Showing 1–1 of 1 results for author: Erofeev, A

Using Echo-State Networks to Reproduce Rare Events in Chaotic Systems

Authors: Anton Erofeev, Balasubramanya T. Nadiga, Ilya Timofeyev

Abstract: We apply the Echo-State Networks to predict the time series and statistical properties of the competitive Lotka-Volterra model in the chaotic regime. In particular, we demonstrate that Echo-State Networks successfully learn the chaotic attractor of the competitive Lotka-Volterra model and reproduce histograms of dependent variables, including tails and rare events. We use the Generalized Extreme V… ▽ More We apply the Echo-State Networks to predict the time series and statistical properties of the competitive Lotka-Volterra model in the chaotic regime. In particular, we demonstrate that Echo-State Networks successfully learn the chaotic attractor of the competitive Lotka-Volterra model and reproduce histograms of dependent variables, including tails and rare events. We use the Generalized Extreme Value distribution to quantify the tail behavior. △ Less

Submitted 22 May, 2025; originally announced May 2025.

MSC Class: 37N99; 68T30