- Research
- Open Access
- Published:
Anti-periodic solutions problem for inertial competitive neutral-type neural networks via Wirtinger inequality
Journal of Inequalities and Applications volume 2019, Article number: 187 (2019)
Abstract
By using the Wirtinger inequality and topology degree theory, we investigate the anti-periodic solutions problem for inertial competitive neutral-type neural networks and obtain the existence of anti-periodic solutions to the above system. Our results are completely new.
1 Introduction
The Wirtinger inequality was first used in Fourier analysis, then was used in 1904 to prove the isoperimetric inequality; see [1, 2]. Since the Wirtinger inequality has been recognized as a powerful tool to estimate the prior bounds of solutions, it has been used in many research areas, such as Hamiltonian system, delay equations, biomathematics, neural networks, partial differential equation; see [3,4,5,6,7,8] and the relevant references therein.
In this paper, we will obtain the existence of anti-periodic solutions for inertial competitive neutral-type neural networks by using the Wirtinger inequality and topology degree theory. In 1996, Meyer, Ohl and Scheich [9] first proposed the competitive neural network, the behavior of this network is characterized by an equation of neural activity as a fast phenomenon and an equation of synaptic modification as a slow part of the neural system. Since by a competitive neural network one can study the dynamics of complex neural networks including the aspects of long- and short-term memory, it has received great attention. Meyer, Pilyugin and Chen [10] studied global exponential stability of competitive neural networks with different time scales. They presented a new method of analyzing the dynamics of a biological relevant system with different time scales based on the theory of flow invariance. After that, Meyer, Reberts and Thmmle [11] further studied the local uniform stability of competitive neural networks with different time scales under vanishing perturbations. Gu, Jiang and Teng [12] obtained the existence and global exponential stability of a unique equilibrium point of competitive neural networks with different time scales and multiple delays by using a nonlinear Lipschitz measure (NLM) method and constructing a suitable Lyapunov functional. Liu et al. [13] obtained the existence of a periodic solution for competitive neural networks with time-varying and distributed delays on time scales.
On the other hand, the anti-periodic solution problem for dynamic systems is an interesting topic which has been investigated by many researchers. Okochi [14] first considered the anti-periodic solution of nonlinear abstract parabolic equations. Then anti-periodic problems of neural networks have attracted much attention by many authors. Li, Yang and Wu [15] discussed an anti-periodic solution for impulsive BAM neural networks with time-varying leakage delays on time scales. Then Li et al. [16,17,18] further studied anti-periodic solutions for different types of neural networks. Xu, Chen and Guo [19] studied anti-periodic oscillations of bidirectional associative memory (BAM) neural networks with leakage delays. The existence and exponentially stability of anti-periodic solutions for neutral BAM neural networks with time-varying delays in the leakage terms have been obtained by Xu and Guo [20]. However, to the best of our knowledge there are only few results on the anti-periodic solutions for neutral-type competitive neural networks with inertial terms.
The above discussions constitute the motivation for the present paper. In this paper, we will study a kind of neutral-type competitive neural networks with inertial terms. Based on the Wirtinger inequality and topology degree theory, we prove the existence of anti-periodic solutions of the above neural network. We list the main contributions of this paper as follows:
-
(1)
We propose a class of inertial competitive neutral-type neural networks which is different from the existing competitive neural networks; see [9,10,11, 13, 15, 19].
-
(2)
Since the model of the present paper contains neutral terms, it is very difficult to estimate an a priori bound. In order to overcome this difficulty, we use the Wirtinger inequality and develop some new mathematical analysis techniques.
-
(3)
A unified framework is established to handle competitive neural networks with neutral-type terms, time-varying delays and inertial terms.
The subsequent sections are organized as follows: In Sect. 2, the description of the model, some useful lemmas and notations are given. In Sect. 3, sufficient conditions are established for existence of anti-periodic solutions of considered neural networks. In Sect. 4, an example is given to show the feasibility of our results. Finally, Sect. 5 concludes the paper.
