Professional Documents
Culture Documents
Jin Xie
College of Mathematical Sciences
Sichuan Normal University
Chengdu 610068, P. R. China
xiejin fighting@126.com
J. Algebra Appl. Downloaded from www.worldscientific.com
Communicated by E. Kwak
In this paper, we study the pullbacks of LPVDs (respectively, t-LPVDs). We also study
the homological properties of LPVDs (respectively, t-LPVDs). More specifically, it is
proved that the only possible weak global dimensions (respectively, w-weak global dimen-
sions) of an LPVD (respectively, a t-LPVD) are 0, 1, 2, and ∞.
Keywords: LPVD; t-LPVD; pullbacks; weak global dimension; w-weak global dimension.
1. Introduction
Throughout this paper, R denotes an integral domain and K denotes the quotient
field of R. We use w.gl.dim(R) to denote the classical weak global dimension of R.
Multiplicative ideal theory is an important tool for characterizing integral
domains. Especially, star operations, which emerged in 1960s, play a key tool to
generalize Noetherian domains. With the appearance of star operations, many clas-
sical results in multiplicative ideal theory can be considered in the framework of the
unified treatment of star operation theory. Thereinto, the classical star operations
include the v-operation and the t-operation, but they cannot be used to deal with
the problems related to homological algebra and the category of modules. Then, in
1997, Wang and McCasland [15] broke it with the introduction of the w-operation
on an integral domain. Star operation theory can be organically combined with
homological algebra theory since then, and the corresponding w-homological the-
ory can be established. Recall that an integral domain R is said to be a Prüfer
v-multiplication domain (PvMD for short) if every nonzero finitely generated ideal
I is w-invertible, that is, there is a fractional ideal B of R such that (IB)w = R. As
2250113-1
2nd Reading
March 6, 2021 16:19 WSPC/S0219-4988 171-JAA 2250113
J. Xie
2250113-2
2nd Reading
March 6, 2021 16:19 WSPC/S0219-4988 171-JAA 2250113
LPVDs. Chang proved that R[X]Nv is an LPVD if and only if R is a t-LPVD and
R is a UMT domain; if and only if R[X] is a t-LPVD [4, Theorem 3.8].
By [9], any PVD is quasi-local. In addition, the name “pseudo-valuation domain”
is justified mainly because if a quasi-local domain (R, M ) is a PVD but not a
VD, then R has a unique valuation overring V = M −1 with maximal ideal M
[9, Theorem 2.7]. This good property of a PVD makes itself a pullback of rings.
by Dr mohamed Khalifa on 06/09/21. Re-use and distribution is strictly not permitted, except for Open Access articles.
a VD. That is, the only possible weak global dimensions of a PVD are 0, 1, 2 and ∞.
Then, it is natural to ask, for the pullbacks and the only possible weak global
dimensions of PVDs above-mentioned, whether LPVDs (resp., t-LPVDs) also have
such good corresponding properties. Note that LPVDs (respectively, t-LPVDs) do
not naturally become pullbacks of rings like PVDs. Therefore, in Sec. 2, we study
the basic diagram pullbacks of LPVDs, the Milnor squares of type D +M of LPVDs
and the general Milnor squares of LPVDs. Let (RDT F, M ) be a Milnor square of
type I. We show that R is an LPVD if and only if D and T are LPVDs and TM is
a VD. Let (RDT F, M ) be a Milnor square of type II. We show that R is an LPVD
if and only if D is a field and T is an LPVD. On the other hand, for the pullbacks
of t-LPVDs, Chang provided t-LPVDs with the Milnor squares of type D + M in
2008, i.e. let V = K + M be a VD and R = D + M , where K is a field, M is a
nonzero maximal ideal of V , and D is a proper subring of K, then R is a t-LPVD
if and only if D is a t-LPVD with quotient field K or D is a field [4, Proposition
3.5(3)]. Here, we further study the basic diagram pullbacks of t-LPVDs and the
general Milnor squares of t-LPVDs. More precisely, let (RDT F, M ) be a Milnor
square of type I. We show that R is a t-LPVD if and only if D and T are t-LPVDs
and TM is a VD. Let (RDT F, M ) be a Milnor square of type II. We show that R
is a t-LPVD if and only if D is a field, T is a t-LPVD and TM is a PVD.
