AKCE International Journal of Graphs and Combinatorics

ISSN: 0972-8600 (Print) 2543-3474 (Online) Journal homepage: www.tandfonline.com/journals/uakc20

On the comaximal graph of a ring

Pravin Gadge, Nilesh Khandekar & Vinayak Joshi

To cite this article: Pravin Gadge, Nilesh Khandekar & Vinayak Joshi (2024) On the comaximal
graph of a ring, AKCE International Journal of Graphs and Combinatorics, 21:2, 143-151, DOI:
10.1080/09728600.2024.2302184

To link to this article: https://doi.org/10.1080/09728600.2024.2302184

© 2024 The Author(s). Published with

license by Taylor & Francis Group, LLC

Published online: 06 Feb 2024.

Submit your article to this journal

Article views: 990

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=uakc20
AKCE INTERNATIONAL JOURNAL OF GRAPHS AND COMBINATORICS
2024, VOL. 21, NO. 2, 143–151
https://doi.org/10.1080/09728600.2024.2302184

On the comaximal graph of a ring

Pravin Gadgea , Nilesh Khandekarb , and Vinayak Joshib
a
GES’s Shri Bhausaheb Vartak Arts, Commerce and Science College, Borivali, Maharashtra, India; b Department of Mathematics, Savitribai Phule Pune
University, Pune, Maharashtra, India

ABSTRACT ARTICLE HISTORY

Received 9 November 2023
In this paper, we prove that the study of the comaximal graph of a finite commutative ring with identity
Accepted 27 December 2023
is the same as the study of the zero-divisor graph of the specially constructed lattice. Also, we introduce
the blow-up of a Boolean lattice using finite chains. Furthermore, we prove that in the case of a reduced KEYWORDS
ring, the comaximal graph and the co-annihilating graph coincide. In the last section, we study the Zero-divisor graph; Boolean
perfectness of the comaximal graphs and characterize the Hamiltonian property of the complement of poset; comaximal graph;
a comaximal graph. reduced ring; perfect graph;
co-annihilating graph
MATHEMATICS SUBJECT
CLASSIFICATION (2020)
05C25; 06A07; 05C17; 13A70

1. Introduction Gaur and Sharma [9] studied the maximal graph of a ring
R whose vertices are the elements of a ring R and two distinct
In [5], Beck introduced the concept of the zero-divisor graph
vertices x and y are adjacent if and only if there exists a maximal
0 (R) of a commutative ring R with identity, whose vertices
ideal of R containing both x and y. They observed that the
are the elements of R and two distinct vertices x and y are
maximal graph of a ring R is the same as the complement of
adjacent if and only if xy = 0. He was mainly concerned with
the comaximal graph of a ring R. On lines of the zero-divisor
the coloring of 0 (R). Let χ (0 (R)) and ω(0 (R)) denote the
graph of rings, Halaš and Jukl [12], Lu and Wu [21] and Joshi
chromatic number and the clique number of 0 (R), respectively.
[14] studied the zero-divisor graph of ordered sets. It is proved
Beck conjectured that χ (0 (R)) = ω(0 (R)), and such graphs
that the zero-divisor graph of ordered sets is weakly perfect.
are called weakly perfect graphs.
Recently, a commendable book on graphs related to rings has
In 1995, Sharma and Bhatwadekar [25] introduced a graph
been authored by D. F. Anderson et al., as documented in [3].
(R) on a commutative ring R with identity, whose vertices are
In this paper, we show that the comaximal graph of a commu-
the elements of R and two distinct vertices x and y are adjacent
tative ring R with identity is nothing but the zero-divisor graph
if and only if Rx + Ry = R. In [25, Theorem 2.3], the authors
of a specially constructed poset; in fact, a 0-1-distributive lattice.
proved that χ ((R)) < ∞ if and only if the ring R is finite, and
It may be noted that in [23], it is shown that the comaximal
in this case, the graph (R) is weakly perfect and χ ((R)) =
graphs are blow-ups of Boolean graphs, the zero-divisor graphs
ω((R)) = t + , where t is the number of maximal ideals
of Boolean rings. From this fact, we can deduce that these graphs
in R and  is the number of units in R. Maimani et al. [22]
are the zero-divisor graphs of ordered sets. However, the con-
named the graph (R) studied by Sharma and Bhatwadekar
struction of the poset is not given. One can observe that the zero-
as the comaximal graph of R. Thus the comaximal graphs are
divisor graphs of 0-1-distributive lattices have many interesting
weakly perfect. Also in [22], the authors studied the subgraphs
properties. In the case of the comaximal graph, we are able to
1 (R), 2 (R) and 2 (R) = 2 (R) \ J(R), where 1 (R) is the
show that it is a zero-divisor graph of a 0-1-distributive lattice.
subgraph of (R) induced on the set of units of R, 2 (R) is the
Hence, we are able to extract many properties of the comaximal
subgraph of (R) induced on the set of non-units of R and 2 (R)
graphs via zero-divisor graphs of ordered sets. Also, we prove
is the subgraph of (R) induced on the set of non-units of R
that in the case of a reduced Artinian ring, the comaximal
which are not  in J(R), the Jacobson radical of R, i.e., 2 (R) =
graph and the co-annihilating graph coincide. By co-annihilating
(R) \ (U(R) J(R)). Moconja and Petrović [23] also studied
graph, denoted by CAR , of a commutative ring R with identity is
the comaximal graph of R and investigated some properties such
the graph with the vertex set μR , the set of all nonzero, non-units
as the center, radius and girth of the comaximal graph. In [28],
of R and two vertices x and y are adjacent whenever ann(x) ∩
authors used graph blow-up (c.f. Definition 4.1) to study the
ann(y) = {0}.
comaximal graph 2 (R).

CONTACT Vinayak Joshi [email protected] Department of Mathematics, Savitribai Phule Pune University, Pune, 411007, Maharashtra, India.
© 2024 The Author(s). Published with license by Taylor & Francis Group, LLC.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution,
and reproduction in any medium, provided the original work is properly cited. The terms on which this article has been published allow the posting of the Accepted Manuscript in a repository by
the author(s) or with their consent.
144 P. GADGE, N. KHANDEKAR, AND V. JOSHI

