The three-state toric homogeneous Markov chain model has Markov degree two (Q480695): Difference between revisions

This paper concerns algebraic statistics and toric ideals. Markov basis are minimal generating sets of toric ideals, it is known from [\textit{P. Diaconis} and \textit{B. Sturmfels}, Ann. Stat. 26, No. 1, 363--397 (1998; Zbl 0952.62088)] that the Markov basis can be used in optimization or in statistics. The main result of this paper is that the three-state toric homogeneous Markov chain model has Markov degree two. In order to be more precise, let \(S,T\) be positive integer numbers, and let \(R_{S,T}\) denote the polynomial ring over a field \(K\) on the set of variables \(\{x_{i_1\ldots i_T} | 1\leq i_j\leq S \;\forall j \;\text{ and } i_j\not= i_{j+1}\}\). The \(S-\)state toric homogeneous Markov \(T-\)chain ideal \(I_{S,T}\) is the kernel of the homomorphism: \(\phi_{S,T}: R_{S,T}\longrightarrow R_{S,2}\) defined by \(\phi_{S,T}(x_{i_1\ldots i_T} )=x_{i_1 i_2} x_{i_2 i_3}\ldots x_{i_{T-1} i_T}\). In this paper, the author proves that if \(S=3, T\geq 3\) the ideal \(I_{3,T}\) is generated by quadratic binomials.