Stability of artificial neural networks with impulses (Q1883183): Difference between revisions

The author studies the existence and uniqueness of the positive equilibrium of the following impulsive problem \[ \frac{dx_i(t)}{dt}=-a_ix_i(t)+\sum_{j=1}^m b_{ij}f_j(x_j(t))+c_i,~~t>t_0,~t\neq t_k, \] \[ x(t_0+)=x_0\in\mathbb{R}^m, \] \[ x_i(t_0+)-x_i(t_k-)=I_k(x_i(t_k-)),~~i=1,2,\dots,m,~~~k=1,2,\dots,~~t_k\rightarrow\infty, \] and its discrete analog \[ y_i(n+1)=\frac{1}{1+a_ih}y_i(n)+\frac{h}{1+a_ih}\sum_{j=1}^m b_{ij}f_j(y_j(n)),~~i=1,2,\dots,m,~~ n>n_0, \] \[ y_j(n_j+1)-y_j(n_j)=I_j(y_j(n_j)), ~~j=1,2,\dots. \] By constructing Lyapunov functions the author obtains explicit stability conditions for this equilibrium. A comparison with known results for nonimpulsive systems is presented.