Sequential sampling of junction trees for decomposable graphs

Abstract The junction-tree representation provides an attractive structural property for organising a decomposable graph . In this study, we present two novel stochastic algorithms, referred to as the expander and collapser , sequential sampling of junction trees graphs. We show that recursive application expander, which expands incrementally underlying with one vertex at time, has full support on space any given number vertices. On other hand, complementary operation removing vertices in tree, while maintaining tree property. A direct proposed algorithms is demonstrated setting Monte Carlo methods, designed from distributions spaces Numerical studies illustrate utility combinatorial computations graphs trees. All methods paper are implemented Python library trilearn