Percy Deift – NYU
Wednesday, June 10, 2026
Zoom opens: 10:30AM EDT
Seminar begins: 10:45AM EDT
Universality for the Toda algorithm to compute the eigenvalues of a random matrix
We consider the Toda algorithm to compute the eigenvalues of a random Hermitian matrix H chosen from a general ensemble E. We show in particular that for a desired eigenvalue accuracy ϵ, the distribution of the (suitably scaled and centered) stopping times of the algorithm to achieve the accuracy ϵ, is universal, independent of the choice of the ensemble E. The universality of the Toda algorithm is just one example of similar behavior for a wide variety of standard numerical algorithms. Joint work of Percy Deift and Tom Trogdon