Difference inequality for stability of impulsive difference equations with distributed delays

In this paper, we consider a class of impulsive difference equations with distributed delays. By establishing an impulsive delay difference inequality and using the properties of "ρ-cone" and eigenspace of the spectral radius of non-negative matrices, some new sufficient conditions for global exponential stability of the impulsive difference equations with distributed delays are obtained. An example is given to demonstrate the effectiveness of the theory.

Difference equations usually appear in the investigation of systems with discrete time or in the numerical solution of systems with continuous time [1]. In recent years, the stability investigation of difference equations has been interesting to many investigators, and various advanced results on this problem have been reported [2,3]. However, almost all available results have been focused on systems with discrete delays. In reality, difference systems with distributed delays become important because it is essential to formulate the discrete-time analogue of the continuous-time system with distributed delays when one wants to simulate or compute the continuous-time one after obtaining its dynamical characteristics. Fortunately, such an issue has been addressed in [4-7].

However, besides the delay effect, an impulsive effect likewise exists in a wide variety of evolutionary processes in which states are changed abruptly at certain moments of time, involving such fields as medicine, biology, economics, mechanics, electronics, and telecommunications. Recently, the asymptotic behaviors of impulsive difference equations have attracted considerable attention. Many interesting results on impulsive effect have been obtained [8-11].

It is well known that distributed delay differential equations with impulses or without impulses have been considered by many authors (see, for instance [12-14]). But, to the best of our knowledge, there is no concerning on the stability of impulsive difference equations with distributed delays in literature. Motivated by the above discussion, we here make a first attempt to arrive at results on the global exponential stability of impulsive difference equations with distributed delays.

Let be the space of n-dimensional (non-negative) real column vectors and denotes the set of m × n (non-negative) real matrices. Usually, E denotes an n × n unit matrix. For A, B ∈ R^{m × n} or A, B ∈ Rⁿ, the notation A ≥ B (A > B) means that each pair of corresponding elements of A and B satisfies the inequality " ≥ (>)". Especially, A ∈ R^{m × n} is called a nonnegative matrix if A ≥ 0, and z ∈ Rⁿ is called a positive vector if z > 0. Z denotes the integer set, Z_∞ = {j ∈ Z |-∞ < j ≤ 0} and . C denotes the set of all bounded functions φ(j) ∈ Rⁿ, j ∈ Z_∞.

For x ∈ Rⁿ, A ∈ R^n×n, φ ∈ C, we define

where , and introduce the corresponding norm for them as follows:

In this paper, we mainly consider the following impulsive difference equations with distributed delays

where 0 < i ≤ n and a_i, b_ij, c_ij are constants. The fixed moments of time m_k ∈ Z, and satisfy . The constants μ_ij(k) satisfy the following convergence conditions:

where λ₀ is a positive constant.

For convenience, we shall rewrite (1) in the vector form:

where x (m) = (x₁ (m), ..., x_n (m))^T, A = diag{a₁, ..., a_n}, B = {b_ij}_{n × n}, C = {c_ij}_{n × n}, f (x) = (f₁ (x₁), ..., f_n (x_n))^T, g(x) = (g₁(x₁), ..., g_n(x_n))^T, μ(k) = (μ_ij(k))_{n × n}, H_m(x(m)) = (H_1m(x(m)), ..., H_nm(x(m)))^T, φ ∈ C, and f(x), g(x), H_m(x) ∈ C[Rⁿ, Rⁿ].

We will assume that there exists one solution of system (2) which is denoted by x(m, 0, φ), or, x(m), if no confusion occurs. We will also assume that g(0) = 0, f(0) = 0 and H_m(0) = 0, m = m_k, for the stability purpose of this paper. Then system (2) admits an equilibrium solution x(m) ≡ 0.

Definition 2.1. The zero solution of Equation 2 is called globally exponentially stable if there are positive constants λ and M ≥ 1 such that for any initial condition φ ∈ C,

Here λ is called the exponential convergence rate.

