A finite-time convergent neural dynamics for online solution of time-varying linear complex matrix equation

This paper proposes and investigates a finite-time convergent neural dynamics (FTCND) for online solution of time-varying linear complex matrix equation in complex domain. Different from the conventional gradient-based neural dynamical method, the proposed method utilizes adequate time-derivative information of time-varying complex matrix coefficients. It is theoretically proved that our FTCND model can converge to the theoretical solution of time-varying linear complex matrix equation within finite time. In addition, the upper bound of the convergence time is derived analytically via Lyapunov theory. For comparative purposes, the conventional gradient-based neural dynamics (GND) is developed and exploited for solving such a time-varying complex problem. Computer-simulation results verify the effectiveness and superiorness of the FTCND model for solving time-varying linear complex matrix equation in complex domain, as compared with the GND model.