Journal of Algebra 416 (2014) 314–332
Contents lists available at ScienceDirect
Journal of Algebra www.elsevier.com/locate/jalgebra
On S-strong Mori domains Hwankoo Kim a , Myeong Og Kim b , Jung Wook Lim b,∗ a
Department of Information Security, Hoseo University, Asan 336-795, Republic of Korea b Department of Mathematics, Kyungpook National University, Daegu 702-701, Republic of Korea
a r t i c l e
i n f o
Article history: Received 12 March 2014 Available online 15 July 2014 Communicated by Kazuhiko Kurano MSC: 13A15 13B25 13E99 13F05 13G05 Keywords: S-w-finite S-w-principal S-strong Mori domain S-factorial domain S-strong Mori module Krull domain
a b s t r a c t Let D be an integral domain, S be a (not necessarily saturated) multiplicative subset of D, w be the so-called w-operation on D, and M be a unitary D-module. As generalizations of strong Mori domains (respectively, UFDs) and strong Mori modules, we define D to be an S-strong Mori domain (respectively, S-factorial domain) if for each nonzero ideal I of D, there exist an s ∈ S and a w-finite type (respectively, principal) ideal J of D such that sI ⊆ J ⊆ Iw ; and M to be an S-strong Mori module if M is a w-module and for each nonzero submodule N of M , there exist an s ∈ S and a w-finite type submodule F of N such that sN ⊆ F ⊆ Nw . This paper presents some properties of S-strong Mori domains, S-factorial domains and S-strong Mori modules. © 2014 Elsevier Inc. All rights reserved.
0. Introduction Throughout this paper, D is an integral domain and S is a (not necessarily saturated) multiplicative subset of D. In [5], the authors introduced the concept of “almost * Corresponding author. E-mail addresses:
[email protected] (H. Kim),
[email protected] (M.O. Kim),
[email protected] (J.W. Lim). http://dx.doi.org/10.1016/j.jalgebra.2014.06.015 0021-8693/© 2014 Elsevier Inc. All rights reserved.
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
315
finitely generated” to study Querre’s characterization of divisorial ideals in integrally closed polynomial rings. Later, Anderson and Dumitrescu [3] abstracted this notion to any commutative ring and introduced the concepts of S-Noetherian rings and S-principal ideal rings. Let R denote a commutative ring with identity and S a (not necessarily saturated) multiplicative subset of R. Then R is called an S-Noetherian ring (respectively, S-principal ideal ring (S-PIR)) if each ideal of R is S-finite (respectively, S-principal), i.e., for each ideal I of R, there exist an s ∈ S and a finitely generated (respectively, principal) ideal J of R such that sI ⊆ J ⊆ I. In [3], Anderson and Dumitrescu tied together several different results which also generalize well-known results on Noetherian rings. Later, Liu studied in [17] when the generalized power series ring is S-Noetherian. Recently, Lim and Oh investigated S-Noetherian properties on composite ring extensions and amalgamated algebras along an ideal in [15] and [16], respectively. The main purpose of this article is to introduce and to investigate the notions of S-strong Mori domains, S-strong Mori modules and S-factorial domains. In fact, an S-strong Mori domain (respectively, S-factorial domain) generalizes both an S-Noetherian domain (respectively, S-principal ideal domain) and a strong Mori domain (respectively, UFD); and an S-strong Mori module generalizes a strong Mori module. (Definitions of S-strong Mori domains and S-factorial domains (respectively, S-strong Mori modules) will be given in Section 1 (respectively, Section 2).) This paper consists of four sections including introduction. In Section 1, we study basic properties of S-strong Mori domains and S-factorial domains. We show that for any maximal w-ideal M of D, D is a (D \ M )-strong Mori domain (respectively, (D \ M )-factorial domain) if and only if DM is a Noetherian domain (respectively, PID) and for every nonzero finitely generated (respectively, principal) ideal J of D, Jw DM ∩ D = Jw : s for some s ∈ D \ M (Proposition 1.4). In Section 2, we study more properties of S-strong Mori domains, which generalize well-known facts on strong Mori domains. More precisely, we prove that if S is an anti-archimedean subset of D, then D is an S-strong Mori domain if and only if the polynomial ring D[X] is an S-strong Mori domain, if and only if the t-Nagata ring D[X]Nv is an S-strong Mori domain, if and only if D[X]Nv is an S-Noetherian domain (Theorems 2.8 and 2.10). We also show that if L is an S-strong Mori submodule of a w-module M such that M/L is an S-strong Mori module, then M is an S-strong Mori module; and if D is an S-strong Mori domain and M is an S-w-finite w-module as a D-module, then M is an S-strong Mori module (Theorem 2.11). Finally, Section 3 is devoted to study the Cohen type theorem for S-factorial domains and to give new characterizations of Krull domains. In fact, we show that D is an S-factorial domain if and only if every prime w-ideal of D (disjoint from S) is S(-w)-principal (Theorem 3.2); and D is a Krull domain if and only if D is a (D \ M )-factorial domain for all maximal w-ideals M of D (Proposition 3.6). In order to help the reader’s better understanding, we review some definitions and notation. Let D be an integral domain with quotient field K, and let F(D) be the set of nonzero fractional ideals of D. For an I ∈ F(D), set I −1 := {x ∈ K | xI ⊆ D}. The mapping on F(D) defined by I → Iv := (I −1 )−1 is called the v-operation on D;
316
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
