OTAgrum documentation

Introduction

The aGrUM library provides efficient algorithms to create and manipulate graphical models. A particular case of such models is the class of Bayesian Networks (BN), which is of first interest in association with OpenTURNS.

A Bayesian network, belief network or directed acyclic graphical model is a probabilistic graphical model that represents a set of random variables and their conditional dependencies via a directed acyclic graph (DAG). In this DAG, edges represent conditional dependencies; nodes which are not connected represent variables which are conditionally independent of each other. Each node is associated with a probability function that takes as input a particular set of values for the node’s parent variables and gives the probability of the variable represented by the node.

The manipulation of a Bayesian network is called inference. Efficient algorithms exist that perform inference and learning of Bayesian networks.

What is otagrum ?

The otagrum module is the link between Bayesian networks built with aGrUM and distributions defined with OpenTURNS.

It offers the ability to:

  • define discretized aGrUM distributions from OpenTURNS distributions
  • extract marginal distributions of aGrUM Bayesian networks as OpenTURNS distributions

Indices and tables