In the last section, we provide applications of the zero-divisor G(P) has at least two vertices, and G(P) is connected with the
graphs of ordered sets to comaximal graphs. diameter at most three [21, Proposition 2.1].
Throughout this paper, all rings are assumed to be non- A poset P is called distributive if, {{a} ∪ {b, c}u } = {{a, b} ∪
local Artinian commutative with identity. Let Max(R) = {a, c} }u holds for all a, b, c ∈ P; see [19]. This definition
{M1 , M2 , . . . , Mn } denotes the collection of all maximal ideals generalizes the usual notion of a distributive lattice (i.e., a lattice
of R. Let U(R) and J(R) denote the set of all units of R and is distributive in the usual sense if and only if it is a distributive
the Jacobson radical of R, respectively. The dual of a lattice L is poset). Moreover, a bounded poset P is called Boolean if P is
denoted by L∂ . distributive and complemented. Clearly, every Boolean lattice
is a Boolean poset, but the converse is not true. A graph G is
a Boolean graph if it is the zero-divisor graph of a Boolean ring.
2. Preliminaries
It is well-known that in a Boolean poset, complementation
The following necessary definitions and terminologies can be coincides with pseudocomplementation (cf. [15, Lemma 2.4]).
found in Devhare et al. [6] and Grätzer [11]. In particular, if P is Boolean, then P is pseudocomplemented,
Let P be a poset. For A ⊆ P, the upper cone of A is Au = and every element x ∈ P has the unique complement x .
{b ∈ P | a ≤ b for every a ∈ A} and the lower cone of A is Sometimes, it is also denoted by x∗ .
A = {b ∈ P | b ≤ a for every a ∈ A}. If a ∈ P, then the sets The direct product of posets P1 , P2 , . . . , Pn is the poset P =
{a}u and {a}l will be denoted by au and a , respectively. By Au , n
Pi with ≤ defined such that a ≤ b in P if and only if ai ≤ bi
we mean {Au } . Dually, we have Au . i=1
A poset P with 0 is called 0-distributive if for a, b, c ∈ P, 
n
{a, b} = {0} = {a, c} implies {a, {b, c}u } = {0}. A lattice L (in Pi ) for every i ∈ {1, 2, . . . , n}. For any A ⊆ Pi , note that
i=1
with 0 is said to be 0-distributive if a ∧ b = 0 and a ∧ c = 0 
n
implies a ∧ (b ∨ c) = 0 for a, b, c ∈ L. Hence, it is clear that if Au = {b ∈ Pi | ai ≤ bi for every a ∈ A and i ∈ {1, 2, . . . , n}}.
i=1
a lattice L is 0-distributive, then L, as a poset, is a 0-distributive 
n
poset. A lattice L with 1 is said to be 1-distributive if a∨b = 1 and Similarly, A = {b ∈ Pi | bi ≤ ai for every a ∈ A and i ∈
i=1
a∨c = 1 implies a∨(b∧c) = 1. An element a = 0 in a lattice L is {1, 2, . . . , n}}.
said to be join-irreducible if and only if whenever a = b ∨ c, then 
n
a = b or a = c. Dually, we have the concept of meet-irreducible Throughout this paper, P = Pi , (n ≥ 2), where Pi ’s are
i=1
element. A representation a = x0 ∨ x1 ∨ · · · ∨ xn−1 is redundant finite bounded posets such that Z(Pi ) = {0}, ∀i and 2 ≤ |P1 | ≤
if and only if a = x0 ∨ x1 ∨ · · · ∨ xi−1 ∨ xi+1 ∨ · · · ∨ xn−1 , for |P2 | ≤ · · · ≤ |Pn |. Since Pi ’s are finite bounded posets, Pi ’s are
some 0 ≤ i < n; otherwise it is irredundant. atomic. Note that Z(Pi ) = 0 if and only if Pi has a unique atom.
Suppose that P is a poset with 0. If ∅ = A ⊆ P, then Further, assume that qi ∈ Pi is the unique atom Pi for every
the annihilator of A is given by A⊥ = {b ∈ P | {a, b} = i. It is not difficult to prove that (q1 , 0, . . . , 0), (0, q2 , 0, . . . , 0),
{0} for all a ∈ A}, and if A = {a}, then we write a⊥ = A⊥ . . . . , (0, . . . , 0, qn ) are the only n atoms of P. Further, if |Pi | = 2,
An element a ∈ P is called an atom if a > 0 and {b ∈ P | 0 < ∀i, then P is a Boolean lattice; in this case, it is denoted by 2n .
b < a} = ∅, and P is called atomic if for every b ∈ P \ {0}, there All the elements of P are denoted by bold letters.
exists an atom a ∈ P such that a ≤ b. Let x and y be elements Let x = (x1 , x2 , . . . , xn ) ∈ P, where xi ∈ Pi . In par-
of P. Then y covers x, written x −< y, if x < y and there is no ticular, q1 , q2 , . . . , qn are all the atoms of P. That is, qi =
element z such that x < z < y. Let P and Q be two posets. The (0, . . . , 0, qi , 0, . . . , 0). Also, (0, 0, . . . , 0), (1, 1, . . . , 1) ∈ P are
linear sum of P and Q is denoted by P ⊕ Q and is defined by the least and the greatest elements of P denoted by 0 and 1,
taking the following order relation on P ∪ Q : x ≤ y if and only respectively.
if x, y ∈ P and x ≤ y in P, or x, y ∈ Q and x ≤ y in Q, or x ∈ P
and y ∈ Q.
A poset P is called bounded if P has both the least element 3. Comaximal graphs
0 and the greatest element 1. An element b of a bounded poset In this section, we observe that the comaximal graphs can be
P is a complement of a ∈ P if {a, b} = {0} and {a, b}u = {1}. studied via zero divisor graphs of ordered sets.
A pseudocomplement of a ∈ P is an element b ∈ P such that Lu and Wu [21, Theorem 3.1] proved that the zero-divisor
{a, b} = {0}, and x ≤ b for every x ∈ P with {a, x} = {0}, see graphs of posets are nothing but the compact graphs, that is,
[13]. It is straightforward to check that any element a ∈ P has at every zero-divisor graph of a poset is a compact graph and vice-
most one pseudocomplement, and it will be denoted by a∗ (if it versa. By a compact graph, we mean a simple undirected, con-
exists). A bounded poset P is called complemented (respectively, nected graph G in which for each pair of non-adjacent vertices
pseudocomplemented) if every element of P has a complement x, y of G, there is a vertex z with N(x) ∪ N(y) ⊆ N(z), where
(respectively, a∗ exists for every a ∈ P). N(x) is the set of all vertices adjacent to x. First, we observe
Let P be a poset with 0. Define a zero-divisor of P to be any that the comaximal graphs are compact graphs. This fact can
element of the set Z(P) = { a ∈ P | there exists b ∈ P\{0} such also be deduced from the fact that the comaximal graphs are the
that {a, b} = {0}}. An element a ∈ P is called dense, if a ∈ / Z(P). blow-up of Boolean graphs, see [23, Theorem 2.4] (see also [28,
The set of all dense elements of P is denoted by D(P). As in [21], Lemma 4.1]).
the zero-divisor graph of P is the graph G(P) whose vertices are
the elements of Z(P) \ {0} such that two vertices a and b are Lemma 3.1. Let R be a commutative ring with identity. Then
adjacent if and only if {a, b} = {0}. If Z(P) = {0}, then clearly 2 (R) is a compact graph.
 AKCE INTERNATIONAL JOURNAL OF GRAPHS AND COMBINATORICS 145

Proof. From [22, Theorem 3.1], it is clear that 2 (R) is con- Proof. The proof of the first part is evident. To prove the more-
nected. Let x and y be any two non-adjacent vertices in 2 (R). over part, it is sufficient to prove that ≈ and  are the same on
Then there exists a maximal ideal Mk such that Rx+Ry ⊆ Mk for R when R is Artinian reduced.
some k ∈ {1, 2, . . . , n}. Let t ∈ N(x)∪N(y). Then Rx+Rt = R or Since R is an Artinian reduced ring, there exist fields
Ry+Rt = R which gives Rx+Ry+Rt = R. Thus xa+yb+tc = 1 F1 , F2 , . . . , Fn such that R ∼ = F1 × F2 × · · · × Fn . Therefore,
for some a, b, c ∈ R. If z = xa + yb, then z.1 + tc = 1. the maximal ideals of R are of the form, Mi = F1 × · · · × Fi−1 ×
Hence Rz + Rt = R. We observe that z ∈ / U(R), otherwise (0) × Fi+1 × · · · × Fn . Let x, y ∈ R. Then x = (x1 , x2 , . . . , xn ) and
Rx + Ry = R. Further, if z ∈ J(R), then 1 = xa + yb + tc ∈ Mt y = (y1 , y2 , . . . , yn ), where xi , yi ∈ Fi for all i ∈ {1, 2, . . . , n}.
for some t ∈ {1, 2, . . . , n} and tc ∈ Mt ; a contradiction. Thus Assume that x ≈ y. Then x, y ∈ Ri1 i2 ...ik for some
z ∈ V(2 (R)). Hence t ∈ N(z). This shows that 2 (R) is a {i1 , i2 , . . . , ik } ⊆ {1, 2, . . . , n}. We have to prove that x  y, i.e.,
compact graph. ann(x) = ann(y). 
As x, y ∈ Ri1 i2 ...ik , we have x, y ∈ Mi and x, y ∈ / Mj ,
From Lemma 3.1, it is clear that the comaximal graph of a i∈{i1 ,i2 ,...,ik }
ring is the zero-divisor graph of an ordered set. Now, we explic- j∈
/ {i1 , i2 , . . . , ik }. Hence,

