Indecomposable integrally closed modules of arbitrary rank over a two-dimensional regular local ring

Abstract: In this paper, we construct indecomposable integrally closed modules of arbitrary rank over a two-dimensional regular local ring. The modules are quite explicitly constructed from a given complete monomial ideal. We also give structural and numerical results on integrally closed modules. These are used in the proof of indecomposability of the modules. As a consequence, we have a large class of indecomposable integrally closed modules of arbitrary rank whose ideal is not necessarily simple. This extends the original result on the existence of indecomposable integrally closed modules and strengthens the non-triviality of the theory developed by Kodiyalam.

Review text

In this review the following notations are used: \((R, \mathfrak m)\) is a two-dimensional regular local ring with infinite residue field, \(K\) its field of quotients. Let \(\mathfrak m:=(x,y)\); \(\mathcal I\) is the set of monomial ideals -- with respect to \(\{x,y\}\); \(\mathcal M\) is the family of finitely generated torsion-free \(R\)-modules. For \(M \in \mathcal M\) let \(\overline M\subset M\otimes_RK\) be the integral closure of \(M\) in the sense of \textit{D. Rees} [Math. Proc. Camb. Philos. Soc. 101, 431--449 (1987; Zbl 0628.13004)] and Chapter 6 in [\textit{I. Swanson} and \textit{C. Huneke}, Integral closure of ideals, rings, and modules. Cambridge: Cambridge University Press (2006; Zbl 1117.13001)]. The author continues his work on modules in \(\mathcal M\) [\textit{F. Hayasaka} and \textit{E. Hyry}, Proc. Am. Math. Soc. 138, No. 2, 545--551 (2010; Zbl 1185.13028); \textit{F. Hayasaka}, J. Algebra 556, 879--907 (2020; Zbl 1440.13038)]. Some of the results proved in these papers are needed in order to prove the main result which is stated below. For \(M\in\mathcal M\) let \(I(M)\) be the ideal \(I(M):=\mathrm{Fitt}_0(M^{**}/M)\); note that \(I(\overline M)=\overline{I(M)}\). \textit{V. Kodiyalam} [Trans. Am. Math. Soc. 347, No. 9, 3551--3573 (1995; Zbl 0848.13007)] and \textit{V. Kodiyalam} and \textit{R. Mohan} [J. Algebra 425, 392--409 (2015; Zbl 1318.13012)] studied modules \(M\) in \(\mathcal M\) and their integral closure \(\overline M\). In particular: Assume that \(M\) has no free summand. In this case Kodiyalam -- in the first paper mentioned above -- showed: \textit{If the ideal \(\overline{I(M)}\) is simple, then \(\overline M\) is indecomposable.} He asked if the converse to this theorem holds also, i.e., let \(M=\overline M\) be an indecomposable \(R\)-module of rank larger than 1. Does this imply that \(I(M)\) is a simple complete ideal? A particular monomial ideal plays an important role. Let \(r\geq 2\) be an integer. Let \( a :=(a_0,\ldots,a_r)\) with \(a_0>a_1>\cdots>a_{r-1}>a_r=0\), \(b:=(b_0,\ldots b_r)\), \( 0=b_0<b_1<\cdots <b_{r-1}< b_r\), and assume that \(a_0\leq b_r\). Now set \(I( a, b ):=\langle x^{a_i}y^{b_i}\mid 0\leq i\leq r\rangle\). \(I\) is an \(\mathfrak m\)-primary complete ideal. In section 2 properties of the map \(\mathcal M\to \mathcal I: M\to I(M)\) are studied. Let \(\mathfrak m=(x,y)\). Let \(r\in\mathbf N\); set \(I:=\langle x^{a_i}y^{b_i} \mid 0\leq i\leq r\rangle\) where \(a_0>a_1>\cdots>a_{r-1}>a_r=0\), \( 0=b_0<b_1<\cdots <b_{r-1}< b_r\) and \(a_0\leq b_r\); The main result of this paper is the following: Let \(e\) be an integer with \(2\leq e\leq r\). Provided some additional conditions are satisfied, then there exists an \textit{indecomposable \(M\in \mathcal M\), with \(M=\overline M\), \(I(M)=I\) and \(\mathrm{rank}(M)=e\).} For \(e=2\) this was proved by the author in the second paper mentioned above. Let \(e\) be an integer with \(2\leq e\leq r\); there is defined an \(R\)-module \(M:=M(I;e)\). In Prop.\ 3.3 the ideal \(I(M)\) is determined; it is a monomial ideal. In Th.\ 3.8 it is shown that \(I(M)\) is integrally closed iff \(M\) is integrally closed. In section 4 the author proves two interesting result on the structure of modules in \(\mathcal M\) which are needed in order to prove the main result. One of these is the following estimate: Let \(M\in\mathcal M\) without a free direct summand, \(M=\overline M\), \(e:=\mathrm{rank}(M)\geq 1\). Then \[ \ell_R(R/I(M))-\ell_R(M^{**}/M)\geq \frac{e(e-1)}{2} . \] Let \(e\) be an integer with \(2\leq e\leq r\); there is defined an \(R\)-module \(M(I;e)=:M\). In Prop.\ 3.3 the ideal \(I(M)\) is determined; it is a monomial ideal. In Th.\ 3.8 it is shown that \(I(M)\) is integrally closed iff \(M\) is integrally closed. Finally, in section 5, the author investigates the indecomposability of the modules \(M(I;e)\), \(2\leq e\leq r\). If \(I(M)\) is complete, then \(M\) is an indecomposable integrally closed \(R\)-module; if \(e=r\), this holds true under an additional condition. In section 6 there are listed some conditions which ensure that \(M(I;e)\) is an indecomposable integrally closed \(R\)-module. Assume that \(M\) has no free summand. In this case Kodiyalam -- in the first paper mentioned above -- showed: If the ideal \(\overline{I(M)}\) is simple, then \(\overline M\) is indecomposable.

Recommendations

Cites work

Cited in

(11)

This page was built for publication: Indecomposable integrally closed modules of arbitrary rank over a two-dimensional regular local ring

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2134658)