Conjecture on the beta-dependent variance of ordered eigenvalues

Let λL\lambda_L be the LL-th ordered eigenvalue in the infinite ordered spectrum. For β=1\beta=1 and 44, define

vβ={π28,β=1,log2+π28,β=4.v_\beta=\begin{cases}-\dfrac{\pi^2}{8},&\beta=1,\log 2+\dfrac{\pi^2}{8},&\beta=4.\end{cases}

Eigenvalue-variance conjecture. As LL\to\infty,

varβ[λL]=2βπ2(log(2πL)+vβ+γ+1)16+O(logLL2).{\rm var}_\beta[\lambda_L]=\frac{2}{\beta\pi^2}\Big(\log(2\pi L)+v_\beta+\gamma+1\Big)-\frac{1}{6}+O\left(\frac{\log L}{L^2}\right).

This conjecture predicts the leading and constant terms in the variance asymptotics for the β=1\beta=1 and β=4\beta=4 ensembles, complementing the proved β=2\beta=2 expansion. The source gives no resolution evidence.

Sources & referencesView supporting material

Primary source

Peng Tian, Roman Riser and Eugene Kanzieper, “On the asymptotic duality of spectral variances in random matrix theory and the "1/6" formula”, arXiv:2604.17150 (2026).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.