Additive and multiplicative functions on random permutations (Q1381648): Difference between revisions

A permutation \(\sigma \in S_n\) has a unique representation as a product of cycles \(\sigma =\kappa _1\dots\kappa _\omega \), and this led V. Goncharov in 1942 to define an analog of the number of prime divisors and to prove the analog of the Erdős-Kac theorem for it. The author considers general additive functions on \(S_n\) which he defines by the requirement \(f(\sigma) = \sum f(\kappa _j)\); such a function is called class dependent if \(f(\kappa)\) depends only on \(l(\kappa)\), the length of the cycle \(\kappa \). Multiplicative functions are defined similarly. In this short paper the author develops the foundations of probabilistic number theory of permutations, including the analogs of the classical theorems of Halász and Kubilius. Most of the results are similar to those of probabilistic number theory, but new phenomena also occur. For example (Theorem 5) the function \(f(\sigma) = \sum l(\kappa _j)^c \) always has a nondegenerate limiting distribution with the proper norming, and for \(c>1\) it has bounded support and so is a fortiori not infinitely divisible.