Snakes and perturbed random walks (Q2446212)

Let \(1\leq L\leq \infty\) be an integer and let \(\{X_k: k\geq1\}\) be a simple symmetric random walk, modified such that the transition probability \(=p\) if the current location equals \(\max_{m=0,1,\dots,L\wedge k}X_{k-m}\), \(=q\) if it equals \(\min_{m=0,1,\dots,L\wedge k}X_{k-m}\) (and \(1/2\) otherwise). \(L\) is interpreted as the length of the memory of the walk. The process thus obtained is called ``a random walk, perturbed at \(L\)-extrema''. For the case \(L=\infty\), the authors obtain i.a.\ results on the time it takes until the length of the visited area reaches a certain value. For the finite case, the walk can be viewed as a Markov chain, and it is, e.g., shown that the chain is recurrent if \(p+q=1\). In addition, a strong law and a central limit theorem are also proved in this case. Some open problems are also mentioned.