Professional Documents
Culture Documents
a r t i c l e i n f o a b s t r a c t
Keywords: From the view of algebra, the mean-square stability of analytic solutions and numerical
Stochastic differential equation solutions for impulsive stochastic differential equations are considered. By the logarithmic
Impulsive norm, the conditions under which the analytic and numerical solutions for a linear impul-
Euler–Maruyama method sive stochastic differential equation are mean-square stable (MS-stable) respectively are
MS-stable
obtained. The conditions are simple and easy to use. Some numerical experiments are
given to illustrate the results.
Ó 2014 Elsevier Inc. All rights reserved.
1. Introduction
Impulsive effects exist widely in many evolution processes in which states are changed abruptly at certain
moments of time, involving such fields as medicine, biology, economics, mechanics, electronics and telecommunica-
tions, etc. (see [1,8]). In recent years, there have been intensive studies on the impulsive differential equations (see
[9–11]). However, besides impulsive effects, stochastic effects likewise exist widely in real systems (see [7]). It has
received considerable attention. For example, the existence and uniqueness of the analytic solution for stochastic dif-
ferential equations (SDEs) has been studied by Gard [5] and Mao [13], the stability of Euler–Maruyama (EM) method
for SDEs has been studied by Cao et al. [3] and Liu et al. [14], the stability and convergence of composite Milstein
method for SDEs has been studied by Omar et al. [15], the convergence of EM method for SDEs has been studied
by Buckwar [2], etc.
Therefore, it is necessary to consider the stability of solutions of impulsive stochastic differential equations. It has
attracted interests of many researchers. Wu and Sun [17], Yang et al. [19,20] and Zhao et al. [21] have studied the pth
moment stability of impulsive stochastic differential equations (ISDEs). Wu and Ding [18] has studied the convergence
and stability of Euler method for ISDEs and Zhao et al. [22] has studied pth moment stability of EM method for ISDEs. Liu
at al. [12] has studied the mean-square stability of semi-implicit method for one-dimensional ISDEs.
Like many other equations, explicit solutions can rarely be obtained for the impulsive stochastic differential equations.
Thus, it is necessary to develop numerical methods and to study the properties of these methods.
Like as ordinary differential equations, we just discuss the test equations. Unfortunately, the diffusion and drift coeffi-
cients can not be diagonalized at the same time.
⇑ Corresponding author.
E-mail address: zgh-hit1108@163.com (G. Zhao).
http://dx.doi.org/10.1016/j.amc.2014.11.098
0096-3003/Ó 2014 Elsevier Inc. All rights reserved.
528 G. Zhao et al. / Applied Mathematics and Computation 251 (2015) 527–538
Hence, in this paper, we consider the stability of both analytic and EM solution of the following equation
8
< dXðtÞ ¼ AXðtÞdt þ BXðtÞdWðtÞ; t – sk ;
> t > 0;
Xðsþk Þ ¼ Hk Xðsk Þ; t ¼ sk ; ð1:1Þ
>
:
Xð0þ Þ ¼ X 0 ;
where A ¼ diagðk1 ; k2 ; . . . ; kn Þ 2 Rnn , B ¼ ðbij Þnn 2 Rnn , Hk ¼ ðHkij Þnn 2 Rnn ðk ¼ 1; 2; 3; . . .Þ are constant matrices, WðtÞ is
T
standard Brownian motion, XðtÞ ¼ ðX 1 ðtÞ; . . . ; X n ðtÞÞ ; Xðt þ Þ is the right limit of XðtÞ; 0 < s1 < s2 < s3 < . . ., limk!1 sk ¼ 1.
Denote s0 ¼ 0, N ¼ f1; 2; . . .g.
By Gard [5] and Mao [13], the Eq. (1.1) has on ð0; 1Þ unique solution
8 n o
2
>
> XðtÞ ¼ exp A 12 b t þ BWðtÞ X 0 ; t 2 ð0; s1 ;
>
>
>
> n o
< 2
XðtÞ ¼ exp A 12 b ðt sk Þ þ BðWðtÞ Wðsk ÞÞ t 2 ðsk ; skþ1 ; k 2 N
>
> k h n oi
>
> Y
>
>
2
Hi exp A 12 b ðsi si1 Þ þ BðWðsi Þ Wðsi1 ÞÞ X 0 :
:
i¼1
In this paper, we always assume that there exist two constants h1 and h2 such that
In Section 2, MS-stability of Eq. (1.1) is analyzed. The results obtained in Section 2 are coincide with existing results. Such as,
Corollary 2.7 is Theorem 1 in [16], Corollary 2.8 is Theorem 2.1 in [12]. In Section 3, conditions of MS-stability of the Euler–
Maruyama scheme corresponding to Section 2 are obtained. The results in Section 3 are also coincide with existing results.