In Sec. 3, the weak global dimensions (respectively, w-weak global dimensions) of
LPVDs (respectively, t-LPVDs) are studied. Let Max(R) (respectively, w-Max(R))
denote the set of all maximal ideals (respectively, w-ideals) of a ring R. It is shown
that LPVDs (respectively, t-LPVDs), even though with weaker conditions than
PVDs, still have such the same possible weak global dimensions (respectively, w-
weak global dimensions) as PVDs, i.e. the only possible values of the weak global
dimensions (respectively, w-weak global dimensions) of an LPVD (respectively, a
t-LPVD) are 0, 1, 2, and ∞, that correspond to the results of Dobbs [5]. More
precisely, let R be an LPVD (respectively, a t-LPVD) but not a Prüfer domain
(respectively, PvMD). Then w.gl.dim(R) (respectively, w-w.gl.dim(R)) = 2 or ∞:
2250113-3
2nd Reading
March 6, 2021 16:19 WSPC/S0219-4988 171-JAA 2250113
J. Xie
For easy reference, we review some definitions and notations. Let (RDT F ) be
the following commutative diagram of rings and ring homomorphisms:
p2
R /T
p1 g2
D /F ,
g1
J. Algebra Appl. Downloaded from www.worldscientific.com
2250113-4
2nd Reading
March 6, 2021 16:19 WSPC/S0219-4988 171-JAA 2250113
(), where each Fi is w-flat. The w-weak global dimension of a ring R is defined by
w-w.gl.dim(R) = sup{w-fdR (M )|M is an R-module}.
For unexplained terminologies and notations, we refer to [4, 7, 9, 10, 14].
to learn from PVDs. We first review the concept of the basic diagram pullbacks in
[6, 8]. Let R be the pullback of D and V over V /P as following:
V
π
D / V /P ,
i
Proof. The necessity follows from [7, Remark 2.4(e)]. We now give the proof of the
sufficiency. Let D be an LPVD. In order to prove that R is an LPVD, we prove: For
any maximal ideal M of R, RM is a PVD. By the hypothesis, (RDV V /P, P ) is a
Milnor square of type I. Hence, by [14, Theorem 8.3.6], M is comparable to P . Since
the maximality of M , P M , by localization at S = R − M and [14, Lemma 8.3.8],
we have a Milnor square (RM DM V V /P, P ), for convenience, we denote (R/P )M/P
by DM because D = R/M . Thus, DM is a PVD since M/P is a maximal ideal of
D = R/P . On the other hand, V /P is also the quotient field of DM since D and
DM have the same quotient field. Moreover, it is clear that DM is not a field.
Then, (RM DM V V /P, P ) is also a pullback of a basic diagram. Thus, RM is a PVD
by [6, Lemma 4.5(v)]. Therefore, R is an LPVD.
In [6], Dobbs also provided PVDs with the Milnor squares of type D + M . Let
V = K + M be a VD and R = D + M , where K is a field, M is a nonzero maximal
2250113-5
2nd Reading
March 6, 2021 16:19 WSPC/S0219-4988 171-JAA 2250113
J. Xie
Proof. We first give the proof of the necessity. Let R be an LPVD, and assume
that D is not a field. Let P be a maximal ideal of D. Then P + M is a maximal
ideal of R by [2, Proposition 2.1(4)], RP +M = DP + M by [3, Theorem 2.1(g)],
and RP +M is a PVD by assumption. Hence, DP is a PVD with quotient field K
by [6, Proposition 4.9(i)]. Therefore, D is an LPVD with quotient field K. Now,
J. Algebra Appl. Downloaded from www.worldscientific.com
Proof. The sufficiency follows from [6, Lemma 4.5(v)]. Thus we prove the necessity.
By [6, Lemma 4.5(i)], D is a PVD. Then by the hypothesis and [9, Proposition 2.6],
RM is a VD. Thus by [14, Exercise 1.17, Lemma 8.3.8, Theorem 8.3.10], T = TM =
RM is a VD.
Lemma 2.5. Let (RDT F, M ) be a Milnor square of type II, where T is a quasi-local
ring. Then R is a PVD if and only if D is a field and T is a PVD.
