Formulas for counting acyclic digraph Markov equivalence classes |
| |
Institution: | 1. School of Environment, Northeast Normal University, Changchun 130117, China;2. Key Laboratory of Groundwater Resources and Environment, Ministry of Education, Jilin University, Changchun 130021, China |
| |
Abstract: | Multivariate Markov dependencies between different variables often can be represented graphically using acyclic digraphs (ADGs). In certain cases, though, different ADGs represent the same statistical model, thus leading to a set of equivalence classes of ADGs that constitute the true universe of available graphical models. Building upon the previously known formulas for counting the number of acyclic digraphs and the number of equivalence classes of size 1, formulas are developed to count ADG equivalence classes of arbitrary size, based on the chordal graph configurations that produce a class of that size. Theorems to validate the formulas as well as to aid in determining the appropriate chordal graphs to use for a given class size are included. |
| |
Keywords: | |
本文献已被 ScienceDirect 等数据库收录! |
|