Corollary 3.8 is coincide with Theorem 2 in [16]. The main result with h ¼ 0 in Section 3 of [12] is coincide with Corollary 3.9.
In Section 4, some numerical experiments confirming our stability analysis in Section 3 are given.
Let (X; F; P) be a probability space with a filtration ðFt ÞtP0 , which satisfies the usual conditions. Let WðtÞ; t P 0 in Eq.
(1.1) be Ft -adapted and independent of F0 . We assume X 0 to be F0 -measurable.
In this section we will give the sufficient conditions under which the analytic solution of (1.1) is MS-stable. First, we will
give the definition of MS-stability.
Definition 2.1 [16]. The zero solution of the system (1.1) is MS-stable if
EðjXðtÞj2 Þ ! 0 as t ! 1;
where j j stands for the Euclidean norm, EðÞ denotes the expectation with respect to P.
Let FðtÞ ¼ EðXðtÞXðtÞT Þ be the n n matrix-valued second moment of the solution of (1.1). By the Ito
^ formula, FðtÞ obeys
the initial value problem of the following impulsive ordinary differential equation
8 T T
< dFðtÞ ¼ ðAFðtÞ þ FðtÞA þ BFðtÞB Þdt; t – sk ;
> t > 0;
T
Fðsþk Þ ¼ Hk Fðsk ÞHk ; t ¼ sk ; ð2:1Þ
>
: þ T
Fð0 Þ ¼ EðX 0 X 0 Þ:
nðnþ1Þ
By virtue of the symmetry of the matrix FðtÞ, we have its governing ordinary differential equation of 2
-dimension.
8
> _ t – sk ;
< YðtÞ ¼ MYðtÞ; t > 0;
Yðsþk Þ ¼ Dk Yðsk Þ; t ¼ sk ; ð2:2Þ
>
:
Yð0þ Þ ¼ Y 0 ;
where
2 2 T
2
YðtÞ ¼ ðEðX 1 ðtÞÞ ; EðX 1 ðtÞX 2 ðtÞÞ; . . . ; EðX 1 ðtÞX n ðtÞÞ; . . . ; EðX i ðtÞÞ ; EðX i ðtÞX iþ1 ðtÞÞ; . . . ; EðX i ðtÞX n ðtÞÞ; . . . ; EðX n ðtÞÞ Þ ;
M ¼ M1 þ diagð2k1 ; . . . ; 2ki ; ki þ kiþ1 ; . . . ; ki kn ; . . . ; 2kn Þ; ð2:3Þ
G. Zhao et al. / Applied Mathematics and Computation 251 (2015) 527–538 529
2
2 2
b 2b11 b12 2b11 b1n b1p 2b1p b1ðpþ1Þ
6 11
6 b11 b21 b11 b22 þ b12 b21 b11 b2n þ b1n b21 b1p b2p b1p b2ðpþ1Þ þ b2p b1ðpþ1Þ
6
6 . .. .. .. .. .. ..
6 .
6 . . . . . . .
6
6b b b11 bn2 þ b12 bn1 b11 bnn þ b1n bn1 b1p bnp b1p bnðpþ1Þ þ bnp b1ðpþ1Þ
6 11 n1
M1 ¼ 6
6 .. .. .. .. .. .. ..
6 . . . . . . .
6
6
6 bi1 bj1 bi1 bj2 þ bi2 bj1 bi1 bjn þ bin bj1 bip bjp bip bjðpþ1Þ þ bjp biðpþ1Þ
6
6 . .. .. .. .. .. ..
6 .. . . . . . .
