Algebra Colloquium 22 : 4 (2015) 655–670 Algebra

DOI: 10.1142/S1005386715000553 Colloquium

c 2015 AMSS CAS
°
& SUZHOU UNIV

C3-Modules
Ismail Amin Yasser Ibrahim
Department of Mathematics, Faculty of Science
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

Cairo University, Giza, Egypt

E-mail: ismail2us@[Link] yfibrahim@[Link]
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

Mohamed Yousif †
Department of Mathematics, The Ohio State University
Lima, Ohio 45804, USA
E-mail: yousif.1@[Link]

Received 3 January 2012

Revised 13 July 2012
Communicated by Nanqing Ding

Abstract. One of the continuity conditions identified by Utumi on self-injective rings is

the C3-condition, where a module M is called a C3-module if whenever A and B are direct
summands of M and A ∩ B = 0, then A ⊕ B is a summand of M . In addition to injective
and direct-injective modules, the class of C3-modules includes the semisimple, continuous,
indecomposable and regular modules. Indeed, every commutative ring is a C3-ring. In
this paper we provide a general and unified treatment of the above mentioned classes of
modules in terms of the C3-condition, and establish new characterizations of several well
known classes of rings.
2010 Mathematics Subject Classification: primary 16D40, 16D50, 16D60; secondary
16L30, 16L60, 16P20, 16P40, 16P60
Keywords: injective and quasi-injective modules, C2-modules and C3-modules

1 Introduction
In his work on continuous rings, Utumi [23] identified three conditions on a ring
that are satisfied if the ring is self-injective. These conditions were extended to
modules by Jeremy [11], Takeuchi [22], and Mohamed and Bouhy [15].
Definition 1.1. A module M is called a C1-module if every submodule is essential
in a direct summand of M . M is called a C2-module if whenever A and B are
submodules of M with A ∼= B and B ⊆⊕ M , then A ⊆⊕ M . M is called a C3-
module if whenever A and B are submodules of M with A ⊆⊕ M , B ⊆⊕ M and
A ∩ B = 0, then A ⊕ B ⊆⊕ M .

†
Corresponding author.
 656 I. Amin, Y. Ibrahim, M. Yousif

A module M is called quasi-injective if any R-homomorphism from a submodule

of M into M can be extended to an R-homomorphism from M into M . It is well
known that every quasi-injective module is a Ci-module for 1 ≤ i ≤ 3, every
C2-module is a C3-module; and there are examples of C1-modules that are not
C3-modules, and of C2-modules that are not C1-modules (see [16, Proposition
2.1]). Also, examples of C3-modules that are not C2-modules are provided in
[16, Proposition 2.1]. C1-modules are also called CS-modules (complements are
summands), and C2-modules are called direct-injective by Nicholson in [17].
Our interest in this paper is in C3-modules. In addition to quasi-injective and
direct-injective modules, examples of C3-modules include continuous and quasi-
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

continuous modules, uniform modules, indecomposable modules, semisimple mod-

ules, regular modules, and SSP -modules (the sum of two summands is again a
summand). It is not difficult to see that every abelian ring is a C3-ring, where a
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

ring is called abelian if its idempotents commute. In this paper we provide a general
and unified treatment of the above mentioned classes of modules in terms of the
C3-condition, and establish new characterizations of several well known classes of
rings in terms of C3-modules.
While C3-modules need not be direct-injective, we show in Proposition 2.3 that
if M = A1 ⊕ A2 is a C3-module, then there is a relative direct-injectivity between
A1 and A2 . More precisely, if M = A1 ⊕ A2 is a C3-module and f : A1 → A2 is
an R-homomorphism with ker f ⊆⊕ A1 , then Imf ⊆⊕ A2 . This result is key to our
work in this paper and is used to characterize many well known classes of rings in
terms of C3-modules. For example, we show in Proposition 2.12 that a ring R is
semisimple if and only if the direct sum of two C3-modules is again a C3-module. In
Theorem 2.13, it is shown that a ring R is regular if and only if every principal right
ideal of M2 (R) generated by a diagonal matrix is a C3-module; equivalently, if every
2-generated submodule of a projective right R-module is a C3-module. Following
ideas of Enochs [7], the notion of C3-cover is introduced, and a ring R is shown to be
right noetherian if and only if every direct sum of injective right R-modules has a C3-
cover; and R is semisimple if and only if every right R-module has a C3-cover. Both
of these results extend the work of Enochs in [7] and are established in Theorems
2.18 and 2.19, respectively. We also provide new characterizations of hereditary,
semiartinian and artinian rings in terms of C3-modules. Following Chen and Shen
[21], a ring R is a right countably Σ-C2-ring if every countable direct sum of copies
of RR is a C2-module, and in this paper R will be called a right countably Σ-C3-
ring if every countable direct sum of copies of RR is a C3-module; or equivalently,
if every projective right R-module is a C3-module. The class of right countably
Σ-C3-rings was investigated and characterized by Bass in his pioneering paper [5],
and recently by Chen and Shen in [21]. It is shown in [5] and [21] that a ring R
is a right countably Σ-C3-ring if and only if R is a right perfect and strongly right
C3-ring, where a ring R is called a strongly right C3-ring if every finite direct sum
of copies of RR is a C3-module. In Theorem 2.28, we provide a slightly different
proof of this result as well as several new characterizations of the class of right
countably Σ-C3-rings. If we replace “projective” by “quasi-projective”, Theorem
2.29 asserts that every quasi-projective right R-module is a C3-module if and only
 C3-Modules 657

if every factor ring of R is a right countably Σ-C3-ring.