Proof. The sufficiency follows from Lemma 2.3. Thus we prove the necessity. It is
clear that T is a PVD since R and T have the same quotient field. If M ∈ Max(R),
then D is a field. If M ∈/ Max(R), then by [9, Proposition 2.6], (RM , M RM ) is a
VD. Thus M RM = M TM = M since T is quasi-local. Moreover, by localization at
2250113-6
2nd Reading
March 6, 2021 16:19 WSPC/S0219-4988 171-JAA 2250113
RM /M RM / F.
by Dr mohamed Khalifa on 06/09/21. Re-use and distribution is strictly not permitted, except for Open Access articles.
T are LPVDs.
Proof. We first prove the necessity. By Lemma 2.6, D and T are LPVDs. By
the hypothesis, there is an ideal P ∈ Max(R) such that M P . Moreover, by
localization at P , we have a Milnor square of type I:
RP / TM
RP /M RP / F,
DS / F.
2250113-7
2nd Reading
March 6, 2021 16:19 WSPC/S0219-4988 171-JAA 2250113
J. Xie
Proof. We only give the proof of the sufficiency. For the proof of the necessity,
see Lemma 2.6. Assume that T is an LPVD. For any P ∈ Spec(R), if M ⊆ P ,
then M = P since D is a field. Thus by [14, Lemma 8.3.8], we have a Milnor
by Dr mohamed Khalifa on 06/09/21. Re-use and distribution is strictly not permitted, except for Open Access articles.
Theorem 2.9. Let (RDT F, M ) be a Milnor square of type II. Then R is an LPVD
J. Algebra Appl. Downloaded from www.worldscientific.com
Proof. The sufficiency follows from Lemma 2.8. Thus we prove the necessity. By
Lemma 2.6, D and T are LPVDs. For any P ∈ Max(D), P = P/M , where P ∈
Max(R), M ⊆ P . Then we have a Milnor square of type II:
RP / TM
DP = DP / F,
Example 2.10. Let D = Q + XC[X], where Q is the field of rational numbers and
C is the field of complex numbers. Thus, we have a Milnor square of type II:
Q + XC[X] / C[X]
Q / C.
Then by [14, Theorem 8.6.2] and Theorem 2.9, D is an LPVD but not a Prüfer
domain. Set R1 = D + XF [[X]] and R2 = D + XF [X], where F is the quotient field
of D. It is clear that F [[X]] is a VD. Hence by Proposition 2.1, R1 is an LPVD.
Since F [X] is not a VD, thus we cannot use Propositions 2.1 or 2.2 to determine
whether R2 is an LPVD. However, by Theorem 2.7, R2 is an LPVD.
Next, we discuss the pullbacks of t-LPVDs. We first study the basic diagram
pullbacks of t-LPVDs.
2250113-8
2nd Reading
March 6, 2021 16:19 WSPC/S0219-4988 171-JAA 2250113
Proof. By the hypothesis, (RDV V /P, P ) is a Milnor square. Hence the neces-
sity follows immediately from Lemma 2.12. Thus we prove the sufficiency. Let
D be a t-LPVD, we will prove that RM is a PVD for any M ∈ w-Max(R). By
the hypothesis and [14, Theorem 8.3.6], we have that M is comparable to P . If
M ⊆ P , by the maximality of M , P = M . Hence, RM = RP = V is a VD. If
P M , then by [14, Lemma 8.3.8], we have a Milnor square (RM DM V V /P, P ).
Then by [14, Theorem 8.3.20], M/P is a maximal w-ideal of D = R/P . Thus
by [14, Exercises 1.37], DM = (R/P )M/P is a PVD. Since D and DM have the
same quotient field V /P and DM is not a field, thus (RM DM V V /P, P ) is also a
basic diagram pullback. Then by [6, Lemma 4.5(v)], RM is a PVD. Hence, R is
a t-LPVD.
Lemma 2.14. Let (RDT F, M ) be a Milnor square. Then M ∈ Max(R) if and only
if M ∈ w-Max(R).
Proof. The necessity follows immediately from Lemma 2.11. Thus we prove the
sufficiency. Let M ∈ w-Max(R) then by [14, Theorem 8.3.20], M/M = 0 ∈ w-
Max(D). Therefore, for any x ∈ D − 0, (0 + Dx)w = D. Thus, there is an ideal
J ∈ GV(D) such that J ⊆ Dx. Thus x−1 ∈ (Dx)−1 ⊆ J −1 = D. Hence D is a field.