4
2 2
bn1 2bn1 bn2 2bn1 bnn bnp 2bnp bnðpþ1Þ
2
3
2b1p b1ðpþ2Þ 2b1p b1n b1n
7
b1p b2ðpþ2Þ þ b2p b1ðpþ2Þ b1p b2n þ b2p b1n b1n b2n 7
7
.. .. .. .. .. 7
. . . . . 7
7
7;
bip bjðpþ2Þ þ bjp biðpþ2Þ bip bjn þ bjp bin bin bjn 7
7
.. .. .. .. .. 7
7
. . . . . 5
2
2bnp bnðpþ2Þ 2bnp bnn bnn
replacing the element bij in M1 by Hkij , we can get the matrices Dk (k ¼ 1; 2; . . .). Denote Y 0 ¼ Yð0þ Þ.
Note that the system (1.1) is MS-stable, if and only if YðtÞ ! 0 as t ! 1.
Definition 2.2 [1]. The function Y : ð0; þ1Þ ! Rnðnþ1Þ=2 is said to be a solution of the system (2.2), if the following conditions
are satisfied:
Denote U j ¼ Dj eMðsj sj1 Þ ; j 2 N. System (2.2) has on ð0; þ1Þ a unique solution
YðtÞ ¼ eMt Y 0 ; t 2 ð0; s1 ;
YðtÞ ¼ eMðtsk Þ U k U k1 U 1 Y 0 ; t 2 ðsk ; skþ1 ; k 2 N:
In the following, jj jj denotes any arbitrary norm on Rn (n 2 N). The same symbol will be used for the matrix norm
subordinate to the given vector norm. Let A ¼ ða i;j Þnn . The logarithmic norm l½A of A is (see [4])
kI þ dAk 1
l½A ¼ limþ ðI is a unite matrix; d 2 RÞ:
d!0 d
For convenience, we introduce some well known identities for our results.
( ) ( )
Xn X
kAk1 ¼ max ij j ;
ja l1 ½A ¼ max aii þ ij j :
ja
16i6n 16i6n
j¼1 j–i
X
k
1
lim sup logðkDj kÞ < l½M: ð2:4Þ
k!1 j¼1
sk
Corollary 2.4. The zero solution of the system (1.1) is MS-stable if there exists a constant q 2 ½0; 1Þ, such that
Proof. Since
! Q
k l½Mðsj sj1 Þ
1 Y
k log j¼1 kDj ke
log kDj k þ l½M ¼ :
sk j¼1
sk
From (2.5), we obtain
!
1 Y
k
k
lim sup log kDj k þ l½M 6 lim sup log q < 0:
k!1 sk j¼1 k!1 kh1
where
8 !2 9 8 !2 9
< X
n = < Xn =
L ¼ max jbip j þ 2ki : i ¼ 1; . . . ; n ; dk ¼ max jH j : i ¼ 1; . . . ; n :
: p¼1
; : p¼1 kip ;
Pn 2 Pn
Proof. Let Q ii ¼ p¼1 jbip j þ p¼1;l>p 2jbip bil j þ 2ki ,
X
n X
n
Q ij ¼ jbip bjp j þ ðjbip bjl þ bil bjp jÞ þ bii bjj þ bij bji þ ki þ kj ; i – j:
p¼1 p¼1;l>p;p–i;l–j
X
n X
n
Q ij ¼ jbip bjp j þ ðjbip bjl þ bil bjp jÞ þ bii bjj þ bij bji þ ki þ kj
p¼1 p¼1;l>p;p–i;l–j
X
n X
n
6 jbip bjp j þ ðjbip bjl þ bil bjp jÞ þ ki þ kj
p¼1 p¼1;l>p
0 !2 1, 0 !2 1,
X
n X
n
6@ jbip j þ 2ki A 2þ@ jbjp j þ 2kj A 2; i – j:
p¼1 p¼1
G. Zhao et al. / Applied Mathematics and Computation 251 (2015) 527–538 531
X
n X
n
Gkij ¼ jHkip Hkjp j þ ðjHkip Hkjl þ Hkil Hkjp jÞ; i – j:
p¼1 p¼1;l>p
!2
X
n X
n X
n
Gkii ¼ H2kip þ 2jHkip Hkil j ¼ jHkip j ;
p¼1 p¼1;l>p p¼1
!2 !2
X
n X
n X
n X
n
Gkij ¼ jHkip Hkjp j þ ðjHkip Hkjl þ Hkil Hkjp jÞ 6 jHkip j =2 þ jHkjp j =2; i – j:
p¼1 p¼1;l>p p¼1 p¼1
In the remainder, we will show that results obtained in this section are coincide with some existing results.
