A weak invariance principle for Hilbert space valued martingales (Q796894)

Let \((X_ n,{\mathcal F}_ n)_{n\geq 1}\) be an adapted sequence of square- integrable random vectors with values in a real separable Hilbert space H, such that \(E(X_ n| {\mathcal F}_{n-1})=0\) and such that the sequence \((V_ n)\) defined by \(V_ n=\sum_{1\leq k\leq n}E(\| X_ k\|^ 2| {\mathcal F}_{k-1})\) converges a.s. to infinity. Similar to the classical case of independent random vectors one can associate to the given sequence \((X_ n)\) a sequence \((Z_ n(.))\) of \(C_ H[0,1]\)- valued random vectors by defining \(Z_ n(k/n)=n^{-1/2}\sum_{j\geq 1}(X_ j1_{[V_ j\leq k]})\) and interpolating linearly between k/n, \((k+1)/n.\) The sequence \((X_ n)\) is said to satisfy the invariance principle, if \((Z_ n)\) converges weakly to a Wiener measure on \(C_ H[0,1]\). It is proved that an invariance principle holds for Hilbert space valued random vectors under some assumptions, which unifies some known theorems for the real valued case. The proof is based on a truncation method, proving for the truncated sequence (\(\bar X_ n)\) that the corresponding sequence (\(\bar Z_ n)\) is tight and that the finite-dimensional distributions converge.