itly construct the desired poset. We consider some equivalence 0 if t ∈ {i1 , i2 , . . . , ik };
relations on the given ring R for that purpose. There are some xt = and
xt (= 0) if t ∈
/ {i1 , i2 , . . . , ik };
known equivalence relations defined on R. We consider a few of 
them here and show that they are the same. 0 if t ∈ {i1 , i2 , . . . , ik };
In [23, Proposition 2.3], the authors have essentially parti- yt =
yt (= 0) if t ∈
/ {i1 , i2 , . . . , ik }.
tioned the vertex setof (R), that is, R, in
 terms Iof thesets
c
VI , where VI = MI (M I )c and MI = Mi , M = Mi It is known that for x ∈ F, x = 0 if and only if ann(x) = F and
i∈I i∈I x = 0 if and only if ann(x) = {0}. Hence, ann(xi ) = ann(yi ) in
for I ⊆ {1, 2, . . . , n} and I c denote the set {1, 2, . . . , n} \ I and Fi for all i ∈ {1, 2, . . . , n}. Hence ann(x) = ann(y), that is, x  y.
c c
(M I )c = R \ M I . One can see that V∅ is the set of all units of R Conversely, assume that x  y. Then ann(x) = ann(y).
and V12...n is the Jacobson radical of R. Now, consider a relation We have to prove that x ≈ y, i.e., x, y ∈ Ri1 ...ik for some
∼ on R as x, y ∈ R, x ∼ y if and only if x, y ∈ VI for some partition Ri1 ...ik of R. Suppose on the contrary, there exist distinct
I ⊆ {1, 2, . . . , n}. Clearly, ∼ is an equivalence relation on R. Let index sets {i1 , i2 , . . . , ik } and {j1 , j2 , . . . , jk } of Ri1 ...ik and Rj1 ...jk ,
[a]∼ denotes the equivalence class of a under the relation ∼. respectively, such that x ∈ Ri1 ...ik and y ∈ R j1 ...jk . Clearly, x = y.
We propose another partition of R influenced by the relation As x ∈ Ri1 ...ik and y ∈ Rj1 ...jk then x ∈ Mi , x ∈ / Mj , j ∈ /
given by Afkhami et al. [1] as follows. i∈{i1 ,...,ik }

Let 1 ≤ i1 < i2 · · · < ik ≤ n. The notation, Ri1 i2 ...ik stands {i1 , i2 , . . . , ik } and y ∈ Mi , y ∈ / Mj , j ∈ / {j1 , j2 , . . . , jk }.
for the set: i∈{j1 ,j2 ,...,jk }

k This gives that
Ri1 i2 ...ik = {x ∈ R | x ∈ Mis and x ∈ / 
 s=1
0 if t ∈ {i1 , i2 , . . . , ik };
Mj } and R∅ = {x ∈ R | Rx = R}. Some- xt = and
j∈{1,2,...,n}\{i1 ,i2 ,...,ik } xt (= 0) if t ∈
/ {i1 , i2 , . . . , ik };
times, we abuse notation of Ri1 i2 ...ik by denoting RI , where I = 
{i1 , i2 , . . . , ik }. Define a relation ≈ on R as : Let x, y ∈ R. Then 0 if t ∈ {j1 , j2 , . . . , jk };
yt =
x ≈ y if and only if x, y ∈ RI for I ⊆ {1, 2, . . . , n}. It is easy to yt (= 0) if t ∈
/ {j1 , j2 , . . . , jk }.
see that R∅ is the set of all units of R and R1...n is the Jacobson
Since index sets {i1 , i2 , . . . , ik } and {j1 , j2 , . . . , jk } of Ri1 ...ik
radical of R. Let [a]≈ denote the equivalence class of a under the
and Rj1 ...jk , respectively, are distinct, without loss of generality,
relation ≈.
we assume that there exists l ∈ {i1 , i2 . . . , ik } such that l ∈ /
Let M(x) = {M ∈ Max(R) | x ∈ M} and D(x) = Max(R) \
{j1 , j2 , . . . , jk }. This gives that xl = 0 and yl = 0. This implies
M(x). Define a relation  on R as x  y if and only if M(x) =
that (0, . . . , 0, 1, 0, . . . , 0) ∈ ann(x), but (0, . . . , 0, 1, 0, . . . , 0) ∈ /
M(y). Clearly,  forms an equivalence relation on R. Let [a]
ann(y), where 1 is at the lth position. This contradicts the fact
denotes the equivalence class of a under the relation .
that ann(x) = ann(y). Hence x ≈ y.
As noted in [4, Remark 2.3], there is an equivalence
relation  on R as x  y if and only if ann(x) =
ann(y). Let [a] denotes the equivalence class of a under the
In view of Theorem 3.2, we have the following observation.
relation .
With this preparation, the following result proves that the Observation 3.3. We note that [x]∼ = [x]≈ = [x] = VI = RI ,
equivalence relations ∼, ≈ and  are the same. Also, we prove where x ∈ VI .
that in the case of an Artinian reduced ring, all the four equiva-
lence relations ∼, ≈,  and  are the same on R. The following result is essentially proved in [23, Proposition
Theorem 3.2. Let R be a commutative ring with identity. Then 2.3 ].
the equivalence relations ∼, ≈ and  are the same. Moreover, if R
Lemma 3.4 (Moconja and Petrović [23, Proposition 2.3 ]). Let R
is an Artinian reduced ring, then the equivalence relations ∼, ,
be a finite commutative ring with identity. Then RI = ∅, for every
≈ and  are the same on R.
I ⊆ {1, 2, . . . , n}.
146 P. GADGE, N. KHANDEKAR, AND V. JOSHI

The following facts are evident from Observation 3.3 and

Lemma 3.4. These are observed in [23].

