An Intemational Joumal

Available online at www.sciencedirect.com computers &

8C I IEN C IE ~___~ D ' I;t IECT " mathematics
with applications
ELSEVIER Computers and Mathematics with Applications 49 (2005) 1295-1301
www.elsevier.com/locate/camwa

Stability of 0-Methods for

Advanced Differential Equations with
Piecewise Continuous Arguments
M. H. SONG
Department of Mathematics, Harbin Institute of Technology
Harbin, 150001, P.R. China
and
Institute of Computational Mathematics and Scientific/Engineering Computing
Chinese Academy of Sciences, Beijing 100080, P.R. China
Z. W . YANG AND M . Z. LIU*
Department of Mathematics, Harbin Institute of Technology
Harbin, 150001, P.R. China
mzliuhope, hit. edu. cn

(Received August 2003; revised and accepted February 2005)

Abstract--This paper deals with the stability analysis of numerical methods for the solution of
advanced differential equations with piecewise continuous arguments. We focus on the behaviour of
the one-leg 0-method and the linear 0-method in the solution of the equation x'(t) = ax(t)+aox([t])+
alx([t+ 1]), with real a, a0, al and [.] designates the greatest-integer function. T h e stability regions of
two 0-methods are determined. The conditions under which the analytic stability region is contained
in the numerical stability region are obtained and some numerical experiments are given. (~) 2005
Elsevier Ltd. All rights reserved.

Keywords--Delay differential equations, Piecewise continuous arguments, Asymptotic stability.

1. I N T R O D U C T I O N
This paper is concerned with the numerical solution of the delay differential equations with
piecewise continuous arguments (EPCA),

x' (t) = f (t, x (t) ,x (al (t)) ,x (a2 (t))), (1.1)

where the arguments ai(t), i = 1, 2, have intervals of constancy.

In recent years, considerable work has been done on differential equations with piecewise con-
tinuous arguments. The first contribution is due to [1,2]. Also, there exists an extensive literature
dealing with EPCA in [3-9]. In their papers, it has been shown that, in genera], the properties of

T h e authors wish to thank the two referees for their valuable comments. Their suggestions are helpful for the
authors to improve the article.
This work is supported by the National Natural Science Foundation of P.R. China (No. 10271036).

0898-1221/05/$ - see front matter (~) 2005 Elsevier Ltd. All rights reserved. Typeset by ~AJ~IS-TEX
doi: 10.1016/j.camwa.2005.02.002
1296 M.H. SONG et al.

solutions to EPCA and the corresponding differential equations without delay are strikingly dif-
ferent. The task of investigating EPCA is also of considerable applied interest since they include,
as particular cases, impulsive and loaded equations of control theory and are similar to those
found in some biomedical models. A typical EPCA contains arguments that are constant on
certain intervals. A solution is defined as a continuous, sectionally smooth function that satisfies
the equation within these intervals. Continuity of a solution at a point joining any two consec-
utive intervals leads to recursion relations for the solution at such points. Hence, the solutions
are determined by a finite set of initial data, rather than by an initial function, as in the case
of general functional differential equation. It seems to us that the strong interest in differential
equation with piecewise continuous arguments is motivated by the fact that describe hybrid dy-
namical systems. They combine the properties of differential equations and difference equations.
As is well known, the stability of numerical solutions is one of the most attracting topics.
In this paper, we consider the advanced equation with piecewise continuous argument,

(t) = ax (t) 4- aox ([t]) 4- a l x (It 4- 1]), t_>0,

(1.2)
z (0) = x0,

where a, ao, al, xo, are real constants and [.] denotes the greatest integer function. In [10], some
properties of the solution of equation (1.2) are investigated.
DEFINITION 1. A solution of equation (1.2) on [0, oc) is a function x(t) that satisfies the following
conditions.
1. x(t) is continuous on [0, oo).
2. The derivative x'(t) exists at each point t E [0, oo), with the possible exception of the
points [t]e [0, o~) where one-sided derivatives exist.
3. Equation (1.2) is satisfied on each interval [n, n + 1) C [0, oo) with integral end-points.

THEOREM 1. If al ~ a/(e a -- 1), then equation (1.2) has on [0, c~) a unique solution,

x (t) = (m0 ({t}) + ~ml ({t})) ~[%0,

where {t} is the fractional part o f t and

mo (t) = e at 4- (e at - 1) a-lao, ~t. 1 (~) = (C at -- 1) a-lal.

