 |
Center for Nonlinear Studies |
The canonical linear-algebraic problem of solving a system of linear equations arises in numerous contexts in the mathematical sciences and engineering. In this talk, we introduce an efficient Gaussian belief propagation (GaBP) solver that does not involve direct matrix inversion. The iterative nature of our approach allows for a distributed message-passing implementation of the solution algorithm. We discuss the properties of the GaBP solver, including convergence,exactness, computational complexity, message-passing efficiency and its relation to classical solution methods. The attractiveness of the proposed solver, in comparison to conventional iterative solution methods, is demonstrated using linear detection applications.
The talk is based on a joint work with Prof. Jack K. Wolf (UCSD), Prof. Paul H. Siegel (UCSD), Dr. Ori Shental (UCSD) and Prof. Danny Dolev (HUJI).
Host: Misha Chertkov, T-13