Some Results on the Lp-convergence (p⩾1) of U-statistics

SU CHUN
DOI: https://doi.org/10.1016/0167-9473(87)90055-7
IF: 2.035
1987-01-01
Computational Statistics & Data Analysis
Abstract:Let X1, X2, … be a strongly stationary and m-dependent random sequence and h(x, y) be a Borel-measurable symmetric function, then Un=1C2n∑1⩽i<j⩽nh(Xi,Xj is called the U-statistic of m-dependent samples with kernel h. Under the condition: b=max2⩽j⩽m+2E∣h(X1,Xj)∣p<∞,p⩾1 we obtain the rate of convergence of E∣Un−θ∣p→0 (n→ ∞), p⩾1, where θ=Eh(X1, X2+m), and the rate is the same as that for iid samples in Chen Xiru's paper (1980).
What problem does this paper attempt to address?