On the convergence of a perturbed one dimensional Mann's process

Abstract

We study a perturbed version of Mann's iterative process (DMP), defined by \[ x_{n+1} = (1 - θ_n)x_n + θ_n f(x_n) + r_n, \] where $f:[0,1]\to [0,1]$ is a continuous function, $\{{\theta }_n\}\subset [0,1]$ is a given sequence, and $\{r_n\}$ represents an error term. We prove that if the sequence $\{{\theta }_n\}$ converges sufficiently slowly to zero and the error term $r_n$ is suitably small at infinity, then any sequence $\{x_n\}\subset [0,1]$ generated by this process converges to a fixed point of $f$. In addition, we investigate the asymptotic behavior of the trajectories $x(t)$ as $t\to \infty$ for a continuous-time version of the process (DMP). We emphasize the parallels between the discrete and the continuous dynamics. Furthermore, through numerical experiments, we analyze the influence of the sequence $\{{\theta }_n\}$ and the error terms on the stability and the convergence rate of the the discrete and the continuous processes. Notably, we observe that the (DMP) algorithm, when affected by stochastic and relatively large error terms, can outperform the bisection method in efficiently identifying the fixed point set of the function $f$.

10 pages