Timing of Transients: Quantifying Reaching Times and Transient Behavior in Complex Systems

TL;DR

This paper introduces two novel metrics, Regularized Reaching Time and AUDIC, to quantify transient behavior in complex systems, addressing issues of non-invariance and interpretability, and providing early-warning signals for bifurcations.

Contribution

The paper proposes two new metrics for transient analysis, overcoming limitations of existing methods, and demonstrates their effectiveness as early-warning signals for bifurcations in complex systems.

Findings

T_{RR} captures critical slowing down as an early-warning indicator.

AUDIC measures reluctance or eagerness of trajectories to reach attractors.

Metrics reveal regularity in chaotic basin structures and sensitivity to prebifurcational changes.

Abstract

When quantifying the time spent in the transient of a complex dynamical system, the fundamental problem is that for a large class of systems the actual time for reaching an attractor is infinite. Common methods for dealing with this problem usually introduce three additional problems: non-invariance, physical interpretation, and discontinuities, calling for carefully designed methods for quantifying transients. In this article, we discuss how the aforementioned problems emerge and propose two novel metrics, Regularized Reaching Time ($T_{RR}$) and Area under Distance Curve (AUDIC), to solve them, capturing two complementary aspects of the transient dynamics. $T_{RR}$ quantifies the additional time (positive or negative) that a trajectory starting at a chosen initial condition needs to reach the attractor after a reference trajectory has already arrived there. A positive or negative value means that it arrives by this much earlier or later than the reference. Because $T_{RR}$ is an analysis of return times after shocks, it is a systematic approach to the concept of critical slowing down [1]; hence it is naturally an early-warning signal [2] for bifurcations when central statistics over distributions of initial conditions are used. AUDIC is the distance of the trajectory to the attractor integrated over time. Complementary to $T_{RR}$, it measures which trajectories are reluctant, i.e. stay away from the attractor for long, or eager to approach it right away. (... shortened for arxiv listing, full abstract in paper ...) New features in these models can be uncovered, including the surprising regularity of the Roessler system's basin of attraction even in the regime of a chaotic attractor. Additionally, we demonstrate the critical slowing down interpretation by presenting the metrics' sensitivity to prebifurcational change and thus how they act as early-warning signals.