2 Model description and Lemmas
Consider the following generalized inertial competitive neutral-type neural networks with time-varying delays:
where \(i,j=1,2,\ldots ,n\); \((A_{i}x_{i})(t)\) is a neutral operator which is defined by
\(c_{i}\) and σ are constants with \(|c_{i}|\neq 1\) and \(\sigma >0\); the second order derivative is an inertial term; \(x_{i}(t)\) is the neuron current activity level; \(\varepsilon >0\) is a fast time scale; \(\alpha _{i}(t)>0\) is a variable coefficient; \(\beta _{i}(t)>0\) is a damping coefficient; \(b_{ij}(t)\) and \(c_{ij}(t)\) represent the connection weight and the synaptic weight of delayed synaptic efficiency; \(m_{ij}(t)\) the is synaptic efficiency; \(d_{j}\) is the constant external stimulus; \(B_{i}(t)\) is the strength of the external stimulus; \(f_{j}(x_{j}(t))\) is the output of neurons; \(\gamma _{ij}(t)>0\) is a transmission delay.
The initial values of system (2.1) are given by
where \(i,j=1,2,\ldots ,n\) and \(s\in [-\tau ,0]\) with \(\tau = \max_{1\leq i,j\leq n} \{\sigma ,\gamma _{ij}(t),t\in \mathbb{R} \}\).
Let \(\varepsilon =1\) and \(z_{i}(t)=\sum_{j=1}^{n}d_{j}m_{ij}(t)\), then (2.1) is transformed into
where \(d=\sum_{j=1}^{n}d_{j}^{2}>0\). Let
then (2.4) can be written as follows:
In view of the initial values of system (2.1), we can obtain the initial values of system (2.6)
where \(i,j=1,2,\ldots ,n\) and \(s\in [-\tau ,0]\).
Now, we give the famous Wirtinger inequality.
Lemma 2.1
([21, 22] Wirtinger inequality)
If u is a \(C^{1}\) function such that \(u(0) = u(T)\), then
where \(\|u'\|_{L_{2}}= (\int _{0}^{T}|u(t)|^{2}\,dt )^{ \frac{1}{2}}\) and \(\bar{u}=\frac{1}{T}\int _{0}^{T}|u(t)|\,dt\).
Remark 2.1
When u is an anti-periodic function, i.e., \(u(t+T)=-u(t)\), \(\forall t \in \mathbb{R}\), then
and the Wirtinger inequality is given by
In this paper, we also need the following lemmas.
Lemma 2.2
([22])
Let X and Y be Banach spaces, and let \(L: \operatorname{Dom}L \subset X \rightarrow Y \) be linear, \(N: X \rightarrow Y \) be continuous. Assume that L is one-to-one and \(K:=L^{-1}N\) is compact. Furthermore, assume there exists a bounded and open subset \(\varOmega \subset X\) with \(0\in \varOmega \) such that the equation \(Lu=\lambda Nu\) has no solutions in \(\partial \varOmega \cup \operatorname{Dom}L\) for any \(\lambda \in (0, 1)\). Then the problem \(Lu = Nu\) has at least one solution in Ω̄.
Lemma 2.3
Define A on \(C_{T}\)
where \(C_{T}=\{x:x\in C(\mathbb{R},\mathbb{R}),x(t+T)\equiv x(t)\}\), c is constant. When \(|c|\neq 1\), then A has a unique continuous bounded inverse \(A^{-1}\) satisfying
Obviously, we have
-
(1)
\(\|A^{-1}\|\le \frac{1}{ \vert 1- \vert c \vert \vert }\);
-
(2)
\(\int _{0}^{T}|[A^{-1}f](t)|\,dt\le \frac{1}{ \vert 1- \vert c \vert \vert }\int _{0}^{T}|f(t)|\,dt\), \(\forall f\in C_{T}\);
-
(3)
\(\int _{0}^{T}|[A^{-1}f](t)|^{2}\,dt\le \frac{1}{ \vert 1- \vert c \vert \vert }\int _{0} ^{T}|f(t)|^{2}\,dt\), \(\forall f\in C_{T}\).
