Seasonal footprints on ecological time series and jumps in dynamic states of protein configurations from a non-linear forecasting method characterization

TL;DR

This paper applies a nonlinear forecasting method to analyze ecological and protein data, revealing how environmental rhythms and mutations influence system dynamics through prediction quality metrics.

Contribution

It introduces a novel application of nonlinear forecasting to characterize dynamic states in ecological and molecular systems, highlighting the impact of environmental cycles and mutations.

Findings

Detection of environmental rhythms in ecological data.

Identification of mutation-induced changes in protein dynamics.

Use of prediction quality to characterize complex system behavior.

Abstract

We have analyzed phenology data and protein configurations from molecular dynamics simulations with the nonlinear forecasting method proposed by May and Sugihara. Our primary focus in this work is to characterize the dynamic state of a system by quantifying prediction quality from data. Full plots of prediction quality as a function of dimensionality $E$ and forecasting time $T_p$, the two basic parameters of the method, give fast and valuable information about Complex Systems dynamics. We detect changes in protein dynamics due to mutations and, regarding ecology data, we show how cycles and {\it rhythms} of the environment manifests in parameter space $(E,T_p)$ for some species.