Corollary 2.6. The system (1.1) is MS-stable if there exists a constant q 2 ½0; 1Þ such that
Corollary 2.7. Assume that n ¼ 2; Hk ¼ I for k ¼ 1; 2; . . .. The system (1.1) is MS-stable, if the following estimation
2 2
maxf2k1 þ ðjb11 j þ jb12 jÞ ; 2k2 þ ðjb21 j þ jb22 jÞ g < 0
holds.
By Corollary 2.6, we can get the following result which is coincide with Theorem 2.1 in [12].
Corollary 2.8. Let A ¼ a 2 R; B ¼ b 2 R; Hk ¼ 1 þ bk 2 R; sk ¼ k. The system (1.1) is MS-stable, if the following estimation
2
j1 þ bk j2 expf2a þ b g 6 q < 1; 8k 2 N
holds. That is
1 2
supfj1 þ bk j : k 2 Ngexp a þ b < 1:
2
In this section, we shall investigate the MS-stability of EM scheme with variable step-size for system (1.1). The adaption
of EM scheme to Eq. (1.1) leads to a numerical solution of the following type
8
< X km;l ¼ ðI þ Ahk þ BDW km;l1 ÞX km;l1 ;
> k ¼ 0; 1; . . . ; l ¼ 1; . . . ; m;
X km;0 ¼ Hk X ðk1Þm;m ; ð3:1Þ
>
:
X 0;0 ¼ X 0 ;
s s
where hk ¼ kþ1m k is a step-size for a positive integer m; I is a unit matrix, the discrete point
t km;l ¼ sk þ lhk ; 0 6 l 6 m; DW km;l ¼ Wðt km;lþ1 Þ Wðtkm;l Þ are independent Nð0; hk Þ-distributed Gaussian random variables,
X km;l Xðtkm;l Þ; 1 6 l 6 m and X km;0 Xðsþk
Þ ¼ limt!sþ XðtÞ. Furthermore X km;l is Ftkm;l -measurable.
k
532 G. Zhao et al. / Applied Mathematics and Computation 251 (2015) 527–538
Theorem 3.1. From Theorem 10.22 in [6], we can obtain that the method (3.1) converges strongly.
Let F km;l ¼ EðX km;l X Tkm;l Þ be the n n matrix-valued second moment of X km;l . F km;l obeys the following difference equation
8
>
> F km;l ¼ ðAF km;l1 þ F km;l1 AT þ BF km;l1 BT Þhk
>
>
< þF þ AF
2
AT h ; l ¼ 1; . . . ; m;
km;l1 km;l1 k
ð3:2Þ
>
> F km;0 ¼ Dk F ðk1Þm;m ;
>
>
:
F 0;0 ¼ EðX 0 X T0 Þ:
Denote U j ¼ Dj M m
j1 j 2 N. From Eq. (3.3) we have
(
Y 0;l ¼ M 0 l Y 0 ; 0 6 l 6 m;
ð3:4Þ
Y km;l ¼ Mlk U k U k1 U 1 Y 0 ; 0 6 l 6 m; k 2 N:
EðjX n j2 Þ ! 0 as n ! 1;
Note that, EðjX n j2 Þ ! 0 as n ! 1, if and only if kY km;l k ! 0, as k ! 1 for all 1 6 l 6 m.
or
(2) the condition (2.4) holds and
X
k
lim sup ðlogðkMj1 kÞ l½Mhj1 Þ 0: ð3:6Þ
k!1 j¼1
Hence
2 Q 3
k Pk
log kY km;l k log j¼1 kDj k log kM j1 k
j¼1
lim sup 6 lim sup4 þ Pk 5
k!1 sk k!1 sk j¼1 hj1
" #
Xk
logðkDj kÞ log kM j1 k
¼ lim sup þ Pk :
k!1 j¼1
sk j¼1 hj1
G. Zhao et al. / Applied Mathematics and Computation 251 (2015) 527–538 533
Remark 3.4. The condition (1) in Theorem 3.3 contains the condition (2). Under condition (2), the numerical solution and
analytical solution are both stable. But, under condition (1), the numerical solution is stable, the analytical solution may be
unstable.