In what follows, for \(u=(u_{1},u_{2},\ldots ,u_{3n})^{\top }\), denote
For \(i,j=1,2,\ldots ,n\), we list the following notations which will be used in this paper:
Throughout this paper, we need the following assumptions:
- (H1):
-
For \(i,j=1,2,\ldots ,n\) and \(t,x\in \mathbb{R}\), \(\alpha _{i},\beta _{i},b_{ij},c_{ij},B_{i},\gamma _{ij}\in C(\mathbb{R}, \mathbb{R})\) with
$$\begin{aligned}& \alpha _{i}(t+\omega )=\alpha _{i}(t),\qquad \beta _{i}(t+\omega )=\beta _{i}(t), \\& B _{i}(t+\omega )=B_{i}(t),\qquad \gamma _{ij}(t+\omega )=\gamma _{ij}(t), \\& b_{ij}(t+\omega )f_{j}(x)=-b_{ij}(t)f_{j}(-x),\qquad c_{ij}(t+\omega )f _{j}(x)=-c_{ij}(t)f_{j}(-x). \end{aligned}$$ - (H2):
-
For \(j=1,2,\ldots ,n\) and \(x,y\in \mathbb{R}\), there exist positive constants \(L_{j}>0\)
$$ \bigl\vert f_{j}(x)-f_{j}(y) \bigr\vert \leq L_{j} \vert x-y \vert . $$ - (H3):
-
For \(i,j=1,2,\ldots ,n\)
$$ 1-\gamma _{ij}'(t)>0,\quad t\in \mathbb{R}. $$
Remark 2.2
From \(1-\gamma _{ij}'(t)>0\), \(t\in \mathbb{R}\), it is easy to see that \(t-\gamma _{ij}(t)\) has the inverse function \(\varGamma _{ij}\) for \(i,j=1,2,\ldots ,n\) and \(t\in \mathbb{R}\). Hence, let \(t-\gamma _{ij}(t)=u _{ij}\), then \(t=\varGamma _{ij}(u_{ij})\).
3 Main results
Let
with the norm
Clearly, X is a Banach space. Let
where \(D(L)=\{u:u\in X,(A_{i}x_{i})',y_{i}',z_{i}'\in X\}\). Let there be a nonlinear operator \(N:X\rightarrow X\):
Clearly,
For \(u=(x_{1},\ldots ,x_{n},y_{1},\ldots ,y_{n},z_{1},\ldots ,z_{n})^{ \top }\in \operatorname{Im}L\), let the inverse operator of L be \(L^{-1}\) as follows:
where
Theorem 3.1
Assume that the assumptions (H1)–(H3) hold. Furthermore, the following assumption holds:
- (H4):
-
$$\begin{aligned} \varTheta =& \min_{1\leq i\leq n} \Biggl\{ 1-\sum _{i=1}^{n} \frac{\omega \vert 1- \alpha _{i}^{-} \vert }{ \vert 1- \vert c_{i} \vert \vert \pi } \\ &{}-\sum_{i=1}^{n} \Biggl[\beta _{i} ^{+}+\sum_{j=1}^{n} \biggl(b_{ij}^{+}L_{j} +c_{ij}^{+}L_{j} \max_{s\in \mathbb{R}}\frac{1}{ \vert 1-\gamma _{ij}'(s) \vert } \biggr) \Biggr] \frac{ \omega ^{2}}{ \vert 1- \vert c_{i} \vert \vert \pi ^{2}} \\ &{}-\sum_{i=1}^{n} \frac{B_{i}^{+}dL _{i}\omega ^{3}}{ \vert 1- \vert c_{i} \vert \vert \pi ^{3}} \Biggr\} >0. \end{aligned}$$
Then system (2.1) has at least one anti-periodic solution.
