Functional limit theorems for the increments of Gaussian samples (Q1780927): Difference between revisions

Let \(X= \{X(t);t\geq 0\}\) be a real-valued, centered Gaussian process with almost surely continuous sample paths and assume that \(X(0)= 0\) and that \(X\) has stationary increments and some continuous covariance function \(R(s, t)\). Put \(\mu={\mathcal L}(X)\) and \(\sigma^2(h)= E(X(t+ h)- X(t))^2\). Put \(\| f\|_\mu= (\int^1_0 |f'(s)|^s ds)^{1/2}.\) Let \(H_\mu\) \((\subset C_0[0,1])\) be the Hilbert space of absolutely continuous functions on \([0, 1]\) with unit ball \[ {\mathcal S}= \{f(t)= \int^t_0 f'(s)\,ds,\;0\leq t\leq 1;\,\| f\|_\mu\leq 1\}. \] Furthermore, for some continuous functions \(A_T\) and \(a_T\) of \(T\), define \[ \Delta_{t,T}(x)= {X(t+ a_T x)- X(t)\over \sigma(a_T)\{2(\log{A_T- a_T\over a_T}+ \log\log A_T)\}^{1/2}}. \] Assuming \(X\) is self-similar in the sense that for \(h> 0\), \(X(h\cdot)\overset {d}= \sigma(h)X(\cdot)\), the author proves functional limit theorems such as \[ \lim_{T\to\infty}\;\sup_{0\leq t\leq A_T- a_T}\;\underset{f\in U}{}{\text{inf}}\|\Delta_{t,T}- f\|_\infty= 0\text{ a.s.,} \] where \(U= \{f\in H_\mu:\| f\|_\mu\leq 1\}\) and \(\|\cdot\|_\infty\) denotes the usual sup-norm on \(C_0[0,1]\). Applying these results the author obtains a functional modulus of continuity theorem and a large increment theorem of a fractional Brownian motion which generalize known results.