OPTIMAL CONVERGENCE PROPERTIES OF KERNEL DENSITY ESTIMATORS WITHOUT DIFFERENTIABILITY CONDITIONS

Let X1, X2,..., X(n) be independent observations from an (unknown) absolutely continuous univariate distribution with density f and let f triple-overdot (x) = (nh)-1 SIGMA-i = 1n K[(x - X(i)/h] be a kernel estimator of f(x) at the point x, - infinity < x < infinity, with h = h(n) (h(n) --> 0 and nh(n) --> infinity, as n --> infinity) the bandwidth and K a kernel function of order r. "Optimal" rates of convergence to zero for the bias and mean square error of such estimators have been studied and established by several authors under varying conditions on K and f. These conditions, however, have invariably included the assumption of existence of the r-th order derivative for f at the point x. It is shown in this paper that these rates of convergence remain valid without any differentiability assumptions on f at x. Instead some simple regularity conditions are imposed on the density f at the point of interest. Our methods are based on certain results in the theory of semi-groups of linear operators and the notions and relations of calculus of "finite differences".