DETECTING NONLINEARITY IN MULTIVARIATE TIME-SERIES

We propose an extension to time series with several simultaneously mea sured variables of the nonlinearity test, which combines the redundanc y-linear-redundancy approach with the surrogate data technique. For se veral variables various types of redundancies can be defined, in order to test specific dependence structures between/among (groups of) vari ables. The null hypothesis of a multivariate linear stochastic process is tested using the multivariate surrogate data. The linear redundanc ies are used in order to avoid spurious results due to imperfect surro gates, The method is demonstrated using two types of numerically gener ated multivariate series (linear and nonlinear) and experimental multi variate data from meteorology and physiology.