This article was downloaded by: [Florida Atlantic University]On: 30 September 2013, At: 21:43Publisher: Taylor & FrancisInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,37-41 Mortimer Street, London W1T 3JH, UK
Communications in AlgebraPublication details, including instructions for authors and subscription information:http://www.tandfonline.com/loi/lagb20
On Generalized Cohen–Macaulay Canonical ModulesNguyen Thi Hong Loan a & Le Thanh Nhan ba Vinh University , Nghe An , Vietnamb Thai Nguyen College of Sciences , Thai Nguyen , VietnamPublished online: 23 Sep 2013.
To cite this article: Nguyen Thi Hong Loan & Le Thanh Nhan (2013) On Generalized Cohen–Macaulay Canonical Modules,Communications in Algebra, 41:12, 4453-4462, DOI: 10.1080/00927872.2012.703737
To link to this article: http://dx.doi.org/10.1080/00927872.2012.703737
PLEASE SCROLL DOWN FOR ARTICLE
Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) containedin the publications on our platform. However, Taylor & Francis, our agents, and our licensors make norepresentations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose of theContent. Any opinions and views expressed in this publication are the opinions and views of the authors, andare not the views of or endorsed by Taylor & Francis. The accuracy of the Content should not be relied upon andshould be independently verified with primary sources of information. Taylor and Francis shall not be liable forany losses, actions, claims, proceedings, demands, costs, expenses, damages, and other liabilities whatsoeveror howsoever caused arising directly or indirectly in connection with, in relation to or arising out of the use ofthe Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematicreproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in anyform 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®, 41: 4453–4462, 2013Copyright © Taylor & Francis Group, LLCISSN: 0092-7872 print/1532-4125 onlineDOI: 10.1080/00927872.2012.703737
ON GENERALIZED COHEN–MACAULAYCANONICAL MODULES
Nguyen Thi Hong Loan1 and Le Thanh Nhan21Vinh University, Nghe An, Vietnam2Thai Nguyen College of Sciences, Thai Nguyen, Vietnam
Let �R��� be a Noetherian local ring which is a quotient of a Gorenstein local ring. Afinitely generated R-module M is called generalized Cohen–Macaulay canonical if thecanonical module K�M� of M is generalized Cohen–Macaulay. In this paper, we givecharacterizations of generalized Cohen–Macaulay canonical modules in term of certainsystems of parameters.
Key Words: Generalized Cohen–Macaulay canonical modules; Residual length; Strict f-sequences.
2000 Mathematics Subject Classification: 13D45; 13H10; 13E05.
1. INTRODUCTION
Throughout this paper, let �R��� be a Noetherian local ring which is aquotient of a n-dimensional Gorenstein local ring �R′��′�. Let M be a finitelygenerated R-module with dimM = d. Denote by Ki�M� the R-module Extn−i
R′ �M�R′�.Then Ki�M� is a finitely generated R-module. Following Schenzel [9], Ki�M� is calledthe ith deficiency module of M for i = 0� � � � � d − 1, and K�M� = Kd�M� is called thecanonical module of M .
Schenzel [9] introduced the notion of Cohen–Macaulay canonical modules: Mis called Cohen Macaulay canonical if the canonical module K�M� of M is CohenMacaulay. The Cohen–Macaulay canonical property is related to some importantquestions. For example, if M is Cohen–Macaulay canonical, then the monomialconjecture raised by Hochster [5] is valid for the ring R/AnnRM . Furthermore, ifR is a domain, then R is Cohen–Macaulay canonical if and only if R posseses abirational Macaulayfication R1, i.e., an extension ring R ⊆ R1 ⊆ Q (where Q is thefield of fractions) such that R1 is finitely generated as an R-module and R1 is aCohen–Macaulay ring, cf. [9, Theorem 1.1].