Proof. From Theorem 3.3, we can easily get the case (2). Next, we will prove the case (1). From (3.5), we have
" # " #
X
k
logðkDj kÞ log kMj1 k Xk
logðkDj kkM j1 km Þ log q
þ Pk ¼ 6 < 0:
j¼1
sk h
j¼1 j1 j¼1
s k h1
Theorem 3.6. The numerical solution (3.1) for (1.1) is MS-stable w.r.t. k k1 , if
Xp
logðdk Þ log krk1 k
ð1Þ lim sup þ Pp < 0; ð3:9Þ
p!1 k¼1 sp k¼1 hk1
or
(2) the condition (2.6) holds and
X
p
lim sup ðlogðrk1 Þ Lhk1 Þ 0; ð3:10Þ
p!1 k¼1
By the similar calculation in Theorem 2.5, we can easily get that rk ¼ kM k k1 . Hence, by Theorem 3.3, we obtain the result. h
dk rm
k1 6 q < 1; 8k 2 N; ð3:11Þ
534 G. Zhao et al. / Applied Mathematics and Computation 251 (2015) 527–538
or
(2) the condition (2.6) holds and
where dk and L have been defined in Theorem 2.5, and rk has defined in Theorem 3.6.
In the remainder, we will show that results obtained in this section are coincide with some existing results.
In particular, If Hk ¼ I. Set sk ¼ k; hk ¼ h, M k ¼ M for all k ¼ 0; 1; . . .. Thus the EM solution (3.1) is stable if and only
if kMk < 1.
Hence we can get Theorem 2 in [16] as following.
Corollary 3.8. Assumed that n ¼ 2; Hk ¼ I. The numerical method (3.1) is MS-stable w.r.t. k k1 if and only if the following
inequality
2 2 2 2
maxfhðjb11 j þ jb12 jÞ þ ð1 þ k1 hÞ ; hðjb21 j þ jb22 jÞ þ ð1 þ k2 hÞ g < 1
holds.
By Theorem 3.3, we can get the following result which is coincide with the main result with h ¼ 0 in section 3 of [12].
2 2 2
ð1 þ ahÞ þ b h 6 eð2aþb Þh
:
4. Numerical experiments
Example 4.1
8
< dXðtÞ ¼ 2XðtÞdt þ XðtÞdWðtÞ; t – k;
>
þ
Xðk Þ ¼ Hk XðkÞ; t ¼ k; ð4:1Þ
>
:
Xð0þ Þ ¼ 1;
where Hk 2 R; k ¼ 1; 2; . . ..
We will confirm the main result in this paper through numerical experiments. First, we choose (4.1) as test equations to
illustrate the convergence.
From [5,13], we can obtain the explicit solution of (4.1).
8
>
> expf1:5t þ WðtÞgXð0þ Þ; for t 2 ð0; 1;
>
>
< expf1:5ðt kÞ þ ðWðtÞ WðkÞÞg
XðtÞ ¼
>
> Yk
> þ
: Hi expf1:5 þ ðWðiÞ Wði 1ÞÞgXð0 Þ for t 2 ðk; k þ 1:
>
i¼1
In Fig. 1, we plot the explicit solution of (4.1) together with the EM method for step-size h ¼ 1=128. Owing to the con-
vergence of the EM method, the figures illustrate that the numerical solution has the same stability property with its ana-
lytical solution. The system (4.1) without impulsive effects that means that Hk ¼ 1; k ¼ 0; 1; . . ..
