The number of zeros in elephant random walks with delays (Q2244448)

The position of the elephant random walk (with delays) after \(n\) steps is \(S_n=X_1+\cdots+X_n\), where \(X_1\) is \(1\) with probability \(p\) and \(-1\) with probability \(1-p\), and, for \(n\geq1\), \[ X_{n+1}=\begin{cases} X_K & \mbox{with probability }p,\\ -X_K & \mbox{with probability }q,\\ 0 & \mbox{with probability }r, \end{cases} \] where \(0<p,q,r<1\) with \(p+q+r=1\), and \(K\) is uniformly distributed over the `memory' set \(\mathfrak{M}_n\subseteq\{1,2,\ldots,n\}\). The authors use conditioning and martingale arguments to establish the large-\(n\) limiting behaviour of the number of delays of the elephant random walk (i.e., the number of the \(X_j\) that are equal to zero up to time \(n\)) in four different cases: (i) \(\mathfrak{M}_n=\{1,2,\ldots,n\}\), (ii) \(\mathfrak{M}_n=\{1\}\), (iii) \(\mathfrak{M}_n=\{n\}\) and (iv) \(\mathfrak{M}_n=\{1,n\}\). These illustrate the very different behaviour that can occur depending on the structure of the memory set.