A critical phenomenon in the two-matrix model in the quartic/quadratic case (Q351217)

The authors consider a model of two random matrices \(M_{1}\) and \(M_{2}\) related to each other, then they study the eigenvalues of \(M_{1}\) after taking the average over \(M_{2}\). More precisely, they consider the so-called kernel that characterizes the eigenvalues of \(M_{1}\), when the dimension goes to infinity. Such kernel has two parameters \(\alpha\) and \(\tau\), and it is possible to say what are the values of those parameters where there is a critical point, namely \(\alpha=-1\) and \(\tau=1\). Then, the authors perform a scaling of the kernel around that critical point, and the asymptotic analysis of the scaling gives rise to a kernel that is written in terms of the solution of a Riemann-Hilbert problem (RH). The RH problem is to find a \(4\times 4\) matrix-valued function with certain properties, and an important ingredient of the problem is a Painlevé II equation, as well as its so-called Hastings-McLeod solution. The two-matrix model is given (after normalizing) by the following probability measure over the set of Hermitian matrices \[ \exp\{ -nTr(V(M_{1})+W(M_{2})-\tau M_{1}M_{2}) \}, \] where \(V(x):=x^{2}/2\) and \(W(x):=y^{4}/2+\alpha x^{2}/2\). Important tools used by the authors are the biorthogonal polynomials, properties of some meromorphic functions, and the steepest descent analysis of the RH problem.