Longest runs in a sequence of \(m\)-dependent random variables (Q1182493)

Let \(\{X_ i, i\geq 1\}\) be a sequence of random 0's and 1's. We define the length \(R_ n\) of the longest head run as follows: \[ R_ n=\hbox{max}\left\{k\leq n:\quad\max_{0\leq i \leq n-k} \text{\textbf{1}} \{X_{i+1} =\dots=X_{i+k}=1\}=1 \right\}. \] The limit law for the distribution of \(R_ n\) as \(n\to \infty\) was found in 1943. The rate of convergence in that limit theorem remained unknown until recently even for Bernoulli independent trials. Supposing that \(\{X_ i, i\geq 1\}\) is a stationary sequence of \(m\)-dependent r.v.'s we give an estimate on the rate of convergence in that limit theorem. This estimate provides the unimprovable rate \(n^{-1}(\ln n)\) for the independent case. A result of \(\lim \inf\) type is obtained, too.