J.Math.Anal.Appl.325(2007)612–623
https://www.wendangku.net/doc/865399564.html,/locate/jmaa
Vector Lyapunov functions for practical stability
of nonlinear impulsive functional differential equations
Ivanka M.Stamova
Department of Mathematics,Bourgas Free University,8000Bourgas,Bulgaria
Received 29September 2005
Available online 9March 2006
Submitted by William F.Ames
Abstract
This paper studies the practical stability of the solutions of nonlinear impulsive functional differential equations.The obtained results are based on the method of vector Lyapunov functions and on differential inequalities for piecewise continuous functions.Examples are given to illustrate our results.
?2006Elsevier Inc.All rights reserved.
Keywords:Practical stability;Impulsive functional differential equations;Vector Lyapunov functions
1.Introduction
Impulsive differential equations arise naturally from a wide variety of applications such as air-craft control,inspection process in operations research,drug administration,and threshold theory in biology.There has been a signi?cant development in the theory of impulsive differential equa-tions in the past 10years (see monographs [3,4,13,20]).Now there also exists a well-developed qualitative theory of functional differential equations [2,9–12].However,not so much has been developed in the direction of impulsive functional differential equations.In the few publications dedicated to this subject,earlier works were done by Anokhin [1]and Gopalsamy and Zhang [8].Recently,some qualitative properties (oscillation,asymptotic behavior and stability)are investi-gated by several authors (see [5–7,18,21,23,24]).
The ef?cient applications of impulsive functional differential equations to mathematical sim-ulation requires the ?nding of criteria for stability of their solutions.
E-mail address:stamova@bfu.bg.
0022-247X/$–see front matter ?2006Elsevier Inc.All rights reserved.
doi:10.1016/j.jmaa.2006.02.019
I.M.Stamova /J.Math.Anal.Appl.325(2007)612–623613
In the study of Lyapunov stability,an interesting set of problems deal with bringing sets close to a certain state,rather than the state x =0.The desired state of a system may be mathemati-cally unstable and yet the system may oscillate suf?ciently near this state that its performance is acceptable.Many problems fall into this category including the travel of a space vehicle between two points,an aircraft or a missile which may oscillate around a mathematically unstable course yet its performance may be acceptable,the problem in a chemical process of keeping the temper-ature within certain bounds,etc.Such considerations led to the notion of practical stability which is neither weaker nor stronger than Lyapunov stability.The main results in this prospect are due to Martynyuk [14,16,17].
It is well known that employing several Lyapunov functions in the investigation of the quali-tative behavior of the solutions of differential equations is more useful than employing a single one,since each function can satisfy less rigid requirements.Hence,the corresponding theory,known as the method of vector Lyapunov functions,offers a very ?exible mechanism [15].
In this paper,we use piecewise continuous vector Lyapunov functions to study practical stabil-ity of the solutions of nonlinear impulsive functional differential equations.The main results are obtained by means of the comparison principle coupled with the Razumikhin technique [14,19].Examples are given to illustrate our results.
2.Statement of the problem.Preliminary notes and de?nitions
Let R n be the n -dimensional Euclidean space with norm |x |=( n i =1x 2i )
1/2,Ωbe a bounded domain in R n containing the origin and R +=[0,∞).Let t 0∈R ,τ>0.Consider the system of impulsive functional differential equations ˙x(t)=f (t,x(t),x t ),t >t 0,t =t k , x(t k )=x(t k +0)?x(t k )=I k (x(t k )),t k >t 0,k =1,2,...,
(1)where f :(t 0,∞)×Ω×D →R n ;D ={φ:[?τ,0]→Ω,φ(t)is continuous everywhere ex-
cept at ?nite number of points ?t
at which φ(?t ?0)and φ(?t +0)exist and φ(?t ?0)=φ(?t )};I k :Ω→Ω,k =1,2,...;t 0
(2)The solution x(t)=x(t ;t 0,?0)of the initial value problem (1),(2)is characterized by the fol-lowing:
(a)For t 0?τ t t 0the solution x(t)satis?ed the initial conditions (2).
