Asymptotics of \(q\)-Plancherel measures. (Q2428504)

From MaRDI portal
Revision as of 02:28, 5 July 2024 by ReferenceBot (talk | contribs) (‎Changed an Item)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
scientific article
Language Label Description Also known as
English
Asymptotics of \(q\)-Plancherel measures.
scientific article

    Statements

    Asymptotics of \(q\)-Plancherel measures. (English)
    0 references
    0 references
    0 references
    26 April 2012
    0 references
    Let \(\mathcal Y_n\) be the set of all partitions of an integer \(n\). On this set there are various natural probability measures, the most famous one being the Plancherel measure, which associates to a partition \(\lambda\) the probability \(\frac{\chi_\lambda(1)^2}{n!}\). \textit{S. V. Kerov} [Funct. Anal. Appl. 26, No. 3, 179-187 (1992; translation from Funkts. Anal. Prilozh. 26, No. 3, 35-45 (1992; Zbl 0796.05092)] introduced a \(q\)-deformation of this measure, which is related to the decomposition of the regular trace in Iwahori-Hecke algebras. In this article the limiting shape of a random partition with respect to this measure is studied. It is shown that if \(\lambda\) is chosen at random with respect to the \(q\)-Plancherel measure for \(q<1\), then for each \(r\) the joint distribution of \(\lambda_1,\dots,\lambda_r\) converges to a Gaussian distribution, in which \(\lambda_i/n\) has mean \((1-q)q^i\), in particular, the first few parts of the partition are containing most of the boxes. This differs greatly from the Plancherel measure or the equidistribution on \(\mathcal Y_n\), where the largest part is of order \(\sqrt n\) and \(\sqrt n\log n\), respectively. The variance of \(\lambda_i\) and the covariance of \(\lambda_i\) and \(\lambda_j\) is also determined. For the proof define for a partition \(\lambda\) the Frobenius coordinates \((a_i,b_i)_{i\leq d}\) by \(a_i=\lambda_i-i\), \(b_i=\lambda'_i-i\), where \(d\) is the size of the diagonal of \(\lambda\). The authors then define the power sums \[ p_k(\lambda)=\sum_{i=1}^d\bigl(a_i(\lambda)+\tfrac{1}{2}\bigr)^k-\bigl(-b_i(\lambda)-\tfrac{1}{2}\bigr)^k, \] and compute the joint distribution of an \(\ell\)-tuple of these functions. This result is then transformed into the result on the row lengths.
    0 references
    random partitions
    0 references
    Young diagrams
    0 references
    limiting shape
    0 references
    Hecke algebras
    0 references
    Plancherel measure
    0 references
    partitions of integers
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references