(1) If the index sets I and J of RI and RJ , respectively, are distinct,

then RI ∩ RJ = ∅ for some I, J ⊆ {1, 2, . . . , n}.
(2) R = · RI .
I⊆{1,...,n}
 
We denote the set [R] = RI | I ⊆ {1, 2, . . . , n} . As observed
in Lemma 3.1, 2 (R) is a compact graph, i.e., it is a zero-divisor
graph of a poset. We now construct the required poset. For this
purpose, we consider a partial ordering relation on [R] defined
below.
Define a relation ≤ on [R] as RI ≤ RJ if and only if I ⊆ J,
where I, J ⊆ {1, 2, . . . , n}. It is easy to prove that ([R], ≤) forms
a poset. Clearly, R∅ ≤ RI and RI ≤ R1...n for all I ⊆ {1, 2, . . . , n}.
Therefore, [R] is a bounded poset with R∅ as the zero element
and R1...n as the greatest element. Since R has n number of
maximal ideals, it is easy to verify that R1 , R2 , . . . , Rn are the
only n atoms of [R]. Also, [R] has 2n elements follows from Figure 1. Construction of the poset from a ring.
Lemma 3.4. Moreover, RI ∧ RJ = RI∩J and RI ∨ RJ = RI∪J for
all I, J ⊆ {1, 2, . . . , n} gives that [R] is a bounded lattice with 2n
As noted earlier, [R] is a Boolean lattice. Now, in the construc-
elements. Further, consider the map f : [R] → P(X), where X =
tion of the poset L from [R], we are replacing elements of [R]
{1, 2, . . . , n} given by f (RI ) = I for RI ∈ [R]. Clearly, f is one-to-
by bounded chains with any predetermined well order. So, it is
one, onto and both f , f −1 are order preserving homomorphisms.
expected to ask which properties of [R] are preserved in L. This
Thus, f is a lattice isomorphism. As a consequence, we have the
question is answered in the following result.
following result.
Lemma 3.8. Let L be a lattice and L be a poset obtained from
Theorem 3.5. The poset [R] forms a Boolean lattice.
L by replacing an element of L with a bounded chain. Then the
The above result also follows from the fact that comaximal following statement holds:
graphs are blow-ups of Boolean graphs [28, Lemma 4.1]. Now, (1) If L is pseudocomplemented, then so is L .
we construct a poset L from [R] as follows: (2) If L is dual pseudocomplemented, then so is L .
Construction 3.6. To construct a poset L from a Boolean lattice Proof. (1) Assume that L is pseudo complemented. Let x ∈ L .
[R], we replace the elements RI of [R] by chains. Let RI ∈ [R], If x ∧ y = 0 for y ∈ L , then we need to show that there exists
where (∅ =)I ⊆ {1, 2, . . . , n}. By considering any predeter- x∗ ∈ L such that x ∧ x∗ = 0 and y ≤ x∗ . Clearly x ∈ Ca for
mined well-order on the elements of RI , replace an element RI some chain Ca and a ∈ L, y ∈ / Ca . Let y ∈ Cb for some chain Cb
of [R] by a bounded chain of elements of RI in L. Clearly, the with a = b ∈ L. Since x ∧ y = 0, we have a ∧ b = 0, because
cardinality of the chain is |RI |. Along with this, the element R∅ of the construction of L from L. As L is pseudocomplemented,
of [R], we are replacing by a single element u in L. Although a has the pseudocomplement, say a∗ ∈ L. Consider the largest
R∅ may contain more than one element, we are replacing it with element of the chain Ca∗ , say x∗ . Then we have x ∧ x∗ = 0 =
only one element, namely, u. Clearly, u is the zero element of a ∧ a∗ . Further, as a ∧ b = 0, and a∗ is the pseudocomplement
L. It is clear that R∅ is the set of all units of R and R1...n is the of a, we get b ≤ a∗ . However, every element, say m, on the Cb is
Jacobson radical of R. If R1...n contains more than one element, always less than or equal to every element on the chain Ca∗ . In
then the element R1...n of [R] will be replaced by a bounded chain particular y ≤ x∗ . This proves that L is pseudocomplemented.
of cardinality |R1...n |. (2) Proof follows by the duality principle.
Henceforth, we use the notation L for the poset con-
As [R] is Boolean, hence it is pseudocomplemented and
structed in Construction 3.6.
dual pseudocomplemented. In view of Lemma 3.8, we have the
The following example illustrates the construction of [R] and
following Corollary 3.9.
the corresponding lattice L as shown in Figure 1.
Corollary 3.9. The constructed poset L from [R] is a pseudocom-
Example 3.7. Consider the ring Z30 ∼ = Z2 × Z3 × Z5 . Let
plemented and dual pseudocomplemented lattice.
M1 = (2), M2 = (3), M3 = (5) be maximal ideals. Then the
construction of [R] = [Z30 ] is as follows: One can see that, R∅ = Remark 3.10. Let x, y be elements of L. Assume that x ∈ RI , y ∈
= {2, 4, 8, 14, 16, 22, 26, 28} =
{1, 7, 11, 13, 17, 19, 23, 29}, R1  RJ for some subsets I, J of {1, 2, . . . , n}. Then
{x ∈ Z30 | x ∈ (2) and x ∈ / (3) (5)}, R2 = {3, 9, 21, 27}, R3 =
{5, 25}, R12 = {6, 12, 18, 24} = {x ∈ Z30 | x ∈ (2) (3) and x ∈ / (1) x ∧ y = z, where z is the greatest element of a bounded chain
(5)}, R13 = {10, 20}, R23 = {15} and R123 = {0}. of elements of RI∩J = RI ∧ RJ and x ∨ y = w, where w is
 AKCE INTERNATIONAL JOURNAL OF GRAPHS AND COMBINATORICS 147

