Strong approximation for sums of asymptotically negatively dependent Gaussian sequences with applications (Q624197): Difference between revisions

Let \(\{X_n,n\geq 1\}\) be a sequence of asymptotically linear negative quadrant dependent (ALNQD) \(N(0,EX_n^2)\)-random variables and set \(S_n=\sum_{i=1}^n X_i,\) \(\sigma_n^2=ES_n^2\) \((n\geq 1).\) Moreover, assume that the variances \(EX_n^2\) are uniformly bounded away from \(0\) and \(\infty\) and that, for some \(r\in (0,3/2),\) \[ \sup_{k\geq 1} \sum_{j:| j-k| \geq n} | \text{Cov}(X_j,X_k)| =O(n^{-r})\quad (n\to\infty). \] Then, the authors' main result proves that the sequence \(\{X_n,n\geq 1\}\) can be redefined on a rich enough probability space, also accomodating a standard Wiener process \(\{W(t),t\geq 0\},\) such that, for any \(0<d<(r-1)/(3r-1),\) \[ S_n-W(\sigma_n^2)=o(\sigma_n^{1-d})\quad a.s.\quad (n\to\infty). \] As applications of the strong invariance principle above, a law of the iterated logarithm, a Chung-type law, and an almost sure central limit theorem for ALNQD sequences can also be derived.