In the following tests, we show that the influence of impulsive effects and the step-size on the MS-stability of the EM
method. The data used in the following figures are obtained by the mean square of data by 10,000 trajectories, that is,
wi : 1 6 i 10; 000, if X n ¼ ðX 1n ; X 2n Þ, then
2
Y 1 ¼ 1=10; 000R10;000
i¼1 ðX 1n ðwi ÞÞ ; Y 2 ¼ 1=10; 000R10;000
i¼1 X 1n ðwi ÞX 2n ðwi Þ;
2
Y 3 ¼ 1=10; 000R10;000
i¼1 ðX 2n ðwi ÞÞ ;
if X n ¼ ðX 1n ; X 2n ; X 3n Þ, then
2
Y 1 ¼ 1=10; 000R10;000
i¼1 ðX 1n ðwi ÞÞ ; Y 2 ¼ 1=10; 000R10;000
i¼1 X 1n ðwi ÞX 2n ðwi Þ;
2
Y 3 ¼ 1=10; 000R10;000
i¼1 X 1n ðwi ÞX 3n ðwi Þ; Y 4 ¼ 1=10; 000R10;000
i¼1 ðX 2n ðwi ÞÞ ;
2
Y 5 ¼ 1=10; 000R10;000
i¼1 X 2n ðwi ÞX 3n ðwi Þ; Y 6 ¼ 1=10; 000R10;000
i¼1 ðX 3n ðwi ÞÞ :
G. Zhao et al. / Applied Mathematics and Computation 251 (2015) 527–538 535
Example 4.2
8 2 3 2 3
>
> 0:5 0 0 0:1 0:5 0:4
>
> 6 7 6 7
> 0:1 0:4 5XðtÞdWðtÞ; t – sk ;
< dXðtÞ ¼ 4 0 0:3 0 5XðtÞdt þ 4 0:5
>
0 0 0:2 0:1 0:4 0:5 ð4:2Þ
>
>
> Xðsk þ Þ ¼ Hk Xðsk Þ;
> t ¼ sk ;
>
>
:
Xð0þ Þ ¼ ð2; 2; 2ÞT ;
where sk ¼ 3 þ 3ðk2Þ
2
if k is even; sk ¼ 1 þ 3ðk1Þ
2
if k is odd.
s s
Example 4.3
8
> 0:2 0 0:2 0:4
>
> dXðtÞ ¼ XðtÞdt þ XðtÞdWðtÞ; t – sk ;
< 0 3:5 0:5 1:5
þ ð4:3Þ
>
> Xðk Þ ¼ Hk XðkÞ; t ¼ sk ;
>
:
Xð0þ Þ ¼ ð1; 1ÞT ;
For Hk ¼ I, by Theorem 3.6, the EM solution is stable. The left of Figure gives a hard evidence. For Hk ¼ 8 5 ð1Þk ;
4 10
dk eLðsk sk1 Þ ¼ 196e0:02 > 1. The right of Fig. 4 shows that the EM solutio of (4.3) is unstable. It shows that the impulsive
Example 4.4
8
>
> 8 0 3 0
>
> dXðtÞ ¼ XðtÞdt þ XðtÞdWðtÞ; t – k;
>
> 0 6 0 3
<
þ 3 1:4 ð4:4Þ
>
> Xðk Þ ¼ XðkÞ; t ¼ k;
>
> 1:4 3
>
>
:
Xð0þ Þ ¼ ð1; 1ÞT :
Fig. 5 show that the EM method for (4.4) is stable when h ¼ hk ¼ 1=100, unstable when h ¼ hk ¼ 1=5. It shows that the
stability of EM method depends on the choice of step-size.
Example 4.5
8
>
> 1:5 0 1 0
>
> dXðtÞ ¼ XðtÞdt þ XðtÞdWðtÞ; t – k;
>
> 0 1:5 0 1
>
< " #
þ
7
e þ ð1Þk 56 e 0 ð4:5Þ
>
> Xðk Þ ¼ 6 XðkÞ; t ¼ k;
>
> 0 0
>
>
>
:
Xð0þ Þ ¼ ð1; 1ÞT :
G. Zhao et al. / Applied Mathematics and Computation 251 (2015) 527–538 537
By calculating, l1 ½M ¼ 2; kDj k1 ¼ 19 e2 when j is odd, kDj k1 ¼ 4e2 when j is even. Clearly the condition (2.7) does not hold
for Eq. (4.5), but the condition (2.6) holds. It shows that Corollary 2.6 is more restrictive than Theorem 2.5. By calculating, we
know that the condition (3.9) holds for m ¼ 10 and m ¼ 4. Fig. 6 gives a hard evidence. At the same, by calculating, we know
that the condition (3.10) and condition (3.11) do not hold for m ¼ 4. It shows that the case (2) in Theorem 3.6 is more restric-
tive than the case (1), the Corollary 3.7 is more restrictive than Theorem 3.6.
Acknowledgements
We would like to thank the professor Xuerong Mao for his useful suggestions.
