Maximum of the characteristic polynomial of random unitary matrices (Q507178)

The authors consider a recent conjecture of \textit{Y. V. Fyodorov}, \textit{G. A. Hiary} and \textit{J. P. Keating} [``Freezing transition, characteristic polynomials of random matrices, and the Riemann zeta-function'', Phys. Rev. Lett., 108, Article ID 170601 (2012; \url{doi:10.1103/PhysRevLett.108.170601})] about the extreme values of the characteristic polynomial for random unitary matrices. Conjecture. For \(N \in \mathbb N\), let \(U_N\) be a random matrix sampled uniformly from the group of \(N \times N\) unitary matrices. Write \(P_N(z)\), \(z \in \mathbb C\), for its characteristic polynomial. Then \[ \begin{aligned} \max_{h \in [0, 2 \pi]} \log | P_N( e^{\mathrm i h} | = \log N - \frac 3 4 \log \log N + \mathcal M_N, \end{aligned} \] where \((\mathcal M_N, N \in \mathbb N)\) is a sequence of random variables that converge in distribution. A necessary condition for this to be true is: Theorem. For any \(\delta > 0\) \[ \begin{aligned} \mathbb P \left( \max_{h \in [0, 2 \pi]} \log | P_N( e^{\mathrm i h}) | \in [(1-\delta) \log N, (1 + \delta) \log N] \right) \to 0, \end{aligned} \] as \(N \to \infty\). And this is the main result of the current work. Additional results on the measure of high points and the freezing of the free energy are also included. The authors very carefully outline their method of proof in broad strokes before providing details. The first part of the proof revolves around connecting the random variable \[ \begin{aligned} \max_{h \in [0, 2 \pi]} \log | P_N( e^{\mathrm i h}) | \end{aligned} \] to an approximate branching random walk. They use techniques such the as the multiscale refinement of Kistler, the asymptotic analysis of Riemann-Hilbert problems and so-called \(K\)-level coarse graining. The second part of the proof revolves around the asymptotics of Toeplitz determinants with Fisher-Hartwig singularities, making use of the methodology of \textit{P. Deift} et al. [Ann. Math. (2) 174, No. 2, 1243--1299 (2011; Zbl 1232.15006)]. This paper presents an impressive confluence of techniques from random matrix theory and mathematical physics.