Original title: Saddle avoidance of noise-induced transitions in multiscale systems
Authors: Reyk Börner, Ryan Deeley, Raphael Römer, Tobias Grafke, Valerio Lucarini, Ulrike Feudel
The article explores transitions between different states in dynamic systems affected by weak Gaussian noise. Normally, these transitions are thought to pass through a critical point called a “saddle.” However, the study reveals that when there’s a difference in timescales, these systems can avoid this critical point. Using models from neuroscience and ecology, researchers find that actual transitions can vary significantly from the predicted paths, even with weak noise. This unexpected behavior is due to a flat quasipotential. They suggest a new approach using the Onsager-Machlup action to better anticipate how these transitions occur. Essentially, this work challenges existing ideas about transition paths in noisy systems and proposes a new method to understand these processes more accurately.
Original article: https://arxiv.org/abs/2311.10231