Vertex-reinforced random walk on \(\mathbb Z\) with sub-square-root weights is recurrent (Q460722)

A nonlinear vertex reinforced random walk (VRRW) on \(\mathbb{Z}\) with a weight function \(w\) on \(\mathbb{Z}\) is considered. The VRRW random walk \(X_0=0,X_1,\dots,X_n,\dots\) is defined via its transition probabilities in the following way: \[ \mathbb P(X_{n+1}=X_n\pm 1 | X_1,\dots,X_n)= \frac{w(Z_n(X_n\pm 1))}{w(Z_n(X_n-1))+ w(Z_n(X_n+1))}, \] where \(Z_n(y)=\# \{m\leq n:X_m=y\}\) is the time spent by the walk in \(y\in \mathbb{Z}\) up to time \(n\). The model with the weight function \(w(k)\sim ck^\alpha,\;k\to \infty,\;c,\alpha>0\) is studied (\(\alpha=1\) corresponds to a linear model). In [J. Theor. Probab. 19, No.~3, 691--700 (2006; Zbl 1107.60068)], \textit{S. Volkov} discovered a surprising phenomenon of phase transition for the model. To describe the phenomenon, let \(R=\{ m\in \mathbb{Z}: X_n=m\text{ for some }n\)