Recursion for the smallest eigenvalue density of \(\beta \)-Wishart-Laguerre ensemble (Q1999411)

The expression of the smallest eigenvalue density of \(\beta\)-Wishart-Laguerre ensemble involves terms such as determinants or Pfaffians. This makes the computation of such expression quite impractical when dealing with large dimension matrices. Following an idea previously developed for \(\beta=1\) or \(2\), the author sets a recurrence scheme to obtain exact and explicit expressions for this density when the exponent \(\alpha\) is a nonnegative integer and \(\beta\) a positive real in the Laguerre weight function \(\lambda^{\alpha} e^{-\beta \lambda/2}\). Both the unrestricted and the fixed trace case are studied. For several values of the parameters, the analytical results are compared to the numerical approximation density obtained by the simulation of 50000 matrices in the setting of the matrix model. Since this smallest eigenvalue density can be expressed in terms of some hypergeometric function of matrix argument, the above recursive scheme may be used to evaluate this function. Letting the size of the rectangular matrices go to infinity, a simple transformation of the smallest eigenvalue density converges to the Tracy-Widom density, and this is graphically verified for some parameter values. Large deviations results are also discussed. The last application concerns the computation of the density of the largest delay times which are eigenvalues of the Wigner-Smith matrix.