POSITIVE MARGINS AND PRIMARY DECOMPOSITION

We study random walks on contingency tables with fixed marginals, cor- responding to a (log-linear) hierarchical model. If the set of allowed moves is not a Markov basis, then there exist tables with the same marginals that are not connected. We study linear conditions on the values of the marginals that ensure that all tables in a given fiber are connected. We show that many graphical models have the posi- tive margins property, which says that all fibers with strictly positive marginals are connected by the quadratic moves that correspond to conditional independence state- ments. The property persists under natural operations such as gluing along cliques, but we also construct examples of graphical models not enjoying this property. Our analysis of the positive margins property depends on computing the primary decomposition of the associated conditional independence ideal. The main technical results of the paper are primary decompositions of the conditional independence ideals of graphical models of the N-cycle and the complete bipartite graph K2,N 2, with various restrictions on the size of the nodes.

[1]  A. Dobra.  Markov bases for decomposable graphical models , 2003 .

[2]  Takayuki Hibi,et al.  Binomial edge ideals and conditional independence statements , 2009, Adv. Appl. Math..

[3]  Seth Sullivant,et al.  Lectures on Algebraic Statistics , 2008 .

[4]  P. Diaconis,et al.  Algebraic algorithms for sampling from conditional distributions , 1998 .

[5]  R. Birkner.  Polyhedra: A package for computations with convex polyhedral objects , 2009 .

[6]  Seth Sullivant.  Toric fiber products , 2006 .

[7]  S. Sullivant,et al.  Sequential importance sampling for multiway tables , 2006, math/0605615.

[8]  Bernd Sturmfels,et al.  Lattice Walks and Primary Decomposition , 1998 .

[9]  Ruriko Yoshida,et al.  Algebraic and Geometric Methods in Statistics: Markov chains, quotient ideals and connectivity with positive margins , 2009 .

[10]  Seth Sullivant,et al.  Multigraded commutative algebra of graph decompositions , 2011, 1102.2601.

[11]  Irena Swanson,et al.  Minimal primes of ideals arising from conditional independence statements , 2011, 1107.5604.

[12]  D. Geiger,et al.  On the toric algebra of graphical models , 2006, math/0608054.

[13]  Thomas Kahle.  Decompositions of Binomial Ideals in Macaulay 2 , 2011 .

[14]  J. Dénes,et al.  Latin squares and their applications , 1974 .

[15]  Thomas Kahle,et al.  Decompositions of binomial ideals , 2009, 0906.4873.

[16]  J. Besag.  Spatial Interaction and the Statistical Analysis of Lattice Systems , 1974 .

[17]  B. Sturmfels.  Gröbner bases of toric varieties , 1991 .

[18]  B. Sturmfels.  Gröbner bases and convex polytopes , 1995 .

[19]  B. Sturmfels.  SOLVING SYSTEMS OF POLYNOMIAL EQUATIONS , 2002 .

[20]  N. Ay,et al.  Robustness and Conditional Independence Ideals , 2011, 1110.1338.

[21]  Ezra Miller,et al.  Decompositions of commutative monoid congruences and binomial ideals , 2011, 1107.4699.

[22]  S. Sullivant,et al.  Markov Bases of Binary Graph Models , 2003, math/0308280.

[23]  Jesús A. De Loera,et al.  Markov bases of three-way tables are arbitrarily complicated , 2006, J. Symb. Comput..

[24]  P. Malkin.  Truncated Markov bases and Gr\"obner bases for Integer Programming , 2006, math/0612615.

[25]  Robin J. Wilson,et al.  An Atlas of Graphs , 1999 .