Equation (1.2) is asymptoticalIy stable O.e., x(t) --~ 0 as t --+ 0 for alI Xo), if and only if

( a(e~" + 1)~
(a+a0+al) al-a0 -j-_--f ] >0. (1.3)

2. THE 0-METHODS
Let h = 1 / m be a given stepsize with integer rn >_ 1 and the gridpoints tn = nh (n = 1, 2 , . . . ),
we consider the linear 8-methods to (1.1),

Xn+l = Xn 4- h {Of ((n -Jc1) h, X n + l , x h (o~1 ((Yt 4- 1) h)), x h (a2 ( ( n + 1) h)))

(2.1)
4- (1 - 8) f (nh, xn, x h (al (nh)), x h (a2 (nh))) } (n = 1, 2 , . . . ) ,

and the one-leg 8-methods applied to (1.1),

Xn+l ~ Xn
(2.2)
+ h f ((n + 8) h , x h ((n O) h ) , x h (c~1 ((n 4- 8) h ) ) , x h (~2 ((n 4- 8) h))) (n = 1,2,...).
 Advanced DifferentialEquations 1297

Here, 0 is a parameter, with 0 < O < 1, specifying the method, xh(~i(t)) (i = 1, 2) denotes an
approximation to x(ai(t)), i = 1,2, and xh(t) is an approximation to x(t) defined by
x h (t) t - nh (n + 1) h - t
=~xn+l+ . xn, fornh<t<(n+l)h, n=0,1,
h h . . . . .
Applying (2.1) and (2.2) to (1.2), we arrive at the following recurrence relations, respectively,
Xn+l = x~ + h {0 (axn+l + aox h ([(n + 1) hi) + a l x h ([(n + 1) h + 1]))
+ (1 - 0) ( a ~ + ~0x ~ ([~h]) + ~1~ ~ ([~h + 1 1 ) ) } , (2.3)
x~+l = x~ + h {a (Ox~+l + (1 - O) x~)
+aox h ([(n + O) h]) + a l x h ([(n + 0) h + 11)}. (2.4)

Let n = k m + l (l --O, 1, 2 , . . . , m - 1 ) . Then, we define x h (It n + 5h]), 0 < 5 < 1, as x km according

to Definition 1. As a result, (2.3) and (2.4) reduce to the same recurrence relation,
Xkm+l+l = aXkm+l + 13Xk,~ + 7X(k+l),~, I = 0, 1, 2 , . . . , m -- 1, (2.5)
where
ha hao hal
a = 1 + 1 - Oh-----'---~' fl = 1 - Oh------~' (2.6)
"Y = 1 - Oh-----~"
In fact, in each interval [n, n + 1), equation (1.2) can be seen as ordinary differential equation,
hence, the &methods for equation (1.2) are convergent of order 1 if 0 ~ 1/2 and order 2 if 0 = 1/2.

3. S T A B I L I T Y ANALYSIS
It is easily seen that (2.5) is equivalent to
~m + (a0/a)(~m - 1)
x(k+l)m = 1- (31/~) ( ~ - 1) x k ~ , (3.1)
Xkrn+'+l"~'(O~l+l+aO(Ogl+X--1))Xkrn+al(ogl+l--l)x(k+l)m, a
0</<m--2.

DEFINITION 2. The process (2.3) or (2.4) for equation (1.2) is caIled asymptotically stable at
(a, ao,al) if and only if there exists a constant M , such that for any given xo, relation (3.1)
defines xn that satis~ xn --+ 0 for n --~ oo whenever m > M .
DEFINITION 3. The set of all triples (a, ao, al) at which the process (2.3) or (2.4) for equation (1.2)
is asymptotically stable is called the asymptotical stability region.
In the following, we take M = lal and denote the asymptotical stability region by So and the
set consisting of all triples (a, a0,31) which satisfy condition (3.1) by H, i.e.,

e 7-1" / > 0} (3.2)

We will investigate which conditions lead to H C_ So. For convenience, we divide the region H
into three parts,
H0 = {(a, ao, al) e H : a = 0},
H1 = {(a, ao,al) C H \ Ho : a + ao + al < 0},
H 2 - - { ( a , ao,al) E H \ H o : a + a o + a l >0}.
In view of (3.1),
IXkm+l+ll <_ N IXkml (0 < l < m - 1), (3.3)
where

max az+l -t- a (~1+1 - 1)l + a.__ ~_1(~__~__~) a~O,

N = o<z_<,~-x a
max {1+/+1 1 a1(1+1)(1+ao) }
0<,<m-1 m a0 + ~-{~::~;y , a = 0.
Hence, we have the following lemma.
1298 M.H. SONGet al.