the mapping on F(D) defined by I → It := {Jv | J is a nonzero finitely generated fractional subideal of I} is called the t-operation on D; and the mapping on F(D) defined by I → Iw := {x ∈ K | xJ ⊆ I for some finitely generated ideal J of D such that Jv = D} is called the w-operation on D. It is easy to see that I ⊆ Iw ⊆ It ⊆ Iv for all I ∈ F(D). An I ∈ F(D) is a v-ideal (or divisorial ideal) (respectively, t-ideal, w-ideal) if Iv = I (respectively, It = I, Iw = I). Clearly, if an I ∈ F(D) is finitely generated, then Iv = It ; and a height-one prime ideal is a t-ideal. For ∗ = v, t or w, a maximal ∗-ideal means a ∗-ideal which is maximal among proper integral ∗-ideals. It was shown in [2, Corollary 2.17] that the notion of maximal t-ideals coincides with that of maximal w-ideals. Let w-Max(D) be the set of maximal w-ideals of D. It is well known that w-Max(D) = ∅ if D is not a field; and a maximal w-ideal is a prime ideal. A w-ideal I ∈ F(D) is of w-finite type if I = Jw for some finitely generated ideal J of D. An I ∈ F(D) is said to be invertible (respectively, w-invertible) if II −1 = D (respectively, (II −1 )w = D). We say that D has finite t-character (respectively, finite w-character) if every nonzero nonunit element in D is contained in only finitely many maximal t-ideals (respectively, maximal w-ideals) of D. Since the notion of maximal t-ideals coincides with that of maximal w-ideals, the concept of finite t-character is precisely the same as that of finite w-character. Recall that an ideal J of D is a Glaz–Vasconcelos ideal (GV-ideal) if J is finitely generated and J −1 = D. Let GV(D) be the set of GV-ideals of D. Then Iw = {x ∈ K | Jx ⊆ I for some J ∈ GV(D)} for all I ∈ F(D). Let M be a (not necessarily torsion-free) module over an integral domain D and let ED (M ) denote the injective envelope (or injective hull) of M . If no confusion arises, we write E(M ) for ED (M ). Set r(M ) := {x ∈ M | (annD (x))w = D}. Following [13], the w-closure of M is defined by Mw = p−1 (r(E(M )/M )), where p : E(M ) → E(M )/M is the canonical projection. Then it is easy to see that Mw = {x ∈ E(M ) | Jx ⊆ M for some J ∈ GV(D)}, Mw is independent of E(M ) (up to isomorphism), and M ⊆ Mw ⊆ E(M ). We say that M is a w-module (or semidivisorial) if M = Mw . Also, M is w-finite type if M is a w-module and M = Bw for some finitely generated submodule B of M . Any undefined terminology used in this article is standard as in [6,12] or will be explained in the course of this paper. 1. Basic results Let D be an integral domain and S a (not necessarily saturated) multiplicative subset of D. We say that a nonzero ideal I of D is S-w-finite (respectively, S-w-principal) if there exist an s ∈ S and a w-finite type (respectively, principal) ideal J of D such that sI ⊆ J ⊆ Iw . We also define D to be an S-strong Mori domain (S-SMdomain) (respectively, S-factorial domain (or S-unique factorization domain)) if each nonzero ideal of D is S-w-finite (respectively, S-w-principal). Clearly, an S-finite (respectively, S-principal) ideal is S-w-finite (respectively, S-w-principal); so S-Noetherian domains (respectively, S-principal ideal domains) are always S-SM-domains (respectively, S-factorial domains). Note that a nonzero ideal I is S-w-finite (respectively,
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
317
S-w-principal) if and only if Iw is S-w-finite (respectively, S-w-principal); so D is an S-SM-domain (respectively, S-factorial domain) if and only if every w-ideal of D is S-w-finite (respectively, S-w-principal). Remark 1.1. (1) Let I be an S-w-finite ideal of D. Then there exist an s ∈ S and a finitely generated ideal J of D such that sI ⊆ Jw ⊆ Iw . Since J is finitely generated, we can find a Q ∈ GV(D) such that QJ ⊆ I (cf. [23, Lemma 1.1]). Note that (QJ)w = Jw and QJ is finitely generated. Hence we may assume that J is a finitely generated subideal of I by replacing J with QJ. Thus we can conclude that D is an S-SM-domain if and only if for each nonzero ideal I of D, there exist an s ∈ S and a finite generated subideal J of I such that sI ⊆ Jw ⊆ Iw . (2) Let I be a nonzero ideal of D. If I is S-w-principal, then there exist an s ∈ S and a principal ideal (d) such that sI ⊆ (d) ⊆ Iw ; so sIw ⊆ (d) ⊆ Iw . Hence Iw is S-principal. Conversely, if Iw is S-principal, then we can find a t ∈ S and a principal ideal (e) of D such that tIw ⊆ (e) ⊆ (Iw )w = Iw ; so tI ⊆ (e) ⊆ Iw . Hence I is S-w-principal. Thus D is an S-factorial domain if and only if every w-ideal of D is S-principal. (3) Let I be an S-w-principal ideal of D. Then there exist an s ∈ S and a principal ideal (d) such that sI ⊆ (d) ⊆ Iw . If d can be chosen in I, then sI ⊆ (d) ⊆ I; so I is S-principal. Before investigating S-w-properties, we first study how w-ideals behave in an integral domain D and its quotient ring DS . While the proofs can be obtained by [20, Theorem 6.7.10], [27, Theorem 8], [22, Proposition 1.2] and [21, Lemma 2.6(1)], we include them for the sake of completeness. Lemma 1.2. Let D be an integral domain, S a multiplicative subset of D and I a nonzero ideal of D. Then the following assertions hold. (1) If I is a w-ideal of D, then IDS ∩ D is a w-ideal of D. (2) If IDS is a w-ideal of DS , then IDS ∩ D is a w-ideal. (3) Iw DS ⊆ (IDS )w and (Iw DS )w = (IDS )w . Proof. (1) Let a ∈ (IDS ∩D)w . Then there exists a J ∈ GV(D) such that Ja ⊆ IDS ∩D. Since J is finitely generated, Jsa ⊆ I for some s ∈ S; so sa ∈ Iw = I. Hence a ∈ IDS ∩D, and thus IDS ∩ D is a w-ideal of D. (2) Let a ∈ (IDS ∩ D)w . Then we can find a GV-ideal J of D such that Ja ⊆ IDS ∩ D. Note that JDS ∈ GV(DS ) (cf. [10, Lemma 3.4(1)]); so aJDS ⊆ IDS . Hence a ∈ (IDS )w = IDS , and thus a ∈ IDS ∩ D. (3) Let a ∈ Iw DS . Then sa ∈ Iw for some s ∈ S; so Jsa ⊆ I for some J ∈ GV(D). Hence aJDS ⊆ IDS , and thus a ∈ (IDS )w . The second assertion follows because by (2), (Iw DS )w ⊆ ((IDS )w )w = (IDS )w ⊆ (Iw DS )w . 2
318
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
Let f : D → E be a ring homomorphism. We say that every nonzero ideal of E is extended from D with respect to f if every nonzero ideal I of E is of the form I = f (A)E for some nonzero ideal A of D. Recall that an integral domain D is a strong Mori domain (SM-domain) if it satisfies the ascending chain condition on integral w-ideals of D, or equivalently, every w-ideal of D is of w-finite type [23, Theorem 4.3]. Clearly, if D is an SM-domain (respectively, UFD), then D is an S-SM-domain (respectively, S-factorial domain). The next proposition collects some properties of our S-w-concepts. Proposition 1.3. Let S be a multiplicative subset of an integral domain D. Then the following statements hold. (1) Let I be a w-ideal of D. If IDS ∩ D is S-w-finite (respectively, S-w-principal), then I is S-w-finite (respectively, S-w-principal) and IDS ∩ D = I : s for some s ∈ S. (2) Let S be the saturation of S. Then D is an S-SM-domain (respectively, S-factorial domain) if and only if D is an S-SM-domain (respectively, S-factorial domain). (3) Let T be a multiplicative subset of D containing S. If D is an S-SM-domain (respectively, S-factorial domain), then D is a T -SM-domain (respectively, T -factorial domain). (4) If S consists of units of D, then D is an S-SM-domain (respectively, S-factorial domain) if and only if D is an SM-domain (respectively, UFD). (5) Let