Identifying early-warning signals of critical transitions with strong noise by dynamical network markers.

ABSTRACT: Identifying early-warning signals of a critical transition for a complex system is difficult, especially when the target system is constantly perturbed by big noise, which makes the traditional methods fail due to the strong fluctuations of the observed data. In this work, we show that the critical transition is not traditional state-transition but probability distribution-transition when the noise is not sufficiently small, which, however, is a ubiquitous case in real systems. We present a model-free computational method to detect the warning signals before such transitions. The key idea behind is a strategy: "making big noise smaller" by a distribution-embedding scheme, which transforms the data from the observed state-variables with big noise to their distribution-variables with small n

SUBMITTER: Liu R 

PROVIDER: S-EPMC4673532 | biostudies-literature | 2015 Dec

REPOSITORIES: biostudies-literature