mathematics

Article
Class of Crosscap Two Graphs Arising from Lattices–I
T. Asir 1, * , K. Mano 2 , Jehan A. Al-Bar 3 and Wafaa M. Fakieh 3

1 Department of Mathematics, Pondicherry University, Pondicherry 605 014, Tamil Nadu, India
2 Department of Mathematics, Fatima College, Madurai 625 018, Tamil Nadu, India
3 Department of Mathematics, Faculty of Science, King Abdulaziz University, Jeddah 21461, Saudi Arabia
* Correspondence: asirjacob75@gmail.com

Abstract: Let L be a lattice. The annihilating-ideal graph of L is a simple graph whose vertex set
is the set of all nontrivial ideals of L and whose two distinct vertices I and J are adjacent if and
only if I ∧ J = 0. In this paper, crosscap two annihilating-ideal graphs of lattices with at most four
atoms are characterized. These characterizations provide the classes of multipartite graphs, which
are embedded in the Klein bottle.

Keywords: crosscap; Klein bottle; lattice; annihilating-ideal graph

MSC: 05C75; 05C25; 05C10; 06A07; 06B99

1. Introduction
According to the well-known theorem of Kuratowski and Wagner, a graph is planar
if and only if it does not contain either of the two forbidden graphs K5 and K3,3 . The Graph
Minor Theorem of Robertson and Seymour [1] can be considered a powerful generaliza-
tion of Kuratowski’s Theorem. In particular, their theorem, which is the “deepest” and
“most important” result in the arena of graph theory [2], implies that each graph property,
no matter what, is characterized by a corresponding finite list of graphs. Thus, for surfaces
Citation: Asir, T.; Mano, K.; (both orientable and non-orientable) in general, it is known that the set of forbidden mi-
Al-Bar, J.A.; Fakieh, W.M. Class of
nors is finite [3]. An analogous characterization for the embedding of graphs on surfaces is
Crosscap Two Graphs Arising from
known for the crosscap one surface (Möbius strip) where 103 forbidden subgraphs (equiv-
Lattices–I. Mathematics 2023, 11, 1553.
alently 35 forbidden minors) are characterized [4,5]. So, an open problem is to determine
https://doi.org/10.3390/
the several forbidden subgraphs for crosscap two surfaces (the Klein bottle). In this se-
math11061553
quel, finding a family of graphs that has a crosscap two is an interesting one. Note that
Academic Editors: Irina Cristea, most of the 103 graphs contain a subgraph that is homeomorphic to K3,3 , and multipartite
Hashem Bordbar and Carsten graphs play a vital role in finding these 103 forbidden subgraphs for the projective plane.
Schneider It is worth mentioning that the crosscap value of bipartite and tripartite graphs are well
Received: 22 December 2022
known (refer to Proposition 1). The main goal of this paper is to identify a large class of
Revised: 4 March 2023 crosscap two r-partite graphs where r ≥ 4.
Accepted: 16 March 2023 Let us introduce the concept of the annihilating-ideal graph of a lattice, a type of multi-
Published: 22 March 2023 partite graph. Note that the annihilating-ideal graph is an extension of the concept of the
zero-divisor graph. The idea of the zero-divisor graph of a ring structure is due to Beck [6].
In 2009, Halaš et al. [7] introduced the zero-divisor graph for a partially ordered set, and,
in 2012, Estaji et al. [8] extended the concept of the zero-divisor graph to an arbitrary
Copyright: © 2023 by the authors. finite bounded lattice. For a clear exposition of the work completed in the area of zero-
Licensee MDPI, Basel, Switzerland.
divisor graphs and their related areas, the reader is referred to the book by Anderson
This article is an open access article
et al. [9]. In 2011, Behboodi et al. [10] defined and investigated the ideal theoretic ver-
distributed under the terms and
sion of the zero-divisor graph, called the annihilating-ideal graph of a ring, and, thereafter,
conditions of the Creative Commons
many facts about zero-divisors were expressed in the language of ideals. The concept of
Attribution (CC BY) license (https://
an annihilating-ideal graph of a ring was extended to an arbitrary lattice by Afkhami et
creativecommons.org/licenses/by/
4.0/).
al. [11] in 2015. The annihilating-ideal graph of a lattice L, denoted by AG(L), is defined to

Mathematics 2023, 11, 1553. https://doi.org/10.3390/math11061553 https://www.mdpi.com/journal/mathematics

Mathematics 2023, 11, 1553 2 of 26

be a simple graph whose vertex set is the set of all non-trivial ideals of L, and whose two
distinct vertices I and J are adjacent if and only if I ∧ J = 0. The hope when studying the
annihilating-ideal graph of a lattice is that the graph theoretic properties of the graph from
the lattice will help us to better understand the lattice theoretic properties of the lattice.
One of the most important topological properties of a graph is its genus, which can
be orientable or non-orientable (crosscap). The genus of graphs associated with algebraic
structures has been studied by many authors (see [12–17]). The planar zero-divisor graph
was first explicitly characterized by Smith [18], and the characterization of commutative
rings with projective zero-divisor graphs was obtained by Chiang-Hsieh [15]. In 2019,
Asir et al. [12] enumerated all commutative rings whose zero-divisor graph has a crosscap
two. The planar and crosscap one annihilating-ideal graph of lattices were characterized
by Shahsavar [19] and Parsapour et al. [20], respectively. Additionally, whether the line
graph associated with the annihilating-ideal graph of a lattice is planar or projective was
characterized by Parsapour et al. [21]. Moreover, the authors of [22] characterized all lat-
tices L whose line graph of AG(L) is toroidal.
Now, this paper aims to classify lattices with a number of atoms less than or equal
to four whose annihilating-ideal graph can be embedded in the non-orientable surfaces
of crosscap two. The main results of this paper are Theorems 2, 3, and 5, in which we
have obtained our classifications. As a result, this classification provides a large class
of r-partite graphs that can be embedded in the Klein bottle. Further, in the proof of the
main theorems, we have shown several minimal r-partite graphs that cannot be embedded
in the Klein bottle. Possibly, these graphs may be realized as forbidden subgraphs for
crosscap two surfaces (refer to Example 1). Further, in order to cover the missing cases in
the proof of Theorem 2.6 [20], which affects the statement of the corresponding theorem,
the modified version is included as Theorem 4.

2. Preliminaries
In this section, we present the definitions and results needed to prove the main re-
sults in the subsequent sections. First, we recall some definitions and notations on lattices.
A lattice is an algebra L = (L, ∧, ∨), where ∧ and ∨ are the binary operations, satisfying
the following conditions: for all a, b, c ∈ L
1. a ∧ a = a, a ∨ a = a;
2. a ∧ b = b ∧ a, a ∨ b = b ∨ a;
3. ( a ∧ b) ∧ c = a ∧ (b ∧ c); a ∨ (b ∨ c) = ( a ∨ b) ∨ c;
4. a ∨ ( a ∧ b) = a ∧ ( a ∨ b) = a.
According to [23] (Theorem 2.1), we can define an order ≤ on L as follows: for any
a, b ∈ L, we set a ≤ b if and only if a ∧ b = a. Then (L, ≤) is an ordered set in which
every pair of elements has the greatest lower bound (glb) and the least upper bound (lub).
Conversely, let P be an ordered set such that, for every pair a, b ∈ P, glb( a, b) and lub( a, b)
belong to P. For each a and b in P, we define a ∧ b = glb( a, b) and a ∨ b = lub( a, b). Then
( P, ∧, ∨) is a lattice. A lattice L is said to be bounded if there are the elements 0 and 1 in L
such that 0 ∧ a = 0 and a ∨ 1 = 1, for all a ∈ L. Clearly, every finite lattice is bounded. Let
(L, ∧, ∨) be a lattice with a least element 0 and I be a non-empty subset of L. Then I is
said to be the ideal of L, denoted by I ⊴ L,
1. For all a, b ∈ I, a ∨ b ∈ I.
2. If 0 ≤ a ≤ b and b ∈ I, then a ∈ I.
In a lattice (L, ∧, ∨) with a least element 0, an element a is called an atom if a ̸= 0, and,
for an element x ∈ L, the relation 0 ≤ x ≤ a implies that either x = 0 or x = a. We denote
the set of all atoms of L by A(L). For basic facts about lattices, we refer the reader to [24].
Next, we recall the following terms regarding graph embedding. For the non-negative
integers ℓ and k, let Sℓ denote the sphere with ℓ handles, and Nk denote a sphere with k
crosscaps attached to it. Note that every connected compact surface is homeomorphic to
Sℓ or Nk for some non-negative integers ℓ and k. The genus γ( G ) of a simple graph G is the
Mathematics 2023, 11, 1553 3 of 26

minimum ℓ such that G can be embedded in Sℓ . Similarly, crosscap number (non-orientable

genus) γ̃( G ) is the minimum k such that G can be embedded in Nk . Note that the projective
space is of crosscap one and the Klein bottle is of crosscap two. If e = xy ∈ E( G ), then the
contraction of e in G, denoted as [ x, y] is the graph obtained from G − xy by identifying ver-
tices x and y to create a new vertex z incident with all edges of G that were incident with
either x or y. We say H is a minor of G, if H can be obtained from G by deleting vertices,
edges, and/or contracting edges. For a graph G, we denote G̃ for the subgraph G − V ′
where V ′ = {v ∈ V | deg(v) = 1}, and we call this graph the reduction of G. For details on
the notion of the embedding of graphs in a surface, we recommend reading [25].
The following three results on the non-orientable embedding of graphs are used fre-
quently in this paper. In what follows, we denote the complete graph with p vertices by
K p , the complete bipartite graph with parts of sizes p and q by K p,q , the complete tripartite
graph with parts of sizes p, q, and r by K p,q,r , and the complete four-partite graph with
parts of sizes p, q, r, and s by K p,q,r,s .

Proposition 1 ([25,26]). Let p, q, r, and s be positive integers greater than or equal to two. Then
{ ⌈ ⌉
( p−3)( p−4)
if p ≥ 3
(a) γ̃(K p ) = 6
3 i f p = 7.
⌈ ⌉
( p−2)(q−2)
(b) γ̃(K p,q ) = 2 .
⌈ ⌉
( p−2)(q+r −2)
(c) γ̃(K p,q,r ) = 2 except for K3,3,3 , K4,4,1 and K4,4,3 . Further,
γ̃(K3,3,3 ) = 3, γ̃(K4,4,1 ) = 4 and⌈ γ̃(K4,4,3 ) = 6. ⌉
( p−2)(q+r +s−2)
(d). If p ≥ q + r, then γ̃(K p,q,r,s ) ≥ 2 .
⌈ ⌉
( p+s−2)(q+r −2)
If p ≤ q + r, then γ̃(K p,q,r,s ) ≥ 2 .

Proposition 2 (([16] Theorem 1.3) (Euler formula)). Let ϕ : G → Nk be a two-cell embedding

of a connected graph G to the non-orientable surface Nk . Then |V | − | E| + | F | = 2 − k, where
|V |, | E|, and | F | are the number of vertices, edges, and faces that ϕ( G ) has, respectively, and k is
the crosscap of Nk .

The following is an easy observation that will be used in the proof of the main theo-
rem.
2| E |
Observation 1. Let G be a simple graph with | E| edges embedded with | F | faces. Then | F|
≥
gr ( G ) where gr ( G ) denotes the length of the shortest cycle in G.

3. Basic Results and Notations

Before going into the classifications, we need to be familiar with the following nota-
tions and observations given by Parsapour and Javaheri in [20].
Notation: ([20]) Let L be a lattice and A(L) = { a1 , a2 , . . . , an } be the set of all atoms. Let
i1 , i2 , . . . , ik be integers with 1 ≤ i1 < i2 < . . . < ik ≤ n. The notation Ui1 i2 ...ik stands for the
following set:
{ }
I ⊴ L : { ai1 , ai2 , . . . , aik } ⊆ I and ai j ∈
/ I for i j ∈ {1, 2, . . . , n} \ {i1 , i2 , . . . , ik } .

The next result provides the structure of AG(L).

Proposition 3. Let L be a lattice with n atoms. Then AG(L) is a 2n − 2-partite graph.

Proof. Let | A(L)| = n. For 1 ≤ i1 < i2 < . . . < ik ≤ n and 1 ≤ j1 < j2 < . . . < jk′ ≤
n, if the index sets {i1 , i2 , . . . , ik } and { j1 , j2 , . . . , jk′ } of Ui1 i2 ...ik and Uj1 j2 ...jk′ respectively,
∪
are distinct, then Ui1 i2 ...ik ∩ Uj1 j2 ...jk′ = ∅. Clearly, V (AG(L)) = Ui1 i2 ...ik .
1≤i1 <i2 <...<ik ≤n
Mathematics 2023, 11, 1553 4 of 26

Therefore, for 1 ≤ i1 < i2 < . . . < ik ≤ n, the set Ui1 i2 ...ik forms a partition of V (AG(L)).
Since 0 ̸= ai1 belongs to every ideal in Ui1 i2 ...ik , no pair of distinct vertices in Ui1 i2 ...ik are
adjacent in AG(L). Note that the number of distinct Ui1 i2 ...ik s is 2n − 1. This, together with
the fact that every vertex in U12...n is isolated in AG(L), implies that AG(L) is a 2n − 2-
partite graph.

According to the abovementioned result regarding the structure of AG(L), in order

to identify the crosscap two r-partite graph or to classify the forbidden r-partite graphs of
a non-orientable surface of order two for some 3 ≤ r ∈ N, one may be interested in finding
all crosscap two annihilating-ideal graphs. This is the main objective of this paper.
We shall also need the following notations:
Notations: Before proving our main results, the following points are assumed for conve-
nience in notations and clarity in proofs. Let us take | A(L)| = n.
• To avoid repetition, we assume |U1 | ≥ |U2 | ≥ . . . ≥ |Un |.
• We denote the vertices of the set Ui1 i2 ...ik by { Ii1 i2 ...ik , Ii′1 i2 ...i , Ii′′1 i2 ...i , . . .}.
k k
• For an integer p, an integer different from p will be denoted by p′ .
• For the sake of convenience, we shall denote U(i1 i2 ...ik )c = Uj1 j2 ...jℓ where
j1 , j2 , . . . , jℓ = {1, 2, . . . , n} \ {i1 , i2 , . . . , ik } and the notation U(i1 i2 ...ik )c exists only when
Ui1 i2 ...ik ̸= ∅.
• The edge between the two vertices I and J is denoted by ( I, J ).
• The notations | F | and f i denote the number of faces and number of i-gons in an em-
bedding of G in Nk , respectively.
• There may be sets Ui1 i2 ...ik such that each vertex of Ui1 i2 ...ik is isolated, ends, or is
adjacent to exactly two ends of an edge in AG(L). In such places, the vertices of
Ui1 i2 ...ik do not affect the crosscap number of AG(L), which leads to ignoring the set
Ui1 i2 ...ik from the corresponding embedding. This fact is used throughout the article
and is sometimes not explicitly pointed out.
• For convenience in any drawing, we provide a particular type of N2 -embedding of
AG(L). This means that instead of drawing graphs for the case Uij with 1 ≤ i ≤ j ≤ 3,
we assume i = 1 and j = 2 in figures. Additionally, the notation · · · is used to denote
the possibility of embedding any number of vertices.
We show a few simple, but useful, properties of a crosscap on AG(L). We now state
and prove the following lemma, which provides a subgraph and super-graph structure of
AG(L).