For , the spectral radius ρ(A) is an eigenvalue of A and its eigenspace is denoted by

which includes all positive eigenvectors of A provided that the non-negative matrix A has at least one positive eigenvector(see [15]).

Lemma 2.1. [16] Suppose that and ρ(M) < 1, then there exists a positive vector z such that

For and ρ(M) < 1, we denote

which is a nonempty set by Lemma 2.1, and satisfying that k₁z₁+k₂z₂ ∈ Ω_ρ(M) for any scalars k₁ > 0, k₂ > 0 and vectors z₁, z₂ ∈ Ω_ρ(M). So Ω_ρ(M) is a cone without vertex in Rⁿ, we call it a "ρ-cone."

Lemma 2.2. Suppose and Q(k) = (q_ij(k))_{n × n}, where q_ij(k) ≥ 0 and satisfy

where λ₁ is a positive constant. Denote and let ρ(P + Q) < 1 and be a solution of the following inequality with the initial condition u(m₀ + m) ∈ C, m ∈ Z_∞,

Then

provided that the initial conditions satisfy

where z = (z₁, z₂, ..., z_n)^T ∈ Ω_ρ(P + Q), m₀ ∈ Z and the positive number λ ≤ λ₁ is determined by the following inequality

Proof. Since ρ(P + Q) < 1 and , then, by Lemma 2.1, there exists a positive vector z ∈ Ω_ρ(P + Q) such that (E - (P + Q))z > 0. Using continuity, there must be a sufficiently small constant λ > 0 such that , i.e., inequality (6) has at least one positive solution λ ≤ λ₁.

Let

Then, from (5), we have

By (3), we have

Since , we derive that

We next show for any m ≥ m₀

If this is not true, then there must be a positive constant m* ≥ m₀ and some integer i such that

By (6), (9), and the second inequality of (11), we obtain that

which contradicts the first inequality of (11). Thus (10) holds for all m ≥ m₀. Therefore, we have

and the proof is completed.

To obtain the global exponential stability of the zero solution of system (2), we introduce the following assumptions.

(A₁) For any x ∈ Rⁿ, there exist non-negative diagonal matrices U and V such that

(A₂) For any x ∈ Rⁿ, there exist non-negative matrices R_k such that

(A₃) Let , and ρ(P + Q) < 1.

(A₄) The set is nonempty.

(A₅) Let

and there exists a constant γ such that

where the positive number λ ≤ λ₀ is determined by the following inequality

Theorem 3.1. Assume that the hypothesis (H) and Conditions (A₁)-(A₅) hold. Then the zero solution of (2) is globally exponentially stable and the exponential convergent rate equals λ - γ.

Proof. Since ρ(P + Q) < 1 and , then, by Lemma 2.1, there exists a positive vector z ∈ Ω_ρ(P + Q) such that (E - (P + Q))z > 0. Using continuity and hypothesis (H), there must be a sufficiently small constant λ > 0 such that , i.e., inequality (14) has at least one positive solution λ ≤ λ₀.

From (2), Conditions (A₁) and (A₃), we have

where m₀ = 0.

For the initial conditions: x(s) = φ(s), -∞ < s ≤ 0, where φ ∈ C, we can get

where

By the property of "ρ-cone" and z ∈ Ω ⊆ Ω_ρ(P + Q), we have d ||φ||∈ Ω_ρ(P + Q). Then, all the conditions of Lemma 2.2 are satisfied by (15), (16), and Condition (A₃), we derive that

Suppose for all q = 1, ..., k, the inequalities

hold, where γ₀ = 1. Then, from Condition (A₂) and (18), we have

Since d ∈ Ω ⊆ W_ρ(R_q), we have R_qd = ρ(R_q)d. Therefore, from (12) and (19), we obtain

This, together with (18), leads to

By the property of "ρ-cone" again, the vector γ₀ ⋯ γ_k-1γ_kd ∈ Ω_ρ(P + Q). It follows from (21) and Lemma 2.2 that

yielding, together with (18), that

By mathematical induction, we can conclude that

Noticing that by (13), we can use (22) to conclude that

which implies that the conclusions of the theorem hold.

