Constructive quantization: approximation by empirical measures (Q376699)

The authors investigate the approximation of a probability measure \(\mu\) on \(\mathbb{R}^d\), \(d \geq 3\), by its empirical measure w.r.t. the minimal \(L_p\)-metric (\(L_p\)-Wasserstein metric). For \(p \in [1, \frac{1}{d})\) and finite absolute \(q\)-th order moment \(\mu_q\) of \(\mu\) they establish an upper bound of the order of the form \( \kappa \, \mu_q \, N^{-1/d}\), a so-called Pierce type estimate. For the uniform distribution on \([0,1)^d\) they obtain an exact limit theorem of order \(N^{-1/d}\) and establish a corresponding limit result for the optimal quantization by \(N\) points. Further it is shown that the order of quantization by empirical measures is optimal under weak distributional assumptions.