For a system of parameters (s.o.p. for short) x = �x1� � � � � xd� of M , set
I�x�M� = �R�M/�x1� � � � � xd�M�− e�x�M��
Received April 14, 2012. Communicated by I. Swanson.Address correspondence to Prof. Nhan Le, Thai Nguyen College of Sciences, Thai Nguyen,
Vietnam; E-mail: [email protected]
4453
Dow
nloa
ded
by [
Flor
ida
Atla
ntic
Uni
vers
ity]
at 2
1:43
30
Sept
embe
r 20
13
4454 NGUYEN AND LE
where e�x�M� is the multiplicity of M with respect to x� It is well known that Mis Cohen–Macaulay if and only if I�x�M� = 0 for all s.o.p x� This is the motivationof the following characterization of Cohen–Macaulay canonical modules in term ofsystems of parameters given in Theorem 4.1 of [7]: M is Cohen–Macaulay canonical
if and only ifd−3∑k=0
Rl(Hd−k−1
� �M/�x1� � � � � xk�M�) = 0 for all s.o.p. �x1� � � � � xd� of M
which are strict f-sequences in sense of Cuong-Morales-Nhan [3]. Here, Rl�−�denotes for the residual length defined by Sharp–Hamieh [11].
The notion of generalized Cohen–Macaulay modules was introduced byCuong–Schenzel–Trung [4]: M is generalized Cohen–Macaulay if there exists anumber I�M� such that I�x�M� < I�M� for all s.o.p x of M . Note that M isgeneralized Cohen–Macaulay if and only if ��Hi
��M�� < � for all i < d� By anobvious way, we say that M is generalized Cohen–Macaulay canonical if the canonicalmodule K�M� of M is generalized Cohen–Macaulay. It is clear that if d � 3 thenM is generalized Cohen–Macaulay canonical. The aim of this paper is to givecharacterizations of generalized Cohen–Macaulay canonical modules in term ofsystems of parameters for the case where d ≥ 4�
For a s.o.p. x = �x1� � � � � xd� of M , set Mx�k = M/�x1� � � � � xk�M for k =1� � � � � d − 3�
Main Theorem. Let d ≥ 4� The following statements are equivalent:
(i) M is generalized Cohen–Macaulay canonical;
(ii) There exists a number c�M� such thatd−3∑k=1
Rl(Hd−k−1
� �Mx�k�)� c�M� for all s.o.p. x
of M which are strict f-sequences;(iii) There exists a s.o.p. x = �x1� � � � � xd� of M which is a strict f-sequence and a
number c�x�M� such thatd−3∑k=1
Rl(Hd−k−1
� �M/�xn11 � � � � � x
nkk �M�
)� c�x�M� for all
n1� � � � � nd−3 ∈ ��
Furthermore, if (i), (ii), (iii) are satisfied, then
d−3∑k=1
Rl�Hd−k−1� �Mx�k�� �
d−3∑k=1
k∑i=0
(ki
)��Hi+2
� �K�M���
for all s.o.p. x of M which are strict f-sequences. The equality holds when x ⊆ �2d−4r �where
r = min�t ∈ � � �tHi��K�M�� = 0 for all i < d��
In Section 2, we outline some properties of strict f-sequences which will beused later. The proof of the main theorem is given in the last section.
2. STRICT f-SEQUENCES
The theory of secondary representation was introduced by Macdonald [6]which is in some sense dual to that of primary decomposition for Noetherianmodules. Note that every Artinian R-module A has a minimal secondary
Dow
nloa
ded
by [
Flor
ida
Atla
ntic
Uni
vers
ity]
at 2
1:43
30
Sept
embe
r 20
13
ON GENERALIZED COHEN–MACAULAY CANONICAL MODULES 4455
representation A = A1 + · · · + An� where Ai is �i-secondary. The set ��1� � � � � �n� isindependent of the choice of the minimal secondary representation of A. This set iscalled the set of attached primes of A� and denoted by AttRA.
Definition 2.1 ([3]). An element x ∈ � is called a strict filter regular element (strictf-element for short) with respect to M if x � � for all � ∈ ⋃d
i=1 AttRHi��M�\���. A
sequence �x1� � � � � xt� of elements in � is called a strict filter regular sequence (strictf-sequence for short) with respect to M if xi+1 is a strict f-element with respect toM/�x1� � � � � xi�M for every i = 0� � � � � t − 1.
