Pseudo-Bifurcations in Stochastic Non-Normal Systems

Virgile Troude, Sandro Lera, Ke Wu, Didier Sornette · 2024

Read Pseudo-Bifurcations in Stochastic Non-Normal Systems by Virgile Troude, Sandro Lera, Ke Wu, Didier Sornette (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-01833.txt.

Introduction

Virgile Troude and co-authors warn that many supposed tipping points may be illusions. Ecosystems, brains, markets, and climate systems sometimes lurch suddenly, and scientists read the wobble as a warning of collapse. Their 2024 paper names a decoy: pseudo-bifurcations in stochastic non-normal systems, where uneven interactions fake instability in a system that is stable long term.

The fakes reproduce every classic early-warning sign — slowing recovery, rising variance, collapsing dimension. Brain data from epileptic seizures supply a living example. Because the alarms can ring far from any real threshold, crisis calls get biased early. The authors urge a rethink of how abrupt shifts are diagnosed everywhere. Systems scientists get a hidden geometry that explains false alarms.

Catalog record

CREATORVirgile Troude, Sandro Lera, Ke Wu, Didier Sornette
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE48 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-01833

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Pseudo-Bifurcations in Stochastic Non-Normal Systems Virgile Troude, Sandro Lera, Ke Wu, Didier Sornette Abstract We challenge the prevailing emphasis on bifurcations as the primary mechanism behind many abrupt surges and transient extremes in complex systems and propose an alternative more universally applicable explanation based on non-normal dynamics. We demonstrate that linear or approximately linear stochastic systems near a dynamical attractor exhibit transient repulsive dynamics – termed pseudo-bifurcations – when interacting components are sufficiently asymmetric and hierarchically organized, i.e., non-normal. These pseudo-bifurcations produce early-warning signals commonly linked to bifurcations, such as dimension reduction, critical slowing down, and increased variance. Furthermore, we show that, as actual bifurcations approach, non-normal transients also emerge, complicating their distinction and potentially creating a bias that suggests the system is much closer to a critical point than it actually is. We support our analytical derivations by empirically demonstrating that the brain exhibits clear signs of such non-normal transients during epileptic seizures. Many systems suspected of approaching critical bifurcation points should be reconsidered, as non-normal dynamics offer a more generic explanation for the observed phenomena across natural, physical, and social systems. Introduction Many of the dynamics driving complex systems in nature and society are marked by abrupt changes that result in transitions to fundamentally different behaviors [ 1 , 2 , 3 , 4 ] . In recent decades, the dominant framework for understanding these transitions has centered around the concept of bifurcations [ 5 , 6 , 4 , 7 ] . A bifurcation is a point at which a small change in a system’s control parameter causes a qualitative shift in its equilibrium state, resulting in a new and distinct configuration of system behavior. There are numerous examples of systems that display dynamics resembling those of bifurcations. For instance in ecology, predator-prey dynamics can lead to the sudden extinctions of a species [ 8 ] , small shifts in climate patterns may induce inter-glacial periods [ 9 ] , vegetation losses above a critical threshold results in desertification [ 10 ] , or nutrient overload can cause a lake to suddenly transition from a clear-water to a algae dominated state [ 11 ] . In social systems, notable examples are financial markets that experience long periods of low volatility followed by abrupt crashes [ 12 ] and growing dissatisfaction with political conditions, which can trigger widespread unrest or revolution [ 13 ] . Health-related examples involve the sudden onset of conditions such as microbiome dysregulation [ 14 ] , transitions between different states that represent distinct illnesses based on varying internal and external influences [ 15 ] , depressive episodes [ 16 ] , or epileptic seizures [ 17 ] . As a system approaches a critical bifurcation threshold, it is often marked by increased variance and autocorrelation in observable metrics. These statistical signatures reveal that the system’s recovery from perturbations is slowing as it approaches bifurcation, alongside an increasing sensitivity to fluctuations in the system’s state–two complementary indicators that signal the proximity to a critical