Decompositions of cellular binomial ideals (Q2835327)

The paper deals with cellular binomial ideals, a certain class of binomial ideals with much simpler combinatorics than that of general binomial ideals. The authors argue that cellular ideals strike just the right balance between computability and combinatorial clarity. Decompositions of arbitrary binomial ideals into cellular binomial ideals exist (and are the same) over any coefficient field, although many choices have to be made in their computation so that cellular components are very far from canonical.NEWLINENEWLINEThe paper describes very explicitly the hull (the intersection over the primary components of minimal primes) and various decompositions of a cellular binomial ideal. The description of the hull is much nicer than the one given by \textit{D. Eisenbud} and \textit{B. Sturmfels} [Duke Math. J. 84, No. 1, 1--45 (1996; Zbl 0873.13021)] as only a monomial ideal needs to be added instead of complicated colons modulo binomials. As the authors argue, this has important ramifications for computations with binomial ideals. A particular feature of the results in this paper is that they are mostly independent of the characteristic of the base field. Finding such unifying results is hard as can be seen from the technicalities in [loc. cit.] and [\textit{T. Kahle} and \textit{E. Miller}, Algebra Number Theory 8, No. 6, 1297--1364 (2014; Zbl 1341.20062)].NEWLINENEWLINETheorem 2.11. in this paper solves one of the problems in [Eisenbud and Sturmfels, loc. cit.] which had remained open for almost 20 years: a decomposition of a cellular binomial ideal as a finite intersection of unmixed cellular binomial ideals in arbitrary characteristic.