Create an MDGM model — mdgm_model • mdgm

Constructs a mixture of directed graphical models for a discrete spatial random field defined on an undirected graph.

Usage

mdgm_model(
  nug,
  dag_type = c("spanning_tree", "acyclic_orientation", "rooted"),
  n_colors = 2L,
  emission = NULL
)

Arguments

nug: A NaturalUndirectedGraph object defining the spatial structure.
dag_type: DAG construction type: "spanning_tree", "acyclic_orientation", or "rooted".
n_colors: Number of categories for the spatial field (default 2).
emission: Optional emission family for hierarchical models: "bernoulli", "gaussian", or "poisson". If NULL (default), creates a standalone model where the spatial field is observed directly.

Value

An MdgmModel object.

Examples

# Standalone model on a triangle graph
edges <- rbind(c(1, 2), c(2, 1), c(2, 3), c(3, 2), c(1, 3), c(3, 1))
nug <- nug_from_edge_list(3, edges, seed = 42L)
model <- mdgm_model(nug, dag_type = "spanning_tree")
model$nvertices()
#> [1] 3
model$has_emission()
#> [1] FALSE