transition. Thus, it has been suggested that these statistical signatures serve as important early-warning signals for the onset of a looming bifurcation point [ 5 , 6 , 4 , 7 ] . Despite the insights offered by bifurcation theory, the framework itself encounters at least two major challenges. First, it relies on the existence of a well-defined control parameter and its proximity to a critical threshold. Second, it assumes the presence of a driving force that pushes the system toward or beyond this threshold. These assumptions are not only challenging to verify empirically, but also conceptually restrictive and lack general applicability. It is therefore not surprising that the question of whether certain systems are undergoing critical bifurcations continues to be a subject of debate. For example, evidence is mixed on whether the onset of epilepsy is truly driven by critical transitions [ 18 , 17 ] . Similarly, it remains uncertain whether components of the Earth’s system, such as the Amazon rainforest or the Atlantic Meridional Overturning Circulation (AMOC), are approaching bifurcation points [ 19 ] . In this work, we present an alternative, more general mechanism that relies solely on a stable equilibrium attractor with linear non-normal stochastic dynamics around it. We demonstrate that small perturbations around the equilibrium produce transient repulsions - we dub them pseudo-bifurcations in this context - with similar statistical signatures to those observed near genuine bifurcations. Much like the effective dimension reduction seen during the onset of a bifurcation, we find that non-normal transients are confined to movements within a two-dimensional subspace. We further find that these transients are accompanied by statistical markers such as critical slowing down and increased autocorrelation. We empirically illustrate our findings by analyzing EEG signals that capture brain activity during epileptic seizures, revealing clear evidence of non-normal dynamics. We further demonstrate that the approach of actual bifurcations is accompanied by non-normal transients, making it difficult to distinguish between the two phenomena and potentially creating a bias that suggests the system is much closer to a critical point than it actually is. Finally, we show that the onset of non-normal transients can be anticipated by measuring the system’s alignment along its non-normal mode , and provide a numerical algorithm for stable identification of these non-normal modes. This opens a new path towards the anticipation and control of instabilities in complex systems. The only requirement to observe such transients is that the system exhibits sufficient non-normality - meaning its components interact within asymmetric hierarchical structures, a feature common in many social and natural systems [ 20 , 21 ] . This characteristic is also found in non-network systems, such as turbulent flows, where non-normal dynamics arise from the interplay of viscous and advective terms, which leads to asymmetric, non-orthogonal interactions among velocity perturbations [ 22 ] . From a mathematical perspective, our findings are generic. To observe pseudo-bifurcation, all one needs to assume is that the Jacobian of the linearized system, around its equilibrium point, forms a non-normal matrix. This is a general condition, as non-normal matrices form a measure-one set in the space of all matrices. Our findings thus suggest that many systems previously thought to be approaching criticality may need to be reconsidered, as they could instead be stable yet non-normal systems. More generally, our results offer a different perspective on how to interpret precursory signals in dynamical systems and highlight the role of non-normal dynamics in improving our understanding and management of a wide range of physical, natural, and social systems. Results Reduced Non-Normal Dynamics We consider a general n -dimensional system dynamics \dot{\mathbf{x}}=\mathbf{A}~\mathbf{x}+B~\boldsymbol{\eta} (1) corresponding to the linearization of a general dynamics in the neighborhood of a stable attractor. The vector \mathbf{x} quantifies the distance from the stable attractor. The matrix \mathbf{A} , which governs interactions of the system, is an n\times n matrix with eigenvalues \lambda_{1},\ldots,\lambda_{n} , all of which are assumed to have negative real parts, ensuring that the equilibrium point \mathbf{x}=0 remains stable. The n -dimensional noise term \boldsymbol{\eta} is assumed to be normally distributed and B quantifies the amplitude of the noise term. For simplicity, we take B as scalar but reintroducing its tensorial nature does not change the main results presented

Read the complete text