Projection-Induced Determinism: A Geometric Constraint on the Observability of Early Warning Signals in Complex Systems

Read full text Summarise with VIRAT
Authors: Bernd von Mallinckrodt
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This paper introduces Projection-Induced Determinism (PID) as a geometric framework for understanding the mechanism-dependent reliability of early warning signals (EWS) in complex dynamical systems. Classical EWS — such as rising variance and increasing lag-1 autocorrelation — are derived from critical slowing down theory and are widely used to anticipate bifurcation-induced transitions. However, empirical evidence shows that these indicators often fail, appear too late, or exhibit sign inversion prior to known transitions. Existing explanations typically attribute these discrepancies to noise, limited data, or insufficient proximity to the critical point. We argue that these explanations are incomplete. The central claim of this work is that EWS failure is, in part, a geometric observability problem rather than solely a statistical one. In real-world systems, the full system state is rarely observable. Instead, measurements correspond to projections from a high-dimensional state space ℝⁿ onto a lower-dimensional observation space ℝᵏ. We show that this projection induces a deterministic distortion of the system’s covariance structure. This effect is formalized as Projection-Induced Determinism (PID): a constraint linking the geometry of the observation map to the observability of dynamical precursors. Using a linearized stochastic dynamical system, we derive the covariance projection identity and demonstrate how alignment between observation directions and the system’s critical eigenvectors governs the detectability of EWS. We provide explicit conditions under which variance signals are attenuated and autocorrelation trends invert sign, despite the presence of critical slowing down in the full system. These results establish an observability bound: there exist configurations in which no amplitude-based indicator can recover early warning signals from projected data, regardless of data quality or sampling effort. Mechanism-dependence of EWS is thus shown to be a geometric consequence of projection, rather than an empirical anomaly. The implications are structural. This work reframes early warning signal research from a problem of signal detection to a problem of observability under projection. It provides a theoretical foundation for interpreting EWS failures, motivates the development of structure-sensitive indicators, and suggests new experimental strategies based on optimizing observation geometry. early warning signals, critical transitions, projection geometry, observability, dimensionality reduction, complex systems, bifurcation theory, covariance structure, critical slowing down, autocorrelation, variance, dynamical systems, nonlinear systems, Lyapunov equation, eigenvectors, system collapse

Fields of Study & Subjects

Ecosystem dynamics and resilience Complex Systems and Dynamics Mathematical and Theoretical Epidemiology and Ecology Models

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
30 March 2026
Language
EN
Document type
Preprint