Some results on the \(L_ p\)-convergence \((p\geq 1)\) of U-statistics (Q1104658): Difference between revisions

Let \(X_ 1,X_ 2,..\). be a strongly stationary and m-dependent random sequence and h(x,y) be a Borel-measurable symmetric function, then \(U_ n=C_ n^{-2}\sum_{1\leq i<j\leq n}h(X_ i,X_ j)\) is called the U- statistic of m-dependent samples with kernel h. Under the condition: \[ b=\max_{2\leq j\leq m+2}E| h(X_ l,X_ j)|^ p<\infty,\quad p\geq 1, \] we obtain the rate of convergence of \( E| U_ n-\theta |^ p\to O(n\to \infty),\) \(p\geq 1,\) where \(\theta =E h(X_ 1,X_{2+m})\).