inferlo.gaussian.inference.gaussian_belief_propagation.gaussian_BP

inferlo.gaussian.inference.gaussian_belief_propagation.gaussian_BP(model, tol=1e-09, max_iter=100000.0)

Inference with Gaussian Belief Propagation.

Exact results are guaranteed only for models which satisfy one of sufficient conditions (such as diagonal dominance, etc.)

For details please refer to: http://ssg.mit.edu/group/willsky/publ_pdfs/185_pub_MLR.pdf

Parameters:

• model – Gaussian graphical model for which to perform inference.

• tol – Tolerance factor for check whether we reached fixed point

• max_iter – Maximal number of iterations, where each iteration is a single message-passing round under fully parrallel schedule.

Returns:

two vectors, where the first vector is a vector of obtained marginal means and the second vector is a vector of obtained marginal variances.