As an application, at the end of our paper, we deal with the open question of
whether right perfect right (or left) self-injective rings are quasi-Frobenius. While
right perfect left self-injective rings are known to be right pseudo-Frobenius, no
major results are known about right perfect right self-injective rings. However, if R
is a right perfect right or left self-injective ring, then J(R) = Z(RR ), and in Theorem
2.30 we provide several characterizations of right perfect rings R with J(R) =
Z(RR ), and show that such rings are indeed right countably Σ-C3. As an immediate
consequence of our work, a ring R is a right perfect right self-injective ring if and
only if R is right countably Σ-C3 and E(RR ) is projective, and R is quasi-Frobenius
if and only if R is right countably Σ-C3 and E(RR ) is countably Π-projective, where
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

a right R-module M is called countably Π-projective if a countable direct product

of copies of M is projective. These results are established in corollaries 2.31 and
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

2.32, respectively. Finally, we highlight our results by providing several examples of

right countably Σ-C3-rings that are neither left countably Σ-C3 nor left (or right)
self-injective.
Throughout this paper every ring R is associative with unity and all modules
are unitary. If M is an R-module, we denote by rad(M ), soc(M ), Z(M ) and E(M )
for the Jacobson radical, the socle, the singular submodule and the injective hull
of M , respectively. If M = R, we write J = J(R) = rad(R). If MR is a right R-
module, we denote by R M ∗ the dual left R-module Hom(MR , RR ) of M . We also
write N ⊆ M if N is a submodule of M , N ⊆ess M if N is an essential submodule
small
of M , N ,→ M if N is a small submodule of M , and N ⊆⊕ M if N is a direct
summand of M . We denote by S = End(MR ) the endomorphism ring of M , Mn (R)
the n × n matrix ring over R, M (I) the direct sum of I-copies of M , M I the direct
product of I-copies of M , and Mod-R the category of right R-modules. If M is a
right R-module, we denote by AnnR (M ) =: {r ∈ R : M r = 0} the annihilator ideal
of M in R, and if X is a subset of R, the right annihilator of X in R is denoted by
r(X) =: {r ∈ R : xr = 0 for all x ∈ X}; and we say that R has the ACC on right
annihilators if every ascending chain of these right ideals is eventually constant.
With similar terminology for left annihilators. We refer to [3], [4], [13] and [25] for
all the undefined notions in this paper.

2 The Results
We start with new characterizations of C3-modules. But first we need the following
lemma which will be used throughout the paper.
Lemma 2.1. [16, Proposition 2.7] A summand of a Ci-module is again a Ci-module
for 1 ≤ i ≤ 3.

Proposition 2.2. The following conditions on a right R-module M are equivalent:

(1) M is a C3-module.
(2) If A ⊆⊕ M , B ⊆⊕ M and A ∩ B = 0, then M = A1 ⊕ B = A ⊕ B1 for
submodules A1 ⊇ A and B1 ⊇ B.
(3) If A ⊆⊕ M , B ⊆⊕ M and A ∩ B ⊆⊕ M , then A + B ⊆⊕ M .
 658 I. Amin, Y. Ibrahim, M. Yousif

Proof. (1)⇒(2) Let A ⊆⊕ M , B ⊆⊕ M and A ∩ B = 0. Since A ⊕ B ⊆⊕ M ,

M = A ⊕ B ⊕ T for a submodule T ⊆ M . If we choose A1 = A ⊕ T and B1 = B ⊕ T ,
then clearly M = A1 ⊕ B = A ⊕ B1 , as required.
(2)⇒(1) Let A ⊆⊕ M , B ⊆⊕ M and A ∩ B = 0. We need to show that
A ⊕ B ⊆⊕ M . By the hypothesis, M = A1 ⊕ B = A ⊕ B1 for submodules A1 ⊇ A
and B1 ⊇ B. Now B1 = B1 ∩ M = B1 ∩ (A1 ⊕ B) = B ⊕ (A1 ∩ B1 ), and so
M = A ⊕ B1 = A ⊕ B ⊕ (A1 ∩ B1 ), as required.
(1)⇒(3) Since A∩B ⊆⊕ M , M = (A∩B)⊕K for a submodule K ⊆ M . Clearly,
A = (A ∩ B) ⊕ (A ∩ K) and B = (A ∩ B) ⊕ (B ∩ K). Since M is a C3-module,
(A ∩ K) ⊆⊕ M , (B ∩ K) ⊆⊕ M and (A ∩ K) ∩ (B ∩ K) = (A ∩ B) ∩ K = 0, we
infer that T =: (A ∩ K) ⊕ (B ∩ K) ⊆⊕ M . Again, since T ⊆⊕ M , A ∩ B ⊆⊕ M and
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

(A ∩ B) ∩ T = 0, we have (A ∩ B) ⊕ T ⊆⊕ M . Now A + B = [(A ∩ B) ⊕ (A ∩ K)]

+[(A ∩ B) ⊕ (B ∩ K)] = (A ∩ B) ⊕ (A ∩ K) ⊕ (B ∩ K) = (A ∩ B) ⊕ T ⊆⊕ M , as
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

required.
(3)⇒(1) Clear. ¤

While C3-modules, as noted above, need not be direct-injective, in the next

proposition we show that if A1 ⊕ A2 is a C3-module, then there is a relative direct-
injectivity between A1 and A2 .
Proposition 2.3. If M is a C3-module, M = A1 ⊕ A2 for submodules A1 and A2 ,
and f : A1 → A2 is an R-homomorphism with ker f ⊆⊕ A1 , then Imf ⊆⊕ A2 .
Proof. We first show that if f : A1 → A2 is an R-monomorphism, then Imf ⊆⊕
A2 . Let T = {a + f (a) : a ∈ A1 } be the graph submodule of M . We claim that
M = T ⊕ A2 . For, if x ∈ M , then x = a + b for some a ∈ A1 and b ∈ A2 . Now
x = a+f (a)−f (a)+b ∈ T +A2 , and so M = T +A2 . If x ∈ T ∩A2 , then x = a+f (a)
for some a ∈ A1 , and hence a = x−f (a) ∈ A1 ∩A2 = 0. Clearly, x = 0, M = T ⊕A2
and T ⊆⊕ M . Next, we show that A1 ∩ T = 0. For, if x ∈ A1 ∩ T , then x = a + f (a)
for some a ∈ A1 , and consequently, x − a = f (a) ∈ A1 ∩ A2 = 0. Now since f is a
monomorphism, a = 0, and hence x = 0. Since M is a C3-module, A1 ⊕ T ⊆⊕ M .
Finally, we show that A1 ⊕ T = A1 ⊕ Imf . For, if x ∈ Imf , then x = f (a) for
some a ∈ A1 , and so x = −a + a + f (a) ∈ A1 + T , and hence A1 ⊕ T = A1 ⊕ Imf .
Since A1 ⊕ T ⊆⊕ M , Imf ⊆⊕ M , and so Imf ⊆⊕ A2 . Now let f : A1 → A2 be an
R-homomorphism with ker f ⊆⊕ A1 . If A1 = ker f ⊕ B for a submodule B ⊆ A1 ,
then M = A1 ⊕ A2 = ker f ⊕ B ⊕ A2 , and the restriction map f |B : B → A2 is a
monomorphism. Since a summand of a C3-module is again a C3-module, we infer
from the preceding argument that Imf = Imf |B ⊆⊕ A2 , as required. ¤

Corollary 2.4. If M is a C3-module, M = A1 ⊕ A2 for submodules A1 and A2 ,

and f : A1 → A2 is an R-monomorphism, then Imf ⊆⊕ A2 .

