Sizes of simultaneous core partitions (Q2237940)

Let \(\lambda\) be an integer partition, that we identify with its Ferrers diagram using English notation. Given a cell in the diagram, its hook is the set of cells below it in the same column, or to its right in the same row. The cardinality of this set is called the hook length. Let \(s\) be a positive integer. The set of \(s\)-cores, integer partitions having no hook length divisible by \(s\), naturally appears in representation theory when \(s\) is a prime number. From the combinatorial point of view, there is a growing interest into \((s,t)\)-cores (that are both \(s\)-cores and \(t\)-cores, by definition) when \(s\) and \(t\) are coprime integers. A fundamental result is the fact that there is a finite number of \((s,t)\)-cores: a bijection by \textit{J. Anderson} [Discrete Math. 248, No. 1--3, 237--243 (2002; Zbl 1001.05015)] with rational Dyck paths show that this number is \(\frac{1}{s+t}\binom{s+t}{s}\). A conjecture by Armstrong (see [\textit{D. Armstrong} et al., Eur. J. Comb. 41, 205--220 (2014; Zbl 1297.05024)]) says that the expected size (number of cells in the Ferrers diagram) of a random \((s,t)\)-core is \(\frac{1}{24}(s+t+1)(s-1)(t-1)\) (under the uniform distribution). Other works (reviewed in the present article) deal with related results. In the work under review, the author finds the limit distribution of the size of an \((s,t)\)-core (chosen under the uniform distribution) when \(s\) and \(t\) both tend to infinity. This answers conjectures by D. Zeilberger (see [\textit{S. B. Ekhad} and \textit{D. Zeilberger}, ``Explicit expressions for the variance and higher moments of the size of a simultaneous core partition and its limiting distribution'', Preprint, \url{arXiv:1508.07637}; \textit{A. Zaleski} and \textit{D. Zeilberger}, J. Difference Equ. Appl. 23, No. 7, 1241--1254 (2017; Zbl 1375.05024)]) and is more precise than several previous works in this direction. Once rescaled by \(st(s+t)/2\), this limit distribution is Watson's \(U^2\) distribution, explicitly given by a convergent sum of independent standard normal variables.