rADMGdist: Generate a distribution from an ADMG model

In rje42/ADMGs2: Functions for Describing and Fitting Discrete Mixed Graphs

Generate a distribution from an ADMG model

Description

Generate a distribution from an ADMG model

Usage

rADMGdist(graph, dims, map, r = TRUE, alpha = 1, new = FALSE)

Arguments

graph	an ADMG object of class mixedgraph
dims	integer vector of dimensions of distribution
map	optionally, the output of maps()
r	logical indicating whether or not recursive factorizations should be used
alpha	parameter to scale Beta distribution down by.
new	logical - should projection method be used?

Details

Under the default method, a random distribution is obtained by starting with the uniform distribution, and adding an independent (scaled) normal random variables to each Moebius parameter. We then scale the move to reach approximately the boundary of the simplex, and then pick a value between the uniform distribution and the boundary point using a Beta distribution. The parameter alpha being larger favours distributions closer to the uniform distribution.

Under the new method, a random dirichlet (with parameters alpha) is mapped to the Moebius parameters, and then these parameters are mapped to probabilities. If any of these are negative, then each district is moved towards the uniform distribution until it becomes positive.

Value

Object of class moebius giving generalized Moebius parameters for a distribution in the model associated with graph.

rje42/ADMGs2 documentation built on Sept. 3, 2024, 7:39 p.m.