Example 2.5. Since the Z-monomorphism 0 → 2Z → Z does not split, the Z-module
M = Z ⊕ 2Z is not a C3-module, while both Z and 2Z are C3-modules.

The next corollary is an immediate consequence of Proposition 2.3; it was es-

tablished in [21], and is attributed to Zhou.
 C3-Modules 659

Corollary 2.6. If M ⊕ M is a C3-module, then M is a C2-module.

Corollary 2.7. If M is a right R-module, then every finite (countable, arbitrary)

direct sum of copies of M is a C3-module if and only if every finite (countable,
arbitrary) direct sum of copies of M is a C2-module.

Proposition 2.8. Let M be a right R-module, and S = End(MR ). If S is a right

C3-ring, then M is a right C3-module. In particular, if Mn (R) is a right C3-ring,
(n)
then RR is a right C3-module.
Proof. Let A ⊆⊕ M , B ⊆⊕ M and A ∩ B = 0. We need to show that A ⊕ B ⊆⊕ M .
Let e2 = e ∈ S, f 2 = f ∈ S such that A = eM and B = f M . Since A ∩ B = 0, we
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

infer that eS ∩f S = 0, and consequently eS ⊕f S ⊆⊕ S. If S = eS ⊕f S ⊕gS for some

idempotent g ∈ S, then M = eM ⊕f M ⊕gM . Therefore, A⊕B = eM ⊕f M ⊆⊕ M ,
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

as required. The last assertion follows from the fact that End(RR ) ∼
(n)
= Mn (R). ¤

Remark 2.9. Being a right C3-ring is not a Morita invariant. Otherwise, M2 (R) is
a right C3-ring whenever R is a right C3-ring, and by Proposition 2.8, R ⊕ R is a
right C3-module. Consequently, by Corollary 2.6, R is a right C2-ring. This is a
contradiction, since Z is an example of a C3-ring that is not a C2-ring.

The next result can be established by combining Corollary 2.7, Proposition 2.8
and [18, Theorem 7.15].
Proposition 2.10. The following conditions are equivalent:
(n)
(1) RR is a right C3-module for all n ≥ 1.
(n)
(2) RR is a right C2-module for all n ≥ 1.
(3) Mn (R) is a right C2-ring for all n ≥ 1.
(4) Mn (R) is a right C3-ring for all n ≥ 1.

It is not difficult to prove that a module M is direct-injective if and only if

given any summand A ⊆⊕ M and any monomorphism f : A → M , there exists a
homomorphism g : M → A such that g ◦ f = i, where i : A → A is the identity map.
Clearly, direct-injective modules are preserved under Morita equivalence. While we
are not certain whether C3-modules are preserved under Morita equivalence or not,
we have the following:

Remark 2.11. Let M be a right R-module, and define condition (∗) on M to be

“If M = A1 ⊕ A2 for some submodules A1 and A2 of M , f : A1 → A2 is an
R-monomorphism and i : A1 → A1 is the identity map, then there exists an R-
homomorphism g : A2 → A1 such that g ◦ f = i.” By Corollary 2.4, every C3-
module satisfies condition (∗). Now since Morita equivalence preserves summands,
monomorphisms, and isomorphisms, it follows that if R and S are Morita equivalent
rings with category equivalence Γ : Mod-R → Mod-S, then a right R-module MR
satisfies condition (∗) if and only if Γ(M )S satisfies condition (∗).

Recall that a ring R is called a right V -ring if every simple right R-module is
injective, equivalently, if rad(M ) = 0 for every right R-module M . In the next
 660 I. Amin, Y. Ibrahim, M. Yousif

proposition we show that the class of C3-modules need not be closed under direct
sums.

Proposition 2.12. The following conditions are equivalent:

(1) R is a semisimple ring.
(2) The direct sum of any two C3-modules is a C3-module.
(3) Every submodule of a projective R-module is a C3-module.
(4) Every submodule of R ⊕ R is a C3-module.
(5) Every submodule of an injective R-module is a C3-module.
Proof. (1)⇒(2), (1)⇒(3)⇒(4) and (1)⇒(5) Clear.
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

(2)⇒(1) Assume that the direct sum of any two C3-modules is a C3-module,
and let K be either a semisimple or indecomposable right R-module. We will show
that in both cases, K is injective. Clearly, K ⊕ E(K) is a C3-module, and if
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

i : K → E(K) is an injection map, then by Corollary 2.4, K ' i(K) ⊆⊕ E(K),

and K is injective. Since every semisimple right R-module is injective, R is a right
noetherian right V -ring (see for example [25]). Now write R = K1 ⊕ K2 ⊕ · · · ⊕ Kn ,
a direct sum of indecomposable right ideals. Since each Ki is injective, it follows
that R is right self-injective, in particular, R/J(R) is a regular ring. But, since R
is a right V -ring, J(R) = 0, and consequently, R is a regular right noetherian ring,
and hence semisimple.
(4)⇒(1) If I is a right ideal of R, then I ⊕ R is a C3-module. By Corollary 2.4,
if i : I → R is the inclusion map, then I ⊆⊕ R, and R is semisimple.
(5)⇒(1) If M is a right R-module, then M ⊕ E(M ), being a submodule of
E(M ) ⊕ E(M ), is a C3-module. Therefore, the canonical monomorphism 0 →
M → E(M ) splits; and so M is injective and R is semisimple. ¤

Theorem 2.13. The following statements are equivalent:

(1) R is a regular ring.
(2) Every principal right ideal of M2 (R) is a C3-module.
(3) Every principal right ideal of M2 (R) generated by a diagonal matrix is a
C3-module.
(4) Every finitely generated submodule of a projective right R-module is a direct
summand.
(5) Every finitely generated submodule of a projective right R-module is a C3-
module.
(6) Every 2-generated submodule of a projective right R-module is a C3-module.
Proof. (1)⇒(2) Since being regular is a Morita invariant property, S = M2 (R) is
a regular ring, and consequently, every principal right ideal of S is a summand of
S. Since every regular ring is a left and right C3-module, and a summand of a
C3-module is again a C3-module, we infer that every principal right ideal of S is a
C3-module.
(2)⇒(3) Clear.
(3)⇒(1) Let S = hM2 (R),
i a ∈ R and I be a principal right ideal of S generated
h by i
a 0 1 0
the diagonal matrix 0 1 . As a right S-module, I is a C3-module. If e = 0 0 ,
 C3-Modules 661

then S and R are Morita equivalent, via M → M e, where M is a right S-module.

Now since Ie ' aR ⊕ R as right R-modules and I satisfies condition (∗), by Remark
2.11, aR ⊕ R satisfies condition (∗) as a right R-module, and the inclusion map
i : aR → R splits. Thus, aR is a summand of R, and R is a regular ring.
(1)⇔(4) This is a well known result of Kaplansky [12, Lemma 4].
(4)⇒(5)⇒(6) Clear.
(6)⇒(1) If a ∈ R, then aR⊕R is a C3-module as a right R-module. By Corollary
2.4, if i : aR → R is the inclusion map, then aR ⊆⊕ R, and R is a regular ring. ¤

A right R-module M is called Π-quasi-injective if every direct product M I of

copies of M is quasi-injective. A ring R is called right hereditary if every submodule
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

of a projective right R-module is projective, equivalently, if every factor module of

an injective right R-module is injective.
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

Theorem 2.14. The following statements are equivalent:

(1) R is right hereditary.
(2) Every factor module of an injective right R-module is a C3-module.
Proof. (1)⇒(2) Clear.
(2)⇒(1) Let M be an injective right R-module and K a submodule of M . We
need to show that M/K is injective. But, E(M/K) ⊕ M/K, being a factor mod-
ule of E(M/K) ⊕ M , is a C3-module. Now by Corollary 2.4, the embedding R-
monomorphism M/K ,→ E(M/K) splits; and M/K is injective, as required. ¤

Proposition 2.15. The following statements are equivalent:

(1) Every factor ring of R is right hereditary.
(2) Every factor module of a Π-quasi-injective right R-module is quasi-injective.
(3) Every factor module of a Π-quasi-injective right R-module is a C3-module.
Proof. (1)⇒(2) By [26, Theorem 6].
(2)⇒(3) Clear.
(3)⇒(1) Let I be an ideal of R and E an injective right R/I-module. By Theo-
rem 2.14, we need to show that every factor module E/N , as a right R/I-module,
is a C3-module. Since I ⊆ AnnR (E), E is injective as a right R/AnnR (E)-module,
where AnnR (E) is the annihilator ideal of E in R. Therefore, by [8, Theorem 1.2],
E is Π-quasi-injective as a right R-module. Now it follows from the hypothesis that
E/N is a C3-module as a right R-module, and hence as a right R/I-module. ¤

A module M is called (countably) Σ-(quasi-)injective if every (countable) direct

sum of copies of M is (quasi-)injective. Analogously, M is called a (countably)
Σ-C3-module if every (countable) direct sum of copies of M is a C3-module. A
result of Faith and Walker [9] asserts that a ring R is right noetherian if and only if
every injective right R-module is Σ-injective, equivalently, as shown by Fuller [10],
if every quasi-injective right R-module is Σ-quasi-injective.

Theorem 2.16. The following conditions are equivalent:

(1) R is right noetherian.
(2) Every injective right R-module is Σ-injective.
 662 I. Amin, Y. Ibrahim, M. Yousif

(3) Every quasi-injective right R-module is Σ-quasi-injective.

(4) Every direct sum of injective right R-modules is a C3-module.
(5) Every countable direct sum of injective right R-modules is a C3-module.
(6) Every injective right R-module is a countably Σ-C3-module.
(7) Every quasi-injective right R-module is a countably Σ-C3-module.
Proof. (1)⇔(2)⇔(3) It follows from [9] and [10].
(1)⇒(4)⇒(5) It follows from [13, Theorem 6.5.1] since every injective module is
a C3-module. L
(5)⇒(1) If M = i∈I Mi is a countable direct sum of injective right R-modules,
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

then M ⊕ E(M ) is a C3-module. By Corollary 2.4, if i : M → E(M ) is an injection

map, then M ∼ = i(M ) is a summand of E(M ). Thus, M ∼ = E(M ) is injective and
R is right noetherian by [13, Theorem 6.5.1].
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

(7)⇒(6) Clear. L
(6)⇒(5) Let M = Q i∈I Mi be a countable direct sum of injective right R-
modules, and set K =: i∈I Mi . Since K is injective, we infer from the hypothesis
that K (I) is a C3-module. As Mi is injective,
¡L ¢ ¡ L K = ¢Mi ⊕ Ti for a submodule Ti ⊆ K
for i ∈ I. Now since K (I) = i∈I Mi ⊕ L i∈I Ti is a C3-module, and summands
of C3-modules are again C3, we infer that i∈I Mi is a C3-module.
(1)⇒(7) By [8]. ¤

In [7], Enochs introduced the notion of injective cover as the dual to the injective
envelope, and proved that a ring R is right noetherian if and only if every right R-
module has an injective cover. In what follows, we adopt the work of Enochs and
introduce the notion of C3-covers. We will show that a ring R is semisimple if and
only if every right R-module has a C3-cover, and R is right noetherian if and only
if every direct sum of injective right R-modules has a C3-cover.

Definition 2.17. An R-homomorphism g : E → M is called a C3-cover of a right

R-module M if E is a C3-module such that any diagram

E
∃α% ↓g
0 θ
E −→ M

with E 0 a C3-module can be completed; and the diagram

E
∃α% ↓g
g
E −→ M

can be completed only by an automorphism α.