Proof
Consider the operator equation
where L and N are defined by (3.1) and (3.2). Let \(u\in D(L)\) be an arbitrary solution of (3.3), then
Multiplying by \(y_{i}'(t)\) on both sides of the second equation of (3.4) and integrating it over \([0,2\omega ]\), we have
From (3.5) and the Hölder inequality, we have
which results in
Consider the fourth term \(\sum_{j=1}^{n}c_{ij}^{+}L_{j} (\int _{0} ^{2\omega } |x_{j}(t-\gamma _{ij}(t))|^{2}\,dt )^{\frac{1}{2}}\) in (3.6). In view of Remark 2.2, we have
Multiplying by \(A_{i}x_{i}'(t)\) on both sides of the first equation of (3.4) and integrating it over \([0,2\omega ]\), we have
which results in
Using Lemma 2.3 and (3.9), we have
The Wirtinger inequality, (3.9) and (3.10) result in
Using again the Wirtinger inequality and (3.11), we have
Multiplying by \(z_{i}'(t)\) on both sides of the third equation of (3.4) and integrating it over \([0,2\omega ]\), we have
which results in
From (3.12) and (3.13), we have
which together with the Wirtinger inequality results in
From (3.8), (3.11), (3.12) and (3.15), we have
By using assumption (H4) and (3.16), we obtain
Equations (3.11) and (3.17) result in
Equations (3.14) and (3.17) result in
Since \(u\in X\) is an ω-anti-periodic function, there exist \(\xi _{i},\eta _{i},\zeta _{i}\in [0,2\omega ]\) such that
which results in
and
Similarly, using (3.17) and (3.19), we also find that there exist positive constants \(\widetilde{M}_{2}\), \(\widetilde{M}_{3}\) such that
Let
From Lemma 2.2, the operator equation \(Lu=Nu\) has at least one ω-anti-periodic solution in X. Thus, system (2.1) has at least one ω-anti-periodic solution. □
Remark 3.1
We very much want to obtain the globally exponential stability of system (2.1) with initial values conditions (2.3). But transforming system (2.6) of system (2.1) contains a neutral term \(A_{i}x_{i}\) which makes constructing the appropriate Lyapinov function very difficult. Hence, we wish that some authors will develop new methods to derive globally exponential stability of system (2.1) in the future.
4 A numerical example
In this section, a numerical example is given to illustrate the effectiveness of the results obtained in this paper.
Example 4.1
Consider the following inertial competitive neutral-type neural networks:
where \(i,j=1,2\),
After a simple calculation, we have
Thus,
and
Then
By Theorem 3.1, system (4.1) has at least a unique π-anti-periodic solution. For the trajectories of \(x_{i}(t)\), \(y_{i}(t)\), \(z_{i}(t)\) and \(m_{ij}(t)\) in system (4.1), see Figs. 1–3.
For \(i,j=1,2\), trajectories of \(x_{i}(t)\) and \(m_{ij}(t)\) in system (4.1)
For \(i,j=1,2\), trajectories of \(y_{i}(t)\) and \(m_{ij}(t)\) in system (4.1)
For \(i,j=1,2\), trajectories of \(x_{i}(t)\), \(y_{i}(t)\) and \(z_{i}(t)\) in system (4.1)
5 Conclusions
In this paper, we study a class of he anti-periodic solutions problem for inertial competitive neutral-type neural networks. By employing the Wirtinger inequality, topology degree theory and some analytic techniques, we have presented some new sufficient criteria for the existence of anti-periodic for the above neural networks. These criteria possess adjustable parameters which are important in some applied fields. Finally, an example is given to demonstrate the effectiveness of the obtained theoretical results. However, there exist many problems for further study such as the problems of the stability and other dynamic properties of anti-periodic solutions to neutral-type neural networks.