Remark 3.1. In Theorem 3.1, we may properly choose the matrix R_k in the condition (A₂) such that Ω ≡ ∅ Especially, when R_k = α_kE (α_k are non-negative constants), Ω is certainly nonempty. So, by using Theorem 3.1, we can easily obtain the following corollary.

Remark 3.2. The conditions (A₁)-(A₅) is conservative. For example, we get the absolute value of all coefficients of (2). Recently, the delay-fractioning or delay-partitioning approach [17,18] is widely used that has shown the potential of reducing conservatism. We will combine delay-partitioning approach with difference inequality approach in our future work to reduce the conservatism.

Corollary 3.1. Assume that (H), (A₁), (A₃), and (A₅) hold. For any x ∈ Rⁿ, there exist non-negative constants α_k such that

And let γ_k ≥ {1, α_k}, where the scalar 0 < λ < λ₀ is determined by (14). Then the zero solution of (2) is globally exponentially stable and the exponential convergent rate equals λ - γ.

Proof. Noticing that (23) is a special case of Condition (A₂). Since ρ(R_k) = α_k, then W_ρ(R_k) = Rⁿ. So, we have . Since the "ρ-cone" Ω_ρ(P + Q) is nonempty by Lemma 2.1, (A₄) obviously holds. Thus we can deduce the conclusion in terms of Theorem 3.1.

Remark 3.3. If H_k(x) = x, then Equation 2 becomes difference equations with distributed delays without impulses in vector form

which contains many popular models such as discrete-time Hopfield neural networks, discrete-time cellular neural networks, and discrete-time recurrent neural networks, and so on.

Corollary 3.2. Assume that (H), (A₁), and (A₃) hold. Then Equation 24 has exactly one equilibrium point, which is globally exponentially stable.

In this section, we will give an example to illustrate the global exponential stability of Equation 1 further.

Example. Consider the following difference equation with distributed delays:

with

and m₁ = 4, m_k = m_k-1 + k for k = 2, 3,.... One can check that all the properties given in (H) are satisfied provided that 0 < λ₀ < 1.

Case 1. If for i = 1, 2 and k = 1, 2,..., then Equation 25 becomes difference equation with distributed delays without impulses. The parameters of Conditions (A₁) and (A₃) are as follows:

where . We can easily observe that ρ(P + Q) = 0.8345 < 1. By Corollary 3.2, Equation 25 has exactly one globally exponentially stable equilibrium (0, 0)^T.

Case 2. Next we consider the case where

We can verify that point (0, 0)^T is also an equilibrium point of the impulsive difference equation with distributed delays (25)-(26) and the parameters of Conditions (A₂) and (A₄) as follows:

So Ω = {(z₁, z₂)^T > 0 |z₂ = z₁} is not empty. Let z = (1, 1)^T ∈ Ω and λ = 0.05 which satisfies the inequality . We can obtain that for k = 1, 2,...

Clearly, all conditions of Theorem 3.1 are satisfied, so the equilibrium (0, 0)^T is globally exponentially stable and the exponential convergent rate is equal to 0.01.

In this paper, we consider a class of impulsive difference equations with distributed delays. By establishing an impulsive delay difference inequality and using the properties of "ρ-cone" and eigenspace of the spectral radius of non-negative matrices, some new sufficient conditions for global exponential stability of the impulsive difference equations with distributed delays are obtained. The conditions (A₁)-(A₅) are conservative. For example, we get the absolute value of all coefficients of (2). We will combine delay-partitioning approach with difference inequality approach in our future work to reduce the conservatism.

The authors declare that they have no competing interests.

Dingshi Li carried out the main proof of the theorems in this paper. Shujun Long carried out the expample. Xiaohu Wang provided the main idea of this paper. All authors read and approve the final manuscript.