Theorem 2.18. The following conditions are equivalent:

(1) R is right noetherian.
(2) Every direct sum of injective right R-modules has a C3-cover.
 C3-Modules 663

Proof. (1)⇒(2) This is clear since over a right noetherian ring, every direct sum of
injective right R-modules
L is injective.
(2)⇒(1) Let M = i∈I Mi be an arbitrary direct sum of injective right R-
modules and g : E → M a C3-cover of M . Consider the following diagram:

E
∃αi % ↓g
πi
Mi −→ M

where πi is the canonical injection for i ∈ I. By Definition 2.17, there exists a

homomorphism
¡ Pn ¢ αi P
: Mi → E such that gαi = πi for i ∈ I. Define α : M → E
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

n
by α m i ¡ P i=1 α
= ¢i (mP
i ) for mi ∈ Mi and i ∈ I. Clearly, α is well defined
i=1
n n Pn Pn
and we have gα i=1 m i = i=1 gαi (mi ) = i=1 πi (mi ) = i=1 mi . Therefore,
gα = 1M , and α : M → E is a split monomorphism. Thus, M ∼ = K ⊆⊕ E, and by
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

Lemma 2.1, M is a C3-module. Now by Theorem 2.16, R is right noetherian. ¤

Theorem 2.19. The following conditions are equivalent:

(1) R is semisimple.
(2) Every right R-module has a C3-cover.
Proof. (1)⇒(2) Clear.
(2)⇒(1) With an argument similar to L
(2)⇒(1) in Theorem 2.18 above, we can
show that an arbitrary direct sum M = i∈I Mi of right C3-modules is again a
C3-module. Now by Proposition 2.12(2), R is semisimple. ¤

Definition 2.20. [2] Let M and N be right R-modules. M is called socle-N -inje-
ctive (soc-N -injective) if any R-homomorphism f : soc(N ) → M extends to N . M
is called strongly soc-injective if M is soc-N -injective for all right R-modules N .

Lemma 2.21. [2, Proposition 16] For a right R-module M , the following conditions
are equivalent:
(1) M is strongly soc-injective.
(2) M is soc-E(M )-injective.
(3) M = E ⊕ T , where E is injective and T has zero socle. Moreover, if M has
non-zero socle then E has essential socle.

A ring R is called right semiartinian if soc(M ) 6= 0 for every non-zero right

R-module M .
Proposition 2.22. The following are equivalent:
(1) R is right semiartinian.
(2) Every strongly soc-injective right R-module is injective.
(3) Every strongly soc-injective right R-module is a C3-module.
Proof. (1)⇒(2) It follows from Lemma 2.21 above.
(2)⇒(3) Clear.
(3)⇒(2) If M is a strongly soc-injective right R-module, then by Lemma 2.21,
M = N ⊕T , where N is injective and soc(T ) = 0. Since M ⊕E(M ) = N ⊕T ⊕E(M )
 664 I. Amin, Y. Ibrahim, M. Yousif

= (N ⊕ E(M )) ⊕ T , it follows from Lemma 2.21 that M ⊕ E(M ) is strongly soc-

injective; and by the hypothesis, M ⊕ E(M ) is a C3-module. Consequently, by
Corollary 2.4, if i : M → E(M ) is an injection map, then M ' i(M ) ⊆⊕ E(M ),
and so M is injective.
(2)⇒(1) Let M be an arbitrary non-zero right R-module. We claim that soc(M )
6= 0. Otherwise, if soc(M ) = 0 and N is an arbitrary non-zero submodule of M ,
then N is strongly soc-injective, and hence injective. Therefore, N ⊆⊕ M , and M
is semisimple, a contradiction. Thus, soc(M ) 6= 0, and R is right semiartinian. ¤

Definition 2.23. [1] Let M and N be right R-modules. M is called simple-N -

injective if for every submodule L of N , every R-homomorphism γ : L → M with
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

γ(L) simple extends to N . M is called strongly simple-injective if M is simple-N -

injective for all right R-modules N .
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

Lemma 2.24.
L
(1) If {Mi : i ∈ I} is a family of right R-modules, then the direct sum i∈I Mi
is strongly simple-injective if and only if each Mi is strongly simple-injective.
L∞
(2) A ring R is right noetherian if and only if i=1 E(Ki ) is injective, whenever
{Ki : i ∈ I} is a (countable) family of simple right R-modules.
Proof. For the proof of (1), we refer to [1, Proposition 1.4]. The proof of (2) is a
well known result on noetherian rings (see for example [13, 6.5.1]). ¤

Proposition 2.25. The following are equivalent:

(1) R is right artinian.
(2) Every strongly simple-injective right R-module is injective.
(3) Every strongly simple-injective right R-module is a C3-module.
Proof. (2)⇒(3) Clear.
(3)⇒(1) With an argument similar to the one provided in (3)⇒(2) of Proposition
2.22, we can show that R is right semiartinian. Since right semiartinian right
noetherian rings are right artinian, it suffices to show that R is right noetherian.
By Lemma 2.24, we need to show that if {Ki : i ∈ I}L is a countable family of simple
∞
right R-modules, then the right R-module M = i=1 E(Ki ) is injective. Now
M ⊕ E(M ) is a strongly simple-injective module, and hence a C3-module. By
Corollary 2.4, if i : M → E(M ) is an injection map, then M ' i(M ) ⊆⊕ E(M ),
and M is injective.
(1)⇒(2) We first show that if M is a strongly simple-injective right R-module
and S is a simple submodule of M , then S is essential in an injective summand of
M . To see this, consider the following diagram:
inc
0 −→ S −→ E(S)
id ↓ .σ
M

Since S ⊆ess E(S), there exists an embedding σ of E(S) in M such that σ(x)
L =x
for all x ∈ S. If F = σ(E(S)), then S ⊆ess F ⊆⊕ M . Now write soc(M ) = i∈I Si ,
 C3-Modules 665

where each Si is a simple submodule of M , and let Si ⊆ess Ei ⊆⊕ ML, where each
Ei is an injective summand of M . Since R is right noetherian, F = i∈I Ei is an
injective summand of M , and M = F ⊕ T for a submodule T ⊆ M with soc(T ) = 0.
Since R is right artinian, T = 0, and M is injective, as required. ¤