References
Dym, H., McKean, H.: Fourier Series and Integrals. Academic Press, San Diego (1985)
Komkov, V.: Euler’s buckling formula and Wirtinger’s inequality. Int. J. Math. Educ. Sci. Technol. 14, 661–668 (1983)
Coti-Zelati, V., Ekland, I., Sere, E.: A variational approach to homoclinic orbits in Hamiltonian systems. Math. Ann. 288(1), 133–160 (1990)
Lv, Y., Tang, C.L.: Existence of even homoclinic orbits for second-order Hamiltonian systems. Nonlinear Anal. 67(7), 2189–2198 (2007)
Du, B.: Stability analysis of periodic solution for a complex-valued neural networks with bounded and unbounded delays. Asian J. Control 20, 1–12 (2018)
Du, B., Lian, X., Cheng, X.: Partial differential equation modeling with Dirichlet boundary conditions on social networks. Bound. Value Probl. 2018, 50 (2018)
Xu, C., Li, P., Pang, Y.: Exponential stability of almost periodic solutions for memristor- based neural networks with distributed leakage delays. Neural Comput. 28(12), 2726–2756 (2016)
Xu, C., Chen, L.: Effect of leakage delay on the almost periodic solutions of fuzzy cellular neural networks. J. Exp. Theor. Artif. Intell. 30(6), 993–1011 (2018)
Meyer-Baese, A., Ohl, F., Scheich, H.: Singular perturbation analysis of competitive neural networks with different time scales. Neural Comput. 8(8), 1731–1742 (1996)
Meyer-Baese, A., Pilyugin, S.S., Chen, Y.: Global exponential stability of competitive neural networks with different time scales. IEEE Trans. Neural Netw. 14(3), 716–719 (2003)
Meyer-Baese, A., Roberts, R., Thmmler, V.: Local uniform stability of competitive neural networks with different time-scales under vanishing perturbations. Neurocomputing 73, 770–775 (2010)
Gu, H., Jiang, H., Teng, Z.: Existence and global exponential stability of equilibrium of competitive neural networks with different time-scales and multiple delays. J. Franklin Inst. 347(5), 719–731 (2010)
Liu, Y., Yang, Y., Liang, T., Li, L.: Existence of periodic solution for competitive neural networks with time-varying and distributed delays on time scales. Adv. Neural Netw. 22, 181–188 (2013)
Okochi, H.: On the existence of periodic solutions to nonlinear abstract parabolic equations. J. Math. Soc. Jpn. 40(1988), 541–553 (2006)
Li, Y., Yang, L., Wu, W.: Anti-periodic solution for impulsive BAM neural networks with time-varying leakage delays on time scales. Neurocomputing 149, 536–545 (2015)
Li, Y., Yang, L.: Anti-periodic solutions for Cohen–Grossberg neural networks with bounded and unbounded delays. Commun. Nonlinear Sci. Numer. Simul. 14(7), 3134–3140 (2009)
Li, Y., Xiang, J.: Existence and global exponential stability of anti-periodic solution for Clifford-valued inertial Cohen–Grossberg neural networks with delays. Neurocomputing 332, 259–269 (2019)
Huo, N., Li, B., Li, Y.: Existence and exponential stability of anti-periodic solutions for inertial quaternion-valued high-order Hopfield neural networks with state-dependent delays. IEEE Access 7, 60010–60019 (2019)
Xu, C., Chen, L., Guo, T.: Anti-periodic oscillations of bidirectional associative memory (BAM) neural networks with leakage delays. J. Inequal. Appl. 2018, 68 (2018)
Xu, C., Li, P.: Existence and exponentially stability of anti-periodic solutions for neutral BAM neural networks with time-varying delays in the leakage terms. J. Nonlinear Sci. Appl. 9(3), 1285–1305 (2016)
Mawhin, J., Willem, M.: Critical Point Theory and Hamiltonian Systems. Springer, Berlin (1989)
Amster, P.: Topological Methods in the Study of Boundary Valued Problems. Springer, New York (2013)
Zhang, M.: Periodic solutions of linear and quasilinear neutral functional differential equations. J. Math. Anal. Appl. 189, 378–392 (1995)
Lu, S., Ge, W., Zheng, Z.: Periodic solutions to neutral differential equation with deviating arguments. Appl. Math. Comput. 152, 17–27 (2004)
Acknowledgements
The author would like to thank the editor and the referees for their valuable comments and suggestions, which improved the quality of our paper.
Availability of data and materials
Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.
Funding
The work is supported by Natural Science Foundation of Jiangsu High Education Institutions of China (Grant No. 17KJB110001).
Author information
Authors and Affiliations
Contributions
Only the author contributed to the writing of this paper. The author read and approved the final manuscript.
Corresponding author
Ethics declarations
Competing interests
The author declares to have no competing interests.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
About this article
Cite this article
Du, B. Anti-periodic solutions problem for inertial competitive neutral-type neural networks via Wirtinger inequality. J Inequal Appl 2019, 187 (2019). https://doi.org/10.1186/s13660-019-2136-1
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s13660-019-2136-1
MSC
- 34K20
Keywords
- Wirtinger inequality
- Anti-periodic solution
- Existence