Simple demo implementation of belief propagation for acyclic graphs using numpy.einsum.

The demo is for inference on the Bayesian network defined by:

For more information, see the accompanying blog post.