In [18, page 172], a ring R was called a strongly right C2-ring if every finite
direct sum of copies of the right R-module RR is a C2-module. In what follows, if
every finite (countable) direct sum of copies of RR is a C3-module, R will be called
a strongly right C3-ring (countably Σ-C3-ring). However, by Corollary 2.7, R is a
strongly right C2-ring if and only if R is a strongly right C3-ring.
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

Theorem 2.26. The following statements are equivalent:

(1) Every finitely generated projective right R-module is a C3-module.
(2) Every finitely generated projective submodule of a projective right R-module
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

is a direct summand.
(3) R is a strongly right C3-ring.
(4) r(I) 6= 0 for every finitely generated proper left ideal I of R.
(5) (P/K)∗ 6= 0, whenever P is a finitely generated projective left R-module and
K is a finitely generated proper submodule.
Proof. (1)⇔(3) Clear.
(3)⇔(4) By Corollary 2.7 and [6, Lemma 2.3].
(2)⇔(4)⇔(5) By [5, Theorem 5.4]. ¤

A result of Faith and Walker [9] asserts that a ring R is quasi-Frobenius if and
only if every projective right R-module is injective; equivalently, if every projective
right R-module is quasi-continuous (see for example Jeremy [11]). Subsequently,
Oshiro [19] considered the rings R, where every projective right R-module is a C1-
module (CS-module), and showed that such rings are left and right artinian, and if
J(R) = Z(RR ), then R is quasi-Frobenius. In what follows, we consider the rings
R where every projective right R-module is a C3-module (C2-module).
According to Chen and Shen [21, Theorem 2.13], a ring R is a right countably Σ-
C2-ring if every countable direct sum of copies of RR is a C2-module. By Corollary
2.7, R is a countably Σ-C2-ring if and only if R is a countably Σ-C3-ring. The
next theorem is essentially due to Bass [5, Theorem 5.4, Lemma 6.1, Theorem 6.3]
and Chen and Shen [21, Theorem 2.13]. For completness we include a slightly short
proof, but first we need the following lemma.

Lemma 2.27. [24, Lemma 2.2] If M is a projective module and N ⊆ M , then

M/N is flat if and only if given x ∈ N , there exists γ : M → N such that γ(x) = x.

Theorem 2.28. The following statements are equivalent:

(1) Every flat right R-module is a C3-module.
(2) Every flat submodule of a flat right R-module is a direct summand.
(3) Every projective right R-module is a C3-module.
(4) Every projective submodule of a projective right R-module is a direct sum-
mand.
(5) R is a right countably Σ-C3-ring.
 666 I. Amin, Y. Ibrahim, M. Yousif

(6) R is a right perfect, strongly right C3-ring.

Proof. (1)⇒(3)⇒(5) and (2)⇒(4) Clear.
(3)⇒(4) Let P be a projective right R-module and Q a projective submodule
of P . Then P ⊕ Q, being projective, is a C3-module. Now if i : Q → P is the
canonical injection, then it follows from Corollary 2.4 that Q ⊆⊕ P .
(4)⇒(3) Let A ∼ = B ⊆⊕ P , where P is projective, A ⊆ P and B ⊆ P . Then A
is projective, and hence a direct summand of P . Therefore, P is a C2-module, and
hence a C3-module.
(5)⇒(6) We only need to show that R is a right perfect ring. By [3, Theorem
28.4], we need to show that R satisfies DCC on principal right ideals of the form
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

Ra1 ⊇ Ra2 a1 ⊇ Ra3 a2 a1 ⊇ · · · . Let F be a free right R-module with basis

x1 , x2 , . . . , xi , . . . (i ∈ N), and G the submodule of F spanned by yi = xi − xi+1 ai
(i ∈ N). By [3, Lemma 28.1], G is free with basis y1 , y2 , . . . , yi , . . . (i ∈ N). Thus,
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

G ⊕ F , being isomorphic to a countable direct sum of copies of R, is a right C3-

module. If σ : G → F is the inclusion map, then it follows from Corollary 2.4 that
G ⊆⊕ F . Then by [3, Lemma 28.2], the chain Ra1 ⊇ Ra2 a1 ⊇ Ra3 a2 a1 ⊇ · · ·
terminates, as required.
(6)⇒(1) Since R is right perfect, every flat right R-module is projective, and
consequently we only need to show that every infinite direct sum of copies of R,
(A) (A)
denoted by RR , is a C2-module, and hence a C3-module. Let F = RR and K
⊕
be a projective submodule of F . We need to show that K ⊆ F . Again, since
R is right perfect, we only need to show that F/K is flat. By Lemma 2.27, it is
enough to prove that given x ∈ K, there exists γ : F → K such that γ(x) = x. As
R is semiperfect, R has a basic set of primitive idempotents {e1 , e2 , . . . , en } such
that {e1 R, e2 R, . . . , en R} is a complete set of representatives of the indecomposable
projective right R-modules, and R = (e1 R)(m1 ) ⊕ (e2 R)(m2 ) ⊕ · · · ⊕ (en R)(mn ) .
σ
Since R is right perfect and K is projective, K = ∼ (e1 R)(A1 ) ⊕ (e2 R)(A2 ) ⊕ · · · ⊕
(en R)(An ) ⊆⊕ F for some, possibly infinite, index sets Ai (1 ≤ i ≤ n). Now
σ(x) ∈ T =: (e1 R)(B1 ) ⊕ (e2 R)(B2 ) ⊕ · · · ⊕ (en R)(Bn ) for some finite index sets Bi
(1 ≤ i ≤ n). Inasmuch as R is a strongly right C2-ring and x ∈ σ −1 (T ) ∼
(I)
= T ⊆⊕ RR ,
where I is finite, we infer that x ∈ σ −1 (T ) ⊆⊕ F . If we define γ = πσ−1 (T ) : F → K
to be the projection map of F onto σ −1 (T ), then γ(x) = x as required.
(6)⇒(2) This is clear since R is right perfect and (6)⇒(4) from above. ¤

Theorem 2.29. The following statements are equivalent:

