Weak convergence of self-normalized random polygonal lines (Q1850764)

Consider a sequence of nondegenerate independent i.i.d. random variables defined on a probability space \((\Omega,\mathcal{F}, P)\). Let \[ S_{0}=0, \quad S_{n}=X_{1}+ \cdots + X_{n}, \quad V^{2}_{n}=X^{2}_{1}+ \cdots +X^{2}_{n}. \] Let \(S_{k}/V_{n}=X_{k}/V_{n}=0\), \(1\leq k \leq n,\) if \(V_{n}=0.\) For any \(n=1,2, \dots \) we divide the segment \([0,1]\) by the points \[ t^{(n)}_{k}=\begin{cases} V^{2}_{k}/V^{2}_{n}, &V_{n}>0,\\ k/n, &V_{n}=0, \end{cases} \] for each \(k=0,1,\dots,n.\) The random polygonal line \(X_{n}(t), t\in [0,1],\) is defined, for \(t\in\) \([t^{(n)}_{k-1}\), \(t^{(n)}_{k}]\), by \[ X_{n}(t)=\frac{S_{k-1}}{V_{n}}+\frac{t-t^{(n)}_{k-1}}{t^{(n)}_{k}-t^{(n)}_{k-1}}\frac{X_{k}}{V_{n}}. \] Let \(P_{n}\) and \(P_{0}\) denote, respectively, the distributions of the process \(X_{n}(t)\) and the Brownian motion \(W(t)\), \(t\in [0,1].\) The distributions \(P_{n}\) and \(P_{0}\) are defined on the Borel \(\sigma\)-algebra of subsets of the Banach space \(C[0,1]\) of continuous functions \(x(t)\), \(t\in [0,1],\) with the norm \(\|x \|= \max_{0\leq t \leq 1} |x(t)|.\) The main theorem of the present article is as follows: The sequence of distributions \(\{ P_{n} \}_{n\geq 1}\) weakly converges to the Wiener measure \(P_{0}\) if and only if the distribution function of the random variable \(X_{1}\) belongs to the domain of attraction of the normal law and the equality \(EX_{1}=0\) holds.