Department Colloquium

Department of Mathematics
Princeton University



February 28th, 2001


The longest paths in random matrices : connection to the eigenvalues of random matrices


In the classical central limit theorem, the sum of i.i.d. random variables has the same limiting fluctuation : the Gaussian distribution. Now we make an M by N matrix with entries of i.i.d. random variables, and take a maximum of sums of entries along a class of up/right paths. Then we ask the question of the limiting fluctuation as M and/or N tend to infinity. This type of problem arises in for example, queueing theory and interacting particle systems. There are a few examples for which one can compute the limiting distribution directly, and it turned out that the role played by the Gaussian distribution in the classical central limit theorem is now played by the largest eigenvalue of a random Hermitian matrix taken from the Gaussian unitary ensemble. We will discuss topics centered around this relation between the longest path in a random matrix and the largest eigenvalue of a random Hermitian matrix.