Then, we had a quite pedagogical and interesting talk by

**Satya Majumdar**on the distribution of the largest eigenvalue of some classes of random matrices. The crux of the matter is that one can compute this exactly, and the result turns out to be exponential in the square of N, matrix size (sometimes linear in N, in the Wishart case) times a constant. The constant depends on the ensemble, and on where one looks at that on the real axis. The fact that seems to have caught the interest of some participants is the fact that the Coulomb gas mapping used to compute finally the distribution is the most clear and simple of all available ones. See recent papers by Satya, including http://xxx.sissa.it/abs/cond-mat/0701193. Olav Tirkkonen pointed out some connections to communications theory.

Of course, the day was finished with wine and cheese (thanks Anne and Laila!) to celebrate the first week.

## No comments:

Post a Comment