Weak invariance of the multidimensional rank statistic with unbounded scores for nonstationary absolutely regular processes (Q1200623)

Let \(X_{ni}\), \(1\leq i\leq n\), \(n\geq 1\), be a nonstationary absolutely regular triangular array of \(\mathbb{R}^ k\)-valued random vectors with continuous distribution function \(F_{ni}(x)\), \(x\in\mathbb{R}^ k\). Let \[ S_{n,m}=\sum^ m_{i=1}c_{ni}a_{n,m}(R^{(1)}_{n,m,i},\dots,R^{(k)}_{n,m,i}), \] where the \(c_{ni}\) are the regression constants defined by a continuous function \(h(x)\) on \((0,1]\) as \[ c_{ni}=h(i/(n+1)),\quad 1\leq i\leq n,\;n\geq 1, \] the \(a_{n,m}(\cdot)\) are the scores defined by a continuous function \(J\) on (0,1) as \[ a_{n,m}(i_ 1,\dots,i_ k)=J(i_ 1/(m+1),\cdots,i_ k/(m+1)) \] and \(R^{(j)}_{n,m,i}\) is the rank of \(X^{(j)}_{ni}\) among \(\{X^{(j)}_{nl},\;1\leq l\leq m\}\). The author proves a weak invariance theorem for the statistic \(S_{n,m}\) when the regression constants and the scores are unbounded. The results obtained are extensions of the results of the author and \textit{M. L. Puri} [J. Multivariate Anal. 36, No. 2, 204-221 (1991; Zbl 0721.60036)].