Therefore M ∈ Max(R).
Proof. We first prove the necessity. By Lemma 2.12, D and T are t-LPVDs.
By assumption, D is not a field, so M ∈/ Max(R), whence M ∈ / w-Max(R) by
Lemma 2.14. Then by [14, Theorem 6.2.14], there is an ideal P ∈ w-Max(R) such
2250113-9
2nd Reading
March 6, 2021 16:19 WSPC/S0219-4988 171-JAA 2250113
J. Xie
DP / F,
by Dr mohamed Khalifa on 06/09/21. Re-use and distribution is strictly not permitted, except for Open Access articles.
J. Algebra Appl. Downloaded from www.worldscientific.com
DS / F.
Proof. We first prove the necessity. By Lemma 2.12, D and T are t-LPVDs. For
any P ∈ w-Max(D), P = P/M , where P ∈ w-Max(R), M ⊆ P . Then we have a
Milnor square of type II:
RP / TM
DP = DP / F,
DM / F,
2250113-10
2nd Reading
March 6, 2021 16:19 WSPC/S0219-4988 171-JAA 2250113
Theorem 3.1. Let R be an LPVD but not a Prüfer domain. Then w.gl.dim(R) = 2
or ∞. More precisely,
Corollary 3.2. Let (RDT F, M ) be a Milnor square of type II, where D is a field
and T is an LPVD. If for any P ∈ Max(R), P = P 2 , then w.gl.dim(R) = 2.
Example 3.3. Let R = Q + XR[X], where Q is the field of rational numbers and
R is the field of real numbers. Thus by [14, Theorem 8.6.2] and Theorem 2.9, R
is an LPVD but not a Prüfer domain. It is easy to see that (Q + XR[X])XR[X] =
Q + X(R[X]XR[X]), and since R[X]XR[X] is a VD, (Q + XR[X])XR[X] is a PVD
but not a VD. It is clear that XR[X] = (XR[X])2 . Hence by Theorem 3.1 and [14,
Exercise 1.35], w.gl.dim(R) = ∞.
Remark 3.4. It is clear that R = Q + XR[X] in Example 3.3 is not a PVD since
R is not quasi-local ring. Therefore, we cannot infer w.gl.dim(R) from the result of
Dobbs [5, Theorem 2.3].
Recall that an integral domain R is called a DW domain if every ideal of R is
a w-ideal [14, Definition 6.3.11]. Then by [14, Theorem 11.8.5; 18, Lemma 3.1.3;
13, Theorem 2.9], every LPVD is a DW domain. Therefore, we do not discuss the
2250113-11
2nd Reading
March 6, 2021 16:19 WSPC/S0219-4988 171-JAA 2250113
J. Xie
w-weak global dimension of an LPVD. Next, we use the w-weak global dimensions
to give homological properties of t-LPVDs.
Theorem 3.5. Let R be a t-LPVD but not a PvMD. Then w-w.gl.dim(R) = 2 or
∞. More precisely,
(1) w-w.gl.dim(R) = 2 if and only if M RM = M 2 RM for any M ∈ w-Max(R)
by Dr mohamed Khalifa on 06/09/21. Re-use and distribution is strictly not permitted, except for Open Access articles.
Proof. According to Theorem 3.5, it suffices to rule out the possibility that w-
w.gl.dim(R) = 2. Then by [14, p. 393] and the hypothesis, RM is a coherent
PVD for any M ∈ w-Max(R). Hence by [5, Corollary 3.4; 16, Proposition 3.1],
w-w.gl.dim(R) = 2. Therefore, the only possible values of w-w.gl.dim(R) are 0, 1
and ∞.
2250113-12
2nd Reading
March 6, 2021 16:19 WSPC/S0219-4988 171-JAA 2250113
References
[1] D. F. Anderson and D. E. Dobbs, Pairs of rings with the same prime ideals, Can. J.
Math. 32 (1980) 362–384.
[2] D. F. Anderson and A. Ryckaert, The class group of D + M , J. Pure Appl. Algebra
J. Algebra Appl. Downloaded from www.worldscientific.com
2250113-13