E be an integral domain and let f : D → E be a ring homomorphism such that every nonzero ideal of E is extended from D with respect to f and for every nonzero ideal I of D, f (Iw )E ⊆ (f (I)E)w . If D is an S-SM-domain (respectively, S-factorial domain), then E is an f (S)-SM-domain (respectively, f (S)-factorial domain). (6) If D is an S-SM-domain (respectively, S-factorial domain), then DS is an SMdomain (respectively, UFD). (7) Assume that for every w-finite type ideal J of D, (JDS )w ∩ D = J : s for some s ∈ S. If DS is an SM-domain, then D is an S-SM-domain. Proof. (1) Since IDS ∩D is S-w-finite (respectively, S-w-principal), there exist an s1 ∈ S and a w-finite type (respectively, principal) ideal J of D such that s1 (IDS ∩ D) ⊆ J ⊆ (IDS ∩ D)w = IDS ∩ D, where the equality follows from Lemma 1.2(1). Since I is a w-ideal and J is of w-finite type (respectively, principal), we can find an s2 ∈ S such that s2 J ⊆ I. Hence we have s1 s2 I ⊆ s1 s2 (IDS ∩ D) ⊆ s2 J ⊆ I, which shows that I is S-w-finite (respectively, S-w-principal) and IDS ∩ D = I : s1 s2 . (2) The “only if” part is obvious. For “if” part, let I be a nonzero ideal of D. Since D is an S-SM-domain (respectively, S-factorial domain), there exist a t ∈ S and a w-finite type (respectively, principal) ideal J of D such that tI ⊆ J ⊆ Iw . Let s ∈ S be a multiple
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
319
of t. Then sI ⊆ J ⊆ Iw ; so I is S-w-finite (respectively, S-w-principal). Thus D is an S-SM-domain (respectively, S-factorial domain). (3) This assertion is clear. (4) The result easily follows from the fact that D is an SM-domain (respectively, UFD) if and only if every w-ideal of D is of w-finite type (respectively, principal) (cf. [11, page 284]). (5) Clearly, f (S) is a multiplicative subset of E. Let A be a nonzero ideal of E. Then A = f (I)E for some nonzero ideal I of D. Since D is an S-SM-domain (respectively, S-factorial domain), there exist an s ∈ S and a finitely generated (respectively, principal) ideal F of D such that sI ⊆ Fw ⊆ Iw ; so by the assumption, we obtain f (s)A = f (s)f (I)E ⊆ f (Fw )E ⊆ f (F )E w ⊆ f (Iw )E w = f (I)E w = Aw , where the second equality comes from the fact that f (Iw )E ⊆ (f (I)E)w if and only if (f (Iw )E)w = (f (I)E)w . Note that f (F )E is a finitely generated (respectively, principal) ideal of E. Hence A is f (S)-w-finite (respectively, f (S)-w-principal), and thus E is an f (S)-SM-domain (respectively, f (S)-factorial domain). (6) The assertion is an immediate consequence of (4), (5) and Lemma 1.2(3). (7) Choose any nonzero ideal I of D. Since DS is an SM-domain, (IDS )w = (JDS )w for some finitely generated subideal J of I; so we have (IDS )w ∩ D = (JDS )w ∩ D = (Jw DS )w ∩ D, where the second equality follows from Lemma 1.2(3). By our assumption, (Jw DS )w ∩ D = Jw : s for some s ∈ S; so we obtain sI ⊆ s (IDS )w ∩ D ⊆ Jw ⊆ Iw . Hence I is S-w-finite, and thus D is an S-SM-domain. 2 Let P be a prime ideal of an integral domain D and set S := D \ P . Then S is a multiplicative subset of D. We say that D is a P -strong Mori domain (P -SM-domain) (respectively, P -factorial domain (or P -unique factorization domain)) if D is an S-SMdomain (respectively, S-factorial domain). Proposition 1.4. Let D be an integral domain and M a maximal w-ideal of D. Then the following assertions are equivalent. (1) D is an M -SM-domain (respectively, M -factorial domain). (2) DM is a Noetherian domain (respectively, PID) and for every nonzero finitely generated (respectively, principal) ideal J of D, Jw DM ∩D = Jw : s for some s ∈ D\M .
320
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
Proof. (1) ⇒ (2) Let A be a nonzero ideal of DM . Then A = IDM for some nonzero ideal I of D. Since D is an M -SM-domain (respectively, M -factorial domain), there exist an s ∈ D \ M and a finitely generated (respectively, principal) ideal J of D such that sI ⊆ Jw ⊆ Iw . Hence we have IDM = sIDM ⊆ Jw DM ⊆ Iw DM . Note that Bw DM = BDM for any nonzero ideal B of D (cf. [25, Theorem 3.9]) because M is a maximal w-ideal of D. Thus IDM = JDM , which indicates that DM is a Noetherian domain (respectively, PID). The second part follows from Proposition 1.3(1). (2) ⇒ (1) Assume that DM is a Noetherian domain (respectively, PID) and let I be a nonzero ideal of D. Then IDM = JDM for some finitely generated (respectively, principal) subideal J of I. Hence we have I ⊆ IDM ∩ D = Jw DM ∩ D = Jw : s for some s ∈ D \ M , which implies that sI ⊆ Jw ⊆ Iw . Therefore I is (D \ M )-w-finite (respectively, (D \ M )-w-principal), and thus D is an M -SM-domain (respectively, M -factorial domain). 2 2. S-strong Mori domains In this section, D is an integral domain, S is a (not necessarily saturated) multiplicative subset of D and a module means a unitary D-module. We start this section with a relation between an S-Noetherian domain and an S-SM-domain. It is clear that if I is an S-finite ideal of D, then Iw is an S-w-finite ideal of D; so I is S-w-finite. Hence every S-Noetherian domain is an S-SM-domain. If D is a DW-domain, then the concept of an S-SM-domain coincides with that of an S-Noetherian domain, because every S-w-finite ideal is S-finite. (Recall that an integral domain D is a DW-domain (or t-linkative domain) if each nonzero ideal of D is a w-ideal, or equivalently, every nonzero prime ideal of D is a w-ideal [18, Proposition 2.2].) Proposition 2.1. The following assertions hold. (1) An S-SM-domain is a w-locally S-Noetherian domain. (2) A w-locally S-Noetherian domain which has finite w-character is an S-SM-domain. Proof. (1) Let M be a maximal w-ideal of an S-SM-domain D and let I be a nonzero ideal of D. If IDM = DM , then we have nothing to prove; so we assume that IDM is a proper ideal of DM . Since D is an S-SM-domain, there exist an s ∈ S and a finitely generated subideal F of I such that sI ⊆ Fw ⊆ Iw . Therefore we have
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
321
sIDM ⊆ (Fw )DM = F DM ⊆ IDM , and hence IDM is S-finite. Thus DM is an S-Noetherian domain. (2) Assume that D is a w-locally S-Noetherian domain which has finite w-character, and let I be a nonzero ideal of D. If I intersects S, then there is nothing to prove (cf. [3, Proposition 2(a)]); so it suffices to consider the case when I ∩ S = ∅. Choose any 0 = a ∈ I. Then a is contained in only finitely many maximal w-ideals of D, say M1 , . . . , Mn . Fix an i ∈ {1, . . . , n}. Since DMi is S-Noetherian, there exist an si ∈ S and a finitely generated subideal Fi of I such that si IDMi ⊆ Fi DMi . Let s := s1 · · · sn and set C := (a) + F1 + · · · + Fn . Then sIDMi ⊆ CDMi . Let M be a maximal w-ideal of D such that M = Mi for all i = 1, . . . , n. Then a is a unit in DM ; so IDM = DM = CDM . Therefore sIDM ⊆ CDM for all maximal w-ideals M of D. Hence we have sIw = s
