Squarefree monomial ideals with constant depth function (Q387402)

By a well-known result of \textit{M. Brodmann} [Math. Proc. Camb. Philos. Soc. 86, 35--39 (1979; Zbl 0413.13011)], the depth of powers of a homogeneous ideal is constant for large enough power. However, the description of the numerical functions which can be the depths of powers of some ideal is an open problem. The paper under review study the ideals which have constant depth function. Let \(k\) be a field, \(S=k[x_1,\ldots,x_n]\) a standard graded polynomial ring over \(k\) and \(I\) a homogeneous ideal of \(S\). By definition, the \textit{depth function} of \(I\) is the numerical function mapping a natural number \(m\) to \(\text{depth}~ S/I^m\). So Brodmann's theorem says that the depth function is a constant for large \(m\). There is an upper bound for \(\lim_{m\to \infty} \text{depth}~ S/I^m\) found by \textit{L. Burch} [Proc. Camb. Philos. Soc. 72, 369--373 (1972; Zbl 0242.13018)], which is given in term of the Rees algebra of \(I\). Let \(\text{Rees}(I)\) be the Rees algebra of \(I\) and \(\text{Rees}(I)/\mathfrak{m}\text{Rees}(I)\) the fiber cone of \(I\), where \(\mathfrak{m}\) is the maximal graded ideal of \(S\). Let \(\ell(I)=\dim \text{Rees}(I)/\mathfrak{m}\text{Rees}(I)\) be the analytic spread of \(I\). Then Burch's result says that \[ \lim_{m\to \infty} \text{depth}~ S/I^m \leq n-\ell(I), \] with equality holds if \(\text{Rees}(I)\) is Cohen-Macaulay (or more generally, if the associated graded ring \(\text{gr}_I(S)\) is Cohen-Macaulay). The main objects of study in the paper are squarefree monomial ideals with constant depth function. The basic example is an ideal generated by variables of \(S\). Then one can build up more examples of such ideals by the following result (Theorem 1.1): let \(I, J\) be monomial ideals generated in disjoint sets of variables, such that \(\text{Rees}(I)\) and \(\text{Rees}(J)\) are Cohen-Macaulay. Then \(\text{Rees}(I+J)\) and \(\text{Rees}(IJ)\) are also Cohen-Macaulay, and the following are equivalent: (i) the depth function of \(I\) and \(J\) are constant; (ii) the depth function of \(I+J\) is constant; (iii) the depth function of \(IJ\) is constant. The main result of the paper is the classification of ideals with constant depth function among edge ideals, matroidal ideals and facet ideals of certain simplicial complexes. All of these ideals can be constructed by repeatedly applying sum or product as in Theorem 1.1. By Theorem 2.2: if \(G\) is a finite simple graph without isolated vertices, then the edge ideal of \(G\) has constant depth if and only if each connected component of \(G\) is a complete bipartie graph. For matroidal ideals, Theorem 2.5 says that if \(I\) is a matroidal ideal such that the minimal monomial generators of \(I\) has no nontrivial common divisor and each variable of \(S\) divides some generator of \(I\), then \(I\) is a product of monomial prime ideals which are generated in disjoint set of variables. Finally for the so-called pure simplicial forests, Theorem 2.6 says that if \(\Delta\) is a pure simplicial forest which is connected in codimension 1, then the depth function of the facet ideal of \(\Delta\) has constant depth if and only if for each connected component \(\Gamma\) of \(\Delta\), the facet ideal \(I(\Gamma)\) is a product of monomial prime ideals such that at most one of them is not principal.