# Andrew Thompson : Constructing optimal subspace packings

This talk is about the problem of packing subspaces of real/complex space so that they are as far apart as possible, sometimes called Grassmann packing. For a particular distance metric, the problem is equivalent to finding a matrix with minimum group coherence. I will give a fundamental lower bound on group coherence, and then describe some matrix constructions based on Kronecker products which achieve the lower bound on group coherence, and which therefore give optimal subspace packings. Then I will talk about results which show that random subspace packings are close to optimal asymptotically. I will also say something about why good subspace packings are currently of interest in signal processing applications. This work is joint with Robert Calderbank and Yao Xie (now at Georgia Tech). The math involved is mainly linear algebra and random matrix theory, but I will take care to make the talk accessible to all.

**Category**: Graduate/Faculty Seminar**Duration**: 01:34:50**Date**: November 22, 2013 at 4:25 PM**Views**: 115-
**Tags:**seminar, Graduate/faculty Seminar

## 0 Comments