1: \begin{abstract}

2: %

3: %  Gaussian belief propagation (GaBP) is an iterative

4: %  message-passing algorithm for inference in Gaussian

5: %  graphical models. It is known that when GaBP converges it

6: %  converges to the correct MAP estimate of the Gaussian

7: %  random vector and simple sufficient conditions for its

8: %  convergence have been established.

9: %

10: %  In this paper we develop a double-loop algorithm for forcing

11: %  convergence of GaBP. Our method computes the correct

12: %  MAP estimate even in cases where standard GaBP would not have converged.

13: %  We further extend this construction to compute least-squares

14: %  solutions of over-constrained linear systems.

15: %  We believe that our construction has numerous applications,

16: %  since the GaBP algorithm is linked to solution of linear systems

17: %  of equations, which is a fundamental problem in computer science and

18: %  engineering.  As a case study, we discuss the linear detection

19: %  problem. We show that using our new construction, we are able to

20: %  force convergence of Montanari's linear detection algorithm, in cases

21: %  where it would originally fail. As a consequence, we are able to

22: %  increase significantly the number of users that can transmit

23: %  concurrently.

24: %

25: %\end{abstract}

26: