We establish consistent estimators of jump positions and jump altitudes of a multi-level step function that is the best $L^2$-approximation of a probability density function $f$. If $f$ itself is a step-function the number of jumps may be unknown.
density estimation, argmin-theorem, step functions, martingale inequalities, multivariate cadlag stochastic processes
62F10, 62G07, 60G44