The growth of the Green function for random walks and Poincaré series (Q6853704)

Let \(\Gamma\) be a finitely generated group with word distance \(|\cdot|\) and let \(\mu\) be an admissible probability measure. For \(0<r\le R\) the paper studies the Green function \(G(x,y\mid r)=\sum_{n\ge 0} r^n p_n(x,y)\) and the spherical sums \(H_r(n)=\sum_{|x|=n} G(e,x\mid r)\). The growth rate \(\omega_\Gamma(r)=\limsup_{n\to\infty}\frac1n\log H_r(n)\) is the critical exponent of the Poincaré series \(\Theta_\Gamma(r,s)=\sum_{n\ge 0} H_r(n)e^{-sn}\).\N\NThe main result exhibits a relatively hyperbolic group \(\Gamma\), endowed with a finitely supported symmetric admissible \(\mu\) and parameters \(1<r_\ast<r_\sharp<R\) such that \(\Theta_\Gamma(r,\omega_\Gamma(r))\) diverges for \(r\le r_\ast\) but converges at \(r=r_\sharp\). This provides an example where convergence at the critical exponent depends on \(r\) and it yields corresponding changes in the parabolic gap property, in Patterson-Sullivan type measures associated with \(\Theta_\Gamma(r,s)\) and in growth tightness for suitable proper distances.\N\NA central argument is a phase transition for random walks on Cartesian products of regular trees, where the asymptotics of \(H_r(n)\) vary with \(r\) and include purely exponential growth as well as polynomial corrections. The construction uses this phenomenon to build a convergent Green-Poincaré series in the relatively hyperbolic setting.