On the link between small ball probabilities and the quantization problem for Gaussian measures on Banach spaces (Q1868455): Difference between revisions

Let \(\mu\) be a Gaussian measure on a Banach space \(E\), \(\mathcal{E}_n\subset E\) be a subset of cardinality \(n\) and \(r>0\). The number \(\delta(n)= \inf\{\int \text{dist}(x,\mathcal{E}_n)\mu (dx): \mathcal{E}_n\subset E\}\) is called the quantization error of \(\mu\). The problem of determining \(\delta(n)\) and the minimizing set \(\mathcal{E}_n\) is of cosiderable importance in engineering. Except for particular cases, explicit expressions for \(\delta(n)\) are unknown. The authors present asymptotics of \(\delta(n)\) when \(n\rightarrow \infty \) and compare the quantization error with the average distance which is obtained when the set \(\mathcal{E}_n\) is chosen by taking \(n\) i.i.d. copies of random elements with law \(\mu\). The approach is based on the study of the asymptotics of \(\mu(B_{\varepsilon})\), where \(B_{\varepsilon}\) is the ball with center in origin and radius \(\varepsilon\), when \(\varepsilon \rightarrow 0\).