(b)For t 0 ˙x(t)=f t,x(t),x t ,t >t 0, x t 0=?0(s),?τ s 0. At the moment t =t 1the mapping point (t,x(t ;t 0,?0))of the extended phase space jumps momentarily from the position (t 1,x(t 1;t 0,?0))to the position (t 1,x(t 1;t 0,?0)+I 1(x(t 1;t 0,?0))). 614I.M.Stamova /J.Math.Anal.Appl.325(2007)612–623 (c)For t 1 y t 1=?1,?1∈D, where ?1(t ?t 1)=????0(t ?t 1),t ∈[t 0?τ,t 0]∩[t 1?τ,t 1], x(t ;t 0,?0),t ∈(t 0,t 1)∩[t 1?τ,t 1],x(t ;t 0,?0)+I 1(t ;t 0,?0),t =t 1. At the moment t =t 2the mapping point (t,x(t))jumps momentarily,etc. The solution x(t ;t 0,?0)of problem (1),(2)is a piecewise continuous function for t >t 0with points of discontinuity of the ?rst kind t =t k ,k =1,2,...,at which it is continuous from the left. Introduce the following notations: I =[t 0?τ,∞);I 0=[t 0,∞);G k = (t,x)∈I 0×Ω:t k ?1 k =1,2,...;G =∞ k =1G k ; φ =sup s ∈[?τ,0] φ(s) is the norm of the function φ∈D.Together with system (1)we shall consider the system ˙u =g(t,u),t >t 0,t =t k , u(t k )=B k (u(t k )),t k >t 0,k =1,2,...,(3) where g :(t 0,∞)×R m +→R m +,B k :R m +→R m +,k =1,2,....Denote by u +(t ;t 0,u 0)the maximal solution of system (3)satisfying the initial condition u +(t 0+0;t 0,u 0)=u 0∈R m + .De?nition 1.System (1)is said to be: (PS 1)practically stable with respect to (λ,A)if given (λ,A)with 0<λ lim t →∞|x(t ;t 0,?0)|=0. Other practical stability notions can be de?ned based on this de?nition.See [14]for details.Introduce in R m a partial ordering de?ned in the following natural way:For u,v ∈R m we will write u v (u >v )if and only if u j v j (u j >v j )for any j =1,2,...,m . De?nition 2.The function ψ:R m +→R m +is said to be monotone nondecreasing in R m +if ψ(u)>ψ(v)for u >v and ψ(u) ψ(v)for u v (u,v ∈R m + ).De?nition 3.The function g :(t 0,∞)×R m +→R m +is said to be quasi monotone nondecreas-ing in (t 0,∞)×R m +if for each pair of points (t,u)and (t,v)from (t 0,∞)×R m +and for j ∈{1,2,...,m }the inequality g j (t,u) g j (t,v)holds whenever u j =v j and u j v j for I.M.Stamova /J.Math.Anal.Appl.325(2007)612–623615 j =1,2,...,m ,i =j ,i.e.,for any ?xed t ∈(t 0,∞)and any j ∈{1,2,...,m }the function g j (t,u)is nondecreasing with respect to (u 1,u 2,...,u j ?1,u j +1,...,u m ). Let J ?R be an interval.De?ne the following classes of functions: PC J,R n = σ:J →R n :σ(t)is continuous everywhere except some points t k at which σ(t k ?0)and σ(t k +0)exist and σ(t k ?0)=σ(t k ),k =1,2,... ; PC 1 J,R n = σ∈PC J,R n :σ(t)is continuously differentiable everywhere except some points t k at which ˙σ(t k ?0)and ˙σ(t k +0)exist and ˙σ(t k ?0)=˙σ(t k ),k =1,2,... ; K = a ∈C [R +,R +]:a(u)is strictly increasing and such that a(0)=0 ;CK = a ∈C [t 0,∞)×R +,R + :a(t,u)∈K for each t ∈[t 0,∞) ;S(α)= x ∈R n :|x |<α . In the further considerations we shall use the class V 0of piecewise continuous auxiliary func-tions V :[t 0,∞)×Ω→R m +which are analogues of Lyapunov’s functions [22].V 0= V :I 0×Ω→R m +:V ∈C G,R m + ,V (t,0)=0for t ∈[t 0,∞),V is locally Lipschitzian in x ∈Ωon each of the sets G k , V (t k ?0,x)=V (t k ,x)and V (t k +0,x)=lim t →t k t>t k V (t,x)exists .We also introduce the following class of functions: Ω1= x ∈PC [I 0,Ω]:V s,x(s) V t,x(t) ,t ?