It should be mentioned that Cuong–Morales–Nhan [3] used strict f-sequencesas a useful tool to study the finiteness for attached primes of local cohomologymodules, to study the polynomial property of the length of generalized fractionsdefined in [11], and to characterize the pseudo generalized Cohen–Macaulaymodules defined in [2]. In [7], strict f-sequences were used to study the monomialconjecture raised by Hochster [5] and to characterize Cohen–Macaulay canonicalmodules.
Lemma 2.2 ([3, Lemma 3.4]). If �x1� � � � � xt� is a strict f-sequence w.r.t. M , then sois �xn11 � � � � � x
ntt � for all positive integers n1� � � � � nt.
Following Cuong–Schenzel–Trung [4], an element x ∈ � is called a filterregular element with respect to M if x � � for all � ∈ AssRM\���� A sequence�x1� � � � � xt� of elements in � is called a filter regular sequence with respect toM if xi+1 is a filter regular element with respect to M/�x1� � � � � xi�M for alli = 0� � � � � t − 1.
Lemma 2.3. Each strict f-sequence w.r.t. M is a filter regular sequence w.r.t. M . Inparticular, if x ∈ � is a strict f-element w.r.t. M then ��0 �M x� < ��
Proof. By [1, 11.3.3], AssRM ⊆ ⋃di=0 AttRH
i��M�� Therefore, if x is a strict f-element
w.r.t. M , then x is a filter regular element w.r.t. M . So, the result follows. �
It is clear by Prime Avoidance that for each integer t > 0, there exists a strictf-sequence w.r.t. M of length t. Also, by Lemma 2.3, if �x1� � � � � xt� is a strict f-sequence w.r.t. M and t � d, then it is a part of a system of parameters of M .
From now on, let R be a quotient of a n-dimensional Gorenstein local ring�R′��′�� Denote by Ki�M� the R-module Extn−i
R′ �M�R′�� Then Ki�M� is a finitelygenerated R-module. Following Schenzel [8, Def. 2.1], Ki�M� is the ith deficiencymodule of M for i < d, and K�M� = Kd�M� is the canonical module of M�
Lemma 2.4. An element x ∈ � is a strict f-element w.r.t. M if and only if x is a filterregular element w.r.t. Ki�M� for all i. In particular, if d > 0 and x is a strict f-elementw.r.t. M , then x is K�M�-regular and ��0 �Ki�M� x� < � for all i < d.
Proof. By the local duality (cf. [1, 11.2.6]), we have an isomorphism
Hi��M� � HomR�K
i�M�� E�R/����
Dow
nloa
ded
by [
Flor
ida
Atla
ntic
Uni
vers
ity]
at 2
1:43
30
Sept
embe
r 20
13
4456 NGUYEN AND LE
where E�R/�� is the injective hull of R/�� Therefore we get by [10, Theorem 2.3]that AssRK
i�M� = AttRHi��M� for all i. Now the result follows. �
Lemma 2.5. Let x ∈ � be a strict f-element w.r.t. M . Then we have an exact sequence
0 −→ Ki+1�M�/xKi+1�M� −→ Ki�M/xM� −→ �0 �Ki�M� x� −→ 0
for each integer i ≥ 1. In particular, Hi�
(K�M�/xK�M�
) � Hi�
(K�M/xM�
)for any
i ≥ 2.