the smallest element of a bounded chain of elements of RI∪J = Proof. (A) Let R be an Artinian reduced ring. Then J(R) = (0).
RI ∨ RJ . Note that such greatest and smallest elements do exist Therefore R123...n = {0} in [R], i.e., 0 is the greatest element of
as chains are bounded. L which is a join-reducible. Conversely, suppose J(R) = (0).
(2) x ∧ y = u if and only if RI ∧ RJ = R∅ . Assume that |J(R)| > 1. Then the elements of R12...n form
(3) The atoms of lattice L are the smallest element of chains of RI , a bounded chain of length |J(R)| with any predetermine well-
I ⊆ {1, 2, . . . , n}. Thus, if the ring R has n maximal ideals, order in L. This implies that the greatest element of L is a join-
then the lattice L has n number of atoms and vice versa. irreducible. This is a contradiction to the fact that the greatest
element is join-reducible. Hence we must have J(R) = (0).
Remark 3.11. Let R be an Artinian reduced ring. Then R = F1 × This, together with R, is an Artinian; we have R as a reduced
F2 × · · · × Fn . Let Mi = F1 × · · · × Fi−1 × (0) × Fi+1 × · · · × Fn ring.
be the maximal ideal of R for 1≤ i ≤ n. One can easily observe (B) Suppose G(L) ∼ = G(L∂ ). We claim that the Jacobson
that |Mi | = |Fj | and RI = Mi \ ( Mj ). This gives |RI | = radical J(R) = (0) and hence R is reduced. Suppose, on the
j=i i∈I j∈I
 /
contrary, that J(R) = (0). Then, by constructing L, the greatest
(|Fj | − 1).
j∈I
/
element of L is a join-irreducible element. But in this case,
V(G(L∂ )) = ∅, a contradiction to V(G(L)) = ∅. This proves
A partial preorder is a reflexive and transitive relation . In that J(R) = (0) and hence R is reduced.
a partial preorder, the relation ≡ defined by x ≡ y if x  y and (C) Let R be a Boolean ring. Then we have |RI | = 1 for all
y  x is an equivalence relation, and  induces a partial order I ⊆ {1, 2, . . . , n}. By Construction 3.6, we have L is a Boolean
on the set of equivalence classes. Then  can be extended to a lattice. Conversely, suppose L is Boolean. Then we have |RI | =
partial order by putting a total order on each equivalence class. 1 for all (∅ =)I ⊆ {1, 2, . . . , n}. We claim that R is Boolean.
LaGrange and Roy [20, Remark 3.4] used the above idea Consider the nonzero, non-unit element x of R. This implies
 that
to define a partial order on a reduced commutative ring with x ∈ RI for some I ⊆ {1, 2, . . . , n}. This means that x ∈ Mi and
identity mentioned as follows: x ≤ y if and only if either  i∈I
x∈ / Mj . By primeness of maximal ideals, x2 ∈ Mi if and only
ann(y)  ann(x), or ann(x) = ann(y) and x is less than or j∈I
/
equal to y in some predetermined well-order on the set {r ∈ if x ∈ Mi for all i. Therefore, we get x2 ∈ RI . As L is Boolean and
R | ann(r) = ann(x)}. In the case of a finite reduced ring R, so |RI | = 1 for all (∅ =)I ⊆ {1, 2, . . . , n}, we have x2 = x for
the poset (in fact a lattice) P constructed from R is essentially x ∈ R\U(R). By Lemma 3.13 and the fact that Z(R) = R\U(R),
the linear sum of Cn and the dual of the lattice L constructed in we get R is a Boolean ring.
Construction 3.6, denoted by L∂ , that is, P = L∂ ⊕ Cn , where
n = |U(R)|− 1. Moreover, if R is an Artinian reduced ring, In [23, Lemma 2.1], Moconja and Petrović (see also Wu et
n
then R = i=1 Fi where Fi ’s are fields. Now, endow a linear al. [27, Theorem 1]) essentially proved the following result in
order on the elements of Fi with 0 as the least element and 1 as different terminology.
the greatest element. The R can be treated as a direct product
chains of length |Fi |. Hence 0 (R) = G(P) = G(L∂ ) =
of  Lemma 3.15 (Moconja and Petrović [23, Lemma 2.1]). For dis-
G( ni=1 Ci ), where Ci ’s are the chains with |Ci | = |Fi |, follows tinct x, y ∈ R, Rx + Ry = R if and only if I ∩ J = ∅, where x ∈ RI
from LaGrange and Roy [20, Remark 3.4], Devhare, Joshi and and y ∈ RJ .
LaGrange [7, Lemma 3.3] and the discussion prior to Corollary
3.5 of Devhare, Joshi and LaGrange [6]. As a consequence, we We close this section by proving the relationship between the
have : comaximal graph of a ring, the zero-divisor graph of a lattice and
the co-annihilating graph of a ring.
Theorem 3.12. Let 0 (R) be the ring-theoretic zero-divisor
graph of a finite reduced commutative ring R with identity. Then Theorem 3.16. Let R be a finite commutative ring with identity.
n zero-divisor graph of L , i.e,
0 (R) equals to the lattice-theoretic ∂
Then 2 (R) = G(L). Moreover, if R be an Artinian reduced ring,
0 (R) = G(P) = G(L ) = G( i=1 Ci ), where Ci ’s are the chains
∂
then G(L) = CAR = 2 (R).
with |Ci | = |Fi |.
Proof. Clearly, V(G(L)) = V(2 (R)) = R \ {U(R) ∪ J(R)}.
We list a few properties of L. For this purpose, we need the Now, let x and y be adjacent in G(L). Clearly, x ∈ RI and y ∈
following result. RJ for some nonempty distinct sets I, J  {1, 2, . . . , n}. From
Lemma 3.15, x and y are adjacent in G(L) if and only if RI and
Lemma 3.13 (Anderson and LaGrange [4, Lemma 2.5]). Let R RJ are adjacent in G([R]) if and only if I ∩ J = ∅ if and only if
be a commutative ring with 1 = 0 and Z(R) = 0. Then R is a Rx + Ry = R if and only if x and y are adjacent in 2 (R). Hence
Boolean ring if and only if x2 = x for every x ∈ Z(R). 2 (R) = G(L).
To prove the moreover part, assume that R is an Artinian
Theorem 3.14. Let R be an Artinian ring with |Max(R)| = n ≥
reduced ring. Hence J(R) = (0) and V(CAR ) = 2 (R) = {x ∈
2. Then
R | x non-zero non-units of R}.
(A) The greatest element of L is join-reducible if and only if R is a Suppose x is adjacent to y in CAR . Then ann(x) ∩ ann(y) =
reduced ring. {0}. On the contrary, assume that x is not adjacent to y in 2 (R),
(B) If G(L) = G(L∂ ), then R is reduced. i.e., Rx+Ry = R. Hence x∧y = u in L, i.e., x∨y = u in L∂ , where
(C) L is Boolean if and only if R is Boolean. u is the only dense element in L∂ . This gives that x ∨ y ∈ Z(L∂ ).
148 P. GADGE, N. KHANDEKAR, AND V. JOSHI

Therefore, there exists some nonzero a ∈ L∂ such that a ∧ (x ∨

y) = 0 in L∂ . This implies that a ∧ x = 0 and a ∧ y = 0 in L∂ .
Hence x is adjacent to a, and a is adjacent to y. By Theorem 3.12,
we have a.x = 0 and a.y = 0. Thus (0 =)a ∈ ann(x) ∩ ann(y),
a contradiction to ann(x) ∩ ann(y) = {0}. Hence x and y are
adjacent in 2 (R).
Now, assume that x is adjacent to y in 2 (R), that is, Rx+Ry =
R. Then x ∧ y = u in L. Hence, x ∨ y = u in L∂ . Suppose
that ann(x) ∩ ann(y) = {0}. Hence, there exists some nonzero
a ∈ ann(x) ∩ ann(y) such that a.x = 0 and a.y = 0. Hence x is
adjacent to a, and a is adjacent to y. By Theorem 3.12, we have
a∧x = 0 and a∧y = 0. Since L∂ is 0-distributive, a∧(x∨y) = 0.
This implies that x ∨ y ∈ Z(L∂ ), a contradiction to x ∨ y = u in
L∂ , where u is the only dense element of L∂ . Thus x is adjacent Figure 2. Blow-up LB of a Boolean lattice L.
to y in CAR .

From Theorem 3.16, it is clear that CAR is connected if R 

k
(2) Let (1 =)x = qij ∈ L, where qij be atoms of L.
is a finite reduced ring, as the zero-divisor graph of ordered j=1
sets is connected with the diameter at most 3 (see [21, Propo- 
k
sition 2.1]). This partially answers the question raised in Replace x ∈ L by the chain Xi1 i2 ...ik of length mj with
j=1
[2, Problem 1].