τ s t,t ∈I 0,V ∈V 0 . Let V ∈V 0.For x ∈PC [I 0,Ω]and t ∈I 0,t =t k ,k =1,2,...,we de?ne the function D ?V t,x(t) =lim inf h →0? h ?1 V t +h,x(t)+hf t,x(t),x t ?V t,x(t) .Introduce the following conditions: (H1)f ∈C [(t 0,∞)×Ω×D,R n ].(H2)The function f is Lipschitz continuous with respect to its second and third arguments in (t 0,∞)×Ω×D uniformly on t ∈(t 0,∞). (H3)f (t,0,0)=0,for t ∈(t 0,∞). (H4)I k ∈C [Ω,Ω],k =1,2,....(H5)I k (0)=0,k =1,2,.... (H6)The functions (I +I k ):Ω→Ω,k =1,2,...,where I is the identity in Ω. (H7)t 0 (H8)lim k →∞t k =∞. In the proof of the main results we shall use the following lemma: 616I.M.Stamova /J.Math.Anal.Appl.325(2007)612–623 Lemma 1.[5,7]Let the following conditions hold : 1.Conditions (H1),(H2),(H4),(H6)–(H8)are met. 2.The function g is quasimonotone nondecreasing,continuous in the sets (t k ,t k +1]×R m +,k ∈N ∪{0}and for each k ∈N ∪{0}and v ∈R m + there exists the ?nite limit lim (t,u)→(t,v)t>t k g(t,u). 3.The functions ψk :R m +→R m +,ψk (u)=u +B k (u),k =1,2,...,are monotone nondecreas-ing in R m +. 4.The maximal solution u +(t ;t 0,u 0)of system (3)is de?ned in the interval I 0. 5.The solution x =x(t ;t 0,?0)of problem (1),(2)is such that x ∈PC [I,Ω]∩PC 1[I 0,Ω]. 6.The function V ∈V 0is such that V t 0,?0(t 0) u 0 and the inequalities D ?V t,x(t) g t,V t,x(t) ,t =t k ,k =1,2,...,V t k +0,x(t k )+I k x(t k ) ψk V t k ,x(t k ) ,k =1,2,..., are valid for t ∈I 0and x ∈Ω1. Then V t,x(t ;t 0,?0) u +(t ;t 0,u 0) for t ∈I 0. 3.Main results Theorem 1.Assume that : 1. The conditions of Lemma 1are satis?ed.2. 0<λ g(t,0)=0for t ∈I 0.4. B k (0)=0,k =1,2,.... 5.There exist functions a,b ∈K such that a |x | L 0(t,x) b |x | (t,x)∈I 0×S(A),where L 0(t,x)= m i =1V i (t,x). 6.b(λ) Proof.1.We shall ?rst prove practical stability of (1).Suppose that (3)is practically stable with respect to (b(λ),a(A)).Then we have m i =1u i 0 I.M.Stamova /J.Math.Anal.Appl.325(2007)612–623617 for some given t 0∈R ,where u 0=(u 01,...,u 0m )T and u(t ;t 0,u 0)is any solution of (3)de?ned in the interval I 0. Setting u 0=V (t 0,?0(0)),we get by Lemma 1,V t,x(t ;t 0,?0) u + t ;t 0,V t 0,?0(0) for t ∈I 0.