(1) Every quasi-projective right R-module is a C3-module.
(2) Every factor ring of R is a right countably Σ-C3-ring.
Proof. (1)⇒(2) Since every projective right R-module is a C3-module, by Theorem
2.28, R is a right perfect ring. If I is an ideal of R, then by Theorem 2.28, we
only need to show that every projective right R/I-module is a C3-module. But
if M is a projective right R/I-module, then M is projective as an R/AnnR (M )-
module. By [10, Theorem 3.3], MR is quasi-projective as a right R-module, and
hence a C3-module by the hypothesis. But then clearly M is a C3-module as an
R/I-module.
 C3-Modules 667

(2)⇒(1) By Theorem 2.28, R is right perfect. Now if M is a quasi-projective

right R-module, then by [10, Theorem 3.3], M is projective as a right R/AnnR (M )-
module. Another application of Theorem 2.28 ensures that M is a C3-module as a
right R/AnnR (M )-module, and hence a C3-module as a right R-module. ¤

A well known result of Osofsky [20] asserts that every right perfect left and
right self-injective ring is quasi-Frobenius, and it is an open question whether right
perfect one-sided self-injective rings are quasi-Frobenius. In the next few results we
show that such rings are indeed right countably Σ-C3-rings, and provide examples
where the converse fails.
By combining [18, Proposition 7.36, Theorem 7.38, Proposition B.59], [27, Propo-
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

sition 3.3] and Theorem 2.28 above, we obtain the next result.
Theorem 2.30. The following statements are equivalent:
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

(1) R is a right countably Σ-C3-ring, and for every a ∈ R, r(a) ⊆ess eR, where
e2 = e ∈ R.
(2) R is a right countably Σ-C3-ring, and for every a ∈ R, aR = P ⊕ S, where
PR is projective and SR is singular.
(3) R is right perfect with J(R) = Z(RR ).
(4) Every submodule Q of a projective right R-module P has a decomposition
Q = X ⊕ Y , where X is a summand of P and Y ⊆ Z(P ).
Since the equality J(R) = Z(RR ) holds in any right self-injective ring, the next
corollary is now an immediate consequence of Theorems 2.28 and 2.30.
Corollary 2.31. The following conditions are equivalent:
(1) R is a right countably Σ-C3-ring and E(RR ) is projective.
(2) R is right perfect right self-injective ring.
A right R-module M is called countably Π-projective if any countable direct
product of copies of M is projective.
Corollary 2.32. The following conditions are equivalent:
(1) R is a right countably Σ-C3-ring and E(RR ) is countably Π-projective.
(2) R is quasi-Frobenius.
Proof. (2)⇒(1) Clear.
(1)⇒(2) By Corollary 2.31, R is right self-injective, and so the right R-module
(N)
RR is countably Π-projective. By Theorem 2.28(4), RR is a summand of the in-
N
jective right R-module RR , and consequently R is right countably Σ-injective. By
a well known result of Faith [8], R is quasi-Frobenius. ¤

A ring R is called left Kasch if every simple left R-module is embedded in R R;

equivalently, if r(L) 6= 0 for every proper left ideal L of R. By Theorem 2.26,
every left Kasch ring is a strongly right C3-ring, and as noted in [21], there are
examples of right countably Σ-C3-rings, and hence of strongly right C3-rings, that
are neither left nor right Kasch. We will provide, below, more examples of such
rings to highlight the relationship between the Kasch condition, the C3-condition,
and perfect rings.
 668 I. Amin, Y. Ibrahim, M. Yousif

A ring R is called a left mininjective ring if every R-homomorphism from a

minimal left ideal of R into R is given by right multiplication by an element of R.
The next corollary is an immediate consequence of [18, Theorem 2.31 and Theorem
3.12] and Theorem 2.28.
Corollary 2.33. The following conditions on a ring R are equivalent:
(1) R is a right perfect left mininjective ring.
(2) R is a right countably Σ-C3-ring and r(L) is simple for every maximal left
ideal L of R.
(3) R is a right countably Σ-C3-ring and the dual of every simple left R-module
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

is simple.
If R satisfies any of the above equivalent conditions, then R is left and right Kasch.
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

Proposition 2.34. Let R be a semiperfect ring with soc(R R) ⊆ess R R and J(R) ⊆
Z(R R). Then R is a right Kasch ring. In particular, R is a strongly left C3-ring.
If in addition either R or R/soc(R R) has ACC on left annihilators, then R is a left
countably Σ-C3-ring.
Proof. Since R is semiperfect, every right ideal T of R can be written as T = eR⊕K,
where e2 = e ∈ R and K ⊆ J(R). In particular, Z(R R) ⊆ J(R), and from
the hypotheses it follows that J(R) = Z(R R). Moreover, since R is semilocal,
soc(R R) ⊆ l(Z(R R)) = l(J(R)) = soc(RR ), and consequently soc(RR ) ⊆ess R R.
Now l(T ) = l(eR) ∩ l(K) = R(1 − e) ∩ l(K) ⊇ R(1 − e) ∩ l(J) = R(1 − e) ∩ soc(RR )
6= 0. Therefore, R is right Kasch, and by Theorem 2.26, R is a strongly left C3-ring.
If we assume in addition that R has ACC on left annihilators, then by a well
known result in [14], Z(R R) is nilpotent, and therefore R is semiprimary. Since R
is a strongly left C3-ring, it follows from Theorem 2.28 that R is a left countably
Σ-C3-ring. Now suppose R = R/soc(R R) has ACC on left annihilators. We will
show that J is left T -nilpotent. To do this, let a1 , a2 , . . . be a sequence of elements
from J(R) = Z(R R). Since soc(R R) ⊆ l(Z(R R)), each ai ∈ r(soc(R R)), and by
[18, Lemma 3.32], we infer that l(a1 a2 · · · an+1 ) = l(a1 a2 · · · an ) for some integer
n ≥ 1. We claim that a1 a2 · · · an = 0. Otherwise, since an+1 ∈ Z(R R), we infer
that l(an+1 ) ∩ Ra1 a2 · · · an 6= 0, and so there exists 0 6= x = ra1 a2 · · · an such that
ra1 a2 · · · an an+1 = 0 for some r ∈ R. But then r ∈ l(a1 a2 · · · an ), and x = 0; a
contradiction. This proves that J(R) is left T -nilpotent. Since R is semiperfect, it
follows that R is left perfect, and hence R is a left countably Σ-C3-ring. ¤
Example 2.35. Let F be a field and R be the ring of all lower triangular, countably
infinite square matrices over F with only finitely many off-diagonal non-zero entries.
Let S be the F -subalgebra of R generated by 1 and J(R). Then S is a left perfect,
left mininjective ring which is neither right perfect nor right mininjective. In par-
ticular, S is not a right countably Σ-C3-ring. Moreover, S is not left self-injective
because it is not left finite dimensional. Inasmuch as S/J(S) ∼ = F , we infer that S
is a right Kasch ring, and consequently S is a strongly left C3-ring. By Theorem
2.28, S is a left countably Σ-C3-ring.
Example 2.36. The following example, due to John Clark, was investigated thor-
 C3-Modules 669

