On almost sure convergence of the quadratic variation of Brownian motion. (Q2574586)

Let \(W = (W_ {t})_ {0\leq t\leq 1}\) be a standard Brownian motion. Let us set \(W(A) = \int ^ 1_ 0 \mathbf 1_ {A}\, dW\) for \(A \subseteq [0,1]\) Borel and denote by \(\lambda \) Lebesgue measure on \([0,1]\). Suppose that \(I^ {n} = \{I^ {n}_ {k},\, 1\leq k\leq k_ {n} \}\), \(n\geq 1\), is a sequence of partitions of \([0,1]\) satisfying \(\max _ {1\leq k\leq k_ {n}}\lambda (I^ {n}_ {k})\to 0\) as \(n\to \infty \). It is well known that \(\sum ^ {k_n}_ {k=1} W^ 2(I^ {n} _ {k})\to 1\) as \(n\to \infty \) in probability. The authors aim at characterizing sequences \(\{I^ {n}\}\) such that \[ \sum ^ {k_n} _ {k=1} W^ 2(I^ {n}_ {k})\to 1 \text{ almost surely} \tag{1} \] as \(n\to \infty \) by using decoupling techniques that convert the problem to a problem about convergence of Gaussian sequences. For example, they prove the following result: Let \(B\) be a Brownian motion independent of \(W\). Then (1) holds if and only if \(\sum ^ {k_n}_ {k=1} W(I^ {n} _ {k})B(I^ {n}_ {k})\to 0\) as \(n\to \infty \) almost surely and if and only if \(\sum ^ {k_n}_ {k=1} \max \{W(I^ {n}_ {k}),0\} B(I^ {n}_ {k})\to 0\) as \(n\to \infty \) almost surely. P. Lévy's theorem stating that (1) holds if the sequence of partitions is refined is also a corollary of the results established in the paper.