LEMMA 1. Xn --~ 0 as n --~ oa i a n d o n l y i f Xkm --* 0 as l~ --* oo.

It is well known by (3.1) that Xkm ~ 0 as k ~ c~ if and only if
I~l < 1, (3.4)
where
a m + (ao/a)(a m - I)
A=
1 - (al/a)(a m - 1) "
Equation (3.4) is equivalent to

(a+ao+al) al-ao-a am- 1 >O. (3.5)

In the similar way, we denote

So = {(a, ao, 31) e s : a = 0 } ,
$1 = {(a, a o , a l ) E S \ So : a + a o ~-al < 0},
~'2 : {(a, a o , a l ) e S \ So : a - ~ a o ~ - a l > 0}.
Therefore, in order to investigate which conditions lead to H C_ So, we only need to prove Hi C_ Si,
i = 0, 1, 2, i.e.,
( 2a 2a ) > 0 . (3.6)
(a + ao + a l ) ej- 1 am-- 1 _

It is a simple matter to verify the following lemma which will be used in the proof of the main
results in the paper.
LEMMA 2. Let ~ ( x ) = Z/x - 1/(e ~ - 1), then ~ ( x ) is a decreasing function and ~ ( - ~ ) = 1,
~(0) = 1/2, and ~ ( + ~ ) = 0.
LEMMA 3. For 311 m k M ,
(1) (1 + a / ( , ~ - Oa)) m >_ e o if and o n l r i f 1 ~ 2 < 0 < 1 for a > O; ~ ( - 1 ) _< 0 _< 1, for a < 0;
(2) (1+ a / ( m - Oa)) m <_ e a i f a n d o n l y i f O < 0 < 1/2, f o r a < O; 0 -< 0 _< ~(1), f o r a > O.
PROOF. It is easily seen that for m > M , m - OR > 0 and 1 + a / ( m - Oa) > O.
(1) Suppose that for all m > M,
( 14
'rig-- ~o,
- a )m > ea" (3.7)
Then, for all m >_ M,
m 1
0> (3.8)
- a e a/m - 1"
According to Lemma 2, (3.8) is equivalent to
1
-<0_<1, for a > 0,
2-
~(-1) _< 0 < 1, for a < 0.
(2) Suppose that for all m >_ M,
( 1 + m ~aO a
)m -< ea"
(3.9)

Then, for all m > M,

m 1
0-< (3.1o)
a ea/m - 1 '

In view of Lemma 2, (3.10) is equivalent to

1
0<0< - for a < O,
- - -2'
o < o < ~(1), for a > O.
 A d v a n c e d Differential Equa t i ons 1299

We note for a E JR,

a
-->0. (3.11)
e~ - 1

For all m > M , we have

a
1 + ~ > 1 , for a > 0,
m - Oa
a
0<1+--<1, for a < O,
rn - Oa

which implies
a
>0.
a m - - 1

THEOREM 2. H I C_ S 1 if and only if

0 < O < qo(1), for all a > O,

(-1) < o < 1, for all a < O.

PROOF. F r o m (a, ao,al) H1, we have a # O, a + a o + a l < 0, and (3.6) reduces to

a a
(3.13)
ea - 1 - (l+a/(m-Oa)) m-l'

which from (3.11) and (3.12) is equivalent to

a
i+ - < e a, for a > O,

1+ - > eat for a < O.

m Oa -

In view of L e m m a 3, the t h e o r e m is proved.

THEOREM 3. H2 C_ $2 i f and only if

1
-<8_<1, for all a > O,
2-
1
0<0<~, for all a < 0.

PROOF. Assume (a, a0, a l ) E H2, we have a ~ 0, a q- a0 + a l > 0, and (3.6) reduces to

a a
> (3.14)
eo - 1 - (1 + a / ( . ~ - 0a)) ~ - 1'

which from (3.11) and (3.12) is equivalent to

1+ a > for a > 0,

m -- 0 a ea

( 1+ m
a
-- Oa
<
- - ca'
for a < O.

