Random Matrices with Prescribed Eigenvalues

Abstract: Classical random matrix theory begins with a random matrix model and analyzes the distribution of the resulting eigenvalues. In this work, we treat the reverse question: if the eigenvalues are specified but the matrix is "otherwise random", what do the entries typically look like? I will describe a natural model of random matrices with prescribed eigenvalues and discuss a central limit theorem for projections, which in particular shows that relatively large subcollections of entries are jointly Gaussian, no matter what the eigenvalue distribution looks like. I will discuss various applications and interpretations of this result, in particular to a probabilistic version of the Schur--Horn theorem and to models of quantum systems in random states. This work is joint with Mark Meckes.

mathematical physicsprobability

Audience: researchers in the discipline

Comments: Please subscribe to our mailing list (https://lists.maths.ox.ac.uk/mailman/listinfo/random-matrix-theory-announce) and Zoom link will be made available the day before.

Oxford Random Matrix Theory Seminars

Series comments: Meeting links will be sent to members of our mailing list (https://lists.maths.ox.ac.uk/mailman/listinfo/random-matrix-theory-announce) in our weekly announcement on Monday.