) = 𝑦𝑛+2 + (𝜏 − + − 3+ 4− )𝑓
12ℎ 54ℎ 2 8ℎ 8ℎ 135 𝑛 1
9𝜏2 7𝜏3 𝜏4 𝜏5 7ℎ 𝑦𝑛+1 = 𝑦𝑛+2 − {23ℎ𝑓𝑛 − 486ℎ𝑓𝑛+1 − 783ℎ𝑓𝑛+2
+( − + − − ) 𝑓𝑛+1 1080
4ℎ2 2ℎ3 20ℎ4 ′′
4ℎ
9𝜏2 5𝜏3 7𝜏4 𝜏5
5
ℎ
+ 249𝑓𝑛+3 − 66ℎ2 𝑦𝑛+3 }
+(− + − + − ) 𝑓𝑛+2 𝑦𝑛+3 = 𝑦𝑛+2 +
1
{7ℎ𝑓𝑛 − 54ℎ𝑓𝑛+1 + 513ℎ𝑓𝑛+2 +
4ℎ 2ℎ2 8ℎ3 10ℎ4 5
13𝜏2 139𝜏3 𝜏4 11𝜏5 11ℎ 1080
′′
+(
12ℎ
−
108ℎ 2 + 2ℎ 3 + 180ℎ 4 −
135 𝑛+3
)𝑓 614ℎ𝑓𝑛+3 − 114 ℎ2 𝑦𝑛+3 }(13)
−𝜏2 11𝜏3 𝜏4 𝜏5 2ℎ
+( + − + − ) 𝑔𝑛+3 (12)
2 18ℎ 4ℎ2 30ℎ3 45 Stability analysis of the new methods
𝑔𝑛+3 =𝒚′′𝒏+𝟑 In this section, we consider the analysis of the newly
constructed methods. Their convergence is determined and
their regions of absolute stability plotted.

FUW Trends in Science & Technology Journal, www.ftstjournal.com

e-ISSN: 24085162; p-ISSN: 20485170; August, 2019: Vol. 4 No. 2 pp. 458 – 461 459
 Three Step Second Derivative Linear Multistep Method for the Approximate Solution……

Convergence multistep methods and the well-known Matlab stiff ODE

The convergence of the new block methods is determined solver ODE15S on the same problems and on the same axes.
using the approach by Fatunla (1991) and Chollomet.al
(2007) for linear multistep methods, where the block methods Problem 1: Irregular Heartbeat and Lidocaine Model
are represented in a single block, r point multistep method of The irregular heartbeat and Lidocaine model is expressed
the form mathematically by the following 𝑖𝑣𝑝
𝑦1′ = −0.09 𝑦1 + 0.038𝑦2
𝑦2′ = 0.066 𝑦1 − 0.038𝑦2
𝑦1 (0) = 𝑦2 (0) = 𝑦0
14 y0 = Maximum Safe Dosage = 3mw/kg 3
0  x  700, h  0.1
Zero Stability of the BBLMM for k=3
To determine the zero stability of the BBLMM we use the 3.5

approach of Ehigie (2007) for linear multistep methods where

he expressed the methods in the matrix form as shown below. 3 y1 BBLMM 3
y2 BBLMM 3
Following the work of Ehigie and Okunuga (2014); we y1 ODE 15s

observed that the seven step block method is zero stable as the 2.5
y2 ODE 15s

roots of the equation

2

y axis
are less than or equal to 1. Since the block method is
consistent and zero-stable, the method is convergent (Henrici, 1.5

1962).
These new methods are consistent since their orders are 4, it is 1

also zero-stable, above all, there are A – stable as can be seen

in Fig. 1. The new three step discrete methods that constitute 0.5
the block method have the following orders and error
constants as shown below. 0
The three step blended block multistep methods has uniform 0 100 200 300
x-axis
400 500 600 700

order of (4, 4, 4)T and error constants of 𝐶6 = Fig. 2: Solution curves of Problem 1 solved by the our
−899 458 −236 𝑇
( , , ) new methods
100000 1000000 1000000

Regions of absolute stability of the methods Problem 2: Van der Pol’s Equations
The absolute stability regions of the newly constructed 𝑦1′ = 𝑦2
blended block linear multistep methods (12) was plotted 𝑦2′ = −𝑦1 − 𝜇𝑦2 (1 − 𝑦12 )
using Ehigie (2007). 𝜇 = 300, 𝑦1 (0) = 2, 𝑦2 (0) = 0 0  x  40, h  0.1
Stiffness ratio3 × 102 .
The Van der Pol’s Equation is an important kind of second-
order non-linear auto-oscillatory equation (Fig. 3). It is a non-
conservative oscillator with non-linear damping.
15

y1 Blended 3
y2 Blended 3
y1 ODE 15S
10 y2 ODE 15s

-5

-10

Fig 1: Absolute stability region For BBLMM For K=3 -15

0 5 10 15 20 25 30 35 40

Fig. 3: Solution curves of Problem 2 solved by the our

This absolute stability region is A –stable since it consist of new methods
the set of points in the complex plane outside the enclosed
figure. Conclusion
Numerical examples Problem 1 is a model of the relationship between Lidocaine
We report here a numerical example on stiff problem taken and Irregular Heart beat. Lidocaine belong to a group of
from the literature using the solution curve. In comparison, we drugs known as anti-arrhythmic which work by preventing
also report the performance of the new blended block linear sodium from being pumped out on the cells of the heart to
help the heart beat normally. From our solution curves, it was
observed that normalcy in the heart beat can be attained with