Proof. Since ��0 �M x� < � by Lemma 2.3, we have Hi��M/�0 �M x�� � Hi
��M�
for all i ≥ 1� Therefore, from the exact sequence 0 −→ M/�0 �M x�x−→ M −→
M/xM −→ 0, we have the exact sequence for all i ≥ 1
0 −→ Hi��M�/xHi
��M� −→ Hi��M/xM� −→ �0 �Hi+1
� �M� x� −→ 0�
By local duality (cf. [1, 11.2.6]), we have the exact sequence for all i ≥ 1
0 −→ Ki+1�M�/xKi+1�M� −→ Ki�M/xM� −→ �0 �Ki�M� x� −→ 0�
In particular, for i = d − 1, we have the exact sequence
0 −→ K�M�/xK�M� −→ K�M/xM� −→ �0 �Kd−1�M� x� −→ 0�
Since �0 �Kd−1�M� x� is of finite length by Lemma 2.4, we have for any i ≥ 2 that
Hi�
(K�M�/xK�M��
) � Hi�
(K�M/xM�
)�
�
Remark 2.6. Let d ≥ 1 and let x ∈ �, be a strict f-element w.r.t. M . Then x isK�M�-regular by Lemma 2.4. Therefore, we have the exact sequence
0 −→ K�M�x−→ K�M� −→ K�M�/xK�M� −→ 0�
This induces the exact sequence
0 −→ Hi��K�M��/xHi
��K�M�� −→ Hi��K�M�/xK�M��� −→ �0 �Hi+1
� �K�M�� x� −→ 0
for all integers i ≥ 0.
Let A be an Artinian R-module. Following Sharp–Hamieh [11], the stabilityindex of A, denoted by s�A�� is the least positive integer s such that �sA = �nAfor all n ≥ s� Denote by Rl�A� the length of A/�s�A�A� Then Rl�A� is finite, and iscalled the residual length of A� It is clear that Rl�A� = 0 if and only if � � AttRA�Moreover, if x � � for all � ∈ AttRA\���, then ��A/xA� � Rl�A� and in this case,��A/xnA� = Rl�A� for all n ≥ s�A��
Dow
nloa
ded
by [
Flor
ida
Atla
ntic
Uni
vers
ity]
at 2
1:43
30
Sept
embe
r 20
13
ON GENERALIZED COHEN–MACAULAY CANONICAL MODULES 4457
Lemma 2.7. Let i ≥ 0 be an integer, and let x be a strict f-regular element w.r.t. M .Then for all n 0, we have
Rl�Hi��M�� = ��H0
��Ki�M��� = ��0 �Ki�M� x
n��
Proof. Note that xn is a strict f-regular element w.r.t. M for any integer n >0. Therefore ��0 �Ki�M� x
n� < � by Lemma 2.4. Since H0��K
i�M�� is the largestsubmodule of Ki�M� of finite length, we have for n 0 that
H0��K
i�M�� = �0 �Ki�M� �n� ⊆ �0 �Ki�M� x
n� ⊆ H0��K
i�M���
So, ��H0��K
i�M��� = ��0 �Ki�M� xn� for n 0� Note that Hi
��M� � D�Ki�M�� by [1,11.2.6], where D�−� is the Matlis dual functor. Since ��0 �Ki�M� x
n� < �� we have
��0 �Ki�M� xn� = ��D�0 �Ki�M� x
n�� = �(D�Ki�M��/xnD�Ki�M��
)= �
(Hi
��M�/xnHi��M�
)�
Because x � � for all � ∈ AttRHi��M�\���� we have �
(Hi
��M�/xnHi��M�
) =Rl�Hi
��M�� for n 0. Therefore, the lemma follows. �
3. PROOF OF MAIN THEOREM
Firstly, we introduce the notion of generalized Cohen–Macaulay canonicalmodules.
Definition 3.1. M is called a generalized Cohen–Macaulay canonical module if thecanonical module K�M� of M is generalized Cohen–Macaulay.
Note that if dimM � 3 or M is generalized Cohen–Macaulay then Mis generalized Cohen–Macaulay canonical. If M/UM�0� is generalized Cohen–Macaulay, where UM�0� is the largest submodule of M of dimension less thand, then M is generalized Cohen–Macaulay canonical. In particular, sequentiallygeneralized Cohen–Macaulay modules and pseudo generalized Cohen–Macaulaymodules defined in [2] are generalized Cohen–Macaulay canonical.
Lemma 3.2. If M is a generalized Cohen–Macaulay canonical module then so isM/xM , for any strict f-regular element x w.r.t. M .