IDM
M ∈w-Max(D)
⊆
sIDM
M ∈w-Max(D)
⊆
CDM
M ∈w-Max(D)
= Cw , where the equalities follow from [10, Proposition 2.8(3)]. Note that C is a finitely generated subideal of I. Hence Iw is S-w-finite, and so I is S-w-finite. Thus D is an S-SM-domain. 2 Proposition 2.2. D is an SM-domain if and only if D is an M -SM-domain for every maximal w-ideal M of D. Proof. The “only if” part is clear. For the converse, let I be a nonzero ideal of D. Since D is an M -SM-domain for every maximal w-ideal M of D, there exist an sM ∈ D \ M and a w-finite type ideal WM of D such that sM I ⊆ WM ⊆ Iw . Let FM be a finitely generated subideal of I such that WM = (FM )w . Note that {sM | M is a maximal w-ideal of D} is not contained in any maximal w-ideal of D; so we can take suitable elements sM1 , . . . , sMn ∈ D such that (sM1 , . . . , sMn )w = D. Therefore we have Iw = (sM1 , . . . , sMn )I w ⊆ (WM1 + · · · + WMn )w = (FM1 + · · · + FMn )w ⊆ Iw , and hence Iw = (FM1 + · · · + FMn )w . Thus D is an SM-domain. 2 Let D be an integral domain, S a (not necessarily saturated) multiplicative subset of D and M a w-module as a D-module. We say that M is S-w-finite if sM ⊆ F for some s ∈ S and some w-finite type submodule F of M ; and M is an S-strong Mori
322
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
module (S-SM-module) if each w-submodule of M is S-w-finite. It is easy to see that M is S-w-finite if and only if there exist an s ∈ S and a finitely generated submodule N of M such that sM ⊆ Nw ⊆ M ; and a w-finite module (respectively, SM-module) is an S-w-finite module (respectively, S-SM-module). Also, note that for a torsion-free D-module M , MP = (Mw )P for all P ∈ w-Max(D) [25, Theorem 3.9]. We next give the S-invariant of the Cohen type theorem for S-SM-domains. To prove this, we need the following lemma. Lemma 2.3. Let M be an S-w-finite w-module as a D-module and N a w-submodule of M which is maximal among non-S-w-finite w-submodules of M . Then (N :D M ) is a prime w-ideal of D with (N :D M ) ∩ S = ∅. Proof. We first show that if (N :D M ) = (0), then (N :D M ) is a w-ideal. While this follows directly from [23, Lemma 1.2], we include a proof for the sake of completeness. Clearly, (N :D M ) is an ideal of D. Let x ∈ (N :D M )w . Then Jx ⊆ (N :D M ) for some J ∈ GV(D); so JxM ⊆ N . Hence xM = (JxM )w ⊆ Nw = N , which says that x ∈ (N :D M ). Thus (N :D M ) is a w-ideal of D. Next, we prove that (N :D M ) is a prime ideal of D. If (N :D M ) = (0), then there is nothing to prove. Assume that (N :D M ) = (0) and deny the conclusion. Then there exist a, b ∈ D \ (N :D M ) such that ab ∈ (N :D M ); so (N + aM )w is S-w-finite by the maximality of N . Hence we can find an s ∈ S and a w-finite type submodule F1 of M such that s(N + aM )w ⊆ F1 ⊆ (N + aM )w . Now, we may assume that F1 = (n1 + am1 , . . . , nl + aml )w for some n1 , . . . , nl ∈ N and some m1 , . . . , ml ∈ M . Let I = {m ∈ M | am ∈ N }. Then I is a D-submodule of M containing N and bM ; so I is also S-w-finite. Hence there are an element t ∈ S and a w-finite type submodule F2 of M such that tI ⊆ F2 ⊆ Iw . Also, we may write F2 = (q1 , . . . , qk )w for some q1 , . . . , qk ∈ I. Let x ∈ N . Then there exists a GV-ideal J1 of D such that sxJ1 ⊆ (n1 + am1 , . . . , nl + aml ). Write J1 = (d1 , . . . , ds ). Then for each i = 1, . . . , s, sxdi = ui1 (n1 + am1 ) + · · · + uil (nl + aml ) for some ui1 , . . . , uil ∈ D. Hence we have sxJ1 = ui1 (n1 + am1 ) + · · · + uil (nl + aml ) | i = 1, . . . , s = (ui1 n1 + · · · + uil nl ) + a(ui1 m1 + · · · + uil ml ) | i = 1, . . . , s . Therefore ({ui1 m1 + · · · + uil ml | i = 1, . . . , s}) ⊆ I. Since tI ⊆ F2 and ({ui1 m1 + · · · + uil ml | i = 1, . . . , s}) is finitely generated, we have tJ2 {ui1 m1 + · · · + uil ml | i = 1, . . . , s} ⊆ (q1 , . . . , qk ) for some J2 ∈ GV(D). Write J2 = (e1 , . . . , ep ). Then for each j = 1, . . . , p, tej (ui1 m1 + · · · + uil ml ) = vij1 q1 + · · · + vijk qk
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
323
for some vij1 , . . . , vijk ∈ D. Hence we have tJ2 {ui1 m1 + · · · + uil ml | i = 1, . . . , s} = {vij1 q1 + · · · + vijk qk | i = 1, . . . , s and j = 1, . . . , p} . Therefore we obtain (ui1 n1 + · · · + uil nl ) + a(ui1 m1 + · · · + uil ml ) | i = 1, . . . , s tJ2 ⊆ (ui1 n1 + · · · + uil nl ) | i = 1, . . . , s tJ2 + a(ui1 m1 + · · · + uil ml ) | i = 1, . . . , s tJ2
stxJ1 J2 =
⊆ (tn1 , . . . , tnl , aq1 , . . . , aqk ). Since x was arbitrary, stN J1 J2 ⊆ (tn1 , . . . , tnl , aq1 , . . . , aqk ) ⊆ N , and hence stN ⊆ (tn1 , . . . , tnl , aq1 , . . . , aqk )w ⊆ Nw = N. This indicates that N is S-w-finite, which is absurd. Thus (N :D M ) is a prime ideal of D. Finally, we show that (N :D M ) ∩ S = ∅. If s1 ∈ (N :D M ) ∩ S, then s1 M ⊆ N . Since M is S-w-finite, there exist an s2 ∈ S and a finitely generated submodule F of M such that s2 M ⊆ Fw . Therefore we have s1 s2 N ⊆ s1 s2 M ⊆ (s1 F )w ⊆ s1 M ⊆ N, which means that N is S-w-finite. This contradicts the choice of N . Thus (N :D M ) ∩ S = ∅. 2 Theorem 2.4. Let M be an S-w-finite w-module as a D-module. Then M is an S-SMmodule if and only if the D-submodules (P M )w are S-w-finite for all prime w-ideals P of D disjoint from S. Proof. (⇒) This follows from the definition of S-SM-modules. (⇐) Suppose to the contrary that M is not an S-SM-module, and consider the set T = {U | U is a w-submodule of M which is not S-w-finite}. Then T = ∅. Let {Uα }α∈Λ be a chain in T and set U := α∈Λ Uα . If a ∈ Uw , then there is a GV-ideal J of D such that Ja ⊆ U . Since J is finitely generated, Ja ⊆ Uβ for some Uβ ; so a ∈ (Uβ )w = Uβ ⊆ U . Hence U is a w-submodule of M . If U is S-w-finite, then there exist an s ∈ S and a w-finite type submodule F of M such that sU ⊆ F ⊆ U . Since F is of w-finite type, F ⊆ Uα for some Uα ; so sUα ⊆ F ⊆ Uα . This contradicts the choice of Uα . Hence U is not S-w-finite. Clearly, U is an upper bound of {Uα }α∈Λ ; so by Zorn’s lemma, we can choose a maximal element N in T . By Lemma 2.3, (N :D M ) is a prime (w-)ideal of D.