k
mj
j=1
elements xi11 i2 ...ik , xi21 i2 ...ik , . . . , xi1 i2 ...ik such that xi11 i2 ...ik −<
4. Blow-up of a Boolean lattice 
k
mj
As observed in [28, Lemma 4.1], the comaximal graphs are j=1
xi21 i2 ...ik −< · · · −< xi1 i2 ...ik , where {i1 , i2 , . . . , ik } ⊆
blow-ups of Boolean graphs. Also, from Theorem 3.16, the {1, 2, . . . , n}.
comaximal graph is the zero-divisor graph of a lattice derived (3) An element 1 ∈ L, replace it by u.
from a Boolean lattice. This motivates us to define the blow-
up of a Boolean lattice. In a Boolean lattice L, atoms are the The blow-up of a Boolean lattice L ∼ = 23 is shown in Figure 2.
only join-irreducible elements, and every nonzero element has The atom qi is replaced by a chain Xi such that |X1 | = 2 and
the unique irredundant representation as the join of atoms, |X2 | = |X3 | = 1. Hence the element q1 ∨ q2 is replaced by the
i.e., every nonzero element of a Boolean lattice L is a join of chain X12 with |X12 | = |X1 ||X2 | = 2 whereas the element q2 ∨q3
atoms (see Grätzer [11] p. 62) and this join is uniquely written. is replaced by the chain X23 with |X23 | = |X2 ||X3 | = 1.
However, this assertion is not true if L is not Boolean. Note that an element qi1 ∨ qi2 ∨ · · · ∨ qik = x ∈ L can be
With this, we are ready to introduce the concept of a blow-up also represented as (x1 , x2 , . . . , xn ) ∈ 2n where xj = 1 if j ∈
of a Boolean lattice L ∼= 2n by using chains. First, we recall the {i1 , i2 , . . . , ik } otherwise xj = 0. Thus, the chain Xi1 i2 ...ik in LB
definition of graph blow-up. uniquely corresponds to the element x = qi1 ∨ qi2 ∨ · · · ∨ qik in
L; equivalently to the element (x1 , x2 , . . . , xn ) ∈ 2n where xj = 1
Definition 4.1 (Ye, Wu, Liu and Yu [28, Definition 2.1 ]). Let G
if j ∈ {i1 , i2 , . . . , ik } otherwise xj = 0.
be a graph and T = (Tx )x∈V(G) be a collection of mutually disjoint
nonempty sets. The blow-up graph of G via T, denoted by G(T), Definition 4.3 ([16] (see also [6], [18])). Let L be a lattice with
is a new graph obtained as follows: . .
0. Define a relation = on L as x = y if and only if x⊥ = y⊥ .
. .
(1) Every vertex x of G is replaced by Tx . Clearly, = is an equivalence relation on L. Let [a]= denotes the
.
(2) For any two vertices x and y, if {x, y} ∈ E(G), then the equivalence class of a under =. The set of equivalence classes
.
subgraph induced on Tx ∪Ty in G(T) is the complete bipartite of L will be denoted by [L]={[a]= | a ∈ L}. Note that [L]
.= . .
graph with parts Tx and Ty . is a meet-semilattice with [a] ∧ [b]= = [a ∧ b] =
.= (see = [16,
.
(3) For any two vertices x and y, if {x, y} ∈ / E(G) then the Lemma 2]) under the partial order given by [a] ≤ [b] if
subgraph induced on Tx ∪ Ty in G(T) is a discrete graph or a and only if b⊥ ⊆ a⊥ . If L is 0-distributive, then [L] is a lattice;
null graph. see [16].
If further, |Tx | < ∞ for all x ∈ V(G), then G(T) is called . .
a finite blow-up of G. In particular, in LB , we observe that, [x12...k
1 ]= = [x12...k
2 ]= =

k
Definition 4.2. The blow-up LB of a Boolean lattice L ∼
= 2n using
mj
.
· · · = [x12...k ]= . Thus, the elements on the chain X12...k have the
j=1
chains is obtained as follows:
same equivalence classes.
(1) For 1 ≤ i ≤ n, replace each atom qi of L by the chain Xi
of length mi with elements xi1 , xi2 , . . . , ximi such that xi1 −< Theorem 4.4 ([18, Theorem 1.1 ]). Let L be a 0-distributive lattice
xi2 −< · · · −< ximi . with finitely many atoms. Then, [L] is a Boolean lattice.
 AKCE INTERNATIONAL JOURNAL OF GRAPHS AND COMBINATORICS 149

Remark 4.5. Let LB denote the blow-up of a Boolean lattice 

n
have G(L∂ ) = G(LB ), where P = Ci with |Ci | = |Fi | and
L ∼= 2n . Then by Theorem 4.4, [LB ] is a Boolean lattice with 2n i=1
elements. Hence [LB ] ∼
=L∼= 2n . [P] = L.

The following result is due to [17, Lemma 2.1]. We need the following observations to prove Theorem 4.10.
It is known that every pseudocomplemented lattice is 0-

n distributive. Further, one can verify that LB is a pseudocomple-
Lemma 4.6 ([17, Lemma 2.1]). Let P = Ci , where Ci ’s are mented and dual pseudocomplemented lattice, and hence a 0-
i=1
finite chains with 2 ≤ |C1 | ≤ |C2 | ≤ · · · ≤ |Cn |. Then P is 1-distributive lattice, by Lemma 3.8. In a Boolean lattice L, the
a pseudocomplemented lattice, and hence [P] is a Boolean lattice. pseudocomplement of an atom is a dual atom. Hence, in LB , the
.  pseudocomplement of an atom is a dual atom.
Moreover, |[x]= | = xi =0 (|Ci |−1) for every nonzero x ∈ P. The

n
Theorem 4.10. Let LB is a blow-up of a Boolean lattice L ∼ =
number of dense elements of P is (|Ci | − 1). Further, deg(q∗1 ) ≤
i=1 2n (n ≥ 3). Then G(LB ) ∼
= G((LB )∂ ) if and only if LB is a Boolean
deg(q∗2 ) ≤ · · · ≤deg(q∗n ) and δ(G(P)) =deg(q∗1 ) = (|C1 | − 1), lattice.
where q∗i is the pseudocomplement of an atom qi of P, ∀i.
Proof. Let LB be a Boolean lattice, and (LB )∂ denotes the dual

n of LB . Clearly, (LB )∂ is also a Boolean lattice. Then it is easy to
Let P = Ci with |Ci | = mi + 1. Then P is a 0-distributive observe that LB ∼ = (LB )∂ under the map x → x∗ , where x∗ is
i=1
lattice with n atoms. By Theorem 4.4, [P] is a Boolean lattice with the pseudocomplement which is also the unique complement of
2n elements. Put [P] = L. Clearly, qi = (0, . . . , 0, qi , 0, . . . 0) ∈ P x ∈ LB . Hence G(LB ) ∼ = G((LB )∂ ).
are the only atoms of P, where qi is the atom of the chain Ci . By Conversely, assume that G(LB ) ∼ = G((LB )∂ ). By Lemma 4.7,
. 
n n
Lemma 4.6, |[qi ]= | = |Ci |−1 = mi for every i. Note that [qi ] are we have G(LB ) ∼ = G( Ci ), where P = Ci . Let 2 ≤ |C1 | ≤
the atoms of L for every i. Now, replace each of the atom [qi ] of i=1 i=1
|C2 | ≤ · · · ≤ |Cn |.
L by the chain of length mi and follow the procedure to replace
We claim that |Ci | = 2 for all i and n ≥ 3.
each nonzero, non-unit element of L by the chains of suitable
Let q1 , q2 , . . . , qn be the all atoms of LB and q∗i are the pseudo-
length mentioned in the Definition 4.2 to get the blow-up LB . It is
complements of qi in LB . It can be observed that q∗i are the dual
an easy verification that G(LB ) ∼
= G(P). Conversely, if the blow-
atoms of LB . Then by the construction of LB and Lemma 4.6, we
up LB of a Boolean lattice L ∼ = 2n is given such that the atoms
have δ(G(LB )) = deg(q∗1 ) = |C1 | − 1.
of L are replaced by the chains of length mi , then there exists a
n 
n−1
poset P = Ci with |Ci | = mi + 1 such that G(LB ) ∼ = G(P). Now, we claim that δ(G((LB )∂ )) = deg(qn ) = (|Ci | − 1).
i=1
i=1
Thus, we have: In (LB )∂ , q∗1 , q∗2 , . . . , q∗n are the atoms of (LB )∂ . Then
by Theorem 4.4, [(LB )∂ ] is also a Boolean lattice, where

n . . .
Lemma 4.7. Let P = Ci with |Ci | = mi + 1 and LB be the [q∗1 ]= , [q∗2 ]= , . . . , [q∗n ]= are the atoms of [(LB )∂ ].
i=1 In the construction of LB , we are replacing the atom qi with
blow-up of a Boolean lattice L ∼ = 2n such that atoms of L are the chain of cardinality |Ci | − 1 and the element q∗i is replaced
replaced by the chains of length mi . Then G(LB ) ∼
= G(P). by the chain of cardinality nj=1,j=i (|Cj | − 1).
As 2 ≤ |C1 | ≤ |C2 | ≤ · · · ≤ |Cn | and by the construction of
. . . .
Example 4.8. The following example illustrates Lemma 4.7 using L , we have |[q∗n ]= | ≤ |[q∗n−1 ]= | ≤ · · · ≤ |[q∗1 ]= | with |[q∗k ]= | =
B
 n
the blow-up of a Boolean lattice L (see Figures 2 and 3). (|Ci | − 1) for 1 ≤ k ≤ n.
i=1,i =k

n

n
Remark 4.9. Let R be a finite reduced ring. Then R = Fi , Since G(LB ) ∼
= G( Ci ), the vertex of G(LB ) can be consider
i=1 i=1
where Fi s are fields. Then by Lemma 4.7 and Theorem 3.12, we as an n-tuple. Let x = (x1 , x2 , . . . , xn ), y = (y1 , y2 , . . . , yn ) ∈
V([(LB )∂ ]). Then x is adjacent to y in G([(LB )∂ ]) if and only if
yi = 1 whenever xi = 0.
By our notation, if an atom q∗i of (LB )∂ is in [q∗i ] =

.= 1 if i = j
[(z1 , z2 , . . . , zn )] , then zj = and the pseu-
0 if i = j
.
docomplement of an atom q∗i of (LB )∂ is qi ∈ [qi ]= =

=
. 0 if i = j
[(y1 , y2 , . . . , yn )] , where yj = for i ∈
1 if i = j
{1, 2, . . . , n}.
Since [(LB )∂ ] is a Boolean, and in a Boolean lattice, pseudo-
complementation is nothing but the unique complementation,
.
we have [qi ]= are the only pendent vertices in G([(LB )∂ ])
Figure 3. Poset and its zero-divisor graph.
150 P. GADGE, N. KHANDEKAR, AND V. JOSHI

.
which are adjacent to [q∗i ]= . Hence in G((LB )∂ ), deg(qi ) = 5. Applications
. 
n
|[q∗i ]= | = (|Cj | − 1). This gives that deg(q1 ) ≥ deg(q2 ) ≥ In this section, we provide some applications of zero-divisor
j=1,j=i graphs of ordered sets to comaximal graphs. In [6, Theorem 4.1],
· · · ≥deg(qn ). Devhare et al. proved the following result.
Let x ∈ V(G(LB )∂ ). Then there exists a dual atom qi of
(L )∂ for some i such that x ≤ qi in (LB )∂ . Thus, deg(x) ≥
B Theorem 5.1 (Devhare, Joshi and LaGrange [6, Theorem 4.1]). If

n−1 P is a finite pseudocomplemented poset, then Gc (P) is a Hamilto-
deg(qi ) ≥ deg(qn ) = (|Ci | − 1). Hence δ(G(LB )∂ ) = nian graph if and only if P has at least three atoms.
i=1

n
deg(qn ) = (|Ci | − 1), where qn is the pseudocomplement Sharma and Gaur [26, Theorem 3.5] proved that if
i=1,i =k |Max(R)| ≥ 3, then (2 (R))c is Hamiltonian. From
of the atom q∗ in (LB )∂ .
n Corollary 3.9, it follows that L is a pseudocomplemented
As G(LB ) ∼
= G((LB )∂ ) and hence δ(G(LB )) = δ(G((LB )∂ )). lattice. Hence, the following result is an easy consequence of the

n−1
Theorems 5.1 and 3.16, which strengthen the result of Sharma
Then we must have |C1 | − 1 = (|Ci | − 1). This gives that
i=1 and Gaur [9, Theorem 3.5].
|C1 | = |C2 | = · · · = |Cn−1 | = 2.
Now, we claim that |Cn | = 2. Suppose |Cn | = m, where Corollary 5.2. Let R be a finite ring. Then (2 (R))c is Hamilto-
m ≥ 2. It is known that in the zero-divisor graph of a Boolean nian if and only if |Max(R)| ≥ 3.
lattice, the set of all atoms forms a unique largest-size clique.
. The following result is an immediate consequence of Theo-
Hence the set of all atoms [qi ]= (i = 1, 2, . . . , n) forms a
complete induced subgraph of G([LB ]), therefore the vertices rems 5.1 and 3.16.
. . .
of equivalence classes [q1 ]= , [q2 ]= , . . . , [qn ]= of G(LB ) forms Corollary 5.3. Suppose R is an Artinian reduced ring. Then
an induced complete n-partite subgraph K1,...,1(n−1 times),m−1 , (CAR )c is Hamiltonian if and only if R has at least three maximal
as |[qi ]| = 1 for i = 1, . . . , (n − 1) and |[qn ]| = m − 1. ideals.
. . .
Also, the vertices of equivalence classes [q∗1 ]= , [q∗2 ]= , . . . , [q∗n ]=
of G((LB )∂ ) form an induced complete n-partite subgraph Theorem 5.4 (Joshi [14, Corollary 2.11]). Let G(P) be the zero-
K(m−1),...,(m−1)(n−1−times),1 , as |[q∗i ]| = m−1 for i = 1, . . . , (n− divisor graph of an atomic poset P with finitely many n atoms.
1) and |[q∗n ]| = 1. As G(LB ) ∼ = G((LB )∂ ) and n ≥ 3. Then we Then χ (G(P)) = ω(G(P)) = n.
must have m − 1 = 1, and this gives that m = 2. Therefore,
we have |C1 | = |C2 | = · · · = |Cn | = 2. Thus, LB is By Theorems 3.16 and 5.4, we have the following corollary.
Boolean.
Corollary 5.5. Let R be a finite commutative ring with identity.
Note that Theorem 4.10 fails in the case of n = 2. Let P = Then χ (2 (R)) = ω(2 (R)) = n, where n is the number of
Cl × Cm . Then clearly, LB ∼ = (LB )∂ and G(LB ) ∼
= G((LB )∂ ) ∼
= maximal ideals of a ring R.
B
K(l−1),(m−1) , but L is not a Boolean lattice.
As 2 (R) = (R) \ U(R) is the subgraph of (R) induced
Corollary 4.11 (Ghoraishi and Samei [10, Theorem 2.2]). Let on the set of non-units of R, and 2 (R) = 2 (R) \ J(R) is the
R be a commutative finite ring with | Max(R)| = n ≥ 2. The subgraph of (R) induced on the set of non-units of R which
following statements are equivalent: are not in J(R), the Jacobson radical of R, we have:

(1) 0 (R) ∼
= 2 (R). Corollary 5.6. Let R be a commutative ring with identity. Then
(2) Either R is a Boolean ring or R ∼
= F1 ×F2 , where Fi ’s are fields. 2 (R) = G(L) + Im , where Im is the complement of the complete
graph Km and m = |J(R)|.
Proof. Suppose R is finite and 0 (R) ∼ = 2 (R). Then
 Corollary 5.7. Let R be
|V(0 (R))| = |V(2 (R))|, which further gives |Z∗ (R)| = a commutative ring with identity. Then
|Z(R) − J(R)|. This gives that J(R) = (0). Thus, R is (R) = (G(L) + Im ) K , where Im is the complement of the
reduced. Then by Theorems 3.12 and 3.16, we have G(L) ∼ = complete Km , where m = |J(R)| and  = |U(R)|.
G(L∂ ). By Remark 4.9 and Theorem 4.10, we have L as a
From Corollaries 5.6 and 5.7, it is clear that 2 (R) and
Boolean lattice. Hence, R is a Boolean ring that follows from
2 (R) are weakly perfect, as G(L) is weakly perfect by The-
Theorem 3.14.
orem 5.4. Hence, Theorem 2.3 of Sharma and Bhatawadekar
Conversely, assume that R is a Boolean ring. By Theo-
follows immediately. Thus it is evident from the above
rem 3.14, L is Boolean. Hence Remark 4.9 and Theorem 4.10
results that many properties of the comaximal graph and co-
yield G(L) ∼ = G(L∂ ). Therefore by Theorems 3.12 and 3.16,
annihilating graph of a commutative ring with identity can be
we get 0 (R) = 2 (R). Now, if R ∼ = F1 × F2 , where
deduced.
Fi ’s are fields. Then clearly, L ∼= L∂ . This implies that
We close the paper by providing a characterization of perfect
G(L) ∼ = G(L∂ ). Hence by Theorems 3.12 and 3.16, we have
∼ comaximal graphs.
0 (R) = 2 (R).
 AKCE INTERNATIONAL JOURNAL OF GRAPHS AND COMBINATORICS 151

Definition 5.8 (Gallai [8]). A graph G is perfect if the clique [2] Amjadi, J., Alilou, A. (2018). The co-annihilating graph of a commu-
number ω(H) is the same as the chromatic number χ (H) for every tative ring. Discrete Math. Algo. Appl. 10(1): 1850013 (1–12).
induced subgraph H of G. [3] Anderson, D. F., Asir, T., Badawi, A., Tamizh Chelvam, T. (2021).
Graphs from Rings. Cham: Springer.
[4] Anderson, D. F., LaGrange, J. D. (2012). Commutative Boolean
The following result essentially follows from [18, Corollary monoids, reduced rings, and the compressed zero-divisor graph. J.
4.11] (see also [24]). Pure Appl. Algebra 216: 1626–1636.
[5] Beck, I. (1988). Coloring of commutative rings. J. Algebra 116: 208–
Theorem 5.9 (Khandekar and Joshi [18, Corollary 4.11]). Let L 226.
be a finite 0-distributive lattice. Then the zero-divisor graph G(L) [6] Devhare, S., Joshi, V., LaGrange, J. D. (2019). Eulerian and Hamil-
is perfect if and only if ω(G(L)) ≤ 4. tonian complements of zero-divisor graphs of pseudocomplemented
posets. Palest. J. Math. 8(2): 30–39.
[7] Devhare, S., Joshi, V., LaGrange, J. D. (2019). On the connectedness
The following result follows from Theorems 3.16 and 5.9. of the complement of the zero-divisor graph of a poset. Quaest. Math.
42(7): 1–13.
Theorem 5.10. Let R be a finite commutative ring with identity. [8] Gallai, T. (1958). Maximum-minimum Sätze über Graphen. Acta
Then the co-maximal graph (R) is perfect if and only if 2 (R) is Math. Acad. Sci. Hungar. 9: 395–434.
perfect if and only if 2 (R) is perfect if and only if G(L) perfect if [9] Gaur, A., Sharma, A. (2013). Maximal graph of a commutative ring.
and only if |Max(R)| ≤ 4. Moreover, if R is an Artinian reduced Int. J. Algebra 7: 581–588.
[10] Ghoraishi, M., Samei, K. (2018). Comparison of graphs associated to
ring with identity, then CAR is a perfect graph if and only if a commutative Artinian ring. Proc. Math. Sci. 128: 56.
|Max(R)| ≤ 4. [11] Grätzer, G. (1998). General Lattice Theory. Basel: Birkhäuser.
[12] Halaš, R., Jukl, M. (2009). On Beck’s coloring of posets. Discrete Math.
Acknowledgments 309: 4584–4589.
[13] Halaš, R. (1993). Pseudocomplemented ordered sets. Arch. Math.
The authors express their gratitude to the referees for their valuable sug- (Brno) 29: 153–160.
gestions that enhanced the paper’s presentation. A segment of the findings [14] Joshi, V. (2012). Zero divisor graph of a poset with respect to an ideal.
from this study was shared by the third author during the ICRAGAA 2023 Order 29: 499–506.
conference held in Thrissur, Kerala, India. The third author extends his [15] Joshi, V., Khiste, A. (2014). The zero divisor graphs of Boolean poset.
thanks to the conference organizers and Ambat Vijayakumar. Math. Slovaca 64: 511–519.
[16] Vinayak Joshi, H. Y. P., Waphare, B. N. (2014). The graph of equiv-
alence classes of zero divisors. Int. Sch. Res. Not. 2014: Article ID
Authors contributions
896270, 7 pp.
All three authors contributed equally to the study of the comaximal graph [17] Khandekar, N., Joshi, V. (2020). Zero-divisor graphs and total coloring
of a commutative ring. All three authors read and approved the final version conjecture. Soft Comput. 24: 18273–18285.
of the manuscript. [18] Khandekar, N., Joshi, V. (2023). Chordal and perfect zero-divisor
graphs of posets and applications to graphs associated with algebraic
structures. Math. Slovaca 73(5): 1099–1118.
Disclosure statement [19] Larmerová, J., Rachůnek, J. (1988). Translations of distributive
and modular ordered sets. Acta. Univ. Palacki. Olomouc. 91: 13–
The authors declare that there is no conflict of interests regarding the 23.
publishing of this paper. [20] LaGrange, J. D., Roy, K. A. (2013). Poset graphs and the lattice of graph
annihilators. Discrete Math. 313(10): 1053–1062.
Funding [21] Lu, D., Wu, T. (2010). The zero-divisor graphs of posets and an
application to semigroups. Graphs Comb. 26: 793–804.
First author: None. Second author: The author is financially supported by [22] Maimani, H. R., Salimi, M., Sattari, A., Yassemi, S. (2008). Comaximal
the Council of Scientific and Industrial Research (CSIR), New Delhi, via graph of commutative rings. J. Algebra 319: 1801–1808.
Senior Research Fellowship Award Letter No. 09/137(0620)/2019-EMR-I. [23] Moconja, S. M., Petrović Yassemi, Z. Z. (2011). On the structure
Third author: The author is financially supported by Science and Engineer- of comaximal graphs of commutative rings with identity. Bull. Aust.
ing Research Board (DST) via Project CRG/2022/002184. Math. Soc. 83:11–21.
[24] Patil, A., Waphare, B. N., Joshi, V. (2017). Perfect zero-divisor graphs.
Discrete Math. 340: 740–745.
Data availability statement [25] Sharma, P. D., Bhatwadekar, S. M. (1995). A note on graphical repre-
sentation of rings. J. Algebra 176: 124–147.
Data sharing does not apply to this article as no datasets were generated or [26] Sharma, A., Gaur, A. (2016). Hamiltonian property of a maximal
analyzed during the current study. graph and the chromatic number of its line graph. JP J. Algebra,
Number Theory Appl. 38:589–607.
References [27] Wu, T., Ye, M., Lu, D., Yu, H. (2013). On graphs related to comaximal
ideals of a commutative ring. ISRN Comb. 2013: 7 pp.
[1] Afkhami, M., Barati, Z., Khashyarmanesh, K. (2012). Planar zero [28] Ye, M., Wu, T., Liu, Q., Yu, H. (2014). Implements of graph
divisor graphs of partially ordered sets. Acta Math. Hung. 137(1– blow-up in co-maximal ideal graphs. Commun. Algebra 42: 2476–
2): 27–35. 2483.