Proof. If d � 2, then the result is clear. Let d � 3� Then dim�M/xM� ≥ 2. Hencedepth�K�M/xM�� ≥ 2 and hence Hi
��K�M/xM�� = 0 for i = 0� 1� By Lemma 2.5,
Hi��K�M/xM�� � Hi
��K�M�/xK�M��
for all i ≥ 2� Note that x is K�M�-regular. Because K�M� is generalized Cohen–Macaulay, so is K�M�/xK�M�. Hence �
(Hi
��K�M�/xK�M��)< � for all i < d − 1�
So, Hi��K�M/xM�� is of finite length for all i < d − 1� Thus, M/xM is generalized
Cohen–Macaulay canonical. �
Dow
nloa
ded
by [
Flor
ida
Atla
ntic
Uni
vers
ity]
at 2
1:43
30
Sept
embe
r 20
13
4458 NGUYEN AND LE
Lemma 3.3. Let d ≥ 4� If M is generalized Cohen–Macaulay canonical, then
Rl�Hd−2� �M/xM�� � ��H2
��K�M���+ ��H3��K�M���
for any strict f-element x w.r.t. M . The equality holds when x ∈ �r , where
r = min�t ∈ � � �tHi��K�M�� = 0 for all i < d��
Proof. Let x be a strict f-element w.r.t. M . Set N = M/xM� Let n ∈ � and let ybe a strict f-element w.r.t. N . Then yn is a strict f-element w.r.t. N . Therefore, byapplying Lemma 2.5 to the module N with the strict f-element yn and the integeri = d − 2, we have the following exact sequence:
0 −→ K�N�/ynK�N� −→ K�N/ynN� −→ �0 �Kd−2�N� yn� −→ 0�
Note that dim�N/ynN� = d − 2 ≥ 2. Therefore, depth(K�N/ynN�
) ≥ 2� It followsthat Hi
��K�N/ynN�� = 0 for i = 0� 1. Note that �
(0 �Kd−2�N� y
n)< � by Lemma 2.4.
So, we have by the above exact sequence that
�0 �Kd−2�N� yn� = H0
��0 �Kd−2�N� yn� � H1
�
(K�N�/ynK�N�
)�
By Lemma 2.7, Rl�Hd−2� �N�� = �
(H0
��Kd−2�N��
) = ��0 �Kd−2�N� yn� for n 0. Hence
Rl�Hd−2� �N�� = �
(H1
��K�N�/ynK�N��
)� (1)
Since dimN = d − 1 ≥ 3, we have depthK�N� ≥ 2. Hence H1��K�N�� = 0. Also, we
can apply Remark 2.6 to the module N with the strict f-element yn and the integeri = 1 in order to obtain the isomorphism
H1�
(K�N�/ynK�N�
) � �0 �H2��K�N�� y
n��
Note that K�N� is generalized Cohen–Macaulay of dimension d − 1 ≥ 3 by Lemma3.2. Therefore, ��H2
��K�N��� < �. Hence �0 �H2��K�N�� y
n� = H2��K�N�� for n 0, and
hence
�(H1
��K�N�/ynK�N��
) = ��H2��K�N���� (2)
It follows by Lemma 2.5 that
H2��K�N�� � H2
��K�M�/xK�M���� (3)
By Remark 2.6, we have the following exact sequence:
0 −→ H2��K�M��/xH2
��K�M�� −→ H2��K�M�/xK�M��� −→ �0 �H3
��K�M�� x� −→ 0�
Dow
nloa
ded
by [
Flor
ida
Atla
ntic
Uni
vers
ity]
at 2
1:43
30
Sept
embe
r 20
13
ON GENERALIZED COHEN–MACAULAY CANONICAL MODULES 4459
Since d ≥ 4, we have by the choice of r that �0 �H3��K�M�� x� = H3
��K�M�� andxH2
��K�M�� = 0 whenever x ∈ �r . Therefore, we have by the above exact sequencethat
�(H2
��K�M�/xK�M��)� �
(H2
��K�M��)+ �
(H3
��K�M��)
(4)
and the equality holds if x ∈ �r . Now, combining (1)–(4) we have
Rl�Hd−2� �N�� = �
(H1
�
(K�N�/ynK�N�
)) = ��H2�
(K�N���
= �(H2
�
(K�M�/xK�M�
))� �
(H2
��K�M��)+ �
(H3
��K�M��)�
and the equality holds when x ∈ �r . �
For a s.o.p. x = �x1� � � � � xd� of M , set Mx�k = M/�x1� � � � � xk�M for k =1� � � � � d − 3.
Lemma 3.4. Suppose that M is generalized Cohen–Macaulay canonical. Let k be aninteger such that 1 � k � d − 3. Then
Rl�Hd−k−1� �Mx�k�� �
k∑i=0
(ki
)��Hi+2
� �K�M���
for any s.o.p. x = �x1� � � � � xd� of M which is a strict f-sequence. The equality holdswhen �x1� � � � � xd� ⊆ �2k−1r , where r = min�t ∈ � � �tHi
��K�M�� = 0 for all i < d�.
Proof. If d � 3, then there is nothing to do. So we assume that d ≥ 4. Let x =�x1� � � � � xd� be a s.o.p of M which is a strict f-sequence. We will show by inductionon d that
Rl�Hd−k−1� �Mx�k�� �
k∑i=0
(ki
)��Hi+2
� �K�M���
for any k = 1� � � � � d − 3, and the equality holds when x1� � � � � xk ∈ �2k−1r . The caseof d = 4 follows immmediately by Lemma 3.3. Let d > 4 and assume that the resultis true for d − 1� Set N = M/x1M . Note that N is generalized Cohen–Macaulaycanonical of dimension d − 1 by Lemma 3.2. Since �x2� � � � � xd� is a strict f-sequences.o.p of N , we have by induction that
Rl�H�d−1�−�k−1�−1� �N/�x2� � � � � xk�N�� �
k−1∑i=0
(k− 1i
)�(Hi+2
� �K�N��)
(5)
for any k = 2� � � � � d − 3, and the equality holds when x2� � � � � xk ∈ �2k−2s, where
s = min�t ∈ � � �t�Hi��K�N�� = 0 for all i < d − 1��
Dow
nloa
ded
by [
Flor
ida
Atla
ntic
Uni
vers
ity]
at 2
1:43
30
Sept
embe
r 20
13
4460 NGUYEN AND LE
Note that N/�x2� � � � � xk�N � Mx�k for all k = 2� � � � � d − 3. On the other hand, byLemma 2.5, we have an isomorphism
Hi��K�N�� � Hi
��K�M�/x1K�M��
for all i ≥ 2. Therefore, from the exact sequence as in Remark 2.6
0 −→ Hi��K�M��/x1H
i��K�M�� −→ Hi
��K�M�/x1K�M�� −→ �0 �Hi+1� �K�M�� x1� −→ 0�
we get that
��Hi��K�N��� = ��Hi
��K�M�/x1K�M��� � ��Hi��K�M���+ ��Hi+1
� �K�M���
for all i = 2� � � � � d − 2. Note that � k−1i �+ � k−1
i−1 � = � ki �. So, we have by (5) that
Rl�Hd−k−1� �Mx�k�� �
k−1∑i=0
(k− 1i
)��Hi+2
� �K�N���
�
k−1∑i=0
(k− 1i
)(��Hi+2
� �K�M���+ ��Hi+3� �K�M���
)
=k∑
i=0
(ki
)��Hi+2
� �K�M���
for all k = 2� � � � � d − 3. The case of k = 1 follows by Lemma 3.3. From the exactsequence as in Remark 2.6
0 −→ Hi��K�M��/xHi
��K�M�� −→ Hi��K�M�/xK�M��� −→ �0 �Hi+1
� �K�M�� x� −→ 0�
we have by the choice of r that �2rHi��K�M�/xK�M��� = 0 for all i < d −
1� Note that depthK�N� ≥ 2 since dimN ≥ 4� Therefore, Hi��K�N�� = 0 for i =
0� 1� Since Hi��K�N�� � Hi
��K�M�/x1K�M�� for all i ≥ 2 by Lemma 2.5, we have�2rHi
��K�N�� = 0 for all i = 2� � � � � d − 2� Hence s � 2r� Therefore, if x1� � � � � xk ∈�2k−1r , then
Rl�Hd−k−1� �Mx�k�� =
k∑i=0
(ki
)��Hi+2
� �K�M����
�
Lemma 3.5. Let x = �x1� � � � � xd� be a s.o.p. of M which is a strict f-sequence. Letc� k be integers such that 1 � k � d − 3 and ��Hk+2
� �K�M��� > c� Then there existpositive integers n1� � � � � nk such that Rl
(Hd−k−1
� �M/�xn11 � � � � � x
nkk �M�
)> c.
