Kernel-based method for joint independence of functional variables.

AbstractThis work investigates the problem of testing whether <italic>d</italic> functional random variables are jointly independent, using a modified estimator of the <italic>d</italic>-variable Hilbert Schmidt Indepedence Criterion (<italic>d</italic>HSIC). We then get the asymptotic normality of this estimator both under the joint independence hypothesis and under the alternative hypothesis. A simulation study shows the good performance of the proposed test on a finite sample. [ABSTRACT FROM AUTHOR]