(5)Let ?0 <λ. (6)Then,because of condition 5of Theorem 1and (6),it follows L 0 t 0,?0(0) b ?0(0) b ?0 b(λ) which due to (4)implies m i =1u +i t ;t 0,V t 0,?0(0) Consequently,from condition 5of Theorem 1,(5)and (7)we obtain a x(t ;t 0,?0) L 0 t,x(t ;t 0,?0) m i =1u +i t ;t 0,V t 0,?0(0) Hence |x(t ;t 0,?0)| m i =1u i 0 for every t 0∈R .We claim that ?0 <λimplies |x(t ;t 0,?0)|t 0such that, x(t ?;t 0,?0) A, x(t ;t 0,?0) a(A) a x t 0;t 0,?0 L 0 t 0,x(t ;t 0,?0) m i =1u +i t 0;t 0,V t 0,?0 t 0?t k 618I.M.Stamova /J.Math.Anal.Appl.325(2007)612–623 The contradiction obtained proves that (1)is uniformly practically stable.The proof is com-plete.2 Note that in Theorem 1,we have used the function L 0(t,x)= m i =1V i (t,x)as a measure and consequently,we need to modify the de?nition of practical stability of (3)as follows:For example,(3)is practically stable with respect to (b(λ),a(A))if (4)is satis?ed for some given t 0∈R .We could use other convenient measures such as L 0(t,x)=max 1 i m V i (t,x),L 0(t,x)=m i =1 d i V i (t,x),wher e d ∈R m +,or L 0(t,x)=Q V (t,x) , where Q :R m +→R +and Q(u)is nondecreasing in u ,and appropriate modi?cations of practical stability de?nitions are employed for system (3). The following example will demonstrate Theorem 1. Example 1.Consider the system ???˙x(t)=n(t)y(t)+m(t)x(t)[x 2(t ?h)+y 2(t ?h)],t =t k ,t >0,˙y(t)=?n(t)x(t)+m(t)y(t)[x 2(t ?h)+y 2(t ?h)],t =t k ,t >0, x(t k )=c k x(t k ), y(t k )=d k y(t k ),k =1,2,...,(12) where x,y ∈R ,h >0,the functions n(t)and m(t)are continuous in (0,∞),?1 c k <0,?1 d k <0,0 y(s)=?2(s),s ∈[?h,0], where the functions ?1and ?2are continuous in [?h,0]. Choose V (t,x,y)=x 2+y 2=r 2(s). Then Ω1= col x(t),y(t) ∈PC R +,R 2 :r 2(s) r 2(t),t ?h s t,t 0 (13)and for t >0,t =t k ,(x,y)∈Ω1we have D ?V t,x(t),y(t) =2m(t)x 2(t)r 2(t ?h)+2m(t)y 2(t)r 2(t ?h) 2m(t)V 2 t,x(t),y(t) . Also V t k +0,x(t k )+c k x(t k ),y(t k )+d k y(t k ) =(1+c k )2x 2(t k )+(1+d k )2y 2(t k ) V t k ,x(t k ),y(t k ) . I.M.Stamova /J.Math.Anal.Appl.325(2007)612–623619 Consider the comparison system ???˙u(t)=2m(t)u 2(t),t =t k ,t >0,u(0)=u 0,u(t k +0)=u(t k ),k =1,2,..., (14)where u ∈R +and u 0=?21(0)+?22(0)=r 2(0).The general solution of system (14)is given by u(t)= u ?10?2t 0m(t)dt ?1.(15) It is clear that the trivial solution of (14)is stable if m(t) 0,t 0.If m(t)>0,t 0,then the trivial solution of (14)is stable when the integral t 0m(t)dt (16) is bounded and unstable when (16)is unbounded.Let A =2λ.We can take a(u)=b(u)=u 2.Suppose that t 0m(t)dt =β>0.It therefore follows from (15)that system (14)is practically stable if β 32. Hence we get,by Theorem 1that system (12)is practically stable if β 32. In Example 1,we have used the single Lyapunov function V (t,x).In this case the function L 0(t,x)=V (t,x). To demonstrate the advantage of employing several Lyapunov functions,let us consider the following example. Example 2.Consider the system ???????˙x(t)=e ?t x(t ?h)+y(t ?h)sin t ?(x 3+xy 2)sin 2t,t =t k ,˙y(t)=x(t ?h)sin t +e ?t y(t ?h)?(x 2y +y 3)sin 2t, t =t k , x(t)=a k x(t)+b k y(t),t =t k ,k =1,2,..., y(t)=b k x(t)+a k y(t),t =t k ,k =1,2,...,(17) where t >0,h >0, a k =12 1+c k + 1+d k ?2 , b k =12 1+ c k ? 