Proof. The proof is by induction on k = 1� � � � � d − 3. Let k = 1� Then��H3
��K�M��� > c� Note that H3��K�M�� is �-torsion, i.e., H3
��K�M�� =
Dow
nloa
ded
by [
Flor
ida
Atla
ntic
Uni
vers
ity]
at 2
1:43
30
Sept
embe
r 20
13
ON GENERALIZED COHEN–MACAULAY CANONICAL MODULES 4461
⋃n≥0
�0 �H3��K�M�� �
n�� Therefore, we can choose an integer n1 such that ��0 �H3��K�M��
xn11 � > c� By Remark 2.6, we have an exact sequence
0 −→ H2��K�M��/x
n11 H2
��K�M�� −→ H2��K�M�/x
n11 K�M��� −→ �0 �H3
��K�M�� xn11 �−→ 0�
It follows that �(H2
��K�M�/xn11 K�M���
)> c� Set M1 = M/x
n11 M� Then by Lemma 2.5,
we have an isomorphism
H2��K�M�/x
n11 K�M��� � H2
��K�M1���
Therefore, ��H2��K�M1��� > c� So, there exists n2 ∈ � such that ��0 �H2
��K�M1��xn22 � >
c� Since xn22 is a strict f-element with respect to M1 by Lemma 2.2, we have by
Remark 2.6 the exact sequence
0 → H1��K�M1��/x
n22 H1
��K�M1�� → H1��K�M1�/x
n22 K�M1��� → �0 �H2
��K�M1��xn22 � → 0�
It follows that �(H1
��K�M1�/xn22 K�M1���
)> c� Also, we have by Lemma 2.5 an exact
sequence
0 −→ K�M1�/xn22 K�M1� −→ K�M1/x
n22 M1� −→ �0 �Kd−2�M1�
xn22 � −→ 0�
Since d ≥ 4� we have dim�M1/xn22 M1� ≥ 2� Therefore, depth�K�M1/x
n22 M1� ≥ 2� Since
0 �Kd−2�M1�xn22 is of finite length, we get by the above exact sequence that
H1�
(K�M1�/x
n22 K�M1�
) � H0�
(0 �Kd−2�M1�
xn22
) = (0 �Kd−2�M1�
xn22
)�
Hence �(0 �Kd−2�M1�
xn22
)> c and hence Rl
(Hd−2
� �M/xn11 M�
)> c by Lemma 2.7. Thus,
the claim is true for k = 1�Let 2 � k � d − 3� Since �
(Hk+2
� �K�M��)> c, we can choose a positive integer
n1 such that ��0 �Hk+2� �K�M�� x
n11 � > c� By Lemma 2.5, we have an exact sequence
0 → Hk+1� �K�M��/x
n11 Hk+1
� �K�M�� → Hk+1� �K�M�/x
n11 K�M��� → �0 �Hk+2
� �K�M�� xn11 � → 0�
It follows that �(Hk+1
� �K�M�/xn11 K�M���
)> c� Since k ≥ 2� we get by Lemma 2.5 an
isomorphism
Hk+1� �K�M�/x
n11 K�M��� � Hk+1
� �K�M/xn11 M���
Set N = M/xn11 M . Then �
(H
�k−1�+2� �K�N��
)> c, where 1 � k− 1 � dimN − 3. So, we
have by induction for the the integer k− 1 with the module N that
Rl(HdimN−�k−1�−1
� �N/�xn22 � � � � � x
nkk �N�
)> c
for some positive integers n2� � � � � nk� It means that Rl(Hd−k−1
� �M/�xn11 � � � � �
xnkk �M�
)> c, and the lemma is proved. �
Dow
nloa
ded
by [
Flor
ida
Atla
ntic
Uni
vers
ity]
at 2
1:43
30
Sept
embe
r 20
13
4462 NGUYEN AND LE
Proof of Main Theorem. The assertion (i)⇒(ii) follows by Lemma 3.4. The claim(ii)⇒(iii) is trivial. Now we prove (iii)⇒(i). Let �x1� � � � � xd� be a s.o.p. of M whichis a strict f-sequence and c�x�M� a number such that
