Asymptotic local uniformity of the quantization error for Ahlfors-David probability measures

Let $\mu$ be an Ahlfors-David probability measure on $\mathbb{R}^q$, namely, there exist some constants $s_0>0$ and $\epsilon_0,C_1,C_2>0$ such that \[ C_1\epsilon^{s_0}\leq\mu(B(x,\epsilon))\leq C_2\epsilon^{s_0},\;\epsilon\in(0,\epsilon_0),\;x\in{\rm supp}(\mu). \] For $n\geq 1$, let $\alpha_n$ be an $n$-optimal set for $\mu$ of order $r$ and $(P_a(\alpha_n))_{a\in\alpha_n}$ an arbitrary Voronoi partition with respect to $\alpha_n$. The $n$th quantization error $e_{n,r}(\mu)$ for $\mu$ of order $r$ is given by $e^r_{n,r}(\mu):=\int d(x,\alpha_n)^rd\mu(x)$. Write \[ I_a(\alpha,\mu):=\int_{P_a(\alpha_n)}d(x,\alpha_n)^rd\mu(x),\;a\in\alpha_n. \] We prove that, $\underline{J}(\alpha_n,\mu):=\min_{a\in\alpha_n}I_a(\alpha,\mu)$, $\overline{J}(\alpha_n,\mu):=\max_{a\in\alpha_n}I_a(\alpha,\mu)$ and the error difference $e^r_{n,r}(\mu)-e^r_{n+1,r}(\mu)$ are of the same order as $\frac{1}{n}e^r_{n,r}(\mu)$. This, together with Graf and Luschgy's work, yields that all the above three quantities are of the same order as $n^{-(1+\frac{r}{s_0})}$.

PDF Abstract