An independence test based on recurrence rates (Q2181733): Difference between revisions

Let \((X_1,Y_1)\), \((X_2,Y_2),\dots, (X_n,Y_n)\) be a sample of independent copies of random element \((X,Y)\), where \(X\in S_X\), \(Y\in S_Y\) and \(S_X\), \(S_Y\) are metric spaces. The authors of the paper propose a new test of independence between random elements \(X\) and \(Y\). The presented test is based on recurrence rates by using information of \[ \boldmath{1}_{\{d_1(X_i,X_j)<r\}}, \mathbf{1}_{\{d_2(Y_i,Y_j)<s\}} \] for any values of \(r\), \(s\). Here \(d_1\), \(d_2\) denote distances in metric spaces \(S_X\), \(S_Y\) and pairs of indices \(\{i,j\}\) are arbitrary from the set \( \{1,2,\ldots,n\}\). The proposed test has very good behaviour under several alternatives.