Bounds for the Gini mean difference of an empirical distribution (Q2488707)

The Gini mean difference for the sequence \(x= (x_1,\dots,x_n)\) and the function \(f\) defined for the points \(x_i\) \((i=1, \dots,n)\), with weights \(p=(p_1, \dots,p_n)\) where \(p_i\geq 0\) and \(\sum^n_{i=1}p_i=1\), is defined by \(M_p(f,x)= \sum_{1\leq i<j\leq n} p_ip_j|f(x_j)-f(x_i)|\). The authors prove certain sharp bounds for these differences for the particular case \(f(x)=x\), as well as for a general \(f\) having bounded or absolutely continuous derivatives on a closed interval.