1+ d k ,?1 0 Suppose that we choose a single Lyapunov function V (t,x,y)=x 2+y 2.Then the set Ω1is given by (13).Hence,using the inequality 2|ab | a 2+b 2and observing that (x 2+y 2)2sin 2t 0,we get D ?V t,x(t),y(t) =2x(t)˙x(t) +2y(t)˙y(t) 4 e ?t +|sin t | V t,x(t),y(t) ,for t 0,t =t k and (x,y)∈Ω1. 620I.M.Stamova /J.Math.Anal.Appl.325(2007)612–623 Also V t k +0,x(t k )+a k x(t k )+b k y(t k ),y(t k )+b k x(t k )+a k y(t k ) = (1+a k )x(t k )+b k y(t k ) 2+ (1+a k )y(t k )+b k x(t k ) 2 V t k ,x(t k ),y(t k ) +2|c k ?d k |V t k ,x(t k ),y(t k ) ,k =1,2,.... It is clear that ˙u(t)=4[|e ?t |+|sin t |]u(t),t =t k ,t >0, u(t k )=2|c k ?d k |u(t k ),k =1,2,..., where u ∈R +,is not practically stable and consequently,we cannot deduce any information about the practical stability of system (17)from Theorem 1,even though system (17)is practi-cally stable. Now,let us take the function V =(V 1,V 2),where the functions V 1and V 2are de?ned by V 1(t,x,y)=12(x +y)2,V 2(t,x,y)=12(x ?y)2so that L 0(t,x,y)=x 2+y 2.This means that we can take a(u)=b(u)=u 2.Then Ω1= (x,y)∈PC R +,R 2+ :V s,x(s),y(s) V t,x(t),y(t) ,t ?h s t,t 0 .Moreover,for t 0and (x,y)∈Ω1the following vectorial inequalities:D ?V t,x(t),y(t) g t,V t,x(t),y(t) ,t =t k ,k =1,2,...,V t k +0,x(t k )+ x(t k ),y(t k )+ y(t k ) ψk V t k ,x(t k ),y(t k ) , k =1,2,...,are satis?ed with g =(g 1,g 2),where g 1(t,u 1,u 2)=2 e ?t +sin t u 1,g 2(t,u 1,u 2)=2 e ?t ?sin t u 2and ψk (u)=u +C k u,k =1,2,...,where C k = c k 00d k .It is obvious that the functions g and ψk satisfy conditions 2and 3of Lemma 1and the comparison system ???˙u 1(t)=2(e ?t +sin t)u 1(t),t =t k ,˙u 2(t)=2(e ?t ?sin t)u 2(t),t =t k , u 1(t k )=c k u 1(t k ), u 2(t k )=d k u 2(t k ),k =1,2,..., is practically stable for any 0<λ Theorem 1implies that system (17)is also practically stable. We have assumed in Theorem 1stronger requirements on L 0only to unify all the practical results in one theorem.This puts burden on the comparison system (3).However,to obtain only nonuniform practical stability criteria,we could weaken certain assumptions of Theorem 1as in the next result. Theorem 2.Assume that the conditions of Theorem 1hold with the following changes in condi-tions 5and 6: 5?.There exist functions a ∈K and b ∈CK,such that a |x | L 0(t,x) b t,|x | (t,x)∈I 0×Ω. 6?.b(t 0,λ) I.M.Stamova /J.Math.Anal.Appl.325(2007)612–623621 Then,the uniform or nouniform practical stability properties of system (3)with respect to (b(t 0,λ),a(A)),imply the corresponding nouniform practical stability properties of system (1)with respect to (λ,A). We shall next consider a result which gives practical asymptotic stability of (1).We will use two Lyapunov like functions. Theorem 3.Assume that :