324
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
Now, we claim that (N :D M ) is disjoint from S. The assertion is clear when (N :D M ) = (0). Assume that (N :D M ) = (0) and (N :D M ) intersects S. Choose any t1 ∈ (N :D M ) ∩ S. Since M is S-w-finite, t2 M ⊆ (b1 D + · · · + bn D)w for some t2 ∈ S and some b1 , . . . , bn ∈ M ; so we have t1 t2 N ⊆ t1 t2 M ⊆ t1 (b1 D + · · · + bn D)w ⊆ N. Therefore N is S-w-finite, a contradiction. Hence (N :D M ) is disjoint from S. Note that (N :D M ) ⊆ N :D (b1 D + · · · + bn D)w ⊆ (N :D t2 M ) = (N :D M ) : t2 = (N :D M ), where the last equality comes because (N :D M ) is a prime ideal of D disjoint from S. Therefore (N :D M ) = (N :D (b1 D + · · · + bn D)w ). Since N is a w-module, (N :D (b1 D + · · · + bn D)w ) = (N :D b1 D) ∩ · · · ∩ (N :D bn D); so (N :D M ) = (N :D b1 D) ∩ · · · ∩ (N :D bn D). Again, since (N :D M ) is a prime ideal, (N :D M ) = (N :D bi D) for some i ∈ {1, . . . , n}. Clearly, bi ∈ / N . By the maximality of N , (N + bi D)w is S-w-finite; so there exist u1 ∈ S, n1 , . . . , np ∈ N and a1 , . . . , ap ∈ D such that u1 (N + bi D)w ⊆ (n1 + a1 bi , . . . , np + ap bi )w . Let N = n1 D + · · · + np D. Then by arguing as in the proof of Lemma 2.3, we obtain u1 N ⊆ N + (N :D M )bi w ⊆ N + (N :D M )M w . By the hypothesis, ((N :D M )M )w is S-w-finite; so u2 ((N :D M )M )w ⊆ Gw for some u2 ∈ S and some finitely generated D-submodule G of ((N :D M )M )w . Hence we have u1 u2 N ⊆ u2 N + (N :D M )M w ⊆ u2 N + G w . Since (u2 N + G)w is a w-finite type submodule of N , N is S-w-finite. However this is absurd. Thus M is an S-SM-module. 2 Note that D itself is an S-w-finite w-module as a D-module; so we have Corollary 2.5. D is an S-SM-domain if and only if every prime w-ideal (disjoint from S) is S-w-finite. Our next work is to study the Hilbert basis theorem for an S-SM-domain when S has a special property. To do this, we need the following lemmas.
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
325
Lemma 2.6. Let M be a nonzero D-module and L be a submodule of M such that M/L is a w-module. If N is a submodule of M containing L, then (N/L)w = Nw /L. Proof. Let a + L ∈ (N/L)w . Then there exists a J ∈ GV(D) such that J(a + L) ⊆ N/L; so Ja ⊆ N . Therefore a ∈ Nw , and hence a + L ∈ Nw /L. Thus (N/L)w ⊆ Nw /L. For the reverse containment, choose any b + L ∈ Nw /L. Then Jb ⊆ N for some GV-ideal J of D; so J(b + L) = Jb + L ⊆ N/L. Hence b + L ∈ (N/L)w , and thus Nw /L ⊆ (N/L)w . 2 Lemma 2.7. Let D be an integral domain and S a multiplicative subset of D. Then the following assertions hold. (1) Let 0 → M → M → M → 0 be a short exact sequence of w-modules as D-modules. Then M is an S-SM-module if and only if M and M are S-SM-modules.
k (2) Let M1 , . . . , Mk be w-modules. Then i=1 Mi is an S-SM-module if and only if each Mi is an S-SM-module. (3) If D is an S-SM-domain, then every w-finite type torsion-free w-module as a D-module is an S-SM-module. Proof. (1) Let φ : M → M and ψ : M → M be D-module homomorphisms in the short exact sequence 0 → M → M → M → 0, and set L := Im(φ) = Ker(ψ). (⇒) Assume that M is an S-SM-module. Since M is isomorphic to a D-submodule of M , M is an S-SM-module. Let N be a w-submodule of M . Then N = N0 /L for some D-submodule N0 of M containing L. Note that N0 is a w-module by Lemma 2.6; so N0 is S-w-finite. Therefore we can find an s ∈ S and a finitely generated D-submodule F of N0 such that sN0 ⊆ Fw . Hence we obtain sN = (sN0 + L)/L ⊆ (Fw + L)/L ⊆ (Fw + L)/L w ⊆ (N0 /L)w = N. Therefore N is S-w-finite, because F/L is finitely generated. Thus M is an S-SMmodule. (⇐) Let N be a w-submodule of M . Then N/(N ∩ L) is a w-module by Lemma 2.6. Note that N/(N ∩L) ∼ = (N +L)/L ⊆ M/L ∼ = M ; so N/(N ∩L) is S-w-finite, because M is an S-SM-module. Hence there exist an s ∈ S and a finitely generated D-submodule H of N containing N ∩ L such that s(N/(N ∩ L)) ⊆ (H/(N ∩ L))w ; so sN ⊆ Hw + (N ∩ L) by Lemma 2.6. Since M is an S-SM-module, L is also an S-SM-module. Hence (N ∩ L)w is S-w-finite; so we can find a t ∈ S and a finitely generated D-submodule F of (N ∩ L)w such that t(N ∩ L)w ⊆ Fw . Therefore we obtain stN ⊆ t Hw + (N ∩ L) ⊆ t Hw + (N ∩ L)w ⊆ tHw + Fw ⊆ (tH + F )w . Since tH + F is finitely generated, N is S-w-finite. Thus M is an S-SM-module. (2) This equivalence comes directly from (1).
326
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
(3) Let M be a w-finite type w-module as a D-module, and let F be a finitely generated D-module such that M = Fw . Since F is torsion-free, we can imbed F into a finitely generated free module H k . Therefore M = Fw ⊆ (H k )w = H k . Note that H k is an S-SM-module by (2). Thus M is an S-SM-module by (1). 2 Recall that a multiplicative subset S of an integral domain D is anti-archimedean if n n≥1 s D ∩ S = ∅ for every s ∈ S. For example, if V is a valuation domain with no height-one prime ideals, then V \ {0} is an anti-archimedean subset of V [4, Proposition 2.1]. We are now ready to prove the Hilbert basis theorem for an S-SM-domain when S is anti-archimedean.
