\(p\)th moment exponential stability of stochastic recurrent neural networks with time-varying delays (Q2371010)

The authors discuss the \(p\)th moment exponential stability of zero solution to the generalized stochastically perturbed neural network model with time-varying defined by the state equation: \[ \begin{aligned} dx_{i}(t) =\left[ -c_{i}x_{i}(t)+\sum_{j=1}^{n}a_{ij}f_{j}(x_{j}(t))+ \sum_{j=1}^{n}b_{ij}g_{j}(x_{j}(t-\tau _{j}(t)))\right] dt \\ \left. +\sum_{j=1}^{n}\sigma _{ij}(t,x_{j}(t),x_{j}(t-\tau _{j}(t)))d\omega _{j}(t),\;i=1,2,,,,n,\right.\end{aligned} \] where \(\tau _{j}(t)>0\) is the transmission delay and \(\omega (t)\) denotes an \(n-\)dimensional Brownian motion on a complete probability space. The following assumptions are given: (1) \(\tau _{j}(t)\) is a differentiable function with a constant \(\alpha >0\) such that \(\frac{d}{dt}\tau _{j}(t)\leq \alpha <1\), (2) \ \(f_{j},g_{j}\) satisfy the Lipschitz condition, (3) \(f(0)=g(0)=\sigma (t,0,0)=0\), \[ \text{trace}\left[ \sigma ^{T}(t,x,y)\sigma (t,x,y)\right] \leq \sum_{i=1}^{n}\left( \mu _{i}x_{i}^{2}+\nu _{i}y_{i}^{2}\right) ,\;\;(t,x,y)\in R\times R^{n}\times R^{n}.\tag{4} \] It is known that under the above conditions there exists a unique global solution to the above state equation. Thus, the authors discuss the \(p\)th moment exponential stability of zero solution. The main theorem is the following. Theorem. If \(\rho \left[ C^{-1}(MM_{1}K+MM_{2}K+NN_{1}+NN_{2})\right] \leq 1,\;\) then the zero solution to the above system is \(p\)th moment exponentially stable. This theorem is a generalization of the theorem on exponential stability proven by \textit{L. Wan} and \textit{J. Sun} [Phys. Lett., 342, No. 4, 331--340 (2005; Zbl 1222.93200)]. The authors also give two numerical examples.