In view of L e m m a 3, the proof is complete. |

1300 M . H . SONG et al.

T H E O R E M 4. F o r P - m e t h o d s w i t h 0 < 0 < 1, we h a v e H o ---- So.

PROOF. A s s u m e a = 0. T h e n , A = (1 + a o ) / ( 1 - a l ) a n d (3.1) r e d u c e s t o

(ao a l ) ( a l -- ao -- 2) > 0, (3.15)

w h i c h is t h e s a m e a s (1.3) w i t h a - - 0. |
R E M A R K 1. F r o m t h e a b o v e d i s c u s s i o n , w e h a v e 1 - Oha > 0 for m > M . I n v i e w o f (3.12), it
is e a s y t o s e e 1 - ( a l / a ) ( a m - 1) 0 i n H1 a n d / / 2 . T h e r e f o r e , p r o c e s s e s (2.3) a n d (2.4) c a n b e
g o i n g o n for m > M .
R E M A R K 2. A s s u m i n g (a, a o , a l ) E H1 a n d M = 1~ t h e n f l ' o m t h e p r o o f s o f L e m m a 3 and
Theorem 2, i t is e a s y t o s e e t h a t t h e s t a b i l i t y c o n d i t i o n s a r e

0<0<~(a), for a > 0,

(a) < 0 < 1, for a < 0. (3.16)

From Lemma 2, T(a) < T ( 1 ) for a > 1 a n d ~ ( a ) > ~ ( - 1 ) for a < - 1 . H e n c e , t h e r a n g e o f 0 is

s m a l l e r t h a n t h a t w h e n M =1 a I.

4. NUMERICAL EXPERIMENTS

In order to give a numerical illustration of the conclusions in the paper, we consider the
following four problems,

x~ (t) - - 5Xl (t) - 7Xl (It]) - 3 x 1 (It -F 11), x1 (0) = 1, (4.1)

x~ (t) = - 1 0 x 2 (t) - 9x2 ([t]) + x2 (It + 1]), x2 (0) = 1, (4.2)
4 (t) = 5x3 (t) - 7x~ ([t]) + 2.1x3 ([t + 1]), x3 (0) = 1, (4.3)
X~ (t) = --0.5X 4 (t) -- X4 (It]) -~- 2X 4 (It -~ 1]), x4 (0) = 1 (4.4)

I t c a n b e s e e n f r o m (1.3) t h a t

(5, - 7 , - 3 ) , (-10,-9, 1) E H i , (5,-7, 2.1), (-0.5, -1, 2) ~ H2.

W e s h a l l u s e t h e 0 - m e t h o d w i t h t h e s t e p s i z e h -- 1 / m t o g e t t h e n u m e r i c a l s o l u t i o n a t t - - 10,
where the true solutions are xl(10) ~ 1.30610137261282E- 2, x 2 ( 1 0 ) ,~ 9 . 9 8 9 9 1 5 7 5 6 6 7 4 9 9 E - 1,
x3(10) ~ 6.088926071159937E - 1, a n d x 4 ( 1 0 ) ~ 9 . 4 2 6 8 6 9 5 1 0 3 9 0 4 4 2 E - 6 from Theorem 1 and
t h e r a n g e o f 0 is a b o u t [0,0.4180] for (4.1), [0.5820, 1] for (4.2), [1/2, 1] for (4.3) a n d {0, 1/2]
for (4.4).
I n T a b l e s 1 a n d 2, w e h a v e l i s t e d t h e a b s o l u t e e r r o r s A E a n d t h e r e l a t i v e e r r o r s R E a t t = 10
of the P-methods, a n d t h e r a t i o o f t h e e r r o r s o f t h e e a s e m = 50 o v e r t h a t o f m = 100. We
can see from these tables that the methods preserve their order of convergence. All numerical
experiments a r e i n a g r e e m e n t w i t h t h e c o n c l u s i o n s in t h i s p a p e r .

Table 1. Problem in the case of a > 0.

