Some results on strong limit theorems for (LB)-space-valued random variables (Q1892957)

The authors show a strong law of large numbers for sums of r.v.s taking values in a strict inductive limit of separable Banach spaces (LB). In Theorem 2.1 they prove a generalization of Komlos theorem to r.v.s taking values in an (LB)-space. This (LB)-space must verify an additional condition. Using this result they obtain that if \(X_n\) is a sequence of independent r.v.s taking values in an (LB)-space \(B\), then: \(P\{(1/m) \sum^m_{i = 1} X_i\), \(m \geq 1\), converges almost surely\}\( =1 \Leftrightarrow\) for some \(n \geq 1\), \(P\{X_1 \in E_n\} = 1\) and \(\mathbb{E} |X^n_1 |< + \infty\) and \((1/m) \sum^m_{i = 1} X_i\), \(m \geq 1\), converges to \(\mathbb{E} X_1\) a.s., where \(E = \bigcup^\infty_{i = 1} E_i\) \((E_i\) Banach spaces) and \(X^n_1 : \Omega_n \to E_1\) such that \(\Omega_n = X^{-1} (E_n)\) and \(X^n_1 (\omega) = X(\omega)\) if \(\omega \in \Omega_n\).