Theorem 2.8. Let D be an integral domain and S an anti-archimedean subset of D. Then D is an S-SM-domain if and only if the polynomial ring D[X1 , . . . , Xn ] is an S-SM-domain. Proof. Note that S is anti-archimedean in every ring containing D as a subring; so it suffices to prove the case n = 1. Set X = X1 . (⇒) Assume that D is an S-SM-domain. Let A be a w-ideal of D[X] and let I be the ideal of D generated by the leading coefficients of the elements in A. Then Iw is an S-w-finite ideal of D; so sIw ⊆ (a1 , . . . , am )w for some s ∈ S and a1 , . . . , am ∈ I. Let f1 , . . . , fm be polynomials in A whose leading coefficients are a1 , . . . , am , respectively, and let n1 , . . . , nm be degrees of f1 , . . . , fm , respectively. Now, let n be the maximum of n1 , . . . , nm , and set B := (f1 , . . . , fm )D[X]. Let f ∈ A with leading coefficient a and degree k. Then sa ∈ (a1 , . . . , am )w ; so Jsa ⊆ (a1 , . . . , am ) for some J ∈ GV(D). Let J = m (d1 , . . . , dt ). Note that for each i = 1, . . . , t, di sa = j=1 rij aj for some ri1 , . . . , rim ∈ D. m If k ≥ n, let gi = di sf − j=1 rij fj X k−nj for each i = 1, . . . , t. Then gi belongs to A and has degree strictly less than k. If we still have deg gi ≥ n for some i ∈ {1, . . . , t}, we repeat the same process. After finitely many steps, we can find a J ∈ GV(D) and an integer q ≥ 1 such that J sq f ⊆ (A ∩M ) +B, where M = D⊕DX ⊕· · ·⊕DX n−1 . Since D is an S-SM-domain, Mw is an S-SM-module by Lemma 2.7(3); so (A ∩M )w is S-w-finite. Therefore t(A ∩ M )w ⊆ (Dh1 + · · · + Dhs )w for some t ∈ S and h1 , . . . , hs ∈ A ∩ M . Note that (Dh1 + · · · + Dhs )w ⊆ ((h1 , . . . , hs )D[X])w . (To see this, let h ∈ (Dh1 + · · · + Dhs )w . Then there exists a GV-ideal J of D such that J h ⊆ (Dh1 + · · · + Dhs ); so J D[X]h ⊆ (h1 , . . . , hs )D[X]. Note that J D[X] ∈ GV(D[X]) [25, Proposition 1.2] (or [9, Lemma 3.1]); so h ∈ ((h1 , . . . , hs )D[X])w .) Let C = (h1 , . . . , hs )D[X] and choose a u ∈ i≥1 si D ∩ S. Then we obtain J D[X]utf ⊆ (A ∩ M ) + B tD[X] ⊆ Cw + B. Note that J D[X] ∈ GV(D[X]) [25, Proposition 1.2] (or [9, Lemma 3.1]); so utf ∈ (Cw + B)w = (C + B)w . Since the choice of C is independent of that of f , it follows that
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
327
utA ⊆ (C + B)w . Since C + B is finitely generated as an ideal of D[X], A is S-w-finite. Thus D[X] is an S-SM-domain. (⇐) Let I be a nonzero ideal of D. Since D[X] is an S-SM-domain, ID[X] is S-w-finite; so there exist an s ∈ S and a finitely generated subideal J of I such that sID[X] ⊆ (JD[X])w ⊆ (ID[X])w . Note that (AD[X])w = Aw D[X] for all nonzero ideals A of D [8, Proposition 4.3]. Hence sI ⊆ Jw ⊆ Iw , and thus D is an S-SM-domain. 2 Remark 2.9. The S-strong Mori property does not carry over to the power series ring. In fact, there is an example of an SM-domain D such that D[[X]] is not an SM-domain [19, Theorem 10]. (This is the case when S consists of units of D.) For an f ∈ D[X], c(f ) denotes the ideal of D generated by the coefficients of f , and for an ideal J of D[X], c(J) stands for the ideal of D generated by the coefficients of elements in J, i.e., c(J) = f ∈J c(f ). Let Nv = {f ∈ D[X] | c(f )v = D}. Then Nv is a (saturated) multiplicative subset of D[X] and the quotient ring D[X]Nv is called the t-Nagata ring of D. It was shown in [20, Theorem 6.10.5] that D[X]Nv is a DW-domain. Theorem 2.10. If S is an anti-archimedean subset of an integral domain D, then the following statements are equivalent. (1) D is an S-SM-domain. (2) D[X]Nv is an S-SM-domain. (3) D[X]Nv is an S-Noetherian domain. Proof. (1) ⇒ (3) Let ID[X]Nv be a nonzero ideal of D[X]Nv , where I is a nonzero ideal of D[X]. Note that D[X] is an S-SM-domain by Theorem 2.8, because D is an S-SM-domain and S is an anti-archimedean subset of D. Therefore we can find an s ∈ S and a finitely generated subideal J of I such that sI ⊆ Jw ⊆ Iw . Hence we have sID[X]Nv ⊆ Jw D[X]Nv = JD[X]Nv ⊆ ID[X]Nv , the equality follows from Lemma 1.2(3) and the fact that D[X]Nv is a DW-domain. Therefore ID[X]Nv is S-finite, and thus D[X]Nv is an S-Noetherian domain. (3) ⇒ (1) Let I be a nonzero ideal of D. Since D[X]Nv is an S-Noetherian domain, there exist an s ∈ S and a finitely generated subideal J of ID[X] such that sID[X]Nv ⊆ JD[X]Nv ⊆ ID[X]Nv . Let a ∈ I. Then sag ∈ J for some g ∈ Nv ; so sa ∈ c(J)w . Therefore sI ⊆ c(J)w ⊆ Iw , because c(J) ⊆ I. Note that c(J) is finitely generated. Hence I is S-w-finite, and thus D is an S-SM-domain. (2) ⇔ (3) It suffices to note that D[X]Nv is a DW-domain. 2 Theorem 2.11. Let D be an integral domain, S a multiplicative subset of D and M a w-module as a D-module. Then the following assertions hold.
