On probabilities of linear sets

Let \(\underline X=(X_ k)\) be a sequence of independent nondegenerate real random variables, which can be seen as a random element in \(\mathbb{R}^ \infty\). Consider the property: if for every linear Borel subspace \(E\) of \(\mathbb{R}^ \infty\) and every \(\underline t \in \mathbb{R}^ \infty\), there exists an integer \(n\), a sequence \(\underline s\) with \(s_ k=0\) for \(k\geq n\), and a linear subspace \(F\) of \(\mathbb{R}^ n\) such that \[ P\bigl( \{\underline X-\underline t \in E\} \triangle \{((X_ k)-(s_ k))_{k \leq n} \in F\} \bigr)=0 \] (where \(\triangle\) denotes the symmetric difference of events). This property implies a number of \(0-1\) laws. It is fulfilled, for example, when the masses of all possible atoms of the sequence \(\underline X\) are uniformly bounded away from 1 (actually, these properties are equivalent). As one consequence, the authors derive the Hoffmann-Jørgensen's \(0-1\) law for nonatomic \(X_ k\)'s.

Recommendations

Cited in

(3)

This page was built for publication: On probabilities of linear sets

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2365596)