The authors would like to thank the referee(s) for his(her) detailed comments and valuable suggestions which considerably improved the presentation of the paper. The study was supported by National Natural Science Foundation of China under Grant 10971147, Scientific Research Fund of Sichuan Provincial Education Department under Grant 10ZA032 and Fundamental Research Funds for the Central Universities 2010SCU1006.

1. Kolmanovskii, VB, Shaikhet, LE: Control of systems with aftereffect. Translations of Mathematical Monographs, American Mathematical Society, Providence, RI (1996)

2. Xu, DY: Asymptotic behavior of nonlinear difference equations with delays. Comput Math Appl. 42, 393–398 (2001). Publisher Full Text

3. Liz, E, Ferreiro, JB: A note on the global stability of generalized difference equations. Appl Math Lett. 15, 655–659 (2002). Publisher Full Text

4. Wang, ZD, Liu, YR, Wei, GL, Liu, XH: A note on control of a class of discrete-time stochastic systems with distributed delays and nonlinear disturbances. Automatica. 46, 543–548 (2010). Publisher Full Text

5. Liu, YR, Wang, ZD, Liang, JL, Liu, XH: Synchronization and state estimation for discrete-time complex networks with distributed delays. IEEE Trans Syst Man Cyber B. 38(5), 1314–1325 (2008)

6. Wang, Z, Liu, Y, Liu, X: Exponential stabilization of a class of stochastic system with Markovian jump parameters and mode-dependent mixed time-delays. IEEE Trans Automat Control. 55, 1656–1662 (2010)

7. Liu, Y, Wang, Z, Liang, J, Liu, X: Stability and synchronization of discrete-time Markovian jumping neural networks with mixed mode-dependent time-delays. IEEE Trans Neural Netw. 20, 1102–1116 (2009). PubMed Abstract | Publisher Full Text

8. Zhu, W, Xu, DY, Yang, ZC: Global exponential stability of impulsive delay difference equation. Appl Math Comput. 181, 65–72 (2006). Publisher Full Text

9. Zhu, W: Invariant and attracting sets of impulsive delay difference equations with continuous variables. Comput Math Appl. 55, 2732–2739 (2008). Publisher Full Text

10. Song, QK, Cao, JD: Dynamical behaviors of discrete-time fuzzy cellular neural networks with variable delays and impulses. J Franklin Inst. 345, 39–59 (2008). Publisher Full Text

11. Xu, HL, Chen, YQ, Teo, KL: Global exponential stability of impulsive discrete-time neural networks with time-varying delays. Appl Math Comput. 217, 537–544 (2010). Publisher Full Text

12. Xu, DY, Zhu, W, Long, SJ: Global exponential stability of impulsive integro-differential equation. Nonlinear Anal. 64, 2805–2816 (2006). Publisher Full Text

13. Zhao, HY: Global asymptotic stability of Hopfield neural network involving distributed delays. Neural Netw. 17, 47–53 (2004). PubMed Abstract | Publisher Full Text

14. Zhang, Q, Wei, XP, Xu, J: Global exponential stability of Hopfield neural networks with continuously distributed delays. Phys Lett A. 315, 431–436 (2003). Publisher Full Text

15. Horn, RA, Johnson, CR: Matrix Analysis. Cambridge University Press, Cambridge (1985)

16. Lasalle, JP: The Stability of Dynamical System. SIAM, Philadelphia (1976)

17. Wang, Y, Wang, Z, Liang, J: A delay fractioning approach to global synchronization of delayed complex networks with stochastic disturbances. Phys Lett A. 372, 6066–6073 (2008). Publisher Full Text

18. Wang, Z, Wang, Y, Liu, Y: Global synchronization for discrete-time stochastic complex networks with randomly occurred nonlinearities and mixed time-delays. IEEE Trans Neural Netw. 21, 11–25 (2010). PubMed Abstract | Publisher Full Text