dirichlet.py

import numpy as np
import math


class Dirichlet():

    def __init__(self, shape=None, params=None):
        
        if params is None:
            self.params = np.ones(shape) # for simplicity we start with all parameters set to 1 (uniform)
        else:
            self.params = params
        self.values = self.params.copy()

    def infer(self, observations):

        self.observations = observations
        if self.values.shape != observations.shape:  # in case of multiple observations
            self.values += observations.sum(axis=0)
        else:
            self.values += observations  # this is the very basic update function to get the posterior

    def forget(self, fr):

        # Only forget what is active
        values_temp = self.values - self.params
        values_temp *= (1 - fr)
        self.values = values_temp + self.params

    def get_MAP_cpd(self):
        return self.values / self.values.sum(axis=0)  # it's also very simple to get a point estimate for the prior over A

    def get_full_cpd(self):  # the MAP (point estimate) is a simplification. When considering the full distribution it gets more complicated...
        pass