oughly in [18, Example 6.6]. It is a commutative ring R with ideal lattice

0 = Rv0 ⊂ Rv1 ⊂ Rv2 ⊂ · · · ⊂ V ⊂ · · · ⊂ Rp2 ⊂ Rp ⊂ R

where p and vi (i ≥ 0) satisfy pvk = vk−1 for all k ≥ 1, and V is the only non-
principal ideal. It is shown that R is local with simple essential socle. Furthermore,
R is a mininjective Kasch ring with J(R) = Z(R), that is neither perfect nor
selfinjective. In particular, R is a strongly C3-ring that is not countably Σ-C3.

Example 2.37. The following example is essentially due to Björk (see [18, Example
2.6]). Let F be a field and assume that a 7→ ā is an isomorphism F → F̄ ⊆ F
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

where the subfield F̄ 6= F . Let R denote the left vector space with basis {1, t}, and
make R into an F -algebra by defining t2 = 0 and ta = āt for all a ∈ F . It is not
difficult to see that R is a local left artinian right mininjective ring. In particular,
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

R is a left and right Kasch, left and right countably Σ-C3-ring. It is also shown
that R is right artinian if and only if F̄ F is finite dimensional. In general, R need
not be quasi-Frobenius; and it can be shown that R is quasi-Frobenius if and only
if dim(F̄ F ) = 1.

Acknowledgement. Part of this work was carried out while the second author was visiting
the Mathematics Department of The Ohio State University. He would like to express his
gratitude to the members of the Mathematics Department at OSU for the hospitality,
warm reception and financial support.

References
[1] I. Amin, Y. Fathi, M.F. Yousif, Strongly simple-injective rings and modules, Algebra
Colloq. 15 (2008) 135–144.
[2] I. Amin, M.F. Yousif, N. Zeyada, Soc-injective rings and modules, Comm. Algebra
33 (2005) 4229–4250.
[3] F.W. Anderson, K.R. Fuller, Rings and Categories of Modules, Springer-Verlag,
Berlin-New York, 1974.
[4] Y. Baba, K. Oshiro, Classical Artinian Rings and Related Topics, World Sci., Singa-
pore, 2009.
[5] H. Bass, Finitistic dimension and a homological generalization of semiprimary rings,
Trans. Amer. Math. Soc. 95 (1960) 466–488.
[6] J.L. Chen, W.X. Li, On Artiness of right CF rings, Comm. Algebra 32 (2004) 4485–
4494.
[7] E.E. Enochs, Injective and flat covers, envelopes and resolvents, Israel J. Math. 39
(1981) 189–209.
[8] C. Faith, Rings with ascending chain condition on annihilators, Nagoya Math. J. 27
(1966) 179–191.
[9] C. Faith, E.A. Walker, Direct sum representations of injective modules, J. Algebra 5
(1967) 203–221.
[10] K.R. Fuller, On direct representations of quasi-injectives and quasi-projectives, Arch.
Math. 20 (1969) 495–502.
[11] L. Jeremy, Sur les modules et anneaux quasi-continus, C. R. Acad. Sci. Paris (Série
A) 272 (1971) 80–83.
[12] I. Kaplansky, Projective modules, Annals of Mathematics (Ser. 2) 68 (1958) 372–377.
 670 I. Amin, Y. Ibrahim, M. Yousif

[13] F. Kasch, Modules and Rings, L.M.S. Monograph No. 17, Academic Press, New York,
1982.
[14] A.C. Mewbor, C.N. Winton, Orders in self-injective semiperfect rings, J. Algebra 13
(1969) 5–9.
[15] S.H. Mohamed, T. Bouhy, Continuous modules, Arabian J. Sci. Eng. 2 (1977) 107–
112.
[16] S.H. Mohamed, B.J. Müller, Continuous and Discrete Modules, Cambridge Univ.
Press, Cambridge, 1990.
[17] W.K. Nicholson, Semiregular modules and rings, Canad. J. Math. 28 (5) (1976)
1105–1120.
[18] W.K. Nicholson, M.F. Yousif, Quasi-Frobenius Rings, Cambridge Tracts in Math.
158, Cambridge Univ. Press, Cambridge, 2003.
Algebra Colloq. 2015.22:655-670. Downloaded from [Link]

[19] K. Oshiro, Lifting modules, extending modules and their applications to QF-rings,
Hokkaido Math. J. 13 (1984) 310–338.
by Dr. Mohamed Yousif on 10/06/15. For personal use only.

[20] B.L. Osofsky, A generalization of quasi-Frobenius rings, J. Algebra 4 (1966) 373–387.

[21] L. Shen, J. Chen, On countably Σ-C2 rings (preprint).
[22] T. Takeuchi, On direct modules, Hokkaido Math. J. 1 (1972) 168–177.
[23] Y. Utumi, On continuous regular rings, Canad. Math. Bull. 4 (1961) 63–69.
[24] R. Ware, Endomorphism rings of projective modules, Trans. Amer. Math. Soc. 155
(1971) 233–256.
[25] R. Wisbauer, Foundations of Module and Ring Theory, Gordon and Breach, Philadel-
phia, 1991.
[26] W. Xue, Characterization of rings using direct-projective modules and direct-injective
modules, J. Pure Appl. Algebra 87 (1993) 99–104.
[27] M.F. Yousif, Y. Zhou, Semiregular, semiperfect and perfect rings relative to an ideal,
Rocky Mountain J. Math. 32 (2002) 1651–1671.

View publication stats