Central limit theorem for random strict partitions (Q5947533)

Let \(n\) be a natural number, \(\mathcal P(n)\) be a set of all partitions of \(n\) into different summands. For \(\lambda \in \mathcal P(n)\) consider the function \[ g_\lambda (t)=\sigma^{(n)}\sum_{k\geq t/\sigma^{(n)}}\rho_k(\lambda),\quad t\geq 0, \] where \(\sigma^{(n)}=\frac{\pi }{2\sqrt{3n}}\), \(\rho_k(\lambda)\) equals 1 if \(k\) is a summand in \(\lambda\), and 0 otherwise. Introducing the uniform probability measure \(\mu^{(n)}\) on \(\mathcal P(n)\), we can interpret \(g_\lambda (t)\) as a random function. \(g_\lambda\) is a renormalization of a function whose subgraph is the Young diagram of a partition. It is known [\textit{A. M. Vershik}, Funct. Anal. Appl. 30, No. 2, 90-105 (1996; Zbl 0868.05004)] that \[ \lim_{n\to\infty}\mu^{(n)}\{ \lambda \in \mathcal P(n):\;|g_\lambda (t)-\log (1+e^{-t})|>\varepsilon \}=0 \] for any \(t\geq 0, \varepsilon >0\). The author gives a refinement of this result considering values of \(g_\lambda\) in several points. It is proved that, after centering the vector of these values by its expectation and a certain renormalization, the resulting vector has an asymptotically normal distribution.