The work was supported by the National Nature Science Foundation of China (No. 11101130). The work was supported by
the foundation of Jiangsu University of Science and Technology of China (Nos. 35050903, 633051205).
References
[1] D.D. Bainov, P.S. Simeonov, Systems with Impulsive Effects: Stability, Theory and Applications, Ellis Horwood, Chichester, 1989.
[2] E. Buckwar, Introduction to the numerical analysis of stochastic differential equations, J. Comput. Appl. Math. 12 (2000) 297–307.
[3] W.R. Cao, M.Z. Liu, Z.C. Fan, MS-stability of the Euler–Maruyama method for stochastic differential delay equations, Appl. Math. Comput. 159 (2004)
127–135.
[4] K. Dekker, J.G. Verwer, Stability of Runge–Kutta Methods for Stiff Nonlinear Differential Equations, North-Holland, Amsterdam, 1984.
[5] T.C. Gard, Introduction to Stochastic Differential Equations, Marcel Dekker, New York, 1988.
[6] P.E. Kloeden, E. Platen, Numerical Solution of Stochastic Differential Equations, Spring, Berlin, 1992.
[7] B. Ksendal, Stochastic Differential Equations: An Introduction with Applications, fourth ed., Spring-Verlag, Berlin, 1995.
[8] V. Lakshmikantham, D.D. Bainov, P.S. Simeonov, Theory of Impulsive Differential Equations, World Scientific, Singapore, 1989.
[9] H. Liang, M.Z. Liu, M.H. Song, Stability of the analytic and numerical solutions for impulsive differential equations, Appl. Numer. Math. 11 (2011) 1103–
1113.
[10] H. Liang, M.Z. Liu, M.H. Song, Extinction and permanence of Euler methods in the numerical solution of a two-prey one-predator system with
impulsive effect, Int. J. Comput. Math. 88 (2011) 1305–1325.
[11] M.Z. Liu, H. Liang, Z.W. Yang, Stability of Runge–Kutta methods in the numerical solution of linear impulsive differential equations, Appl. Math.
Comput. 192 (2007) 346–357.
538 G. Zhao et al. / Applied Mathematics and Computation 251 (2015) 527–538
[12] Mingzhu Liu, Guihua Zhao, M.H. Song, Stability of the semi-implicit Euler method for a linear impulsive stochastic differential equation, Dyn. Contin.
Dis. Ser. B 18 (2011) 123–134.
[13] X.R. Mao, Stochastic Differential Equations and their Applications, Horwood Publication, Chichester, 1997.
[14] Wei Liu, Xuerong Mao, Strong convergence of the stopped Euler–Maruyama method for nonlinear stochastic differential equations, Appl. Math.
Comput. 223 (2013) 389–400.
[15] M.A. Omar, A. Aboul-Hassan, S.I. Rabia, The composite Milstein methods for the numerical solution of stratonovich stochastic differential equations,
Appl. Math. Comput. 215 (2009) 727–745.
[16] Y. Saito, T. Mitsui, Mean-square stability of numerical schemes for stochastic differential systems, Vietnam J. Math. 30 (2002) 551–560.
[17] S.J. Wu, J.T. Sun, P-Moment stability of stochastic differential equations with impulsive jump and Markovian switching, Automatica 42 (2006) 1753–
1759.
[18] Wu. Kaining, Xiaohua Ding, Convergence and stability of Euler method for impulsive stochastic delay differential equations, Appl. Math. Comput. 229
(2014) 151–158.
[19] Z.G. Yang, D.Y. Xu, L. Xiang, Exponential p-stability of impulsive stochastic differential equations with delays, Phys. Lett. A 359 (2006) 129–137.
[20] J. Yang, S.M. Zhong, W.P. Luo, Mean-square stability analysis of impulsive stochastic differential equations with delays, J. Comput. Appl. Math. 216
(2008) 474–483.
[21] Guihua Zhao, Mingzhu Liu, Wanjin Lv, Exponential p-stability of impulsive stochastic differential equations with delays, J. Nat. Sci. Heilongjiang Univ.
26 (2009) 722–727.
[22] Guihua Zhao, Minghui Song, M.Z. Liu, Exponential stability of Euler–Maruyama solutions for impulsive stochastic differential equations with delay,
Appl. Math. Comput. 215 (2010) 3425–3432.