bayespy.inference.vmp.nodes.gaussian_markov_chain.GaussianMarkovChainMoments.get_converter

GaussianMarkovChainMoments.get_converter(moments_to)

Finds conversion to another moments type if possible.

Note that a conversion from moments A to moments B may require intermediate conversions. For instance: A->C->D->B. This method finds the path which uses the least amount of conversions and returns that path as a single conversion. If no conversion path is available, an error is raised.

The search algorithm starts from the original moments class and applies all possible converters to get a new list of moments classes. This list is extended by adding recursively all parent classes because their converters are applicable. Then, all possible converters are applied to this list to get a new list of current moments classes. This is iterated until the algorithm hits the target moments class or its subclass.