Date:	Tue, February 24, 2009
Time:	17:15
Place:	West Hall 8

Abstract: The discrete uncertainty principle asks if there exists a vector x with support T and support of its DFT in W for two subsets T and W of the indices (1,...,N). This question can also be phrased in terms of the norm of the DFT submatrix with rows corresponding to W and columns corresponding to T. We take this as our motivation for considering the spectral asymptotics of random DFT submatrices. Using an approach recently developed in information theory, we are able to determine the limiting empirical spectral distribution. While this is not yet a statement on the asymptotic norm of such matrices, it is an important step in that direction.