Inferring the connectivity of coupled oscillators from time-series statistical similarity analysis

摘要: A system composed by interacting dynamical elements can be represented a network, where the nodes represent that constitute system, and links account for their interactions, which arise due to variety of mechanisms, are often unknown. popular method inferring connectivity (i.e., set among pairs nodes) is performing statistical similarity analysis time-series collected from dynamics nodes. Here, considering two systems coupled oscillators (Kuramoto phase Rossler chaotic electronic oscillators) with known controllable coupling conditions, we aim at testing performance this inference method, using linear non measures. We find that, under adequate network perfectly inferred, i.e., no mistakes made regarding presence or absence links. These conditions perfect require: i) an appropriated choice observed variable analysed, ii) interaction strength, iii) thresholding matrix. For units considered here measure performs, in general, better than non-linear ones.