Two-frequency self-oscillations in a FitzHugh-Nagumo neural network (Q2357122)

The authors derive a mathematical model describing a one-dimensional array of FitzHugh-Nagumo neuron models which are diffusively coupled. It reads \[ \begin{aligned} {du_j\over dt} &= v_j+\varepsilon\alpha\Biggl(u_j- {u^3_j\over 3}\Biggr),\\ {dv_j\over dt} &= -u_j+ d(u_{j+1}- 2u_j+ u_{j-1})-\varepsilon w_j- \varepsilon\beta(v_j- w_j),\\ {dw_j\over dt} &= -u_j-\varepsilon w_j,\end{aligned}\tag{\(*\)} \] where \(u_0= u_1\), \(u_{N+1}= u_N\), \(0\leq j\leq N\), \(0<\varepsilon<< 1\), \(\alpha\), \(\beta\), d are positive parameters. The authors prove that \((*)\) has coexisting stable two-dimensional invariant tori whose number increases with increasing \(\alpha\) and \(N\).