(4.1) with 0 ----0 (4.3) with ~ ----0.5 (4.3) with 0 = 1

m
AE RE AE RE AE RE
2 1.2708E - 2 9.7298E - 1 4.3305E - 3 7.1121E - 3 3.0458E - 1 5.0022E - 3
3 1.1466E - 2 8.7784E - 1 4.4704E - 3 7.3418E - 3 1.2219E - 1 2.0068E - 1
5 8.6306E - 3 6.6076E - 1 1.9774E - 3 3.2476E - 3 - -
10 4.7962E - 3 3.6722E - 1 5.2414E - 4 8.6081E - 4 4.3049E - 3 7.0701E - 3
20 2.4207E - 3 1.8534E - 1 1.3274E - 4 2.1801E - 4 2.6819E - 3 4.4045E - 3
50 9.5394E - 4 7.3073E - 2 2.1314E - 5 3.5004E - 5 1.1980E - 3 1.9675E - 3
100 4.7266E - 4 3.6188E - 2 5.3311E - 6 8.7554E - 6 6.1951E - 4 1.0174E - 3
Ratio 2.0182 2.0182 3.9980 3.9980 1.9337 1.9337
 A d v a n c e d Differential E q u a t i o n s 1301

T a b l e 2. P r o b l e m in t h e case of a < 0.

(4.1) w i t h 0 = 0 (4.3) w i t h 0 = 0.5 (4.3) w i t h ~ = 1

Tr~
AE RE AE RE AE RE
2 4.6913E - 1 4.6960E - 1 1.5362E - 6 1.6296E - 1 1.4862E - 4 1.5731E -b 1
3 2.4057E - 1 2.4081E - 1 6.5817E - 7 6.9818E - 2 6.2374E - 5 6.6166E + 0
5 8.6730E - 2 8.6817E - 2 2.3254E - 2 2.4668E - 2 2.5323E - 5 2.6863E -~ 0
10 2.0480E - 2 2.0501E - 2 5.7679E - 8 6.1185E - 3 9.3612E - 6 9.9304E - 1
20 5.6542E - 3 5.6599E - 3 1.4391E - 8 1.5266E - 3 410192E - 6 4.2636E - 1
50 1.4307E - 3 1.4322E - 3 2.3013E - 9 2.4412E - 4 1.4672E - 6 1.5564E - 1
100 6.0296E - 4 6.0357E - 4 5.7529E - 10 6.1026E - 5 7.1163E - 7 7.5490E - 2
Ratio 2.3729 2.3729 4.0003 4.0003 2.0618 2.0618

REFERENCES
1. K.L. C o o k e a n d J. W i e n e r , R e t a r d e d differential e q u a t i o n s w i t h piecewise c o n s t a n t delays, J. Math.Anal. 99,
265-297, (1984).
2. S.M. S h a h a n d J. W i e n e r , A d v a n c e d differential e q u a t i o n s w i t h piecewise c o n s t a n t a r g u m e n t deviations,
Internat. J. Math. Sci. 6, 671-703, (1983).
3. Y. M u r o y a , Persistence, c o n t r a c t i v i t y a n d global s t a b i l i t y in logistic e q u a t i o n s w i t h piecewise c o n s t a n t delays,
J. Math. Anal. Appl. 270, 602-635, (2002).
4. J. W i e n e r , Differential e q u a t i o n s w i t h piecewise c o n s t a n t delays, In Trends in the Theory and Practice of
Nonlinear Di~erential Equations, ( E d i t e d by V. L a k s h m i k a n t h a m ) , Marcel Dekker, N e w York, (1983).
5. J. W i e n e r , Di~erential Equations, ( E d i t e d by I.W. K n o w l e s a n d R . T . Lewis), pp. 571-580, N o r t h - H o l l a n d ,
N e w York, (1984).
6. J. W i e n e r a n d L. D e b n a t h , B o u n d a r y value p r o b l e m s for t h e diffusion e q u a t i o n w i t h piecewise c o n t i n u o u s
t i m e delay, Int. J. Math. Sa. 20, 187-195, (1997).
7. J. W i e n e r a n d L. D e b n a t h , A s u r v e y of p a r t i a l differential e q u a t i o n s w i t h piecewise c o n t i n u o u s a r g u m e n t s ,
Int. J. Math. Sci. 18, 209-228, (1995).
8. J. W i e n e r a n d W . Heller, Oscillatory a n d periodic s o l u t i o n s to a diffusion e q u a t i o n of n e u t r a l type, Int. J.
Math. Math. Sei. 22, 313-349, (1999).
9. J. W i e n e r a n d V. L a k s h m i k a n t h a m , Excitability of a s e c o n d - o r d e r delay differential e q u a t i o n , Nonlinear Anal.
38, 1-11, (1999).
10. J. W i e n e r , Generalized Solutions of Di]yerential Equations, W o r l d Scientific, Singapore, (1993).