Lemma 1. Let L be a lattice, | A(L)| = n, and n ≥ k ∈ N. Let αi1 i2 ...ik = |Ui1 i2 ...ik |, λ =
max{αi1 i2 ...ik } for all 1 ≤ i1 < i2 < . . . < ik ≤ n. Then
(a). Kα1 ,α2 ,...,αn is a subgraph of AG(L).
(b). K(2n −2)(λ) is a super-graph of AG(L).
∪
Proof. Let H be the induced subgraph of AG(L), induced by the vertex subset in=1 Ui . It
is clear that no two distinct vertices in Ui are adjacent, and every vertex in Ui is adjacent
to all of the vertices of Uj for i ̸= j in AG(L). Thus H = Kα1 ,α2 ,...,αn .
∪
The second part follows from the facts that V (AG(L)) = Ui1 i2 ...ik ; the number of
n n n
1) = 2 − 2; and
vertex subsets Ui1 i2 ...ik , except U12...n , in V (AG(L)) is ( 1 ) + ( 2 ) + . . . + (n− n

λ = max{αi1 i2 ...ik }.

We are now in the position to provide a lower bound for the crosscap of AG(L).
Applying Proposition 1c,d in the first part of the above lemma, we obtain the following
result.

Theorem 1. Let L be a lattice, | A(L)| = n ≥ 3, and |U1 | ≥ |U2 | ≥ . . . ≥ |Un |.

Mathematics 2023, 11, 1553 5 of 26
Mathematics 2023, 1, 0 5 of 25

⌈ ⌉
(|U |−2)(|U2 |+|U3 |−2)
(a). If n = 3, then γ̃(AG(L)) ≥ ⌈ 1 2 . ⌉Moreover, the equality holds whenever
(|U1 |−2)(|U2 |+|U3 |−2)
(a). UIf ijn ==∅3,for
then ≤ i ≤ j ≤≥
all 1γ̃(AG(L)) 3. . Moreover, equality holds whenever
⌈ 2 ⌉
Uij = ∅ for all 1 ≤ i ≤ j ≤ 3. (|U1 |−2)(|U2 |+|U3 |+|U4 |−2)
⌈ 2 ⌉ i f |U1 | ≥ |U2 | + |U3 |
(b). If n ≥ 4, then γ̃(AG(L)) ≥ ⌈ (|U1 |−2)(|U2 |+|U3 |+|U4 |−2) ⌉ i f |U1 | ≥ |U2 | + |U3 |
 (|U1 |+|U4 |−2)(| 2 U2 |+|U3 |−2)
(b). If n ≥ 4, then γ̃(AG(L)) ≥ ⌈ 2 ⌉ i f |U1 | < |U2 | + |U3 |.
 (| U1 |+| U 4 |− 2 )(| U2 |+|U3 |−2)
i f |U1 | < |U2 | + |U3 |.
2
We now enter into the core part of the paper. We first observe that AG(L) is totally
disconnected
We now enter wheninto | A(L)| = 1, part
the core and of AG(L)
the paper.contains We K 7 asobserve
first a subgraph when | A
that AG(L) is(L)| ≥
totally
disconnected when | A(L)| = 1 and AG(L) contains K7 as a7 subgraph when | A(L)| ≥ 7.
7. Further, according to Proposition 1a, the crosscap of K is three. Thus, one obtains
the following
Further result, which
by Proposition provides
1a, the crosscapa of bound
K7 is 3. forThus
the number of atoms
one gets the resultLwhich
in lattice
following with
γ̃(AG(L))
provides = 2. for the number of atoms in lattice L with γ̃(AG(L)) = 2.
a bound

Proposition 4. Let L be
be aa lattice.
lattice. IfIfthe
thecrosscap
crosscapofofannihilating
the annihilating-ideal AG(L)
ideal graphgraph AG(L) is then
is two, two,
2 ≤ |2A≤
then | A(L)|
(L)| ≤ 6. ≤ 6.

We start the the characterization

characterization by
by analyzing
analyzing the simple case
a simple case that | A(L)|
= 2. If | A(L)| = 2, then Theorem 2.6 [20] implies that AG(L) ∼ = K|U11 |,|U22 | ,and
andso
so
⌈ ⌉
(|U1 | − 2)(|U2 | − 2)
γ̃(AG(L)) =
2

whenever |U1 |, |U2 | ≥ 2. Now,

Now a simple calculation has yielded thethe following
followingresult,
result which
characterized lattice L with acrosscap
crosscaptwo AG(L)inincase
twoAG(L) theof
case of | A=
| A(L)| 2. = 2.
(L)|

Theorem 2. Let L be a lattice and | A(L)| = 2. Then γ̃(AG(L)) = 2 if and only if |U1 | =
|U2 | = 4 or |Ui | = 3 and |Uj | ∈ {5, 6} where i, j ∈ {1, 2} with i ̸= j.

To finish this section we

we show
show two
two results
results,that willwill
which be used to prove
be used the main results.
to prove
The graphs given in in Figures
Figures 11 and
and22play
playaavital
vitalrole
roleinincharacterizing
characterizinglattice
a lattice with
with cross-
crosscap
cap
two two annihilating-ideal
Annihilating-ideal graphs,
graphs andand, therefore,
so we draw the we drawwith
graph the graph with its embedding
its embedding in the first
in the first result.
result.

Lemma 2. For the graphs H1 and H22, as

as shown in Figures 1 and 2, we have γ̃( H1 ) = γ̃( H2 ) = 2.
shown in

u 1 u 2 u
b
3 u 4 ub 3 ub 4
b b b

v1 v3 ub 1
b b b v2

v4 v1
v2 b b b b

b b b b
u2
v1 v2 v3 v4
b b
u3 u4

(a). The graph H1 (b). An N2 -embedding of H1

Figure 1. The graph H11 and its N22 -embedding.
 Mathematics2023,
Mathematics
Mathematics 2023,1,1,11,
2023, 00 1553 666of
of25
of 26
25

uu44 uu 55 uu66 uu44

uu11 uu22 uub 3b 3 uu44 b b uub 5b 5 uu66 b b b b b b b b
b b b
b b b

uu22 uu11
vv11 b b b b b b b b vv22

uu33 vv33
vv22 b b b b b b b b vv11
b b b b b b
vv11 vv22 vv33 b b b b b b b b
uu44 uu55 uu66 uu44

(a).
(a). The graph H
The graph H22 (b). An N
(b). An N22-embedding of H
-embedding of H22
Figure 2. The
Figure2. graph H
Thegraph H22 and its N
andits N22-embedding.
-embedding.
Figure 2. The graph H2 and its N2 -embedding.
The graphs H
The graphs and H
H33 and H44 given
given in
in Figure
Figure 33 play
play aa vital
vital role
role in
in our
our main
main theorems.
theorems.
The graphs H3 and H4 given in Figure 3 play a vital role in our main theorems.
Lemma 3. For
Lemma 3. For the
the graphs
graphs H
H33 and
and H
H44 as
as shown
shown in
in Figure
Figure 3,
3, we
we have H33)) ≥
have γ̃γ̃((H ≥ 33 and H44)) ≥
and γ̃γ̃((H ≥ 3.
3.
Lemma 3. For the graphs H3 and H4 , as shown in Figure 3, we have γ̃( H3 ) ≥ 3 and γ̃( H4 ) ≥ 3.

ub ub 11 ub ub 22 ub ub 33 ub ub 44 ub ub 11 ub ub 22 ub ub 33 ub ub 44

b b b b b b b b b b b b b b b b

vv11 vv22 vv33 vv44 vv11 vv22 vv33 vv44

(a). Graph H
(a). Graph H33 (b). Graph H
(b). Graph H44
Figure 3.3.The
Figure3.
Figure The graphsH
Thegraphs
graphs andH
333and
HH and 44.4. .
HH

′′ ′ ∼
∼
Proof.(a).
Proof.
Proof. (a). Consider
(a). Considerthe
Consider the subgraphH
thesubgraph
subgraph HH ′′ ′ =
333 = =H HH333−−−{{{uuu11}1}}.. . Clearly
ClearlyH
Clearly HH 333 ==∼
=KKK777− −−eeewherewhereeee=
where ==
((u(uu22,2,u,uu33)3))and
, andthere thereare are1313faces
facesin anyN
ininany ofH ′′ ′
and there are 13 faces any NN 22-embedding
-embeddingof
2-embedding of HH333in ofwhich
in which12
which 12are
12 aretriangular
are triangular,and
triangular and
and
one1 isis
one is rectangular.
rectangular.Now,
rectangular. Now
Nowwe we
wetry try
trytoto recover
recover an an NN222-embedding
-embedding of of H H33 by insertinguuu111with
by inserting
inserting withits
with its
its
edges.
edges.Since
edges. Sinceuuu111is
Since isisadjacent
adjacentto
adjacent tofour
to fourvertices
four verticesof
vertices ofofH
HH ′′,′, u
,u u should
should
should be
be
be inserted
inserted
inserted into
into
into the
the
the rectangular
rectangular
rectangular
333 1 11
face
faceof
face ofof H H3′3′.. However,
But
But every
everyall vertex
vertex
vertices of H
of H3′3′ H
of is3′ adjacent
is adjacent
are adjacent to
to eacheach
to each other
other other, except
exceptexcept uu22 for
and
and u2uuand
33 so
southat
that
3 , so
thetherectangular
the rectangularface
rectangular faceof
face of HH3′3′ must
must contain
contain either eitheruuu222or oruuu33,3,,which
or aa contradiction
contradiction
is in contradiction to uu22 and
to andtouuu332not
not
and
belonging
belonging
u3 not belonging to
to the
the neighborhood
neighborhood
to the neighborhood set
set ofof uu11set .. Therefore
Therefore
of u1 . Therefore, H33)) ≥
γ̃γ̃((H ≥γ̃3. ( H3 ) ≥ 3.
3.
(b).
(b).Apply
(b). Applyaaasimilar
Apply similarargument
similar argumentas
argument asin
as inin(a)(a)for
(a) forthe
for subgraphH
thesubgraph
the subgraph HH′′ ′ =
444= =H HH444−−−{{{uuu11}1}}∼
=∼
=∼
=KKK777−−−2e.
2e.
2e.
Here
Here,notice noticethat thatthe thelargest
largestface facein inany anyN ofH ′′ ′ anduuu111
Here notice that the largest face in any NN 22-embedding
-embeddingof
2-embedding of HH444is isaaaunique
is uniquepentagon
unique pentagon,and
pentagon and
is
isisadjacent
adjacentto
adjacent tothe
to thefive
the verticesvvv11,1,v,vv22,2,v,vv33,3,v,vv444and
fivevertices
five vertices , anduu4u
and 4..4 .

4.
4.4.The
TheCase
The When||A
CaseWhen
Case When |AA(L)|
(L)|=
(L)| ==33 3
Let
Letus
Let usstart
us startthe
start theclassification
the classificationresult
classification resultwith
result with lattice
lattice containing
containing
a lattice containingexactly
exactly 33 atoms.
exactly atoms.
three Note
Note that
atoms. that
Note
the
the following
thatfollowing theorem
theorem
the following provides
provides
theorem aa class
providesclassaof
of multipartite
multipartite
class graphs, which
which are
graphs,graphs,
of multipartite are embedded
whichembedded in
in the
are embedded the
Klein
in thebottle,
Klein Kleinrefer
bottle, refer (refer to11Example
bottleExample
Example for
for illustration.
illustration.
1 for an illustration).

Theorem
Theorem3.
Theorem 3.3.LetLetL
Let LLbe
bebeaalattice
alattice with
with
lattice ||A
with A|(L)| ==33and
(L)|=
(L)|
A and
3, let 11 ≤
letlet
and ≤ i ̸=
1 i≤ j ̸=
̸=i j̸= k≤
̸=j k̸= ≤ ≤
3.Then
k 3. Then γ̃γ̃(AG(L))
3. Then γ̃(AG(L))
(AG(L)) = 2=
= 2
ifif2and
and only
only
if and ifif one
only one of
of the
if one the following
of thefollowing conditions
conditions
following hold:
conditionshold:
hold:
∪
∪∪
(i). |||
(i).
(i). 333
nn= 111U
n== nn|n||=
UU 9,9;there
==9, thereisisisU
there UU with||U
ii iwith
with |Uii||i |=
U ==666and
andU
and ==∅.
UUjkjkjk= ∅.
∅.
∪
∪∪
(ii). |||
(ii).
(ii). 333
nn= 111U
n== nn|n||=
UU ==888,and
andone
and oneof
one ofofthethefollowing
the followingcases
following casesisisissatisfied:
cases satisfied:
satisfied:
[a]There
[a]
[a] ThereisisisU
There UU with
ii i with
with ||U|
U U| |
ii i |
= =
= 666and
and
and||U|
U U |
jkjkjk| |
= =
= 1.
1.
1.
Mathematics 2023, 11, 1553 7 of 26

