Further probabilistic results on the symmetric p-groups (Q923122)

A Sylow p-subgroup \(P_ n\) of the symmetric group of degree \(p^ n\) is well known to be an iterated wreath product \(Z_ p wr...wr Z_ p wr Z_ p\) of n cyclic groups of order p. These are the groups which are called symmetric p-groups by the authors of this paper. In an earlier paper [Acta Math. Hung. 41, 137-150 (1983; Zbl 0521.20053)], the authors had shown that if \(P_ n\) has \(a_{nk}\) complex irreducible characters of degree \(p^ k\), and if characters are assigned the uniform probability \(1/h_ n\), where \(h_ n\) is the total number of complex irreducible characters of \(P_ n\) (so that \(h_ n=\sum_{k}a_{nk}=\) number of conjugacy classes of \(P_ n)\), then the distribution of the variable k approximates to a normal distribution (whose expectation and variance were explicitly calculated) as n gets large. The paper under review falls into three parts. In the first the authors deal with the analogous distribution where a character of degree \(p^ k\) is assigned the probability \(p^{2k}/| P_ n|\). (The authors, following S. V. Kerov, call this the Plancherel measure.) They show that again the distribution of the variable k approximates to a normal distribution as n grows large and they give expressions for its expectation and variance. In the second part they compare the distribution with respect to the uniform measure and the distribution with respect to `the Plancherel measure'. The third part of the paper is concerned with the statistics of a distribution arising from \(P_ n\) itself rather than from its characters. The subject of the investigation is the random variable which assigns to an element g of \(P_ n\) the number m, where \(p^ m\) is the order of the centralizer of g in \(P_ n\). The authors are able to show that if the group \(P_ n\) is given the uniform probability distribution, so that elements g occur with probability \(1/| P_ n|\), then there is a limit distribution as \(n\to \infty\), although they are unable to identify exactly what it is. On the other hand, if the conjugacy classes of \(P_ n\) are chosen with equal probability then the distribution again approximates a normal distribution as n becomes large.