FUW Trends in Science & Technology Journal, www.ftstjournal.com

e-ISSN: 24085162; p-ISSN: 20485170; August, 2019: Vol. 4 No. 2 pp. 458 – 461 460
 Three Step Second Derivative Linear Multistep Method for the Approximate Solution……

the use of Lidocaine within the correct dosage. Our solution Ezzeddine, A.K. and Hojjati, G., (2012). Third derivative
curves coincide with the solutions of ODE 15s. multistep methods for stiff systems. Int. J. Nonlinear
Van der Pol’s equation in Problem 2 is a non conservative Sci., 14: 443-450.
oscillator with non linear damping energy dissipated at high Gamal AF, Ismail K & Iman HI 1999. A new efficient second
amplitude. From the solution curves the trajectories traces the derivative multistep method for stiff system. Applied
motion of a single point through a flow with a limit circle Mathematical Modeling, 23: 279–288.
where the trajectories spiral into or away from the limit circle. Henrici P 1962. Discrete Variable Methods in Ordinary
Our solution curves compete favourably with ODE 15s. Differential Equations (p. 407). John Willey, New York.
It can be seen clearly from the curve that our new methods Holling CS 1959. Some characteristic os simple type of
perform favourably better than the well known ODE15S for predator and pasitism. Canadian Entomologist, 91: 385-
the problems solved in Problems 1 and 2. It was also 398. http:www.hindawi.com/Journals/aaz/2013/127103/
observed that the new methods have better stability regions Kumleng GM, Sirisena UW & Chollon JP 2012. A class of a-
than the conventional Adams Moulton method for step stable order four and six linear multistep methods for
number 3. stiff initial value problems. Mathematical Theory and
Modeling, 3(11): 94–102.
Conflict of Interest Kumleng GM, Sirisena UWW & Dang BC 2013. A ten step
Authors declare that there is no conflict of interest reported on block generalized adams method for the solution of the
this work. Holling Tanner and Lorenz Models. Afr. J. Natural Sci.,
16: 63–70.
References Lie I & Norset R 1989. Super convergence for multistep
Butcher JC 1966. On the convergence of numerical solutions collocation. Mathematics of Computation, 52(185): 65–
to ordinary differential equations. Math. Comp., 20: 1 – 79.
10. Lotka AJ 1925. Element of Physical Biology. Baltimore,
Chollom JP 2005. A study of Block Hybrid Methods with link Williams and Wilkins Company.
of two-step Runge Kutta Methods for first order Mehdizadeh M, Khalsarai N, Nasehi O & Hojjati G 2012. A
Ordinary Differential Equations. PhD Thesis class of second derivative multistep methods for stiff
(Unpublished) University of Jos, Jos Nigeria. systems. Journal of Acta, 15: 209–222.
Chollom JP, Ndam JN & Kumleng GM 2007. Some Onumanyi P, Awoyemi DO, Jator SN & Sirisena UW 1994.
properties of block linear multistep methods. Science New linear multistep methods with continuous
World Journal, 2(3): 11-17. coefficients for first order IVPs. Journal of NMS, 31(1):
Chollom JP, Olatunbusin IO & Omagwu S 2012. A Class of 37–51.
a-stable block explicit methodsfor the solution of Onumanyi P, Sirisena UW & Jator SN 1999. Continuous
ordinary differential equations. Res. J. Math. and Stat., finite difference approximations for solving differential
4(2): 52-56. equations. Int. J. Comp. and Math., 72(1), 15–27.
Ehigie JO, Okunuga SA, Sofoluwe AB & Akanbi MA 2010. Rosenzweng ML & MacArthur RH 1963. Graphical
Generalized 2-step continuous linear multistep method of representation and stability conditions of pradator-prey
hybrid type for the integration of second order ordinary interactions. A.M. Nat., 97: 209-223.
differential equations. Scholars Res. Library (Archives of Sahi RK, Jator SN & Khan NA 2012. A Simpson’s Type
Appl. Sci. Res.), 2(6): 362–372. Second Derivative Method for Stiff Systems. Int. J. Pure
Enright WH 1972. Numerical Solution of Stiff Differential and Appl. Math., 81(4): 619–633.
Equations (pp. 321–331). Dept. of Computer Science, Tanner JT 1975. The stability and intrinsic growth rates of
University of Toronto, Toronto, Canada. prey and predator populations. Ecology, 56: 855-867.
Enright WH 1974. Second derivative multistep methods for Volterra V 1927. Variation and fluctuations in number of
stiff ordinary differential equations. SIAM Journals on coexisting animal species in F. M. Scudo.
Numerical Analysis, 11(2): 376-391.

FUW Trends in Science & Technology Journal, www.ftstjournal.com

e-ISSN: 24085162; p-ISSN: 20485170; August, 2019: Vol. 4 No. 2 pp. 458 – 461 461