d−3∑k=1
Rl�Hd−k−1� �M/�x
n11 � � � � � x
nkk �M�� � c�x�M�
for all n1� � � � � nd ∈ �. Suppose that M is not generalized Cohen–Macaulaycanonical. Since K�M� satisfies the condition Serre �S2� and K�M� isequidimensional, it follows by [8, Lemma 3.2.1] that ��H1
��K�M�� < � for alli � 2� Therefore, there exists an integer k such that 1 � k � d − 3 and Hk+2
� �K�M��is of infinite length. It follows that ��Hk+2
� �K�M��� > c�x�M� for some integer1 � k � d − 3. So, there exist by Lemma 3.5 integers n1� � � � � nk such thatRl
(Hd−k−1
� �M/�xn11 � � � � � x
nkk �M�
)> c�x�M�. This gives a contradiction. �
ACKNOWLEDGMENT
The authors are supported by the Vietnam National Foundation for Scienceand Technology Development (Nafosted) under grant numbers 101.01-2011.20 and101.01-2011.49.
REFERENCES
[1] Brodmann, M., Sharp, R. Y. (1998). Local Cohomology: An Algebraic Introduction withGeometric Applications. Cambridge University Press.
[2] Cuong, N. T., Nhan, L. T. (2003). On pseudo Cohen–Macaulay and pseudogeneralized Cohen–Macaulay modules. J. Algebra 267:156–177.
[3] Cuong, N. T., Morales, M., Nhan, L. T. (2004). The finiteness of certain sets ofattached prime ideals and the length of generalized fractions. J. Pure Appl. Algebra189(1-3):109–121.
[4] Cuong, N. T., Schenzel, P., Trung, N. V. (1978). Verallgemeinerte Cohen–Macaulaymoduln. Math. Nachr. 85:57–75.
[5] Hochster, M. (1973). Contracted ideals from integral extensions of regular rings.Nagoya Math. J. 51:25–43.
[6] Macdonald, I. G. (1973). Secondary representation of modules over a commutativering. Symposia Mathematica. p. 11.
[7] Nhan, L. T. (2006). A remark on the monomial conjecture and Cohen–Macaulaycanonical modules. Proc. Amer. Math. Soc. 134:2785–2794.
[8] Schenzel, P. (1982). Dualisierende Komplexe in der lokalen Algebra und BuchsbaumRinge. Lecture Notes in Math., Vol. 907. Berlin- Heidelberg-New York: Springer-Verlag.
[9] Schenzel, P. (2004). On birational Macaulayfications and Cohen–Macaulay canonicalmodules. J. Algebra 275:751–770.
[10] Sharp, R. Y. (1975). Some results on the vanishing of local cohomology modules. Proc.London Math. Soc. 30:177–195.
[11] Sharp, R. Y., Hamieh, M. A. (1985). Lengths of certain generalized fractions. J. PureAppl. Algebra 38:323–336.
Dow
nloa
ded
by [
Flor
ida
Atla
ntic
Uni
vers
ity]
at 2
1:43
30
Sept
embe
r 20
13