Random permutations with cycle weights (Q627247)

The authors study the cycle distribution in models of weighted random permutations. For a permutation \(\pi\) on \(n\) elements, set \(P(\pi):= \frac{1}{n!h_n}\prod_{j\geq 1} \theta_j^{r_j(\pi)}\) where \(\theta_1, \theta_2,\dots,\) are nonnegative numbers, \(r_j(\pi)\) denotes the number of \(j\)-cycles in \(\pi\) and \(h_n\) is the normalization. This model was introduced by \textit{V. Betz} and \textit{D. Ueltschi} [Probab. Theory Relat. Fields 149, No.~1--2, 191--222 (2011; Zbl 1226.82003)]. Such models are used in the investigation of large systems of quantum boson particles (see, e.g., [\textit{V. Betz} and \textit{D. Ueltschi}, Commun. Math. Phys. 285, No.~2, 469--501 (2009; Zbl 1155.82022)]). The case of constant \(\theta_j=\theta\) is known as the Ewens distribution appearing in mathematical biology (see [\textit{W. J. Ewens}, Theor. Population Biology 3, 87--112 (1972; Zbl 0245.92009)]). The relevant random variables are the lengths \(l_j=l_j(\pi)\) of the cycle containing the index \(j\in\{1,\dots,n\}\). Special attention is paid to \(\theta_j = e^{-\alpha_j}\) where \(\alpha_j\sim j^{\gamma}\) and \(\gamma\in\mathbb{R}\). Depending on the weights, the typical cycle length grows like the total number of elements \(n\), or a fraction of \(n\) or a logarithmic power of \(n\).