Universality of a double scaling limit near singular edge points in random matrix models (Q2384765)

For unitary random matrix ensembles, the limiting mean eigenvalue distribution equals a suitable equilibrium measure. If the confining properties of the spectrum need to be enhanced, external fields are introduced. The paper investigates both the equilibrium measure in external fields and correlation functions for the random matrix eigenvalues, in various scaling limits. Earlier results due to \textit{P. Deift} et al. [J. Approximation Theory 95, No.~3, 388--475 (1998; Zbl 0918.31001)] on logarithmic potentials in external fields are invoked. An emphasis is put on the singular behavior of boundary points in the support of the inferred equilibrium density. The paper addresses the type III singular (edge) points around which the pertinent density, and the correlation kernel, show up a universal scaling (critical) behavior with an exponent 5/2. The major technical input is an asymptotic analysis (Deift-Zhou steepest descent method) of the Riemann-Hilbert problem for orthogonal polynomials.