328
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
(1) If L is an S-SM-submodule of M such that M/L is an S-SM-module, then M is an S-SM-module. (2) If D is an S-SM-domain and M is an S-w-finite torsion-free D-module, then M is an S-SM-module. Proof. (1) Let N be a w-submodule of M . If N ⊆ L, then there is nothing to prove, because L is an S-SM-module. Assume that N L. Note that N/(N ∩ L) is a w-module by Lemma 2.6. Since N/(N ∩ L) ∼ = (L + N )/L ⊆ M/L and M/L is an S-SM-module, N/(N ∩L) is S-w-finite. Hence there exist an s ∈ S and a finitely generated D-submodule H of N containing N ∩ L such that s(N/(N ∩ L)) ⊆ (H/(N ∩ L))w = Hw /(N ∩ L), where the equality follows from Lemma 2.6. Therefore sN ⊆ Hw + (N ∩ L). Also, since L is an S-SM-module, there exist a t ∈ S and a finitely generated D-submodule F of (N ∩ L)w such that t(N ∩ L)w ⊆ Fw . Therefore we have stN ⊆ t Hw + (N ∩ L) ⊆ tHw + Fw ⊆ (tH + F )w . Note that tH + F is a finitely generated submodule of N . Hence N is S-w-finite, and thus M is an S-SM-module. (2) Let N be a w-submodule of M . Since M is an S-w-finite D-module, there exist an s ∈ S and a finitely generated D-submodule F of M such that sM ⊆ Fw ; so sN ⊆ Fw . Note that Fw is an S-SM-module by Lemma 2.7(3); so we can find a t ∈ S and a finitely generated D-submodule H of sN such that tsN ⊆ Hw . Hence N is S-w-finite, and thus M is an S-SM-module. 2 Let M be a unitary D-module. We say that M is w-faithfully flat if M is w-flat and (M/P M )w = 0 for all maximal w-ideals P of D. (Recall that M is w-flat if MP is flat for all maximal w-ideals P of D.) It was shown that M is w-faithfully flat if and only if M is w-locally faithfully flat, i.e., the DP -module MP is faithfully flat for all maximal w-ideals P of D [14, Proposition 2.5]; and that for an extension D ⊆ E of integral domains, if E is a w-faithfully flat D-module, then (IE)w ∩ D = Iw for any nonzero ideal I of D [14, Lemma 2.10]. The following result shows that the S-strong Mori property descends into a w-faithfully flat ring extension. Proposition 2.12. Let D ⊆ E be an extension of integral domains such that (IE)w ∩ D = Iw for each nonzero ideal I of D, and let S be a multiplicative subset of D. If E is an S-SM-domain, then so is D. Proof. Let I be a nonzero ideal of D. Since E is an S-SM-domain, there exist an s ∈ S and g1 , . . . , gn ∈ I such that s(IE) ⊆ ((g1 , . . . , gn )E)w . Therefore by the assumption, we have
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
329
sI ⊆ (sI)w = (sIE)w ∩ D ⊆ (g1 , . . . , gn )E w ∩ D = (g1 , . . . , gn )D w ⊆ Iw , and hence I is S-w-finite. Thus D is an S-SM-domain. 2 3. S-factorial domains Throughout this section, D always denotes an integral domain and S is a (not necessarily saturated) multiplicative subset of D. Our first theorem is a generalization of the Cohen type theorem due to Isaacs. To do this, we need the following lemma. Lemma 3.1. A w-ideal M of D maximal among non-S-w-principal ideals is a prime ideal of D. Proof. Let M be a w-ideal of D maximal among non-S-w-principal ideals of D, and suppose to the contrary that M is not a prime ideal. Then we can find suitable elements a, b ∈ D \ M such that ab ∈ M ; so M + (a) is S-w-principal by the maximality of M . Hence there exist an s ∈ S and a principal ideal (c) of D such that s(M + (a)) ⊆ (c) ⊆ (M + (a))w . Let I := {d ∈ D | dc ∈ M }. Then I is an ideal of D containing M and b; so I is also S-w-principal by the maximality of M . Therefore we can find a t ∈ S and an ideal (e) of D such that tI ⊆ (e) ⊆ Iw . Let x ∈ M . Then sx = cy for some y ∈ I; so sM ⊆ Ic ⊆ M . Hence we have stM ⊆ tIc ⊆ (ce) ⊆ (Ic)w ⊆ Mw = M, which means that M is S-w-principal. However, this is absurd. Thus M is a prime ideal of D. 2 Note again that a nonzero ideal I is S-w-principal if and only if Iw is S-principal (Remark 1.1(2)). Theorem 3.2. D is an S-factorial domain if and only if every prime w-ideal of D (disjoint from S) is S(-w)-principal. Proof. (⇒) This implication is clear. (⇐) Suppose that D is not an S-factorial domain, and let Ω be the set of non-S(-w)-principal w-ideals of D. Then Ω is nonempty and is partially ordered un der inclusion. Let {Pα }α∈Λ be a chain in Ω and set P := α∈Λ Pα . If a ∈ Pw , then Ja ⊆ P for some J ∈ GV(D). Since J is finitely generated, Ja ⊆ Pα for some Pα ; so a ∈ (Pα )w = Pα ⊆ P . Hence P is a w-ideal of D. If P is S(-w)-principal, then there exist an s ∈ S and a principal ideal (d) of D such that sP ⊆ (d) ⊆ P ; so (d) ⊆ Pβ for some Pβ . Therefore sPβ ⊆ (d) ⊆ Pβ . This is impossible, because Pβ is not S(-w)-principal. Hence P is not S(-w)-principal. Note that P is an upper bound of {Pα }α∈Λ ; so by Zorn’s
330
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
lemma, there exists a maximal element M in Ω. By Lemma 3.1, M is a prime ideal, which contradicts the hypothesis. Thus D is an S-factorial domain. 2 Let D be an integral domain and S a multiplicative subset of D. Recall that D is a Prüfer v-multiplication domain (PvMD) if every nonzero finitely generated ideal of D is w-invertible (or equivalently, DP is a valuation domain for all prime w-ideals P of D [7]); and D is well behaved if for each prime t-ideal P of D, P DP is a prime t-ideal of DP . (Note that P DP is not generally a t-ideal of DP for a prime t-ideal P of D [26, Proposition 2.5].) It was shown in [26, Corollary 1.3] that if P is a prime t-ideal of a well behaved domain D such that P ∩ S = ∅, then P DS is a prime t-ideal of DS . Note that w = t in a PvMD; and that PvMDs are well behaved [26, page 202]. Hence if P is a prime w-ideal of a PvMD D such that P ∩ S = ∅, then P DS is also a prime w-ideal of DS . Corollary 3.3. Let P be a subset of a PvMD D consisting of representatives for the principal primes of DS . If DS is a UFD, then D is an S-factorial domain if and only if pDS ∩ D is S-w-finite for each p ∈ P. Proof. The necessary condition is clear. For the converse, let P be a prime w-ideal of D. If P ∩ S = ∅, then P is S-principal [3, Proposition 2(a)]; so we assume that P is disjoint from S. Since D is a PvMD, P DS is a prime w-ideal of DS . Since DS is a UFD, P DS = pDS for some p ∈ P (cf. [11, page 284]). Note that P = P DS ∩ D = pDS ∩ D. By the hypothesis, there exist an s ∈ S and a w-finite type ideal F of D such that sP ⊆ F ⊆ P . Since F is of w-finite type, stP ⊆ tF ⊆ (p) ⊆ P for some t ∈ S. Hence P is S(-w)-principal. Thus by Theorem 3.2, D is an S-factorial domain. 2 Let D be an integral domain and S a (saturated) multiplicative subset of D. We say that S is a t-splitting set of D if for each 0 = d ∈ D, (d) = (AB)t for some integral ideals A, B of D such that At ∩ (s) = sAt (or equivalently, (A, s)t = D (cf. [1, Lemma 2.4])) for all s ∈ S and Bt ∩ S = ∅. It was shown in [1, Corollary 2.3] that S is a t-splitting set of D if and only if for all 0 = d ∈ D, dDS ∩ D is w-invertible in D; so if S is a t-splitting set of D, then for all 0 = d ∈ D, dDS ∩ D is of w-finite type [10, Proposition 2.6], and hence S-w-finite. Thus by Corollary 3.3, we have Corollary 3.4. Let D be a PvMD such that DS is a UFD. If S is a t-splitting set of D, then D is an S-factorial domain. Recall that D is a Krull domain if there exists a family {Vα }α∈Λ of rank-one essential discrete valuation overrings of D such that D = α∈Λ Vα and this intersection has finite character, i.e., each nonzero nonunit in D is a nonunit in only finitely many of valuation overrings Vα . It was shown in [24, Theorem 2.8] that D is a Krull domain if and only if D is both a PvMD and an SM-domain. We also note that an SM-domain is always an S-SM-domain; and every multiplicative subset of a Krull domain is t-splitting [1, page 8].
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
331
Hence as immediate consequences of Proposition 1.3(6) and Corollaries 3.3 and 3.4, we obtain Corollary 3.5. The following assertions hold. (1) If D is both a PvMD and an S-SM-domain, then DS is a UFD if and only if D is an S-factorial domain. (2) If D is a Krull domain, then D is an S-factorial domain if and only if DS is a UFD. Note that D is a Krull domain if and only if every nonzero ideal of D is w-invertible (cf. [11, Theorem 3.6]); so D is a Krull domain if and only if D is an SM-domain and DM is a PID for all maximal w-ideals M of D (cf. [10, Proposition 2.6]). We are closing this article with new characterizations of Krull domains. Proposition 3.6. D is a Krull domain if and only if D is an M -factorial domain for all maximal w-ideals M of D. Proof. (⇒) Assume that D is a Krull domain, and let M be a maximal w-ideal of D. Then D is an SM-domain; so by Proposition 2.2, D is an M -SM-domain. Note that DM is a PID. Thus by Proposition 1.4, D is an M -factorial domain. (⇐) Clearly, any M -factorial domain is an M -SM-domain; so D is an M -SM-domain for all maximal w-ideals M of D. Hence D is an SM-domain by Proposition 2.2. Also, DM is a PID by Proposition 1.4. Thus D is a Krull domain. 2 For a prime w-ideal P of an integral domain D, the w-height of P , denoted by w-ht(P ), is the supremum of lengths of chains of prime w-ideals between (0) and P . (Here, we regard (0) as a prime w-ideal of D.) The w-dimension of D, denoted by w-dim(D), is defined to be the supremum of {w- ht(P ) | P is a prime w-ideal of D}. Note that if D is a Krull domain, then w-dim(D) = 1, because in a Krull domain, every w-ideal is a v-ideal and every height-one prime ideal is a maximal v-ideal. Corollary 3.7. D is a Krull domain if and only if for all prime w-ideals P ⊆ Q of D, bP ⊆ (a) for some a ∈ P and b ∈ D \ Q. Proof. (⇒) Let P ⊆ Q be prime w-ideals of D. Since D is a Krull domain, w-dim(D) = 1; so P = Q and P is a maximal w-ideal of D. Again, since D is a Krull domain, by Proposition 3.6, D is a P -factorial domain; so the result holds. (⇐) Let Q be any maximal w-ideal of D, set S := D \Q, and choose any prime w-ideal P of D. If P ∩ S = ∅, then P is S-principal [3, Proposition 2(a)]; so P is S-w-principal. If P ∩ S = ∅, then our assumption forces P to be S-w-principal; so D is a Q-factorial domain by Theorem 3.2. Thus D is a Krull domain by Proposition 3.6. 2
332
H. Kim et al. / Journal of Algebra 416 (2014) 314–332
Acknowledgments First of all, we would like to thank the referee for several valuable comments and suggestions which resulted in an improved version of the paper. The first author was supported by Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education, Science and Technology (2010-0011996). The third author was supported by Kyungpook National University Research Fund, 2013. References [1] D.D. Anderson, D.F. Anderson, M. Zafrullah, The ring D + XDS [X] and t-splitting sets, Arab. J. Sci. Eng. Sect. C Theme Issues 26 (2001) 3–16. [2] D.D. Anderson, S.J. Cook, Two star-operations and their induced lattices, Comm. Algebra 28 (2000) 2461–2475. [3] D.D. Anderson, T. Dumitrescu, S-Noetherian rings, Comm. Algebra 30 (2002) 4407–4416. [4] D.D. Anderson, B.G. Kang, M.H. Park, Anti-archimedean rings and power series rings, Comm. Algebra 26 (1998) 3223–3238. [5] D.D. Anderson, D.J. Kwak, M. Zafrullah, Agreeable domains, Comm. Algebra 23 (1995) 4861–4883. [6] R. Gilmer, Multiplicative Ideal Theory, Queen’s Papers in Pure Appl. Math., vol. 90, Queen’s University, Kingston, Ontario, 1992. [7] M. Griffin, Some results on v-multiplication rings, Canad. J. Math. 19 (1967) 710–722. [8] J.R. Hedstrom, E.G. Houston, Some remarks on star-operations, J. Pure Appl. Algebra 18 (1980) 37–44. [9] C.J. Hwang, J.W. Lim, A note on ∗w -Noetherian domains, Proc. Amer. Math. Soc. 141 (2013) 1199–1209. [10] B.G. Kang, Prüfer v-multiplication domains and the ring R[X]Nv , J. Algebra 123 (1989) 151–170. [11] B.G. Kang, On the converse of a well-known fact about Krull domains, J. Algebra 124 (1989) 284–299. [12] I. Kaplansky, Commutative Rings, Polygonal Publishing House, Washington, New Jersey, 1994. [13] H. Kim, Module-theoretic characterizations of t-linkative domains, Comm. Algebra 36 (2008) 1649–1670. [14] H. Kim, T.I. Kwon, Module-theoretic characterizations of strongly t-linked extensions, Kyungpook Math. J. 53 (2013) 25–35. [15] J.W. Lim, D.Y. Oh, S-Noetherian properties of composite ring extensions, Comm. Algebra (2014), http://dx.doi.org/10.1080/00927872.2014.904329, in press. [16] J.W. Lim, D.Y. Oh, S-Noetherian properties on amalgamated algebras along an ideal, J. Pure Appl. Algebra 218 (2014) 1075–1080. [17] Z. Liu, On S-Noetherian rings, Arch. Math. (Brno) 43 (2007) 55–60. [18] A. Mimouni, Integral domains in which each ideal is a w-ideal, Comm. Algebra 33 (2005) 1345–1355. [19] M.H. Park, Power series rings over strong Mori domains, J. Algebra 270 (2003) 361–368. [20] F. Wang, Foundations of commutative ring theory, preprint. [21] F. Wang, On w-projective modules and w-flat modules, Algebra Colloq. 4 (1997) 111–120. [22] F. Wang, w-Dimension of domains, II, Comm. Algebra 29 (2001) 2419–2428. [23] F. Wang, R.L. McCasland, On w-modules over strong Mori domains, Comm. Algebra 25 (1997) 1285–1306. [24] F. Wang, R.L. McCasland, On strong Mori domains, J. Pure Appl. Algebra 135 (1999) 155–165. [25] H. Yin, F. Wang, X. Zhu, Y. Chen, w-Modules over commutative rings, J. Korean Math. Soc. 48 (2011) 207–222. [26] M. Zafrullah, Well behaved prime t-ideals, J. Pure Appl. Algebra 65 (1990) 199–207. [27] S. Zhao, F. Wang, H. Chen, Flat modules over a commutative ring are w-modules, J. Sichuan Norm. Univ., Nat. Sci. 35 (2012) 364–366 (in Chinese).