[b] There exist Ui and Uj such that |Ui | ∈ {5, 4} and |Uj | = 2 with Ujk = ∅.
[c] There exist Ui and Uj such that |Ui | = 4 and |Uj | = 3 with Uik = Ujk = ∅.
[d] There exist Ui and Uj such that |Ui | = |Uj | = 3 with Uij = Uik = Ujk = ∅.
∪3
(iii). | n=1 Un |= 7, and one of the following cases is satisfied:
[a] There is Ui with |Ui | ∈ {5, 4} and |Ujk | = 1.
[b] There exist Ui and Uj such that |Ui | = |Uj | = 3 with either |Uik | ∈ {1, 2} and
Ujk = ∅ or Uik = ∅ and |Ujk | ∈ {1, 2}.
[c] There exist Ui and Uj such that |Ui | = 3, |Uj | = 2 with |Ujk | ∈ {1, 2}. Further, if
|Ujk | = 1, then either Uij = ∅ or Uik = ∅ and, if |Ujk | = 2, then Uij = Uik = ∅.
∪3
(iv). | n=1 Un | = 6, and one of the following cases is satisfied:
[a] There is Ui with |Ui | = 4 and |Ujk | = 2.
[b] There is Ui with |Ui | = 3 and |Ujk | ∈ {2, 3}.
∪3
(v). | n=1 Un | = 5; there is Ui with |Ui | = 3 and |Ujk | ∈ {3, 4}.
∪
Proof. Assume that γ̃(AG(L)) = 2. First of all, if | 3n=1 Un | ≤ 4, then AG(L) is planar
∪3
⌈ [19]). Suppose | ⌉ n=1 Un | ≥ 10. If |U2 | ≥ 2, then by Theorem 1 we have γ̃(AG(L)) ≥
(see
(|U1 |−2)(|U2 |+|U3 |−2)
2 ≥ 3, which is a contradiction. Suppose |U2 | = 1. Then |U3 | = 1.
Note that every vertex in U12 , U13 , and U23 is adjacent to all of the vertices of U3 , U2 , and
U1 , respectively. So, if U23 = ∅, then clearly AG(L) is planar. If not, the vertices in U1 are
adjacent to all of the vertices of U2 ∪ U3 ∪ U23 . Since |U1 | ≥ 8, K8,3 is a subgraph of AG(L)
∪
that has a crosscap of more than three, refer to Proposition 1a. Thus, 5 ≤ | 3n=1 Un | ≤ 9.
∪3
Case 1 Let | n=1 Un | = 9. Then, clearly, |U1 | ≤ 7. If |U1 | = 7, then a slight modifi-
cation to the discussion made in the above paragraph would show that AG(L) is planar
whenever U23 = ∅ and the graph AG(L) contains K7,3 as a subgraph when U23 ̸= ∅.
If |U1 | = 6, then |U2 | = 2 and |U3 | = 1. Now, if U23 ̸= ∅, then AG(L) contains K6,4
as a subgraph, which is a contradiction. So, U23 = ∅. Here, all of the vertices in U12
are adjacent to a single vertex of U3 , and, therefore, the vertices in U12 do not affect the
crosscap. In Figure 4a, we provide the canonical representation of the embedding of the
resulting graph in N2 so that, in this case, γ̃(AG(L)) = 2. Next, if |U1 | = 5 or 4, then
|U2 | + |U3 | ≥ 4, and so, by Theorem 1a, we obtain γ̃(AG(L)) ≥ 3. Thus, |U1 | = 3, and,
therefore, |U2 | = |U3 | = 3. Here, K3,3,3 is a subgraph of AG(L), and, therefore, according
to Proposition 1c, we have γ̃(AG(L)) ≥ 3.
∪
Case 2 Let | 3n=1 Un | = 8.
If |U1 | = 6, then |U2 | = |U3 | = 1. Clearly, by [19], AG(L) is planar in the case that
U23 is empty. If |U23 | ≥ 2, then the partite sets X = U1 and Y = U2 ∪ U3 ∪ U23 form
K6,4 as a subgraph in AG(L), which is a contradiction. Therefore, |U23 | = 1. In this case,
the vertices in U13 ∪ U12 are all end vertices, and, therefore, it does not affect the crosscap.
Thus, the resulting graph is K6,3 ∪ {( I2 , I3 )}, which is a subgraph of a graph given in
Figure 2a, and, therefore, γ̃(AG(L)) = 2.
Suppose |U1 | ∈ {5, 4}. Then, according to Theorem 1a, we have γ̃(AG(L)) ≥ 2.
If U23 ̸= ∅, then the sets X = U1 and Y = U2 ∪ U3 ∪ U23 form K5,4 as a subgraph of AG(L),
and so γ̃(AG(L)) ≥ 3. Therefore, U23 = ∅. Let |U2 | = 2, |U12 | ≥ 0, and |U13 | ≥ 0. For
the embedding of AG(L) in N2 , in the case of |U1 | = 5, we can obtain help from Figure 4a
because the number of vertices and edges of AG(L) is less than that of in Figure 4a. Fur-
ther, Figure 4b provides an N2 -embedding of AG(L) in the case of |U1 | = 4. Here, notice
that the open neighborhood of each vertex in U13 is { I2 , I2′ }, and, in Figure 4a,b, there is a
face in an N2 -embedding of AG(L) that contains both I2 and I2′ so that every vertex of U13
can be embedded in N2 no matter what its cardinality may be. Let |U2 | = 3. This implies
that |U1 | = 4. If U13 = ∅ (recall that U23 = ∅), then AG(L) is a subgraph of the graph
H1 in Figure 1, and, therefore, according to Lemma 2, γ̃(AG(L)) = 2. If not, consider
that the subgraph AG(L) − {( I3 , I1 ), ( I3 , I1′ ), ( I3 , I1′′ ), ( I3 , I1′′′ )} contains K3,6 . By Euler’s
formula, any embedding of K3,6 in N2 has nine faces. Further, by solving the equations
 Mathematics 2023, 11, 1553 8 of 26

Mathematics 2023, 1, 0 8 of 25
2| E| = 4 f 4 + 6 f 6 and | F | = f 4 + f 6 , we have all the faces as rectangular faces in any
N2 -embedding of K3,6 . Now we try to recover the embedding of AG(L) by inserting all
edges ( I3 , I1 ), ( I3 , I1′ ), ( I3 , I1′′ ), ( I3 , I1′′′ ) into the embedding of K3,6 . ′Since deg ( I ) = 3,
to recover the embedding of AG(L) by inserting all edges ( I3 , I1 ), ( I3 , I1 ), ( I3 , I1′′ )K,3,6( I33, I1′′′ )
the vertex I3 is in the boundary of three rectangular faces of any N2 -embedding of K3,6 .
into the embedding of K3,6 . Since degK3,6 ( I3 ) = 3, the vertex I3 is in the boundary of three
In addition, note that, at the maximum, each rectangular face can adopt one edge incident
rectangular faces of any N2 -embedding of K3,6 . Also note that at the maximum each rect-
with I3 . So, we cannot insert all four edges of I3 into N2 without crossing, which is a
angular face can adopt one edge incident with I3 . So we cannot insert all four edges of I3
contradiction. Thus, γ̃(AG(L)) ≥ 3.
in N2 without crossing, a contradiction. Thus γ̃(AG(L)) ≥ 3.

I1′′′ I1′′′′ I1′′′′′ I1′′′ I1′′ I1′′′

b b b b b b

b
I1 I1b
I2′ b b
I1′′ b b I3 I2′ b I13 b b
I3
I13 I2
I1′ I1′ I3′
b

I3 b b b b b
I2′ I3 b b b b I2′
′
I13 b
I2 I12
b b b b b b

I1′′′ I1′′′′ I1′′′′′ I1′′′ I1′′ I1′′′

∪3 ∪3
(a). | n=1 Un | = 9, |U1 | = 6 and U23 = ∅ (b). | n=1 Un | = 8, |U1| = 4, |U2| = 2 and U23 = ∅

Figure 4. N2 -embedding of AG(L).

Suppose |U1 | = 3. If Uij = ∅ for all 1 ≤ i < j ≤ 3, then, by Proposition 1c, we have
γ̃(AG(L))
Suppose = |U2.1 | Next,
= 3. If our Uijclaim= ∅ isfor that all U 1 ij≤=i ∅ <for j ≤all3,1then ≤i< ≤ 3.
byj Proposition 1c, we have
γ̃(AG(L)) Assume that U12 ̸= ∅. Then the
= 2. Next, our claim is U = ∅ for
ij minor subgraph is all 1 ≤ i < j ≤ 3.
Assume that U12 ̸= ∅. Then the minor subgraph
AG(L) − {( I1 , I3′ ), ( I1′ , I3′ ), ( I1′′ , I3′ ), ( I2 , [ I3 , I12 ]), ( I2′ , [ I3 , I12 ]), ( I2′′ , [ I3 , I12 ])} ∼ = K4,4
AG(L) − {( I1 , I3′ ), ( I1′ , I3′ ), ( I1′′ , I3′ ), ( I2 , [ I3 , I12 ]), ( I2′ , [ I3 , I12 ]), ( I2′′ , [ I3 , I12 ])} ∼ = K4,4
with the partite sets X = U2 ∪ {[ I3 , I12 ]} and Y = U1 ∪ { I3′ }. By Euler’s formula, any
with partite sets Xof =
N2 -embedding K4,4 U2has ∪ {[eight
I3 , I12rectangular
]} and Y =faces. U1 ∪ {Next, I3′ }. By we Euler attempt formula,
to obtain anyan N2N - 2-
embedding
embedding of4,4 of K AG(L) from any N2 -embedding of K4,4 . For this, we try to2 embed the six
has eight rectangular faces. Next, we attempt to get an N -embedding
AG(L) edges
of omitted from any of AG(L) N2 -embedding into an arbitrary of K4,4 . NFor that, we try to embed the six omitted
2 -embedding of K4,4 . First, to embed the three
edges of AG(L) ′ ′
into ′an arbitrary ′′
edges ( I1 , I3 ), ( I1 , I3 ), and ( I1 , I3 ),2 three rectangular ′N -embedding of K4,4 . First,
faces areto embed the
required, three as
denoted edges F1 , F2 ,
( I1and ′
, I3 ),F(3I,1all′ ′
, I3of ) and ′′
( I1contains′
, I3 ), it requires ′ ′ F1 , F2 and F3 ,
which I3 (refer three to Figure rectangular
5a). Sincefaces, degK4,4name ( I3 ) =it 4,asexactly one more
′ I3′ ) label
allface
of which shouldcontains have I3′ ;I3it (refer is denoted Figure as5a). Since degK4,4 (we
F4 . Intentionally, = 4,the exactly one more
diagonals of F4 face as the
should
vertices have I2 andI3′ , denote as F4 . Intentionally,
[ I3 , I12 ]itbecause F4 can adopt label one diagonalthe diagonals edge that of Fcan 4 bybe the
used verticesto embed I2
and [ I , I ] because F can adopt one diagonal edge
the fourth edge ( I2 , [ I3 , I12 ]). Finally, to embed the rest of the two edges ( I2 , [ I3 , I12 ]) and
3 12 4 that can be used to embed ′ the fourth
edge( I2′′(, [I2I3, ,[ II312, ])
I12, ]) . Finally,
two distinct tofaces
embed arethe rest of the
required, denotedtwo edges by F5(and I2′ , [ IF3 6, ,I12 ( I2′′ , [ I3have
]) andshould
which , I12 ])the
,
it requires two distinct faces,
vertex [ I3 , I12 ]. Note that, in any Nk -embedding, denote it by F 5 and F , which should have
6 every edge of a graph is in exactly the vertex [ I , I
3 12two ] .
Note that,
faces. Since the edge in any N k -embedding, every edge of a graph is in ′ exactly
( I1 , [ I3 , I12 ]) is in F2 and the edge ( I1 , [ I3 , I12 ]) is in F4 , the common two faces. Since the
′
edgeedge ( I1between
, [ I3 , I12 ])F5isand in FF26 and mustthe ( I1′′ , [ I(3I,1I,12[ I])3 ,. INow,
be edge 12 ]) isthe F4 , thefor
in choice common edge between
the unlabelled vertex of
F5 Fand F I ′′ I ′I F ′ for
5 and 6 F must
6 is be
either ( I11 , [
or 3 ,
I1 . ]) .
Without
12 Now a the
loss choice
of for
generality, the unlabelled
we label I1
vertex
for F 5
of
and I5 and F6F6(refer
is
either I1 or I5b). ′ . Without loss of generality, we label I for F and I ′ for F (refer 1Figure 5b).
to Figure 1 Since any N 2 -embedding of K 4,4 has 1 eight 5 faces, 1 there 6 are two more faces,
Since
lets any say N F72 -embedding
and F8 , that have of K4,4 to has eight faces,
be formed usingthere all ofare thetwo more faces,
remaining lets say
vertices and Fedges 7 and of
F8 ,Khave to be formed using all remaining
4,4 . Notice that, in any N2 -embedding of K4,4 , each vertex is present
vertices and edges of K 4,4 . Notice
in exactly fourany
that, in faces,
N2and-embeddingeach edge is 4,4 of K present in exactly two faces. Since the vertices I2 ∈ Xedge
, each vertex is present in exactly four faces and each andinI1′exactly ∈ Y are
two used faces. twice Since in thethe faces
vertices F1 , .I.2. ∈, F6X, theandfaces I1′ ∈ FY7 and are used F8 must twice sharein the edge F(1I,2., .I.1′ ,)F(refer
thefaces 6 , the to
′ ′
faces
Figure F7 and 5c).F8Now, mustthe share the edge
choices for the ( I2 , third
I1 ) (refer and Figure fourth 5c). Nowof
vertices theF7choices
and F8 for arethe I2′′ ∈ X
I2 , third
′′ ′ ′′ ∈ Xhave ′′
I1 ∈ Ydistinct
and and fourthI1 , I1vertices ∈ Y, respectively.
of F7 and F8 are I2 , I2 we
Clearly, and to I1 ,select respectively. vertices Clearly,
for F7 we andhave F8 , in
to which
select one distinct is from { I2′ , I2′′for
vertices } andF7 and F8 , iniswhich
the other from {one ′′ }from
I1 , I1is { I2′ , I2′′ } andtoanother
. A contradiction this factfrom is that
, I1′′edges
{ I1the }. A contradiction
( I2′ , I1 ) and ( Ito ′′ the′′ fact is that the edges ( I ′ , I1 )Fand ( I ′′ , I ′′ ) are used twice in
2 , I1 ) are used twice in the faces 2 1 , . . . , F26 . 1
the faces F1 , . . . , F6 .
 Mathematics 2023, 11, 1553 9 of 26
Mathematics 2023, 1, 0 9 of 25

Mathematics 2023, 1, 0 I1
b
I2′ I1′ I1 I2′ 9 of 25
b b b b b b

F1 F2 F5
F7
I2′′ I1b I2′ [ I3 ,II′12 ] Ib 1 I2′ ′′
b b b1 [ I3 , I12 ] b bb I
1 I2 bb b I1′
F1 I3′ F2 F5
F3 F4 F87
[ I3 , b I12 ] F6
I2′′ bb bb b [ I3 , I12 ] bb bb
I1′′ I2 bb
bb I1′
I1′′ I2 I ′ I1 I1′ I2′′
3
F3 F4 F8
F6
b
(a) b b b
(b) b b (c) b

I1′′ representation
Figure 5. Faces
I2 I1
of N -embedding
I′
of K 1 . I2′′
2 faces of N2 -embedding
Figure 5. Representation of 4,4 of K4,4 .

(a)
̸= ∅Ufor i ∈ {1, 2i}∈ AG(L) − { Ii3 , ( I− (b) (c)
AssumeAssume that Ui3 that i3 ̸ = ∅ for
some some {1, 2}the
. Then subgraph
. Then, the subgraph AG(L) i , I{3I)i3
, , ( Ii , I3 ),
Ii′ , I3 ),5.((IFaces
(Figure Ii′′i′,, II33)} Ii′′ , I3 )} contains
, (contains
)representation K4,4of−Ne2 -embedding
K4,4 −
with partite
e withofsets
Kthe
4,4
X. = U
partite
i ∪ { I
sets3 }
X and
= Y
Ui ∪
= {
U Ii3 ∪
′} { I
and3
′ } where
Y = U i′ ∪ { I3 }
′
′
i ∈ {1, 2where } \ {i }i and ′ ′ ′
∈ {e1,=2}(\I3{, Ii3}).andBy eProposition
= ( I3 , I3 ). 2,
Byany N2 -embedding
Proposition 2, anyof 4,4 − e has one
N2K-embedding of K4,4 − e
hexagonal hasand
Assume that
one six U
hexagonal i3 ̸ = ∅ for
rectangular andfaces.some i ∈ {that
Note
six rectangular 1, 2}the hexagonal
. faces.
Then the
Note subgraph face the
that AG(L)
shouldhexagonalhave − {either
Iface I,3I3or
i3 , ( Iishould ), have
(I3I′ i′ and Ii′′ ,vertex
, I3 ), the
(either Ii3I3is
I3 )}I3contains
or ′ , and
adjacent
K4,4 the −vertexetowith Ii3i′′ is
{ Ii′ ,Ipartite , Ii′′adjacent
′ }sets⊂ Y. X Soto U
= Ii i∪
{Ii3 I{i′′I,3I}i′′′its
′ ,with }⊂
and edges
Y
Y.=So, must { I3′ inserted
UiI′ i3∪with
be } its
where edges must
i′ ∈ {the
into 1, 2be} \inserted
hexagonal{i } andface e =which
into I3′ )implies
( I3 , hexagonal
the . By Proposition face,I3which
that is in the
2, any
implieshexagonal
N2 -embedding that I3face. is inof Since
theK4,4 − eKhas
deg
hexagonal I3 ) Since
(face.
4,4 − e one
= 3, exactly
hexagonal and two six
degK4,4 −e ( I3 ) rectangular
rectangular faces
faces. contain
Note I
that3 inthe which
hexagonal it is notface possible
should to
haveembed either all I the
3 or
′ ′ , I ) and ( I ′′ , I ), a ′contradiction. ′′ ∅ ∈ { 3}. all of
three
I3 and the edges= vertex( I ,
3, exactlyI )
i 3 Ii3 is, ( I two adjacent
3 rectangular i {3Ii′ ,faces
to } ⊂ Y. So
Ii′ , Ii′ contain I3 in Thus
Ii3 which U
withijitsit edges = for all
is not must i,
possiblej 1,
be inserted 2,
to embed
∪3 i ′ ′′
into the Case the Let | edges
hexagonal
3 three n=1 U(nI|i ,=
face which I37. ), implies
( Ii , I3 ), and that (II3i is , I3in), the
which hexagonal is a contradiction. face. Since Thus, degK4,4U −ije (= I3 )∅ for all
= 3, Suppose i, j ∈ two
exactly {|1, | 3∈}.{5, 4∪}. Clearly
U12,rectangular faces contain AG(L)I3isineither which it is not
planar possible towhen
or projective embed U23all=the ∅
′ , I ) and ′′ , I ), a contradiction. Thus U = ∅ for all i, j ∈ {1, 2, 3}.
three edges
(refer [19] and Ii , [20])
(Case I3 ),3(Let Iand 3| K 3
n =1 n
5,4 is (
U I
a | =
subgraph
3 7. of the contraction of
ij AG(L) when | U 23 | ≥ 2.
∪3 i i
Therefore Case 3 Let | U |
23 | n=1 Un1| = 7.
Suppose will be | U 1.| ∈Then { 5, 4 }
AG(L). Clearly, is a AG(L)
subgraph is either
of the planar graph or projective
given in Figurewhen 4a U23 = ∅
whenever (refer
Suppose |U1 | U=1 | 5∈and
to [19,20]), {5, AG(L)
4}. Clearly
and K isAG(L) is of
is a subgraph
5,4 a subgraph eitherof the
the graph contraction
planar given or projective of
in Figure AG(L)
when U23 = ∅23 | ≥ 2.
when
4b whenever | U
|U1 | =[19]
(refer 4Therefore,
soand that[20]) γ̃(AG(L))|Uand 23 | will
K5,4 =be is2. aone. subgraph Then, AG(L) is a subgraph
of the contraction of AG(L) of the graph when given |U23 | ≥ in 2. Figure 4a
Therefore when
Assume |U23||UUwill 11|| =be 5,=and1. |Then AG(L)
U 2 | AG(L) =is a 3. subgraph
is a subgraph Then of the AG(L) of graph the is given
graph in
projectiveFigure
given 4b
inwhen
Figure when U4a i3|U1 | = 4
= ∅ forso
whenever Ui1 |=γ̃
all|that =(AG(L))
1,52and andAG(L) the = 2. graph is a AG(L) subgraph of the K
contains 3,7 asgiven
graph a subgraph in Figure when |Ui3 | ≥ 3
4b whenever
|U1 some
for | = 4 so i =that 1, 2.γ̃Suppose
Assume (AG(L)) thatU=13 |U2.1̸=| ∅= and|U U23 2 | ̸= = ∅. Now 3. the Then, graph AG(L)
AG(L)is− {projective I3 } is isomor- when Ui3
to K=4,4∅−for {|eU }all { ′ , I ′′AG(L) } { ′ ,K ′′ , I as}awhere |)U. i3 | ≥ 3
phic Assume 1 |
with i =
= 1,
bipartite |
2, U and
2 | sets the
= graph
I , 3.I
1 1 1 13 , I
Then and AG(L)
contains I , I
2 2 23,7 23 I is subgraph
projective e =
when (when
I ,
13 23i3IU
= ∅ for all i =
Note thatfor γ̃ (some
K − {
= e })
4,4 1, 2 and the graph AG(L)
i 1, 2.
= 2 Suppose
and there U are
13 ̸ = 7contains K233,7 as2 a subgraph when4,4|Ui3 | ≥ 3− { I3 } is
∅faces and inU any ̸ = N ∅. Now,
-embedding the graph
of K AG(L)
− { e } in
which
for some isomorphic
6 arei =rectangular
1, 2. Suppose to K and
4,4U13 − one{ e }
̸=is∅hexagonal.
with and U23 ̸= Since
the ∅. Now
bipartite γ̃sets
(Kthe4,4 {)Igraph
1= , I1′2, Iand ′′ , I } and { I , I ′ , I ′′ , I } where
1AG(L) every
13 − face
{ I3 }2in is2any 2 N23
isomor- 2-
embedding
phic to Ke4,4 =− ( I{13Ke,}4,4
of I23
with ). rectangular,
is Note
bipartite that γ̃sets (K the 4,4 −
1, { I1′e,})
{ Ihexagonal I1′′ ,=I132,face
}and {any
andofthere I2 , I2are ′N, I22′′-embedding
, I23 } where
seven faces in e any
of =K4,4( IN13−2, -embedding
I{23e}).
must that
Note of γ̃K
have the −−{e{}e,})ofI=
K4,4vertices
(4,4 13which
2 andI23
and six areare
. Now
there rectangular,
we try toinrecover
7 faces and
any N one 2an isN2hexagonal.
-embedding -embedding of K4,4 of−AG(L)
Since {eγ̃}(Kin4,4 ) = 2
from an
which 6 and
are every face and
N2 -embedding
rectangular in Kany
of one−Nis{2 -embedding
4,4 } by inserting
ehexagonal. of IK
Since γ̃with
3 4,4(Kis4,4its rectangular,
) =edges. 2 andHere theI3hexagonal
every face in any face
is adjacent Nto2 - of any
six vertices
embedding N2 -embedding
I1 , K
of I1′4,4
, I1′′is ′ and
, Iof
, I2rectangular,
2
K 4,4 −
I2
′′ .{ eBut
the } must the
hexagonal have
hexagonal the
face vertices
face
of any of NI
K 13
24,4 −
and {
-embedding I
e }
23 . Now,
does of not
Kwe4,4 −
try
contain { eto} recover
two ofhave
must an N
them theso
2 -embedding
that
vertices γ̃ (AG(L))
I 13 of
and AG(L) ≥I 23 3.
. from
Therefore
Now an
we N
try -embedding
either
2 to U
recover 13 = an ∅
of K
or
N 4,4U − {=e
-embedding
2 23 } ∅.
by inserting
Now of AG(L)
by I
the 3 with its
′ , I ′′ , I , I ′ , and I ′′ . However, the hexag-
from an N2 -embedding3 of K4,4 − {e} by inserting I3 with
help of edges.
Figure 6,Here,
we haveI is γ̃ adjacent
(AG(L)) to= the 2 six
whenever vertices 1 ≤I ,| UI
1 1its | ≤
i3 1edges. 2 for unique i
2 2 Here I23 is adjacent to ∈ { 1, 2 } .
six vertices onalI1face , I1′ , Iof ′′ K , I ′−and
1 , I24,4 2
{e}I2does ′′ . But not thecontain hexagonal two of face them of Kso 4,4 that does not ≥
− {eγ̃}(AG(L)) 3. Therefore,
contain
′ U I = ∅ orI U =I ′∅. Now, with the help of Figure 6, we have γ̃ (AG(L)) = 2 when
either
two of themb 2so that I 13 3 γ̃ (AG(L)) 2 23 ≥ 2 3. Therefore either U13 = ∅ or U23 = ∅. Now by the
b b b
help of Figure 1 ≤ |U6, i3 |we ≤ 2have for aγ̃unique(AG(L)) i ∈={1, 2}.
2 whenever 1 ≤ |Ui3 | ≤ 2 for unique i ∈ {1, 2}.

I1 I2b ′ I3b ′ Ib
I23 2 I2′b I1′
b
I23b b b

I2′′
I1′ b b b I1′′ b I1′
I1 b b ′
I23 b b I1
I23
b
Ib 2′′ b
′′
b

I1′ I2′b I3b Ib 2I1 I2′b I1

∪3
Figure 6. | nb=1 Un | b = 7 with |U1 | = |U2 | = 3, U13 = ∅ and |U23 | = 2.
b b

I2′ |U |I3= 3 and I2 |U | = I′

Assume 1 2 2 2. If |U23 | ≥ 3, then AG(L) contains K3,7 as a subgraph
∪
and if 6.UFigure ∅, ∪3 AG(L) |U23 | = 2.
Figure | 3n=
23 Un||then
=1 6.
by| Theorem
=n=71with
U n = |U71 |with 2.4iii
= |U|2U | =3, |[20],
|1= U 2| = ∅
U13 and
3, U13 =
is∅
|U projective.
= 2.|U23 | = Suppose
23 |and 2.
If U1j ̸= ∅ for j = 2 or 3, then consider a subgraph G1 = AG(L) − { I1j , I23 ′ ,e ,e ,e ,e }
1 2 3 4
Assume |U1 | = 3 and |U2 | = 2. If |U23 | ≥ 3, then AG(L) contains K3,7 as a subgraph
and if U23 = ∅, then by Theorem 2.4iii [20], AG(L) is projective. Suppose |U23 | = 2.
If U1j ̸= ∅ for j = 2 or 3, then consider a subgraph G1 = AG(L) − { I1j , I23 ′ ,e ,e ,e ,e }
1 2 3 4
 Mathematics 2023, 11, 1553 10 of 26

Mathematics 2023, 1, 0 Assume that |U1 | = 3 and |U2 | = 2. If |U23 | ≥ 3, then AG(L) contains10Kof 3,725as a sub-
graph, and, if U23 = ∅, then, by Theorem 2.4iii [20], AG(L) is projective. Suppose |U23 | =
2. If U1j ̸= ∅ for j = 2 or 3, then consider a subgraph G1 = AG(L) − { I1j , I23 ′ ,e ,e ,e ,e }
1 2 3 4
where e I , I , e′ I , ′
I ′ , e I ′ , I ′
, ′
and e I ′ , I ′ . Clearly, G
= ( ) = ( ) = (
where e1 = ( I2 , I13 ), e2 =2 ( I32 , I32), e3 = 2( I23, I3 )3and e4 2= (3 I2 , I3 ). Clearly ) 4 = ( G2 contains
3 ) K3,5 with 1 contains
K ′ ′′ X { ′
I I ′ , I′ ′′ } and Y = { I , I ′ , I , 1I ′ , I }. Note
partite sets with the partite sets = ,
3,5X = { I , I , I } and Y = { I2 , I 1, I31, I 1, I23 }. Note that2any
1 1 1 2 3 2 N 3 2 -embedding
3 23 of K3,5 any N2 -
that
embedding
has one hexagonal and six of K has one hexagonal and six
3,5 rectangular faces. Now we try to recover an N2 -embedding rectangular faces. Now, we try to recover
of an
of AG(L) from ′ is adjacent ′
AG(L) N 2 -embedding
from any N2 -embedding of K3,5 any N2 -embedding
. Since I23 of K3,5to. all
Since theI23 three is adjacent
verticestoofall three
′ X, I ′ requires ′ K ′′ have Ithat ′ ′′
vertices of the embedding of
X, to embed I23 it requires the hexagonal face of K3,5 to have I1 , I1 and I1 . Notice
23 the hexagonal face of 3,5 to 1 , I1 , and I1 .
Notice that
each rectangular each
face may rectangular
adopt at most face may one edge adoptinto at most
it. Soone edge into
to insert e f ’s, it. forSo,
1≤ ≤ 4, e f s, for
to finsert
1 ≤ f ≤ 4, into any N 2 -embedding of K 3,5
in any N2 -embedding of K3,5 , it requires four rectangular faces with diagonals as the end , four rectangular faces with diagonals as the end
vertices of each e are required.
vertices of each e f . At lastf to insert I1j , it requires a rectangular At last, to insert I 1j , a rectangular face
face with diagonals Ij′ and with the diagonals
′ ′ I and I ′ for j′ ∈ {2, 3} \ { j } is required. Therefore, it requires one hexagonal face with
Ij′ for j ∈j {2, 3}j′\ { j}. Therefore it requires one hexagonal face with five rectangular faces
′

five rectangular faces ′ ′ ′

′ the vertices I2 , I , I3 , and I in at least three different faces.
that contains the vertices I2 , Icontaining
2 , I′ 3 and I3 in′ at least three2 different 3faces. Since the degree
′ Since the ′
of I2 , I2 , I3 and I3 in K3,5 is 3, all2 the four Ivertices
degree of I2 , I , I3 , and 3 in K3,5 are is three,
placedallinfour exactlyvertices threeare placed
faces of any in exactly
three faces
N2 -embedding of K3,5 . So the of any N -embedding of K . So, the
2 sixth rectangular3,5face of K3,5 could not be formed using sixth rectangular face of K 3,5thecould not
be formed using the only left-out vertex in X (namely I ), which is a contradiction. Thus,
12 = U13 = ∅ and an
only one left out vertex in X(namely I23 ), a contradiction. Thus 23 U
U
N2 -embedding 12 = U of = ∅, and an N -embedding
13 AG(L) for this 2case is provided in Figure 7a. of AG(L) for this case is provided in Figure 7a.
Suppose |U | = |1.
Suppose U23 If|U= 1.̸=If∅Ufor
23 1j ̸ =
1jj= ∅ 2for and j =3,2then and the 3, thenminor thesubgraph
minor subgraph is

AG(L)
G2 = −
G2 = AG(L) { I13 , e− , e , e5 }−∼
{ I e , ,ee1 , ,ee2 , }e3∼ ={K , − { e },
e}4,4 (1)
1 , e2 ,133 4 5 = 4K4,4 (1)
with the ′′ { I1 , I1′ , {I1I′′ ,,II3′}, [and
{ I1 , I1′ , Isets
bipartite ′ { I , I2′ ,}[ Iwhere
′
3 , I12 ], eI123=} where ′
( I1 , I3 )e,1e2==( I(1I,1′ I,3I)3,)e,2e3==( I1 , I3 ),
with bipartite sets
′′ 1 , I3 } and ′ 2 2 I3 , I12′]2, I23 ′
( I1′′ , I3 ), ee34 = = ((I I1, [,II′3,)I, e4]),=e (=
2 3 12
I , [ I′ , I ′ ,])I , ])
5 2 ( I23, [ I12
e5 = ( I , [ I , I12 ]), and e = ( I3 , I23 ). Note that any N2 -
3 12 and e 2= (3I3 , I23 ). Note that any N2 -embedding
embedding of K −
of K4,4 − {e} has 6 rectangular faces and a hexagonal
4,4 { e } has six rectangular faces face,andand a hexagonal
the hexagonal face, and facethe must hexagonal
face must have the vertices I3 and I23 .
have the vertices I3 and I23 . Let us denote the 6 rectangular faces by F1 , . . . , F6 and theLet us denote the six rectangular faces by F1 , . . . , F6
the hexagonal face by F
hexagonal face by F7 . Now let us try to recover an N2 -embedding of AG(L) by inserting by in-
and 7 . Now, let us try to recover an N 2 -embedding of AG(L)
serting thethe vertex I13eand the edges e for all i = 1, . . . , 5. If we embed the edge e4 , the edge
the vertex I13 and edges i for all i = 1, . .i. , 5. If we embed the edge e4 or the edge e5 or
e , or the
5 I together withvertex I together with intoitsF7edges, theninto F7 , theninsertwe cannot insert e1 the
or eedges e , e , or
the vertex 13 13 its edges we cannot the edge 2 or e3 1 2
e 3 into F 7 . Since deg ( I 3 ) = 3, the vertex I3 is in exactly three faces of an N 2 -embedding
into F7 . Since deg G2 ( I3 ) = 3,G2the vertex I3 is in exactly three faces of an N2 -embedding of
G2 . So, in of suchG2 . So,cases, in such
the edges cases,e the , e edges
and e ecannot
1 , e2 andbe eembedded
3 cannot beinembedded two rectangular in twofaces rectangular
1 2 3
faces which contains I 3 . Therefore we have
which contains I3 . Therefore we have to add at least one of the edges e1 , e2 or e3 into F7 . For to add at least one of the edges e 1 , e 2 or e3 into
F 7 . For the best possibility, say e 1 and
the best possibility, say e1 and e2 are embedded in F7 . Then e3 has to be embedded into one e 2 are embedded in F 7 . Then, e 3 has to be embedded
of the two rectangular faces that contains I3 , say F1 . Notice that there are two rectangularthere are
into one of the two rectangular faces that contains I3 , for example, F 1 . Notice that
two F2 rectangular
and F3 , thatfaces, say IF2 and F3 , that shouldI23
onecontain , inembed
whichany oneofshould e4 or enot embed any
faces, say contains 23 in which not 5 or I13
with its edges.of e , e ,
4 5So the or I with its edges.
13edges e4 and e5 has to be embedded So, the edges e 4 and e have to be embedded
5 different rectangular faces,
into into different
rectangular
say F4 and F5 respectively. Therefore, faces, say F 4 and F , respectively. Therefore, after
5after embedding the edges from e to e5 nicely, we embedding the edges from
1
are left-out e to
1 with e nicely, we are left with the
5 the single rectangular face F6 that could not be formed single rectangular face F 6 that could not be
using the diagonal formed using
the diagonal ′ vertices I and I ′ . Thus, γ̃ (AG(L)) ≥ 3. Hence, either U = ∅ or U = ∅.
vertices I2 and I2 . Thus γ̃(AG(L)) ≥ 3. Hence either U12 = ∅ or U13 = ∅. For
2 2 12 this case, 13
In this case, with
with the help of Figure 7b, we get γ̃(AG(L)) = 2.the help of Figure 7b, we obtain γ̃ (AG(L)) = 2.

′
I23 I2 I3′ ′
I23 I3′ I23
b b b b b b

I1′′ I23 I2 I3b

b b
I2′ b b I1′ I1 b b b
I1′
I13 b

I1
I1′ b b b b
I1′′ I1′ b b b
I1′′ b I1
I3 I2′
b b b b b b
′ I3′ ′
I23 I2 I23 I3′ I23
(a). U12 = U13 = ∅ and |U23 | = 2 (b). U12 = ∅, |U13 | ≥ 0 and |U23 | = 1
∪3
| n=17.Un| |∪=
Figure 7.Figure 3 7 with |U | = 3 and |U2 | = 2..
n=1 Un | = 71 with |U1 | = 3 and |U2 | = 2.
∪
Case 4 Let | 3n=1 Un | = 6. Suppose |U1 | = 4. If |U23 | ≥ 3, then K4,5 is contained
in AG(L) and if |U23 | = 1, then AG(L) is projective. Therefore |U23 | = 2. Clearly
AG(L)(except end vertices) is a subgraph of the graph H1 given in Figure 1a and so
Lemma 2 implies γ̃(AG(L)) = 2.
Mathematics 2023, 11, 1553 11 of 26

∪
Case 4 Let | 3n=1 Un | = 6. Suppose |U1 | = 4. If |U23 | ≥ 3, then K4,5 is contained
in AG(L), and if |U23 | = 1, then AG(L) is projective. Therefore |U23 | = 2. Clearly,
AG(L)(except for the end vertices) is a subgraph of the graph H1 given in Figure 1a, and
Mathematics 2023, 1, 0
so Lemma 2 implies γ̃(AG(L)) = 2. 11 of 25

Suppose |U1 | = 3. Then AG(L) contains K3,7 when |U23 | ≥ 4, and AG(L) is projec-
tive when |U23 | ≤ 1. Thus, 2 ≤ |U23 | ≤ 3. Then, AG(L) − {U13 } is a subgraph of the
graph H2 (seeSuppose Figure |2a), U1 | so= 3. Then
that AG(L) contains
γ̃(AG(L) − {U13 }) K3,7=when |U23 |that
2. Note ≥ 4every
and AG(L)
vertexisin projec-
U13
tive when |U23 | ≤ 1. Thus 2 ≤ |U23 | ≤ 3. Then AG(L) − {U13 } is a subgraph of the
is adjacent to exactly two vertices of U2 in AG(L). Therefore, replace the labels u4 and u5
graph H2 , see Figure 2a so that γ̃(AG(L) − {U13 }) = 2. Note that every vertex in U13 is
with I2 and I2′ , respectively,
adjacent
in the N -embedding of H2 provided
to exactly two vertices2 of U2 in AG(L). So
in Figure 2b, and then
replace the labelling u4 and u5 as I2
label all of the ′ other vertices accordingly. Now,
and I2 respectively in the N2 -embedding of H2 provided in Figure we can insert any number
2b, andofthenvertices
label of
all
U13 into atheface that contains both I and I ′ so that γ̃ (AG(L)) = 2.
other vertices accordingly. 2 Now2 we can insert any number of vertices of U13 in a face
Moreover, if |U1 |both
that contains = 2,I2thenand IAG(L) ′
2 so that γ̃ (AG(L)) = 2.
is either planar or projective (refer to [19,20]).
∪
Case 5 Let | 3n=1 U
Moreover |
if n|U
∪31
|
= =5.2, Then
then AG(L)
AG(L) isiseither
planar or projective
planar or projective (refer|U
when 1| =
[19] and2.[20]).
This
Case 5 | n | = AG(L)
implies that |U1 | = 3. If n|=U123 | ≥ 5, then AG(L) contains K3,7 , and, if |U23 | 1≤ 2, then
Let U 5. Then is planar or projective when | U | = 2. This
implies | 1 | = 3. If | 23 | ≥ 5, then AG(L)
AG(L) is projective. Thus, |U23 | = 3 or 4. Then, clearly, AG(L) is a subgraph of the graph
U U contains K 3,7 and if | U23 | ≤ 2, then AG(L) is
projective. Thus | U
H1 , as in Figure 2a, so that γ̃(AG(L)) = 2.
23 | = 3 or 4. Then clearly AG(L) is a subgraph of the graph H 1 as in
Figure 2a so that γ̃(AG(L)) = 2.
All of the results proved in this paper have a similar structure to that of those given
All the results proved in this paper have a similar structure to that of those given in
in the statement of Theorem 3. To familiarize readers with the connection between the
the statement of Theorem 3. In order to make the readers more familiar with the connec-
multipartite graph
tion betweenandthethe statementgraph
multipartite and the3,statement
of Theorem we illustrate two four-partite
of Theorem graphs,
3, we illustrate two
with γ̃( G
G and H,4-partite ) = 2 and γ̃ ( H ) ̸ = 2, respectively, in the following example.
graphs G and H with γ̃( G ) = 2 and γ̃( H ) ̸= 2 respectively, in the following
example.
Example 1. Consider Case (iii)[c] in Theorem 3. Let |U1 | = 3, |U2 | = 2, |U3 | = 2, and
|U23 | = 1.Example
If |U12 | 1.= Consider
k ∈ Z+ and the case = ∅, then
U13 (iii)[c] in Theorem 3. Let |U1 | = four-partite
the corresponding 3, |U2 | = 2, |Ugraph
3 | = 2G and
is
| U 23 | = 1. If | U12 | = k ∈ Z + and U
13 = ∅,
a crosscap two, which is given in Figure 8a. Additionally, if |U12 | = 1 and |U13 | = 1, then
then the corresponding 4-partite graph G
is crosscap two which is given in Figure 8a. Also, if | U | =
the crosscap of the corresponding four-partite graph H, given in Figure 8b, is not equal
12 1 and | U13 | = 1, then the
crosscap of corresponding 4-partite graph H, given in Figure 8b, is not equal to two. It
to two. It is worth mentioning that the four-partite graph H in Figure 8b is minimal with
is worth to mention that the 4-partite graph H in Figure 8b is minimal with respect to
respect toγ̃γ̃ ( H ) ̸= 2; that is, there exists an edge e in H such that γ̃( H − e) = 2. Further,
( H ) ̸= 2; that is there exists an edge e in H such that γ̃( H − e) = 2. Further, the graph H
the graphmayH may be realized
be realized as oneasofone
the of the forbidden
forbidden subgraphs
subgraphs for crosscapfor two
a crosscap
surface. two surface.

U2 U1 U3 U2 U1 U3

b b
b b b b

b b
b b b b
b b

b b b b b b b
I23 ′
I12 k
I12
I12 I12 I13 I23

(a) A crosscap two 4-partite graph G (b) A minimal 4-partite graph H with crosscap ̸= 2
Figure 8. 4-partite graphs.
Figure 8. Four-partite graphs.
By using the proof of Theorem 3, we establish the following points which will be used
By using the proof of Theorem 3, we establish the following points, which will be
in the subsequent results.
used in the subsequent results.
Remark 1. If a graph G is isomorphic to K6,3 ∪ (K4 − e) or K4,5 − e where e is an edge, then
Remark 1. If)a≥graph
γ̃( G 3. G is isomorphic to K6,3 ∪ (K4 − e) or K4,5 − e where e is an edge, then
γ̃( G ) ≥ 3.
5. The Case when | A(L)| = 4
5. The Case When | Afix
Next we (L)|
the = 4
number of atoms as 4. As mentioned in the introduction, for 1 ≤ i ̸=
Next,j ≤we
4, we
fix denote U(ij)c =ofUatoms
the number k ℓ where ℓ ∈ {1,As
ask,four. 4} \ {i, j} and
2, 3,mentioned in the
thenotation U(ij)c exists
introduction, for
only when Uij ̸= ∅. Before going into the characterization of crosscap two AG(L) with
1 ≤ i ̸= j ≤ 4, we denote U(ij)c = Ukℓ where k, ℓ ∈ {1, 2, 3, 4} \ {i, j}, and the notation
Mathematics 2023, 11, 1553 12 of 26

Mathematics 2023, 1, 0 12 of 25
U(ij)c exists only when Uij ̸= ∅. Before going into the characterization of the crosscap two
AG(L) with | A(L)| = 4, we provide modifications for Theorem 2.6 [20]. To be precise,
the missing
| A(L)| = 4,cases and the modifications
we provide correspondingfor
conditions
Theoremfor 2.6the projectiveness
[20]. To be precise, of AG(L)
the missing are
given below.
cases and the corresponding conditions for the∪projectiveness of AG(L) are given below.
(i)
(i)First
First of
ofall,
all, consider the missing case| ∪
missingcase | 4 4n=1UU|n |==4.4. |Ui ||U=i | 1=for
Then, 1 for all 1 ≤
consider the ∪n=1 n Then ∪ all 1 ≤ i ≤ 4.
≤ 4. Clearly,
i Clearly AG(L) ∪ U ∅. ∪ c ̸ = ∅.
is planar
AG(L) is planar whenever wheneverU = ∅. c =
(ij) ThereforeTherefore,U ̸= U∅.
ij)ijc̸=∅
(U
(ij)If |U ∪ (ijU)cij ̸=∅ ij
Uij ̸=∅ Uij ̸=∅
IfU|Uij |∪≥U4(ijwith
)c | ≥U4, with Uij , U ij)c ̸=the∅,subgraph
then theinduced
subgraph byinduced by=the ∪ UjX∪ U
Ui sets =
(ij)c c ̸ = ∅, (then
ij U(ij)∪ the sets X ij
Uand ∪ U ∪ U∪ and Y = U ∪ U contains K or K as a subgraph. This implies
i Yj = ij Uk ∪ U(ij)ck̸=contains
i,j
k c
K(ij4,4
) or K
3,5 as a subgraph. This implies γ̃ (AG(L)) ≥ 2.
4,4 3,5
k̸=i,j
(AG(L)) 2≥≤2.|U
γ̃Therefore Therefore, 2 ≤ |Uij ∪ U(ij)c | ≤ 3 if Uij , U(ij)c ̸= ∅ for 1 ≤ i ̸= j ≤ 4.
ij ∪ U(ij)c | ≤ 3 if Uij , U(ij)c ̸ = ∅ for 1 ≤ i ̸ = j ≤ 4.
Suppose | U ∪ U |
Suppose |Uijij ∪ U(ij(ij)c)c | =33for
= someUUijij, ,UU(ij(ij)c)c̸≠=∅∅with
forsome with11≤≤i i̸≠=j j≤≤4.4.IfIfUUkℓk,ℓ ,UU(k(ℓ) c ̸=
k ℓ)c ̸ =
forkkℓℓ ̸≠= ij,
∅∅for ij, then
then the
the subgraph
subgraph∪ AG(L)
AG(L)− −{{U Uij ∪∪U U((ijij))cc}} contains
contains KK3,3 with the partite
3,3 with partite sets
sets X U ∪ U ∪ U and Y ∪ Um ∪ U ijc . Note γ̃ K
=
X = U ∪ U ∪ U and Y =
k
k
ℓ k
ℓ
ℓ = U ∪U
kℓ
( k ℓ) . Note that γ̃(K ) = 1. Now,
m̸=k,ℓ m
that
(kℓ)c
( 3,3 ) = 1. Now wewe try trytoto
3,3
m̸=k,ℓ
embed
embedall allofthe
thevertices
vertices of of U Uijij ∪∪UU((ijij))cc with
withtheir its edges edgesin in any
any NN1-embedding
1
-embedding of of KK3,3 . Since
3,3 . Since
|U ij ∪ U c | = 3, either | U
|Uij ∪ U(ij(ij) )c | = 3, either |Uij | = 2 or |Uij | = 2 or | U (ij) c | = 2. Without loss of generality, let |U || =
c | = 2. Without a loss of generality, let | U ij =2.2.
Since the vertex I ∈ U is adjacent
(ij)
to I , I ′ ∈ U , all of the three vertices I , I ′ij, and
Since the vertex (Iij()ij)c ∈ (U c c
ij) c is adjacent to ij I ij, I ′ ∈ij U , all the three vertices ijI ,ijI ′ and
(ij) ij ij ij ij ij
I(Iij)c must must be embedded
be embedded intointo a single
a single face of
face the
of N1-embedding
N -embedding of of KK 3,3 ,, denoted
name it as
as FF1 .. Now,
(ij) c
′ 1 3,3 1 Now
draw
drawthe pathIijIij−−I(Iij(ij)c)c−−IijIij′into
thepath intoF1F1and andthenthendraw drawthe theedges ),)(,I(ijI, I,nI),)(,I(ij′I,′ I,mI),)and
edges( I(ijI,ijI,mIm ij n ij m and
((Iij′I ′, ,InI ))where m, n ∈
/ { i, j } . Now, the edges ( I c , Ii ) and ( I(ij )c , I j ) cannot be embedded
ij n where m, n ∈ / {i, j}. Now the edges ( I∪ ( ij )
(ij)c , Ii ) and ( I(ij)c , Ij ) cannot be embedded
∪
intoFF11. .Therefore,
into (AG(L)) ≥
Therefore γ̃γ̃(AG(L)) ≥ 2.
2. Thus,
Thus kℓ̸=ij,(ij)c ;U ̸=∅U U(kℓ)c =
(kℓ)c
= ∅. ∅.
k ℓ̸=ij,(ij) ;Ukℓ ̸=∅
c kℓ

Suppose|U
Suppose |Uijij∪∪UU(ij(ij)c)|c |==22for forall allUUijij, U, U(ij(ij)c)c̸≠=∅∅with with11≤≤i i̸≠=j j≤≤4.4.Then, ThenFigure Figure99
guarantees thatγ̃γ̃(AG(L))
guaranteesthat (AG(L))==1.1.

I1 I14 I2
b b b

I34
I24 b b
I3
b

I12 I4
I13 b b b

b b
I23 b

I2 I14 I1
∪∪
Figure9.9.| | n4n==1 1UUnn| |==44with
with|U
|Uijij∪∪UU(ij(ij)c)|c |≤≤22for , U(ij(ij)c)c̸≠=∅.
4
Figure allUUijij, U
forall ∅.
∪∪
(ii)
(ii)LetLet| | n=n=
44
n |n |==5. 5.
1 1UU Then|U|iU
Then, | i=
| =2 for
2 for some some 1 ≤1 i ≤≤ i4,≤and 4 andthe condition
the condition for the for
projectiveness
the projectiveness of AG(L)
of AG(L) givengiven in Theorem 2.6i [20]
in Theorem 2.6iis[20]thatis|Uthat
jk | =| U1jk |
or =2, 1in orwhich
2 in at
which most at
one of the U s has exactly two elements for 1 ≤ i ̸ = j ̸ = k ≤
most one ofjkthe Ujk ’s has exactly two elements for 1 ≤ i ̸= j ̸= k ≤ 4. But,jkif |Ujk | = 2 4. However, if | U | = 2 with
Uwith
( jk)c U̸=( jk∅, ̸= ∅,the
)c then setsthe
then = UiX∪ =
X sets UℓU ∪i U∪(U
jk )ℓc ∪ U( jkY)c =and
and ∪U
Uj Y =k ∪ UjU∪jkU , where ∈
k ∪ U jkℓ, where/ {i, j, ℓk}∈ /,
contain 4,4 in AG(L)
{i, j, k},Kcontains K4,4 insoAG(L)
∪∪
that we obtain
so that weγ̃get γ̃(AG(L))
(AG(L)) ≥ 2.≥In2. fact, if |ifU|jkU| jk=
In fact, | =2 2forforsome some
j,j,kk ̸≠= i,i,then
then UU( pq
( pq)c)c == ∅.∅.Otherwise,
Otherwise,the setsXX == UUj j∪∪UUk k∪∪UUjkjk∪∪[ I[ pq
thesets I pq, ,I(Ipq
( pq)c)]c ]
p,q
p,q̸≠=i;U
i;Upqpq̸≠=
∅∅
andYY==UU1 1∪∪UUℓ ,ℓ ,where
and whereℓ ℓ∈/∈/{{i,i,j,j,k}k } formKK
, ,form 5,3 , so
5,3 sowe
wecan canconclude
concludethat (AG(L)) ≥≥2.2.
thatγ̃γ̃(AG(L))
∪∪
Further, if | | ≤ 1 for all ̸
Further, if |Ujkjk | ≤ 1 for all j, k ̸= i, then |
U j, k = i, then | UU( pq | ≤ 1. For if |
( pq)c)c | ≤ 1. For if |U
U ( pq)c)|c |≥
( pq ≥2,2,then
then
p,q
p,q̸≠=i;U
i;Upqpq̸≠=
∅∅
the setsXX == UUp p∪∪UUq q∪∪UUpqpqand
thesets andYY == UUi i∪∪UUr r∪∪UU( pq wherer r ∈
)c),c ,where
( pq /∈/ {{i,i,p,p,q}q} formKK
, ,form 3,5 ,
3,5
and ifif ||U
and, U((pq
pq))cc||,, ||UU((pp11qq11))cc|| =
= 1
1 for
for some
some 1
1 ≤≤ p p 1 1 ̸ ̸
= = qq11 ≤ ≤ 44 with
with pp1 qq
1 11 ̸ ≠ = pq,
pq, then
then the
the sets
sets
XX==UUpp∪∪UUq q∪∪UUpqpq∪∪{[{[I pI1pq11q,1 ,I(Ip( 1pq11q)1c)]} andYY==UUi i∪∪UUr r∪∪UU( pq
c ]}and formKK4,4
( pq)c)cform 4,4−−{{ee}}ininAG(L)
AG(L)
whererr∈
where /{{i,i,p,p,qq}}. .
/∈
∪∪
(iii) Let| | 4n4n==11UUnn| |==6.6.IfIfthere
(iii)Let exists|U
thereexists |Ui |i |==33for forsomesome11≤≤i i≤≤4,4,then thenthethestatement
statement
([20]Theorem
ofof([20] Theorem2.6(ii)(a))2.6(ii)(a))says saysthat Ujk
thatif if Uℓjk=
ℓ = ∅ for∅ for 1 ≤1 i≰=i j ̸= ̸= jk ̸≠= kℓ ≤ ̸= 4, ℓ |≤
Ujk4,| ≤
|U1,
jk | and
≤ 1
atand
most one ofone
at most theofUthe jk s has exactly
Ujk ’s one element,
has exactly one element, then AG(L)then AG(L) is projective. However,
is projective. But,for for
instants, if |Ujk | = 1 with U( jk)c = Uiℓ ̸= ∅, then the partite sets X = Ui ∪ Uℓ ∪ Uiℓ and
Y = Uj ∪ Uk ∪ Ujk , contains K5,3 as a subgraph of AG(L) so that γ̃(AG(L)) ≥ 2. Therefore
the condition U( jk)c = ∅ has to be added in the statement of ([20] Theorem 2.6iia).
Mathematics 2023, 11, 1553 13 of 26

instance, if |Ujk | = 1 with U( jk)c = Uiℓ ̸= ∅, then the partite sets X = Ui ∪ Uℓ ∪ Uiℓ and
Y = Uj ∪ Uk ∪ Ujk contain K5,3 as a subgraph of AG(L) so that γ̃(AG(L)) ≥ 2. Therefore,
the condition U( jk)c = ∅ has to be added to the statement of ([20] Theorem 2.6iia).
As a result of the above remarks (i), (ii), and (iii), we modify the statement of ([20]
Theorem 2.6) as follows.

Theorem 4. Let L be a lattice with | A(L)| = 4. Let 1 ≤ i ̸= j ̸= k ̸= ℓ ≤ 4 and 1 ≤ p ̸= q ≤ 4.

Then γ̃(AG(L)) = 1 if and only if one of the following conditions hold:
∪
(i). | 4n=1 Un | = 4; there exist two non-empty sets Uij and U(ij)c such that 2 ≤ |Uij ∪ U(ij)c | ≤
∪
3. Moreover, if |Uij ∪ U(ij)c | = 3, then U( pq)c = ∅.
pq̸=ij,(ij)c ;U pq ̸=∅
∪4 ∪
(ii). | n=1 Un | = 5; there is Ui with |Ui | = 2, | U pq | ≤ 4 in which at most one of the
p,q̸=i
∪
U pq s has a maximum of two elements, and | U( pq)c | ≤ 1. Moreover, if |U pq | = 2, then
U pq ̸=∅
∪ ∪
U( pq)c = ∅, and, if U pq = ∅, then Ujkℓ ̸= ∅.
U pq ̸=∅ p,q̸=i
∪
(iii). | 4n=1 Un |= 6, and one of the following is satisfied:
[a] There is Ui with |Ui | = 3. If |Ujkℓ | = 1, then Ujk = Ujℓ = Ukℓ = ∅ and if
Ujkℓ = ∅, then |Ujk ∪ Ujℓ ∪ Ukℓ | ≤ 1. Moreover, U( pq)c = ∅ whenever U pq ̸= ∅.
[b] There exist Ui and Uj such that |Ui | = |Uj | = 2 with |Ukℓ | ≤ 1. Additionally,
U( pq)c = ∅ whenever U pq ̸= ∅. Moreover, if |Uik |, |Uiℓ | ≤ 1 or |Ujk |, |Ujℓ | ≤ 1, then
|Ukℓ | ≤ 1. Furthermore, if |Uik | = |Ujk | = 1 or |Uiℓ | = |Ujℓ | = 1, then Ukℓ = ∅.
∪4
(iv). | n=1 Un |= 7 and one of the following is satisfied:
[a] There is Ui with |Ui | = 4 and Ujkℓ = Ujk = ∅.
[b] There exist Ui and Uj such that |Ui | = 3 and |Uj | = 2. Additionally, Ukℓ = ∅, and
Ujkℓ = ∅ whenever Uik = Uiℓ = Ujk = Ujℓ = ∅.

We are now in the position to state and prove the second result which classifies all
lattices L with four atoms whose AG(L) has a crosscap two.

Theorem 5. Let L be a lattice with | A(L)| = 4. Let 1 ≤ i ̸= j ̸= k ̸= ℓ ≤ 4 and 1 ≤

p, q, r, s, t ≤ 4. Then γ̃(AG(L)) = 2 if and only if one of the following conditions hold:
∪4
(i). | n=1 Un | = 9; there is Ui with |Ui | = 6 and Ujk = Ujℓ = Ukℓ = Ujkℓ = ∅.
∪4
(ii). | n=1 Un |= 8, and one of the following cases is satisfied:
[a] There is Ui with |Ui | = 5 and Ujk = Ujℓ = Ukℓ = Ujkℓ = ∅.
∪
[b] There exist Ui and Uj such that |Ui | = 4, |Uj | = 2 and U pq = Ujkl = ∅.
∪pq̸=ij
[c] There exist Ui and Uj such that |Ui | = |Uj | = 3 and U pq = Uikℓ = Ujkℓ = ∅.
pq̸=ij ∪
[d] There exist Ui , Uj , and Uk such that |Ui | = 3, |Uj | = |Uk | = 2, and U pq =
∪
U pqr = ∅ for 1 ≤ p ̸= q ̸= r ≤ 4.
pqr ̸=ijk
∪
(iii). | 4n=1 Un | = 7, and one of the following cases is satisfied:
∪
[a] There is Ui with |Ui | = 4 and | U pq ∪ Ujkℓ | = 1. Moreover, U( pq)c = ∅ when-
p,q̸=i
ever |U pq | = 1 for p, q ̸= i.
∪
[b] There exist Ui and Uj such that |Ui | = 3, |Uj | = 2 and | U pq ∪ Ujkℓ | ≤ 1.
p,q̸=i
∪
Moreover, if | U pq ∪ Ujkℓ | = 1, then U( pq)c = ∅ and Uik = Uiℓ = Uikℓ = ∅, and if
p,q̸=i
∪
U pq ∪ Ujkℓ = ∅, then |Uik ∪ Uiℓ | ∈ {1, 2}.
p,q̸=i ∪
[c] There exist Ui , Uj , and Uk such that |Ui | = |Uj | = |Uk | = 2 with | U pq | ≤ 2, in
which at most one of the U pℓ s has exactly one element, and, also, at most two distinct sets’
Mathematics 2023, 11, 1553 14 of 26

Urst s are non-empty for all rst ̸= ijk. Moreover, if |U pq | = 2 or |U pℓ | = 1 for p, q ̸= ℓ, then
at most one of the Urst s is non-empty.
∪4
(iv). | n=1 Un |= 6, and one of the following cases is satisfied:
∪
[a] There is Ui with |Ui | = 3, | U pq ∪ Ujkℓ | ∈ {2, 3} in which |U pq | ≤ 2, and
p,q̸=i
∪ ∪
| U( pq)c | ≤ 1. Moreover, if |U pq | ∈ {1, 2} with |Ujkℓ | = 2, then U( pq)c = ∅.
U pq ̸=∅ U pq ̸=∅
[b] There exist Ui and Uj such that |Ui | = |Uj | = 2 and |Uij ∪ Ukℓ | ≤ 3 with
∪
|Uij |, |Ukℓ | ≤ 2. Additionally, if |Uij | = 2, then |Ukℓ | ≤ 1 and U pq = Uikℓ =
pq̸=ij,k ℓ
∪
Ujkℓ = ∅, and, if |Uij | = 1, then |Ukℓ | ≤ 1 and | U pq | ≤ 1. Moreover, in the case of
pq̸=ij,k ℓ
Uij = ∅, one of the following hold:
∪
[b1] If |Ukℓ | = 2, then | U pq | ≤ 2 in which |U pq | ≤ 1 and
pq̸=ij,k ℓ
∪
U( pq)c = ∅.
U pq ̸=∅
[b2] If |Ukℓ | = 1, then |Urs | ≤ 3 with U(rs)c = ∅ where |Urs | = max |U pq | and
pq̸=ij,k ℓ
∪
| Umn | ≤ 1.
mn̸=ij,k ℓ,rs,(rs)c
∪
[b3] If Ukℓ = ∅, then | U pq | ≤ 4 in which at most three U pq s are non-empty.
pq̸=ij,k ℓ
Furthermore, if |U pq | ∈ {2, 3}, then U( pq)c = ∅.
∪4 ∪
(v). | n=1 Un | = 5; there exists Ui such that |Ui | = 2 and 1 ≤ | U pq | ≤ 6 in which
p,q̸=i
|U pq | ≤ 4. Moreover,
∪ ∪
[a] If |U pq | = 4, then U( pq)c = ∅, | Urs | ≤ 1, and U(rs)c = ∅.
r,s̸=i;rs̸=∪
pq Urs ̸=∅
[b] If |U pq | = 3, then U( pq)c = ∅, | Urs | ≤ 2 and U(rs)c = ∅ whenever
r,s̸=i;rs̸= pq
|Urs | = 2.
[c] In the case of |U pq | = 2, one of the following holds
∪ ∪
[c1] If | Urs | = 4, then U(rs)c = ∅.
r,s̸=i;rs̸= pq Urs ̸=∅
∪ ∪ ∪
[c2] If | Urs | ∈ {2, 3}, then | U(rs)c | ≤ 1. In addition, | U(rs)c | =
r,s̸=i;rs̸= pq Urs ̸=∅ Urs ̸=∅
∪
1 whenever | Urs | = 2 in which exactly two Urs s are non-empty.
r,s̸=i;rs̸= pq
∪ ∪
[c3] If | Urs | ≤ 1, then either U( pq)c = ∅ with 1 ≤ | U(rs)c | ≤ 2 or
r,s̸=i;rs̸= pq Urs ̸=∅
U(rs)c = ∅ with |U( pq)c | ≤ 1.
∪
[d] If |U pq | ≤ 1 for all 1 ≤ p ̸= q ̸= i ≤ 4, then 2 ≤ | U( pq)c | ≤ 3 in which at
U pq ̸=∅
most two distinct U( pq)c s are non-empty.
∪
(vi). | 4n=1 Un | = 4; there exist two non-empty sets Uij and U(ij)c such that 2 ≤ |Uij ∪ U(ij)c | ≤
5, and one of the following cases is satisfied:
[a] If |Uij ∪ U(ij)c | = 5, then either |Uij | = 4 or |U(ij)c | = 4. Further,
∪
U( pq)c = ∅.
pq̸=ij,(ij)c ;U pq ̸=∅
[b] If |Uij ∪ U(ij)c | = 4, then |U pq ∪ U( pq)c | = 2 whenever U pq , U( pq)c ̸= ∅ for
pq ̸= ij. Further, if |Uij | = |U(ij)c | = 2, then at most one pair of U pq , U( pq)c is nonempty
for all pq ̸= ij.
[c] If |Uij ∪ U(ij)c | = 3, then |U pq ∪ U( pq)c | ∈ {2, 3} whenever U pq , U( pq)c ̸= ∅ for
pq ̸= ij. Further, if U(rs)c ̸= ∅ for 1 ≤ r ̸= s ≤ 4 and rs ̸= pq, ij, then |Urs ∪ U(rs)c | ∈
{2, 3} with |(U pq ∪ U( pq)c ) ∪ (Urs ∪ U(rs)c )| ∈ {4, 5}.
Mathematics 2023, 11, 1553 15 of 26

∪
Proof. Assume that γ̃(AG(L)) = 2. Then, by Theorem 1b, we have | 4n=1 Un | ≤ 9. So,
∪
4 ≤ | 4n=1 Un | ≤ 9.
∪
Case 1 Let | 4n=1 Un | = 9. Then, by Theorem 1b, γ̃(AG(L)) = 2 implies |U1 | = 6.
If Uij ̸= ∅ or Uijk ̸= ∅ for some i ̸= 1, then the sets X = U1 and Y = V (AG(L)) \ U1
contain K6,4 , which has a crosscap four. So, Uij , Uijk = ∅ for all i ̸= 1. Here, remember
that every vertex in U1jk is an end vertex, and every vertex in U1j is of degree two. Let G12
∪
be the induced subgraph of AG(L) induced by the vertex subset 4n=1 Un . It is clear that
G12 ∼= K6,1,1,1 , and G12 is a subgraph of the graph H2 given in Figure 2a with the labels
uℓ ∈ U1 (for ℓ = 1, . . . , 6), I2 = v1 , I3 = v2 , and I4 = v3 . By Figure 2b, the N2 -embedding
of G12 contains three different faces with vertices I2 , I3 ; I3 , I4 ;, and I2 , I4 , respectively. So,
any number of vertices in U1j can be embedded into the N2 -embedding of G12 without
edge-crossing. Thus, γ̃(AG(L)) = 2.
∪
Case 2 Let | 4n=1 Un | = 8.
Case 2.1 Suppose |U1 | ∈ {5, 4}. If Uij ̸= ∅ or Uijk ̸= ∅ for some i ̸= 1, then AG(L)
contains K5,4 as a subgraph, which is a contradiction. Therefore, Uij = ∅ and Uijk = ∅
for all i ̸= 1. Now, if |U1 | = 5, then AG(L) is a subgraph of the annihilating-ideal graph
in Case 1 with |U1 | = 6 so that γ̃(AG(L)) = 2. Suppose |U1 | = 4. Here, |U2 | = 2. If
∪
I ∈ Uij ∪ U234 , then AG(L) contains a copy of K4,5 where the partite sets are U1 and
i ̸ =1
U2 ∪ U3 ∪ U4 ∪ { I } so that γ̃(AG(L)) ≥ 3. If U1j ̸= ∅ for some j ∈ {3, 4}, then AG(L)
contains K5,4 − e as a subgraph with the partition sets U1 ∪ U1j and U2 ∪ U3 ∪ U4 so that,
∪
by Remark 1, we have γ̃(AG(L)) ≥ 3. Therefore, Uij = ∅ and U234 = ∅. In this case,
ij̸=12
one can retrieve an N2 -embedding of AG(L) from Figure 4b by changing the label I3′ to I4
and its related edges such that γ̃(AG(L)) = 2.
Case 2.2 Suppose |U1 | = 3. Let |U2 | = 3. If Uij ̸= ∅ or Uijk ̸= ∅ for ij ̸= 12, then
AG(L) contains K4,5 − e, which is a contradiction. Therefore, Uij = ∅ and Uijk = ∅ for
all ij ̸= 12. In this case, the crosscap of AG(L) is same as the crosscap of K3,3,1,1 so that
∪
γ̃(AG(L)) = 2. Let |U2 | = 2 and I ∈ Uij ∪ Uijk .
ijk̸=123

• In the case that I ∈ Uij for ij ∈ {12, 13}, the contraction of AG(L) induced by the
partite sets X = Ui ∪ U4 and Y = Uj ∪ { Ik , [ Ik′ , Iij ]}, where k ∈ / {i, j, 4}, forms a copy
of H4 .
• In the case that I ∈ Uij for ij ∈ {14, 23, 24, 34}, the graph AG(L) contains K5,4 with
the partite sets Ui ∪ Uj ∪ Uij and Uk ∪ Uℓ where k, ℓ ∈ / {i, j}.
∪ ∪
• In the case that I ∈ Uijk , the contraction of AG(L) induced by ( 4n=1 Un \
ijk ̸=123
{ Iℓ }) ∪ {[ Iℓ , I ]} forms H4 where ℓ is the least integer in {1, 2, 3, 4} \ {i, j, k}.
∪
Thus, Uij ∪ Uijk = ∅, and, so, the crosscap of AG(L) is the crosscap of K3,2,2,1 , which
ijk̸=123
is two.
Case 2.3 Suppose |U1 | = 2. Then, K2,2,2,2 is a subgraph of AG(L). Suppose
γ̃(K2,2,2,2 ) = 2. Then, by Euler’s formula, the number of faces in an N2 embedding of
K2,2,2,2 is 16 so that all the faces are triangular, which contradicts the fact that K2,2,2,2 has
no triangular embedding (see [27]). Thus, γ̃(AG(L)) ≥ 3.
∪
Case 3 Let | 4n=1 Un | = 7.
∪
Case 3.1 Suppose |U1 | = 4. If | Uij ∪ Uijk | ≥ 2, then AG(L) contains K4,5 with
i ̸ =1
one partite set X = U1 , and, so, γ̃(AG(L)) ≥ 3. Further, by Theorem 4iv, AG(L) is
∪
projective whenever Uij = Uijk = ∅ for all i ̸= 1. Therefore, | Uij ∪ Uijk | = 1, and let
i ̸ =1
∪
I ∈ Uij ∪ Uijk . Now, if U1j = ∅ for all 2 ≤ j ≤ 4, then it is easy to verify that AG(L)
i ̸ =1
is isomorphic to a subgraph of the graph H1 (see Figure 1a). Therefore, by Lemma 2, we
have γ̃(AG(L)) = 2. So, let U1j ̸= ∅ for some 2 ≤ j ≤ 4. Suppose Ukℓ = ∅ for 2 ≤ j ̸=
k ̸= ℓ ≤ 4. Here, the open neighbor of each vertex in U1j is Ik and Iℓ in AG(L). Let G13 be
Mathematics 2023, 11, 1553 16 of 26

∪
the induced subgraph of AG(L) induced by the vertex subset 4n=1 Un ∪ { I }. Clearly, G13
Mathematics 2023, 1, 0 is a subgraph of the graph H1 given in Figure 1a with the labels uℓ ∈ U1 (for ℓ = 1, 16 . . .of, 4),
25
v1 = I2 , v2 = I3 , v3 = I4 , and v4 = I. Since ( I3 , I4 ), ( I2 , I4 ), ( I2 , I3 ) ∈ E(AG(L)), any
number of vertices in U1j (for 2 ≤ j ≤ 4) can be embedded in the N2 -embedding of G13
(for 2 ≤ edge-crossing,
without j ≤ 4) can beand, therefore, γ̃(AG(L)) = 2. Now, take U ̸= ∅ for 2 ≤ j ̸=
embedded in N 2 -embedding of G13 withoutk ℓedge crossing and
k ̸ = ℓ ≤ set U
therefore γ̃(AG(L)) = 2. Nowkℓ take Ukℓ ̸= ∅ for 2singleton
4. Note that the is nothing but the ≤ j ̸= kset̸={ Iℓ}.≤Now, 4. Noteconsider
that the the
subgraph
set G14 = AG(L)
Ukℓ is nothing but the−singleton
{ I1j , ( Ij , Iset ( IIk}, .IℓNow
k ), { ), ( Ij ,consider
Iℓ ), ( I, Ij )} , which
the subgraphis isomorphic
G14 = AG(L) to K4,4 −
with the partition sets X = U 1 and Y = { I j , Ik , Iℓ , I }
{ I1j , ( Ij , Ik ), ( Ik , Iℓ ), ( Ij , Iℓ ), ( I, Ij )} which is isomorphic to K4,4 with partition sets X = Uof1
. Note that any N 2 -embedding
G14 has Y =eight { Ij , Irectangular faces so that each face shares exactly two vertices from X and Y.
and k , Iℓ , I }. Note that any N2 -embedding of G14 has 8 rectangular faces so that
AG(L)
each face share exactly two adjacent
In , the vertex I1j is vertices from to three X and verticesY. InofAG(L) Y, namely Ik , Iℓ , and
, the vertex I1j isI.adjacent
Therefore, to
one cannot insert
three vertices of Y namely I1j with its edges into
Ik , Iℓ and I. So one N without crossing, which is
2 cannot insert I with its edges in N without a contradiction.
∪ 1j 2
Casea 3.2
crossing, Suppose |U1 | = 3. Then, |U2 | = 2. If | Uij ∪ Uijk | ≥ 2, then it is easy to
contradiction.
∪ i ̸ =1
checkCase theSuppose
that3.2 |U1 | =
of 3. Then |contains
AG(L) U2 | = 2. If | UKij ∪ U ijk | ≥ 2,3,6
then
∪ (it
K4is−easy to acheck
contraction either
i ̸ =1 4,5 − e or K∪ e) as sub-
graph, and, so, by Remark 1, we have γ̃ (AG(L)) ≥ 3. Therefore, |
that the contraction of AG(L) contains either K − e or K ∪ (K − e)ijas a subgraph U ∪ Uijk | ≤ 1. and 4,5 3,6 4 i ̸ =1
∪
Remark|1,∪we
so byAssume = 1. If Uij≱=3. ∅,
∪ Uijkγ̃|(AG(L))
Uijhave then U(ij|)c =Uij∅;∪otherwise,
Therefore Uijk | ≤ 1. the graph induced
i ̸ =1
i∪
̸ =1
Assume | sets Uij ∪XUijk =| U = 1. If Uij ̸= ∅, then U )c = ∅ otherwise the graph induced
by the partition
i ̸ =1 1 ∪ U3 and Y = U2 ∪(ijU 4 ∪ [ Iij , I(ij)c ] form H4 in AG(L) so
thatthe
by partition ≥
γ̃(AG(L)) 3. X
sets = U1 if∪ IU∈
Further, U13 Y
3 and
∪ U=14 U ∪ U∪134 U,4 then∪′ [ Iij′′consider
, I(ij)c ] form the H graph AG(L)so
in AG(L) −
{ I, e , e , e , e , e } ∼
= K − e with the bipartite
2
sets { I , I , I , I } and { I , I ′4, I , I } where
that γ̃(AG(L)) ≥ 3. Further, if I ∈ U13 ∪ U14 ∪ U134 , then
1 2 3 4 5 4,4 1 j
1 1consider the graph i i k AG(L) −
ijk
{e1I, =
e1 , (eI21, ,eI3j,)e,4e,2e5= }∼ ( I1′K, Ij ),−e3e =
= with( I1′′bipartite
, Ij ), e4 =sets ( Ii {, IIk ), ,Ie′ 5, I=′′ , I( I}i′ ,and
Ik )}{, Iand
, I ′ , eI = , I ( I}j ,where
Iijk ). Now,
e1 =
4,4 1 1 1 j i i k ijk
(aI∪similar
, I ) , e argument
= ( I ′ , I ), egiven = ( I ′′ , I G
for ) 2e(refer
, = ( I to
, I Equation
) , e = ( I ′1)
, I leads
)} and to eγ̃= (AG(L))
( I , I ) .≥Now 3. Therefore,
a similar
1 j 2 1 j 3 1 j 4 i k 5 i k j ijk
| Uij ∪ Ugiven ∪
argument ijk | = for 1 with G U U14 = U134
13 = equation
(refer 2
= ∅. to
1) leads In γ̃ this(AG(L))case, with ≥ 3. theTherefore
help of Figure | U10a, ∪ ij
i ̸ =1
i ̸ =1
we obtain γ̃(AG(L)) = 2. Notice that in Figure 10a, we take |U34 | = 1.
U ijk | = 1 with∪U13 = U14 = U134 = ∅. In this case, with the help of Figure 10a, we get
Assume
γ̃(AG(L)) Uij ∪ Uthat
= 2. Notice ijk =in∅. If |U1j10a,
Figure | ≥ we3 fortake
some j∈
|U34 | ={3,1.4}, then the sets X = U2 ∪ Uj′
i∪
̸ =1
and Assume
Y = U ∪U Uij∪∪UUijk, where
= ∅. Ifj′|U∈1j |{≥
3, 43}for
\ {some j ∈ {K3, 4.},So,
j}, form ≤ 2Xfor
then|Uthe| sets =U j 2=∪ 3,
Uj4.
′
1 i ̸ =1 j 1j 3,7 1j
and Y = |U
Suppose ∪ U14 | ≥ 3. Let |U |′ ≥ 2 and |U { j|}≥
1 ∪ U j ∪ U1j′, where′′1jj ∈ {3, 4} \ 1k
U13 1 for Kj, k .∈ So
, form 3,7
{3, |4U}. Then, the subgraph
1j | ≤ 2 for j = 3, 4.
AG(L) − { I , (
Suppose |U131k∪ U114 |j ≥ 3. I , I ) , ( I ,
1 LetI ) ,
j |U ( I , I )} contains
11j | j ≥ 2 and |U1k |3,6 K with the partite sets
≥ 1 for j, k ∈ {3, 4}. Then the X = ∪ Uk and
U2subgraph
Y = U
AG(L)1− { I1k ∪ U ∪ U . Since ′ deg ′′ ( I ) = 3,
, Ij ), ( I1 , Ij ), (KI3,61 , Ijj )} contains
j , ( I11j I j is contained
K3,6 with partite sets X = U2 ∪ Uk andfaces
in exactly three rectangular Y=
{ ′ , I ′′ , I } , ( I1′ , Iin
in any N -embedding of K . Since
U1 ∪ Uj ∪2 U1j . Since degK3,63,6( Ij ) = 3, I1j is1 contained I , I 1 j ⊂ Y, in to embed
exactly the edges
three ( I1 , Ij )faces
rectangular j ),
and I ′′ , I ), the vertices I , I ′ , and I ′′ on
(
any N21-embedding j of K3,6 1 . 1Since { I1 , I ′ , I ′′ , I } ⊂ Y, to embed the edges ( I , I ), ( I ′ , I )
the diagonals of the three rectangular faces that
1 1 1 j 1 j 1 j
contain ′′ I j , respectively, are required.I Now, ′ after ′′ embedding the three edges, I is in exactly
and ( I1 , Ij ) it requires the vertices 1 , I1 and I1 on the diagonals of the threej rectangular
six triangular
faces that contains faces,Iallrespectively.
of which were Now formedafter by using twothe
embedding vertices
three from
edges,Y and
Ij is one vertex
in exactly
j
from X. Therefore, the vertex I cannot be embedded
six triangular faces all of which were formed by using two vertices from Y and one vertex
1k because it is adjacent to I j as well
as two vertices from X.
from X. Therefore the vertex I1k13cannot So, | U ∪ U | ≤ 2. However, AG(L) is projective
14 be embedded because it is adjacent to I along if U 13 ∪ U 14 =
with
j
∅. Thus, 1 ≤ |
two vertices from 13X. So 14 U ∪ U | ≤ 2. Now, one can obtain help
|U13 ∪ U14 | ≤ 2. But AG(L) is projective if U13 ∪ U14 = ∅. Thus from Figure 10b to say that
γ̃ (AG(L)) = 2.
1 ≤ |U ∪ U | ≤ 2. Now one can get help from Figure 10b, to say that γ̃(AG(L)) = 2.
13 14

I1 I4 I1′ I1 I1 I1′ I4 I1
b b b b b b b b

I1′′ I14
I2′ b b b I34 b I3 I2′ b b
I1′′ b b
I2
b
I134 I3
I3 b b I2 b I2′ I2 b b b b I2′
′
I14
b b b b b b b b

I1 I4 I1′ I1 I1 I1′ I4 I1
∪ ∪
(a). | Uij ∪ Uijk | = 1 and U13 = U14 = U134 = ∅ (b). Uij ∪ Uijk = ∅ and 1 ≤ |U13 ∪ U14 | ≤ 2
i ̸ =1 i ̸ =1
∪
∪44
Figure 10. | U | = 7 with |U11| = 33.̸
=11 nn
nn=

Case 3.3 Suppose |U1 | = 2.

Claim A: At most two distinct Uij ’s are non-empty in which at most one Ui4 is non-
empty for 1 ≤ i ̸= j ≤ 4. Also at most two distinct Uℓmn ’s are non-empty for ℓmn ̸= 123.
Assume on the contrary that at least three Uij ’s are non-empty for 1 ≤ i, j ≤ 4, say
Ui1 i2 , Ui3 i4 and Ui5 i6 are non-empty. Let p ∈ {1, 2, 3} \ {i1 , i2 }, q ∈ {1, 2, 3} \ { p, i3 , i4 } and
Mathematics 2023, 11, 1553 17 of 26

Case 3.3 Suppose |U1 | = 2.

Claim A: At most two distinct Uij s are non-empty in which at most one Ui4 is non-
empty for 1 ≤ i ̸= j ≤ 4. Additionally, at most two distinct Uℓmn s are non-empty for
ℓmn ̸= 123.
Assume on the contrary that at least three Uij s are non-empty for 1 ≤ i, j ≤ 4; say,
Ui1 i2 , Ui3 i4 and Ui5 i6 are non-empty. Let p ∈ {1, 2, 3} \ {i1 , i2 }, q ∈ {1, 2, 3} \ { p, i3 , i4 }
and r ∈ {1, 2, 3} \ { p, q, i5 , i6 }. If r exists, then the minor subgraph induced by the vertices
[ I p , Ii1 i2 ], I p′ , [ Iq , Ii3 i4 ], Iq′ , [ Ir , Ii5 i6 ], Ir′ , and I4 forms K7 in AG(L), which is a contradiction. If r
does not exist, then take r as {1, 2, 3} \ { p, q} and form a minor of AG(L) with the partite
sets X = { Ir , Ir′ , I4 , Ir4 } and Y = {[ I p , Ii1 i2 ], I p′ , [ Iq , Ii3 i4 ], Iq′ }, which is isomorphic to either
H3 or H4 , as in Figure 3. So, by Lemma 3, we have γ̃(AG(L)) ≥ 3. Therefore, only at
most two distinct Uij s can be non-empty for 1 ≤ i ̸= j ≤ 4. Further, if Um4 , Un4 ̸= ∅ for
some 1 ≤ m ̸= n ≤ 4, then the subgraph induced by the sets X = Um ∪ Um4 ∪ { Ik } and
Y = Un ∪ U4 ∪ {[ Ik′ , In4 ]}, where k ̸= m or n, form H4 which has a crosscap of at least three.
Note that all the vertices in U123 are end vertices in AG(L). If Uijk , Uℓmn , and U pqr are
non-empty for ijk, ℓmn, pqr ̸= 123, then the minor subgraph induced by
{[ I(ijk)c , Iijk ], I(′ ijk)c , [ I(ℓmn)c , Iℓmn ], I(ℓ ′
mn)c
, [ I( prq)c , I pqr ], I(′ pqr)c , I4 } is K7 , which is a contradic-
tion. Therefore, at most two distinct Uℓmn s are non-empty for ℓmn ̸= 123.
Claim B: |Uij | ≤ 2 and |Ui4 | ≤ 1 for all 1 ≤ i < j ≤ 3.
If |Uij | ≥ 3 for some 1 ≤ i, j ≤ 3, then AG(L) contains K7,3 as a subgraph with the
partite sets X = Ui ∪ Uj ∪ Uij and Y = Uk ∪ U4 where k ∈ {1, 2, 3} \ {i, j}. Additionally,
if |Ui4 | ≥ 2 for some 1 ≤ i ≤ 3, then AG(L) contains K5,4 as a subgraph with the partite
sets X = Ui ∪ U4 ∪ Ui4 and Y = Uj ∪ Uk where j, k ∈ {1, 2, 3} \ {i }. Thus, |Uij | ≤ 2 and
|Ui4 | ≤ 1 for all 1 ≤ i < j ≤ 3.
Assume |Uij | = 2 for some 1 ≤ i, j ≤ 3. Suppose Ukℓ ̸= ∅ for some 1 ≤ k < ℓ ≤ 4
and k ℓ ̸= ij. Let us take j ∈ / {k, ℓ} ∩ {i, j}. Then, AG(L) contains K6,3 ∪ (K4 − e) with the
partite sets X = { Ii , Ii′ , Ij , [ Ij′ , Ikℓ ], Iij , Iij′ } and Y = Um ∪ U4 where m ∈ {1, 2, 3} \ {i, j}. So,
by Remark 1, γ̃(AG(L)) ≥ 3. Therefore, Ukℓ = ∅. In this case, the number of Uijk cannot
be more than one because here AG(L) contains K6,3 ∪ (K4 − e). For the remaining cases,
by Figure 11a, we obtain γ̃(AG(L)) = 2.
Assume |Uij | ≤ 1 for all 1 ≤ i, j ≤ 3. Suppose |Uk4 | = 1 for some 1 ≤ k ≤ 3. If there
are two Uℓmn s that are non-empty for ℓmn ̸= 123, then it is not hard to verify that AG(L)
contains a subgraph similar to the structure of H3 , which has a crosscap of at least three.
For all the remaining cases, that is |Uij | = |Uk4 | = 1 with unique Uℓmn ̸= ∅ or |Uij | ≤ 1
and |U pq | ≤ 1 with at most two Uℓmn s that are non-empty for 1 ≤ i, j, k, p, q ≤ 3 and
ℓmn ̸= 123, one can use Figure 11b to obtain γ̃(AG(L)) = 2.

∪4
Figure 11. | n=1 Un | = 7 with |U1 | = 2.
∪4
Case 4 Let | n=1 Un | = 6.
Mathematics 2023, 11, 1553 18 of 26

Case 4.1 Suppose |U1 | = 3. Note that each vertex of Uij for i = 1 is adjacent to exactly
two vertices Ii′ and Ij′ for i′ , j′ ∈/ {i, j} and ( Ii′ , Ij′ ) ∈ E(AG(L)), so we do not want to
bother about U1j and U1jk for all 2 ≤ j < k ≤ 4. If |Uij | ≥ 3 for some i ̸= 1, then AG(L)
contains K4,5 as a subgraph with the partite sets X = U1 ∪ Uk and Y = Ui ∪ Uj ∪ Uij where
k ∈ {2, 3, 4} \ {i, j}, which is a contradiction. So, |Uij | ≤ 2 for all i ̸= 1.
((i).Assume |Uij | = 2 for some i ̸= 1. If U(ij)c ̸= ∅, then the sets X = Ui ∪ Uj ∪ Uij
and Y = U1 ∪ Uk ∪ U(ij)c form K4,5 in AG(L), and, if Ukℓ ̸= ∅ for some k ̸= 1 with
kℓ ̸= ij or U234 ̸= ∅, then AG(L) contains K4,5 − e so that γ̃(AG(L)) ≥ 3. If not, that is
U(ij)c , Ukℓ , U234 = ∅ for all k ̸= 1 with k ℓ ̸= ij, then by Figure 12a, we have γ̃(AG(L)) = 2.
(ii). Assume |Uij | ≤ 1 for all i ̸= 1. If U(i1 j1 )c ̸= ∅ and U(i2 j2 )c ̸= ∅ for some
Ui1 j1 ̸= ∅ and Ui2 j2 ̸= ∅, then the sets X = Ui1 ∪ Uj1 ∪ Ui1 j1 ∪ {[ Ii2 j2 , I(i2 j2 )c ]} and Y =
U1 ∪ Um ∪ U(i1 j1 )c , where m ̸= i1 , j1 , contains K4,5 − e in AG(L). Additionally, if |U(ij)c | ≥ 3,
then the sets X = Ui ∪ Uj ∪ Uij and Y = U1 ∪ Um ∪ U(ij)c , where m ̸= i, j, form K3,7
in AG(L), which is a contradiction. So, at most one of the sets U(ij)c is non-empty with
|U(ij)c | ≤ 2.
∪
Let |U(ij)c | = 2. If I ∈ Ukℓ ∪ U234 , then the sets X = { Ii , Ij , Iij } and Y =
k ℓ̸=ij
{ I1 , I1′ , [ I1′′ , I ], Im , I(ij)c , I(′ ij)c }, where m ̸= i, j, form K3,6 ∪ (K4 − e) so that, by Remark 1,
∪
γ̃(AG(L)) ≥ 3. Therefore, Ukℓ ∪ U234 = ∅. For this case, readers can verify the
k ̸=1;k ℓ̸=ij
N2 -embedding of AG(L).
∪
Let |U(ij)c | = 1. If I, J ∈ Ukℓ ∪ U234 with |Ukℓ | ≤ 1, then the sets { