You are on page 1of 18

This article was downloaded by: [Queensland University of Technology]

On: 31 October 2014, At: 14:58


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Communications in Algebra
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/lagb20

Valuation and Pseudovaluation Subrings


of an Integral Domain
a b
Ahmed Ayache & Othman Echi
a
Faculty of Sciences, Department of Mathematics , University of
Bahrain , Isa Town, Kingdom of Bahrain
b
Faculty of Sciences of Tunis, Department of Mathematics ,
University of Tunis–El Manar , Tunis, Tunisia
Published online: 01 Feb 2007.

To cite this article: Ahmed Ayache & Othman Echi (2006) Valuation and Pseudovaluation Subrings of
an Integral Domain, Communications in Algebra, 34:7, 2467-2483, DOI: 10.1080/00927870600650515

To link to this article: http://dx.doi.org/10.1080/00927870600650515

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the
“Content”) contained in the publications on our platform. However, Taylor & Francis,
our agents, and our licensors make no representations or warranties whatsoever as to
the accuracy, completeness, or suitability for any purpose of the Content. Any opinions
and views expressed in this publication are the opinions and views of the authors,
and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content
should not be relied upon and should be independently verified with primary sources
of information. Taylor and Francis shall not be liable for any losses, actions, claims,
proceedings, demands, costs, expenses, damages, and other liabilities whatsoever or
howsoever caused arising directly or indirectly in connection with, in relation to or arising
out of the use of the Content.

This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden. Terms &
Conditions of access and use can be found at http://www.tandfonline.com/page/terms-
and-conditions
Communications in Algebra® , 34: 2467–2483, 2006
Copyright © Taylor & Francis Group, LLC
ISSN: 0092-7872 print/1532-4125 online
DOI: 10.1080/00927870600650515

VALUATION AND PSEUDOVALUATION SUBRINGS


OF AN INTEGRAL DOMAIN

Ahmed Ayache
Faculty of Sciences, Department of Mathematics, University of Bahrain,
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

Isa Town, Kingdom of Bahrain

Othman Echi
Faculty of Sciences of Tunis, Department of Mathematics,
University of Tunis–El Manar, Tunis, Tunisia

Let R S be two rings. We say that R is a valuation subring of S (R is a VD in S, for


short) if R is a proper subring of S and whenever x ∈ S, we have x ∈ R or x−1 ∈ R.
We denote by NuR the set of all nonunit elements of a ring R. We say that R is
a pseudovaluation subring of S (R is a PV in S, for short) if R is a proper subring
of S and x−1 a ∈ R, for each x ∈ S\R, a ∈ NuR. This article deals with the study
of valuation subrings and pseudovaluation subrings of a ring; interactions between the
two notions are also given. Let R be a PV in S; the Krull dimension of the polynomial
ring on n indetrminates over R is also computed.

Key Words: Krull dimension; Pseudovaluation domain; Valuation domain.

2000 Mathematics Subject Classification: Primary 13G05, 13F30, 18A40; Secondary 18A20, 18A32.

INTRODUCTION
Many commutative rings naturally occurring in algebraic geometry are
not Noetherian, but in compensation have other properties, which make them
manageable. Valuation rings are among the well-known rings used in geometry.
There is a class of rings closely related to valuation rings; namely pseudovaluation
domains; a type of quasilocal domains introduced by Hedstrom and Houston
(1978a).
This article is a contribution to the study of valuation and pseudovaluation
domains with a new point of view.
We adopt the conventions that all rings are commutative with unit and
integral. An inclusion (extension) of rings signifies that the smaller ring is a subring
of the larger one and possesses the same multiplicative identity. Throughout this
article, qfR denotes the quotient field of an integral domain R and for an extension

Received February 20, 2005; Revised April 16, 2005. Communicated by A. Facchini.
Address correspondence to Othman Echi, Faculty of Sciences of Tunis, Department of
Mathematics, University Tunis–El Manar, Campus Universitaire, 2092 El Manar, Tunis, Tunisia;
E-mail: othechi@yahoo.com and othechi@math.com

2467
2468 AYACHE AND ECHI

of integral domains R ⊆ S, tdS  R is the transcendence degree of qfS over


qfR. Also, we assume that a subring R of S is always properly contained in S.
Recall that an integral domain R is said to be a pseudovaluation domain (or,
in short, a PVD) (Hedstrom and Houston, 1978a) if each  ∈ SpecR satisfies the
following condition: whenever x y ∈ qfR are such that xy ∈ , then either x ∈
 or y ∈ . A considerable amount of attention has been paid over the years to
pseudovaluation domains (see for example Anderson et al., 2000; Badawi, 1995,
2000a,b, 2001, 2002; Badawi and Houston, 2002; Badawi et al., 1997; Cho, 1996,
2001; Dobbs, 1978; Dobbs et al., 1982; Gebru, 1998; Hedstrom and Houston,
1978a,b; Kang and Park, 2002; Matsuda and Sugatani, 1995).
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

An integral domain R is said to be a valuation domain if it satisfies the


following property: whenever x ∈ qfR, we have x ∈ R or x−1 ∈ R. It is natural to
think about domains R satisfying the above property with replacing the field qfR
by a given domain S containing R; thus the following definition is legitimated.

Definition 0.1. Let R S be two rings. We say that R is a valuation subring of S


(R is a VD in S, for short) if R is a subring of S and whenever x ∈ S, we have x ∈ R
or x−1 ∈ R.

Now, we propose a definition of pseudovaluation subring of a ring.

Definition 0.2. Let R be a ring. We denote by NuR the set of all nonunit
elements of R. We say that R is a pseudovaluation subring of S (R is a PV in S, for
short) if R is a subring of S and x−1 a ∈ R, for each x ∈ S\R, a ∈ NuR.

Our purpose in this article is the study of valuation subrings and


pseudovaluation subrings of a ring; interactions between the two notions are also
given. Let R be a PV in S; the Krull dimension of the polynomial ring on n
indeterminates over R is also computed.
We close this introduction by noting that a number of results namely
Propositions 1.4, 1.8, 3.3, 3.6 and Theorem 5.3 are slight extensions of the
corresponding ones valid for pseudovaluation domains given by Hedstrom and
Houston (1978a). For the convenience of the reader, we include the proofs which
are mostly identical to those of Hedstrom and Houston (1978a).

1. GENERALITIES
Remark 1.1. Let R be a ring.

0 If R is a PV (or a VD) in S, and R ⊂ T ⊂ S, then R is also a PV (respectively,


a VD) in T .
1 R is a PV in S = qfR means that R is a PVD.
2 Recall that an extension of rings R ⊆ S is said to be residually algebraic if for
each prime ideal  of S, S/ is algebraic over R/ ∩ R (Dobbs and Fontana,
1984). The pair R S is said to be residually algebraic if for each ring T between
R and S the extension R ⊆ T is residually algebraic (Ayache and Jaballah, 1997).
VALUATION AND PSEUDOVALUATION SUBRINGS 2469

Let R be a local ring which is integrally closed in S. If R S is residually


algebraic, then for x ∈ S\R, we have according to Ayache and Jaballah (1997,
Theorem 2.5), x−1 ∈ R, so R is a VD in S.

3 If R is a PV in S, then qfR = qfS.

Indeed, if x ∈ S\R, a ∈ NuR, then ax−1 = r ∈ R\0 , so x = a


r
∈ qfR.

Let R be a subring of a domain S. A prime ideal  of R is said to be S-strong


if whenever x y ∈ S satisfy xy ∈ , then either x ∈  or y ∈  (Dobbs et al., 1982).
If each  ∈ SpecR is S-strong, we say that S is a strong extension of R (Dobbs
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

et al., 1982).

Lemma 1.2. Let R be a PV in S. Then S is a strong extension of R.

Proof. Let  be a prime ideal of R and a b ∈ S such that ab ∈ . If a b ∈ R, then


there is nothing to prove. Suppose that one of the two elements a b is not in R;
say for instance b  R. As ab ∈ NuR, we get b−1 ab = a ∈ R. If a is a unit of R,
then b = a−1 ab ∈ , a contradiction, so that a ∈ NuR. Thus b−1 a ∈ R. Therefore,
a2 = b−1 aab ∈  and consequently, a ∈ . 

Remark 1.3. The converse of the above lemma does not hold. It suffices to take a
non-quasilocal domain R and S = RX. Then S is a strong extension of R; however,
R is not PV in S.

Proposition 1.4. Let R be a subring of a domain S. Then the following statements


are equivalent:

(i) R is a PV in S;
(ii) for each x ∈ S\R and each,  ∈ SpecR, we have x−1  ⊆ ;
(iii) for each x ∈ S\R and each a ∈ NuR, x + aR = xR.

Proof. i ⇒ ii Let p ∈  and x ∈ S\R. Then, x−1 p ∈ R, so p = ax for some
a ∈ R. As p = ax ∈  and x  , then a ∈ , by Lemma 1.2. Thus x−1 p ∈ .

ii ⇒ iii Let a ∈ NuR, and x ∈ S\R. Then there exists a prime ideal 
of R such that aR ⊆ . Hence x+a
a
∈ , since x + a ∈ S\R. But x+a
x
= 1 + x−1 a ∈ 1 +
 ⊆ R, so that x + aR ⊆ xR. On the other hand, x+a = 1 − x+a ∈ 1 +  ⊆ R, and
x a

consequently, xR ⊆ x + aR.
iii ⇒ i Let x ∈ S\R and a ∈ N . Then x + aR = xR. Hence x+a
x
∈ R,
i.e., 1 + ax−1 ∈ R. Thus ax−1 ∈ R. 

Recall that a prime ideal  of a ring R is said to be divided if R =  (Dobbs,


1976). Note that if  is a divided prime ideal of a ring R, then it is comparable to
any ideal of R (Akiba, 1967).
The following result links VD subrings to divided prime ideals and valuation
rings.
2470 AYACHE AND ECHI

Theorem 1.5. Let R ⊆ S be an extension of rings. Then the following statements are
equivalent:
(i) R is a VD in S;
(ii) there is a divided prime ideal  of R such that S = R and the quotient R/ is a
valuation domain.

Proof. i ⇒ ii Let  be a maximal ideal of S. Then, we have  ∩ S\R =


∅, so that   R. Hence  is contained in the conductor R  S of R in S, and
consequently,  = R  S. It follows that S is a quasilocal domain with maximal
ideal  = R  S.
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

Clearly, the integral domain R/ may be considered as a subring of the field
K = S/. Let u = s +  ∈ K, with s ∈ R. Since s ∈ R or s−1 ∈ R, we get u = s +  ∈
R/ or u−1 = s−1 +  ∈ R/. Therefore R/ is a valuation ring with quotient field K.
Since S is a quasilocal domain with maximal ideal , it is easy to see that R ⊆ S.
Now, if s ∈ S\R, then s−1 ∈ R, so that s−1  . It follows that s = s−1
1
∈ R . Thus
S = R and  = S = R .
ii ⇒ i Of course, we have qfR/ = R /R = S/. Now, let x ∈ S; then
x +  ∈ S/ = qfR/. As R/ is a valuation ring, we get x +  ∈ R/ or x + −1 ∈
R/.
If we suppose that x +  ∈ R/, then x ∈ R.
If x + −1 ∈ R/, then there is y ∈ R such that xy − 1 ∈ . Thus x−1 = 1+p
y
,
−1
for some p ∈ . Therefore, x ∈ R =  ⊆ R, proving that R is a VD in S. 

Corollary 1.6. Let R ⊆ S be an extension of rings such that R is a VD in S. Then R


and S are quasilocal.

Proof. The proof of the previous theorem shows that S is quasilocal. Furthermore,
there is a divided prime ideal  of R such that the quotient R/ is a valuation
domain, by Theorem 1.5. Since each ideal of R is comparable to  (by Akiba, 1967)
and R/ is quasilocal, it is easily seen that R is also quasilocal. 

Proposition 1.7. If R is PV in S, then R is quasilocal.

Proof. Suppose that R is not quasilocal. Let  and  be two distinct maximal
ideals of R. Then  +  = R. Hence there exist m ∈ , m ∈  such that m +
m = 1. For x ∈ S\R, we have x−1 m x−1 m ∈ R. Thus x−1 ∈ R. It follows that R is a
valuation subring of S; against our assumption, by Corollary 1.6. 

Let  be a prime ideal of a ring R and let S be an overring of R. Then  S is


called a comparable pair if  and xR are comparable for each x ∈ S (Okabe, 1986).

Proposition 1.8. Let R ⊆ S be an extension of rings such that R is quasilocal with


maximal ideal . Then the following statements are equivalent:
(i) R is a PV in S;
(ii) for each x ∈ S\R x−1  ⊆ ;
(iii)  S is a comparable pair.
VALUATION AND PSEUDOVALUATION SUBRINGS 2471

Proof. i ⇒ ii Straightforward.


ii ⇒ iii Let x ∈ S\.
If x ∈ R, then x is a unit of R and  ⊆ xR = R.
If x  R, then x−1  ⊆ , so  ⊆ x ⊆ xR.
iii ⇒ i Let x ∈ S\R. Then  ⊆ xR. Hence x−1  ⊆ R. It follows that

x−1 a ∈ x−1  ⊂ R for every a ∈ NuR

On the other hand, for each m ∈ , we have xx−1 m = m ∈ . Thus, as x 


 and  is S-strong, we get x−1 m ∈ . Therefore R is a PV in S. 
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

Corollary 1.9. Let R be a PV in S and  the maximal ideal of R. Then for each prime
ideal 
=  of R, the localization R is a valuation subring of S .

Proof. Let z ∈ S . Then z =


s with s ∈ S and
∈ R\; this implies that z
∈ S.
If z
∈ R, then z = z

∈ R .
If z
 R, then z
−1  ⊆ , by Proposition 1.8. Let m ∈ \; then z−1 =
z
 m
m−1 ∈ Rm−1 ⊆ R .
−1


If R is a PV in S and  is the maximal ideal of R, then two cases have to be


considered; the first (resp. second) case is when S
= S (resp. S = S). In the first
(resp. second) case, we say that we have a first (resp. second) type of PV subrings.
The next two sections deal with these two types.

2. FIRST TYPE OF PSEUDOVALUATION SUBRINGS OF A RING


Let B be an integral domain, I an ideal of B and D a subring of B/I. Consider
the pullback construction of commutative domains:

R −→ D
↓ ↓
B −→ B/I

Following Cahen (1988, 1990), we say that R is the domain of the B I D
construction and we set R = B I D. Under these notations, we have a
characterization of pseudovaluation subrings of first type.

Theorem 2.1. Let R be a quasilocal subring of a ring S and  be the maximal ideal
of R. Suppose that S
= S. Then the following statements are equivalent:

(i) R is a PV in S;
(ii)  is a divided prime ideal of S;
(iii)  is a prime ideal of S and SpecR = SpecS ;
(iv)  is a prime ideal of S and for each z ∈ S \R, z−1 ∈ S ;
(v)  is a prime ideal of S and R is a PV in S ;
(vi)  is a prime ideal of S and for each z ∈ S − R  = z;
(vii)  is a prime ideal of S and R is the ring of S  S  R/ construction.
2472 AYACHE AND ECHI

Proof. i ⇒ ii Let  be a maximal ideal of S containing S. Then  ∩ R = .


Let s ∈ S; if we suppose that s  R, then ms  R for some m ∈ . As ms ∈ S,
we get ms−1 m ∈ . Hence s−1 ∈ , so that 1 = ss−1  ∈ , against the maximality
of . It follows that s ⊆ R.
Now, we will prove that s ⊆ . Indeed, if s contains a unit of R, then
s = R. Consequently,  = Rs−1 , and so s−1 ∈ , this gives, as previously, 1 ∈ .
Therefore, s ⊆ ; and  is an ideal of S. As  is S-strong, by Lemma 1.2,  is
a prime ideal of S.
It remains to show that S = . Let x ∈ S  x = mq with m ∈  and
q ∈ S\.
If q ∈ R, then q is a unit of R, so that x = mq −1 ∈ .
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

If q  R, then q −1  ⊆ , i.e.,  ⊆ q. In particular, there is a m ∈  such


that m = qm . Thus x = m ∈ .

ii ⇒ iii Now, R ⊆ S is a pair of rings with the same maximal ideals
MaxR = MaxS  =  . Thus SpecR = SpecS , by Anderson and Dobbs
(1980).
iii ⇒ iv Let z ∈ S \R. Since S =  ⊂ R, we have z  S . But S
is quasilocal with maximal ideal S . Thus z is a unit of S .
iv ⇒ ii Let z ∈ S . Then z−1  S . It follows that z ∈ R, so that z ∈
S ∩ R = .
ii ⇒ i Let z ∈ S\R. Then z  . Hence 1z  ⊆ S = .
Thus we have proven the following equivalences:

i ⇐⇒ ii ⇐⇒ iii ⇐⇒ iv

v ⇒ i Straightforward.


iv ⇒ v Let z ∈ S \R. Then z−1 ∈ S and consequently, z−1 S ⊆ S .
But  is a divided prime ideal of S; thus z−1  ⊆ .
vi ⇒ i Let z ∈ S\R, then  = z, so z−1  = .
i ⇒ vi Let z ∈ S\R, we have z−1  ⊆ . But  is an ideal of S, so z ⊆
. Therefore,  = z.
ii ⇐⇒ vii Straightforward. 

Corollary 2.2. Let R ⊂ S be an extension of ring such that R is a quasilocal with


maximal ideal . If S
= S and R ⊂ S satisfies INC, in particular, if S is integral over
R, then the following statements are equivalent:

(i) R is a PV in S;
(ii) SpecR = SpecS;
(iii) S is quasilocal with maximal ideal .

Proof. ii ⇐⇒ iii By Anderson and Dobbs (1980).


iii ⇒ i By Theorem 2.1.
VALUATION AND PSEUDOVALUATION SUBRINGS 2473

i ⇒ iii Since  is a divided prime ideal of S, it is comparable to each


prime ideal of S. If  is any maximal ideal of S, then  ⊆ , i.e.,  = , by INC.


Remark 2.3. In Corollary 2.2, ii ⇐⇒ iii ⇒ i are always satisfied. Then, we
can give two easy examples of R PV in S, as follows.
1 If R is a valuation subring of S, then it is a pseudovaluation subring of S.
2 If R ⊆ S is an extension of rings such that SpecR = SpecS, then R is a
pseudovaluation subring of S.
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

Example 2.4. Let R be a quasilocal domain with maximal ideal . If R is PV in


S and S
= S, then S is not necessarily quasilocal:
Let k be a field and K = kY be the quotient field of the polynomial ring kY.
Set T = K X and  = XT . Let R S be the two domains defined by pullbacks;
S = T  kY and R = S  k. Thus we have the following two cartesian
squares:

R −→ R/ = k
↓ ↓
S −→ kY
↓ ↓
T = K X −→ K = kY

Clearly, S is not quasilocal since S/  kY, and  is a divided prime ideal of S.
Therefore, R is PV in S.

We give here (easiest possible) examples of rings R which are PV (respectively,


VD) in some S, but are not pseudovaluation domains (respectively, valuation
domains) in the usual sense. (The two following examples are communicated to us
by the anonymous referee.)

Example 2.5. Let F ⊂ K be a proper field extension, and let X be an indeterminate


over K. Let R = F + X 2 KX and S = K + X 2 KX. Then R is PV in S, by
Corollary 2.2 and Remark 2.3. Moreover, X ∈ qfR\R and X 2 ∈  = X 2 KX
implies that R is not a pseudovaluation domain.

Example 2.6. Let V be a nontrivial valuation domain of the field K, and let X Y
be indeterminates over K. Let S = KX Y and let R consist of those formal power
series of S whose constant terms lie in V . Then R is VD in S, but R is not a valuation
domain.

3. SECOND TYPE OF PSEUDOVALUATION SUBRINGS OF A RING


This section is devoted to the study of pseudovaluation subrings R of a ring S
such that S = S, where  is the maximal ideal of R.

Lemma 3.1. Let R be a PV in S of the second type. Then for each x ∈ S\R, x−1 ∈ S.
2474 AYACHE AND ECHI


Proof. Let  be the maximal ideal of R. As S = S, we get ni=1 mi si = 1 for
−1
some mi ∈  and si ∈ S. If x ∈ S\R, then x mi ∈  for each 1 ≤ i ≤ n. Thus
x−1 = ni=1 x−1 mi si ∈ S. 

Notation 3.2. Let R  be a quasilocal domain and S be a domain containing R.


Define

  S = x ∈ S  x ⊆ 

and
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

−1
S = x ∈ S  x ⊆ R 

Note that   S is strictly contained in S, when S = S while


  S = .

Proposition 3.3. Let R  be a quasilocal domain. Then the following statements
are equivalent:
(i) R is PV in S;
(ii)   S is a valuation subring of S with maximal ideal .
If one of the two statements is satisfied, then   S is the unique intermediate
ring between R and S that is a valuation subring of S with maximal ideal .

Proof. i ⇒ ii As R is PV in S, then for each x ∈ S\R, x−1  ⊆ , so that


x−1 ∈   S . Thus   S is a valuation subring of S.
To show that  is the maximal ideal of   S , let x ∈   S be not unit.
If x  , then x  R, so x−1  ⊆ ; thus x−1 ∈   S , a contradiction.
ii ⇒ i If   S is a valuation subring of S of maximal ideal , then
SpecR = Spec  S  by Anderson and Dobbs (1980). Let x ∈ S\R.
If x ∈   S , then x ∈   S \R. Hence x−1  ⊆ , since Corollary 22
implies that R is a PV in   S .
If x    S , then x−1 ∈   S since   S is a valuation subring of S.
Thus x−1  ⊆ .
It follows that R is PV in S (see Remark 2.3).
For the uniqueness, let T be a subring of S containing R such that T is a
valuation subring of S with maximal ideal . Since T ⊆   S , T is a valuation
subring of   S . Now, by Theorem 1.5, there is a divided prime ideal  of T
such that T =   S ; hence  =  and   S = T = T . 

Corollary 3.4. Let R ⊆ S be an extension of domains. Then the following statements


are equivalent:
(i) R is VD in S with maximal ideal ;
(ii) R is PV in S with maximal ideal  and R =   S .

Proof. ii ⇒ i By Proposition 3.3.


VALUATION AND PSEUDOVALUATION SUBRINGS 2475

i ⇒ ii We know that each VD in S is PV (Remark 2.3). Furthermore,


it is clear that R ⊆   S . Conversely, if x ∈ S\R, then x−1 ∈ R, so x−1 ∈ . As
xx−1 = 1  , we get x  . Therefore, x    S . 

Lemma 3.5. Let R  be a quasilocal domain which is PV in a domain S. If R has


a nonzero principal prime ideal, then R =   S .

Proof. Set T =   S . According to Proposition 3.3, T is quasilocal with


maximal ideal . Hence SpecR = SpecT, by Anderson and Dobbs (1980).
Suppose that R has a nonzero prime ideal  = rR, then  is a nonzero prime
ideal of T . Let t ∈ T ; then, as  = T  = rT , we get tr ∈ , so tr =
r for some
∈ R.
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

Hence t =
∈ R.
Therefore, R = T . 

Proposition 3.6. Let R ⊂ S be an extension of domains such that R is quasilocal with


maximal ideal  and R
=   S . Then the following statements are equivalent:

(i) R is PV in S;
(ii) −1
S is a valuation subring of S with maximal ideal .

Proof. It is sufficient to show that −1S =   S and then apply Proposition 3.3.
It is clear that   S ⊆ R  S . Conversely, let x ∈ R  S ; then x ⊆ R.
Hence x is an ideal of R. If we suppose that x  , then x = R, so that  =
Rx−1 is a principal ideal of R and consequently, R =   S by Lemma 3.5.


Now, we are in a position to give a characterization of pseudovaluation


subrings of a ring of the second type.

Theorem 3.7. Let R ⊂ S be an extension of domains such that R is quasilocal with


maximal ideal  and S = S. Then the following statements are equivalent:

(i) R is PV in S;
(ii) there is a divided prime ideal  of R such that S = R and R/ is a PVD;
(iii) S = R for a prime ideal  of R and R ⊂ S is strong;
(iv) S = R for a divided prime ideal  of R and  is S-strong;
(v) For z ∈ S\R, z−1 ∈ S, and  is S-strong.

Proof. i ⇒ ii Set T =   S ; then we have R ⊆ T  S and T is a


valuation subring of S of maximal ideal . By Theorem 1.5, there exists a divided
prime ideal  of T such that S = T and T/ is a valuation domain. As SpecR =
SpecT, then p ⊂  is a prime ideal of R and R = T = S (see Anderson and
Dobbs, 1980, Proposition 3.5). Hence R = T =  (i.e.,  is a divided prime ideal
of R). It remains to show that R/ is a PVD. The quotient field of R/ is qfR/ =
S/. Let z = z +  ∈ qfR/\R/, where z ∈ S. Then z ∈ S\R; so z−1  ⊆ .
Since S is quasilocal with maximal ideal , z−1 ∈ S. Let  S → S/ be
the canonical epimorphism. Then z−1  = z−1   ⊆ . Hence z−1 / ⊆
/. Thus R/ is PV in its quotient field; that is R/ is a PVD.
2476 AYACHE AND ECHI

ii ⇒ iii This can be derived easily from Anderson and Dobbs (1980).
iii ⇒ iv As R ⊆ R = S is a strong extension, then  is divided, since 
is R -strong.
iv ⇒ v Let z ∈ S\R. Then z   = R . Hence z is a unit of R = S.
v ⇒ i Let x ∈ S − R and m ∈ . We have x−1 mx ∈ . Now, as
−1
x ∈ S, x   and  is S-strong, we get x−1 m ∈ . Therefore, R is PV in S. 

Proposition 3.8. Let R be a PV in a domain S,  be the maximal ideal of R and C


be the conductor of R in S. Then the following properties hold.
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

(i) If S
= S, then for each prime ideal  of S such that   C, R/ ∩ R is PV
in S/.
(ii) If S = S, then for each prime ideal  of S such that  ⊆ C, R/ ∩ R is PV
in S/.

Proof. (i) In this case, we have SpecR = SpecS , C = . Hence  ∈ SpecR


and S = . As /S/
= S/ and /S// = /S / = /; this
yields R/ is PV in S/, by Theorem 2.1.
(ii) In this case, C =  is a prime ideal of R, S = R , R =  and R/ is a
PVD.
If  = , then R/ is PV in S/ = qfR/.
If   , then  ∈ SpecR, R = . Hence, we have /S/ = S/, S/ =
R /R  R// , /R// = /R /R  = / and finally R/// 
R/ is a PVD. Thus R/ is PV in S/, by Theorem 3.7. 

Proposition 3.9. Let R be an integral domain. Then the following statements are
equivalent:

(i) R is a PVD;
(ii) R is PV in a valuation overring.

Proof. i ⇒ ii Let V be the valuation ring associated with R. As R is PV in


qfR, then R is PV in V .
ii ⇒ i Suppose that R is PV in a valuation overring V of R. Let  be
the maximal ideal of R. Let x ∈ qfR\R. Either x ∈ V , then x ∈ V \R, so x−1  ⊆ ;
or x  V , so x−1 ∈ V . We consider two cases:

(a) If V = , then  is a prime ideal of V , so that x−1  ⊆  (in V ).


(b) If V
= V , then there exists a divided prime ideal  of R such that V = R and
so x−1 ∈  ⊂ R. Thus x−1  ⊆ . 

The following result gives links between valuation domains and completely
integrally closed domains which are PV in an overring.

Proposition 3.10. Let R be a completely integrally closed domain. If R is PV in S,


then R is a 1-dimensional valuation domain.
VALUATION AND PSEUDOVALUATION SUBRINGS 2477

Proof. Let  be the maximal ideal of R. We have necessarily S = S. By Theorem


3.7, there is a prime ideal  of R such that  ⊂ , R , S = R and R/ is a PVD.
Hence  is the conductor of R in S. Thus  = 0, S = R0 = qfR and R/ = R
is PVD. The associated valuation ring of R/ is V =    = R; and so R is a
valuation ring. Now, since R is completely integrally closed, then dim R = 1. 
A similar proof leads to the following proposition.

Proposition 3.11. Let R be a 1-dimensional domain which is PV in S. Let  be the


maximal ideal of R and suppose that S = S. Then R is a PVD.
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

4. KRULL DIMENSION
In this section, we are concerned with the prime spectrum of pseudovaluation
subrings of a domain; we are in particular concerned with the Krull dimension of
these rings.

Proposition 4.1. Let R be a pseudovaluation subring of domain S, C be the conductor


of R in S and  be the maximal ideal of R.
(1) If S
= S, then the following properties hold:
a)  = C and C is a divided prime ideal of S;
b) SpecS = SpecR ∪ Q ∈ SpecS  Q ⊇ C ;
c) dim S = dim R + dimS/C.
(2) If S = S, then the following properties hold:
a) C ⊆  and C is a divided prime ideal of R;
b) SpecR = SpecS ∪ Q ∈ SpecR  Q ⊇ C ;
c) dim R = dim S + dimR/C.

Proof. (1) Suppose that S


= S.
(a)  = C from Theorem 2.1.
(b) Set X = Q ∈ SpecS  Q ⊇  . Let  be a prime ideal of R. As SpecR =
SpecS , by Theorem 2.1, then we get  =  S for a prime ideal  of S. Clearly
we have  =  . Thus SpecR ∪ X ⊆ SpecS. The reverse containment follows
immediately from the fact that  is comparable to any prime ideal of S, since it
is a divided prime ideal of S.
(c) A consequence of (a) and (b).
(2) Suppose that S = S. Then, by Theorem 3.7, there is a divided prime ideal  of
R such that S = R .
(a) It is clear that C = p, by Theorem 3.7.
(b) Set Y =  ∈ SpecR   ⊇  . Let  be a prime ideal of S. Then  ⊆ R =
, so that  is a prime ideal of R. Thus SpecS ∪ Y ⊆ SpecR. For the
reverse containment, let  ∈ SpecR. Since  is a divided prime ideal of S,
two cases have to be considered:
(i) If  ⊇  then  ∈ Y
2478 AYACHE AND ECHI

(ii) If  ⊆ , then there is  ∈ SpecS such that  ∩ R = . But we have just


seen that SpecS ⊆ SpecR, then  =  ∈ SpecS.

(c) Combine (a) and (b). 

Corollary 4.2. Let R be PV in S and  be the maximal ideal of R. Let T be an


intermediate ring between R and S. If T = T or R T is INC, then the following
properties hold:

(i) SpecT ⊆ SpecR;


(ii) T is PV in S.
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

Proof. (i) From Proposition 4.1.


(ii) Let x ∈ S\T ; then x ∈ S\R. If Q ∈ SpecT, then Q ∈ SpecR, so x−1 Q ⊆
Q since R is PV in S. 

5. INTEGRAL CLOSURE AND NOETHERIAN CASE


Proposition 5.1. Let R be PV in S and  be the maximal ideal of R. If R is the
integral closure of R in S, then R ⊆   S .

Proof. The result is trivial when R is integrally closed in S. If R


= R , then
R
= R . By Proposition 4.1, SpecR ⊆ SpecR . Thus  is a prime ideal of R
and R ⊆   S . 

Lemma 5.2. If R is Noetherian and PV in S and  is the maximal ideal of R, then


R =   S .

Proof. Let x ∈   S ; then x ∈ S and x ⊆ . As  is finitely generated, x is


integral over R (by using an argument of determinant); so that x ∈ R . 

The following result characterizes Noetherian subrings which are PV in a given


domain.

Theorem 5.3. Let R be a Noetherian quasilocal subring of a domain S. Then the


following statements are equivalent:

(i) R is a PV in S;
(ii) x−1 ∈ R , whenever x ∈ S\R.

Proof. i ⇒ ii Let  be the maximal ideal of R. Then R =   S , by


Lemma 5.2.
If R = S, then R ⊂ S is an integral extension and S is quasilocal with maximal
 (see Corollary 2.2). In this case, for x ∈ S\R, we have x  , so that x is a unit
of R .
Now suppose that R ⊂ S. Then, by Proposition 3.3, R is VD in S. Hence, if
x ∈ S\R, we get either x ∈ R and consequently x   and x is a unit if R , or x  R ,
which implies x−1 ∈ R .
VALUATION AND PSEUDOVALUATION SUBRINGS 2479

ii ⇒ i In view of Proposition 1.8, it suffices to show that x−1  ⊆  for
all x ∈ S\R. This happens since, by hypothesis, x−1 ∈ R =   S . 

Theorem 5.4. Let R be a Noetherian quasilocal subring of a domain S with maximal


ideal . Then the following statements are equivalent:
(i) R is PV in S;
(ii) R ⊂ S is an integral extension and S is quasilocal with maximal , or S = qfR
and R is a rank one discrete valuation ring.
Under these conditions, every intermediate ring T between R and S is Noetherian
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

and PV in S.

Proof. When S = qfR, this result holds, by Hedstrom and Houston (1978a).
Suppose that S
= qfR. In this case, ii ⇒ i follows immediately from
Corollary 2.2.
i ⇒ ii As in the proof of the previous Theorem, either S = R or R is
PV in S. In both cases, R  S is a residually algebraic pair. Hence R S is also
residually algebraic. Thus R S is an INC pair (that is for each R ⊂ T ⊂ S, R ⊂ T
is INC); so that T is PV in S, by Corollary 4.2. In particular SpecS ⊆ SpecR.
Let Q ∈ SpecS, Q
= 0 (this is possible since S < qfR). Let x ∈ Q\0 ; then xS ⊆
Q ⊆ R; so S ⊆ x1 R (also T ⊆ S ⊆ x1 R). As R is Noetherian, S (or T ) is a finite type
R-module. Thus S is integral over R and T is Noetherian. 

6. PULLBACK CONSTRUCTIONS
It is always of interest to study pullback constructions; in order to provide
examples and counter examples.

Theorem 6.1. Let S be a ring,  a maximal ideal of S, K = S/, D a subring of K


and R be the domain of S  D construction. Then R is PV in S if and only if either:
(i) SpecR = SpecS, or
(ii) S = R , R = , and D is PVD.

Proof. Suppose that R is PV in S. Then R is quasilocal with maximal ideal . As


 is a common ideal of R and S, we have  ⊆ C, where C is the conductor of R
in S, so  = C.
If S
= S, then C =  = . Thus SpecR = SpecS, by Theorem 2.1.
If S = S, then C ⊂  so S = R , R =  and R/  D is PVD, by Theorem
3.7. Conversely, if SpecR = SpecS, it is clear that R is PV in S. If S = R , R =
 and D is PVD, then R/  D is PVD and R is quasilocal with maximal ideal  ⊇ 
such that S = S. Finally, R is PV in S, by Theorem 3.7. 

The following result gives condition making PV property transitive.

Proposition 6.2. Let R be PV in T and  the maximal ideal of R. If T


= T , and
if T is PV in S, then R is PV in S.
2480 AYACHE AND ECHI

Proof. By Theorem 2.1,  is a prime ideal of T . Let z ∈ S\R.


If z ∈ T , then z ∈ T \R, and so z−1  ⊆  since R is PV in T .
If z  T , then z ∈ S\T , so z−1  ⊆  since T is PV in S. 

The following example shows that without the condition T


= T , Proposition
6.2 does not hold.

Example 6.3. This example is similar to that given in Dobbs et al. (1982, p. 177).
Let K be a field, V1 = KX Y Ztt = KX Y Z + , where  = tV1 , and let
T = KX Y + . Then T is a PVD with quotient field KX Y Z t, which is not a
valuation ring.
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

Let V2 = KXYY ; then V2 is a valuation ring with a maximal ideal  = YV2 .


The domain D = K +  is a PVD.
Now, let us consider R = D +  = K +  with  =  +  , R is a pullback
construction determined by the following Cartesian square:

R −→ D
↓ ↓
T −→ T/  KX Y

As qfD = KX Y, then R = , T = R , and R/  D is a PVD. Hence, by


Theorem 6.1, R is PV in T with T = T . According to Hedstrom and Houston
(1978a, Proposition 2.6), the localization of a PVD under a nonmaximal prime ideal
is a VD. But    and R = T is not a valuation ring; thus R is not a PVD.
Therefore, R is PV in T and T is a PVD with T = T but R is not PVD.

Corollary 6.4. Let R be a quasilocal ring with maximal ideal . Suppose that the
integral closure R of R is strictly contained in S. Then R is PV in S if and only if R is
PV in S with maximal ideal .

Proof. If R is PV in S, then R is PV in R , as well, since R is an intermediate ring.


Hence R is quasilocal with maximal ideal . As R ⊂ R is INC, then R is PV in
S, by Corollary 4.2. Conversely, suppose that R is PV in S with maximal ideal ;
then SpecR = SpecR , and consequently, R is PV in R . Finally, as R
= R , we
conclude that R is PV in S, by Proposition 6.2. 

Proposition 6.5. Let R be a quasilocal ring with maximal ideal . Suppose that R is
PV in S, then each ring T between R and S is comparable with the ring   S .

Proof. The result is clear if   S is equal to R or S.


Let us suppose that R    S  S. Two cases arise:
(a) If T
= T , then  ∈ SpecT,by Theorem 2.1, and consequently T ⊆   S ;
(b) If T n= T , then there exist t1  t1      tn ∈ T and m1  m2      mn ∈  such
that
n i=1 mi ti = 1. For each z ∈   S , we have z ⊆ . Then z = 1z =
i=1 m i zti ∈ T = T . Thus   S ⊆ T . 

Next, we investigate pairs of rings R S such that each ring between R and S
is PV in S.
VALUATION AND PSEUDOVALUATION SUBRINGS 2481

Theorem 6.6. Let R be a quasilocal ring with maximal ideal . Then the following
statements are equivalent:
(i) Every ring between R and S is a PV in S;
(ii) R =   S and R is PV in S;
(iii) S is integral over R and quasilocal with maximal ideal  or R is a valuation
subring of S with maximal ideal , where R is the integral closure of R in S.

Proof. i ⇒ ii Clearly, R is PV in S. It remains to show that R =   S .


We know, from Proposition 5.1, that R ⊆   S and R is quasilocal with
maximal ideal  (Corollary 2.2). Set k = R /, we get the pullback
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

R −→ k
↓ ↓

  S −→   S /

Suppose that k ⊂   S /M is not algebraic. Let x ∈   S such that t =


x +  ∈   S / is transcendental over k. We have k ⊆ kt2  ⊆ kt ⊆   S .
Hence −1 kt2  = T is an intermediate ring strictly between R and   S . As
T/  kt2  is not quasilocal, then T is not PV in S, a contradiction. It follows that
k ⊂   S / is algebraic, and  is the maximal ideal of   S . Moreover,
  S is integral over R . But R is integrally closed in S, this yields R =   S .
ii ⇒ iii Either S =   S = R , then S is integral over R with
SpecR = SpecS, or R =   S  S, then R is a valuation ring of S with
maximal ideal , by Proposition 3.3.
iii ⇒ i Suppose that S is integral over R, and quasilocal with maximal
ideal . Then SpecR = SpecS, and so R is PV in S. Let T be a ring between R
and S; since R ⊂ T is INC, then T is PV in S, by Corollary 4.2.
Suppose now that R  S is a valuation subring of S, then R is PV in S and
S = S. By Corollary 6.4, R is PV in S. Now, using Proposition 3.3,   S is the
unique intermediate ring between R and S that is a valuation subring of S. Thus R =
  S . Let T be an intermediate ring between R and S. Then by Proposition 6.5:
(a) either T ⊆ R , so R ⊂ T is INC, and so T is PV in S, by Corollary 4.2; or
(b) or R  T ; In this case, it is clear that T is a valuation subring of S, thus T is
PV in S. 

The following result deals with the Krull dimension of the polynomial ring on
n indeterminates over a ring R which is PV in a domain S. Let us first fix some
notations. Let R be a ring, m an integer and  be a prime ideal of R; we denote by
Rm the polynomial ring on m indeterminates over R and m the prime ideal of
Rm whose elements are polynomials over .
Proposition 6.7. Let R be PV in a domain S, and  be the maximal ideal of R. Then
the following properties hold:
(i) If S = S, then for each integer m, we have
dim Rm = dimR/Cm − m + dim Sm

where C is the conductor of R in S;


2482 AYACHE AND ECHI

(ii) If S
= S, then

dim Sm + m − dimS/m ≤ dim Rm


≤ dim Sm + m − dimS/m + Infm d

where d is the transcendental degree of S/ over R/.

Proof. If S = S, then C is a divided prime ideal of R such that S = RC and R/C is


PVD. Hence R is the pullback construction given by the following Cartesian square
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

R −→ R/C
↓ ↓
S = RC −→ qfR/C

Thus, by Anderson et al. (1988), we have dimRm = dimR/Cm − m + dimSm.


Now, suppose that S
= S, then we have the two following pullbacks:

R −→ R/
↓ ↓
S −→ S/
↓ ↓
S −→ qfS/

Again, using Anderson et al. (1988), we get

dim Sm = dim S m + dimS/m − m

Thus

dim Sm = htSm m + m + dimS/m − m = htSm m + dimS/m

By using Cahen’s (1990) formulas, we get

dim Rm ≤ dimR/m + htSm m + Infm d

and

dim Rm ≥ dimR/m + htSm m ≥ m + dim Sm − dimS/m 

ACKNOWLEDGMENT
The authors wish to thank the referee for his valuable suggestions, comments,
and corrections. The second author thanks the DGRST (03/UR/15-03) for its
partial support.
VALUATION AND PSEUDOVALUATION SUBRINGS 2483

REFERENCES
Akiba, T. (1967). A note on AV -domains. Bull. Kyoto Niv. Education Ser. B 31:1–3.
Anderson, D. F., Dobbs, D. E. (1980). Pairs of rings with the same prime ideals. Canad. J.
Math. 32:362–384.
Anderson, D. F., Bouvier, A., Dobbs, D. E., Fontana, M., Kabbaj, S. (1988). On Jaffard
domains. Expo. Math. 5:145–175.
Anderson, D. F., Badawi, A., Dobbs, D. E. (2000). Pseudovaluation rings II. Boll. Unione
Mat. Ital. Sez. B Artic. Ric. Mat. 3(8):535–545.
Ayache, A., Jaballah, A. (1997). Residually algebraic pairs of rings. Math. Z. 225:49–65.
Badawi, A. (1995). A visit to valuation and pseudovaluation domains. In: Zero-Dimensional
Commutative Rings (Knoxville, TN, 1994). Lecture Notes Pure Appl. Math. 171. New
Downloaded by [Queensland University of Technology] at 14:59 31 October 2014

York: Dekker, pp. 155–161.


Badawi, A. (2000a). On -pseudovaluation rings II. Houston J. Math. 26:473–480.
Badawi, A. (2000b). On chained overrings of pseudovaluation rings. Comm. Algebra
28:2359–2366.
Badawi, A. (2001). On -chained rings and -pseudovaluation rings. Houston J. Math.
27:725–736.
Badawi, A. (2002). Pseudovaluation domains: a survey. In: Mathematics and Mathematics
Education (Bethlehem, 2000). River Edge, NJ: World Sci. Publishing, pp. 38–59
Badawi, A., Houston, E. (2002). Powerful ideals, strongly primary ideals, almost
pseudovaluation domains, and conducive domains. Comm. Algebra 30:1591–1606.
Badawi, A., Anderson, D. F., Dobbs, D. E. (1997). Pseudovaluation rings. In: Commutative
Ring Theory (Fes, 1995). Lecture Notes Pure Appl. Math. 185. New York: Dekker,
pp. 57–67.
Cahen, P.-J. (1988). Couples d’anneaux partagéant un idéal. Arch. Math. (Basel) 51:505–514.
Cahen, P.-J. (1990). Construction (B, I, D) et anneaux localement ou residuellement de
Jaffard. Arch. Math. 54:125–141.
Cho, Y. H. (1996). Pseudovaluation domains. Commun. Korean Math. Soc. 11:281–284.
Cho, Y. H. (2001). Pseudovaluation rings. Honam Math. J. 23:21–28.
Dobbs, D. E. (1976). Divided rings and going-down. Pacific J. Math. 67:353–363.
Dobbs, D. E. (1978). Coherence, ascent of going-down, and pseudovaluation domains.
Houston J. Math. 4:551–567.
Dobbs, D. E., Fontana, M. (1984). Universally incomparable ring homomorphism. Bull. Aust.
Math. Soc. 29:125–141.
Dobbs, D. E., Fontana, M., Huckaba, J. A., Papick, I. J. (1982). Strong ring extensions and
pseudovaluation domains. Houston J. Math. 8:167–184.
Fontana, M. (1980). Topologically defined classes of commutative rings. Ann. Mat. Pura Appl.
123:331–355.
Gebru, H. (1998). Krull’s conjecture, pseudovaluation domains and PF-dimension. Math. J.
Toyama Univ. 21:153–161.
Hedstrom, J. R., Houston, E. G. (1978a). Pseudovaluation domains. Pacific J. Math.
75:137–147.
Hedstrom, J. R., Houston, E. G. (1978b). Pseudovaluation domains. II. Houston J. Math.
4:199–207.
Kang, B. G., Park, M. H. (2002). Completion of a globalized pseudovaluation domain.
Houston J. Math. 28:701–710.
Lang, S. (1993). Algebra. 3rd ed. Reading, Mass: Addison-Wesley.
Matsuda, R., Sugatani, T. (1995). Cancellation ideals in pseudovaluation domains. Comm.
Algebra 23:3983–3991.
Okabe, A. (1986). Some ideal-theoretical characterizations of divided domains. Houston J.
Math. 12:563–577.

You might also like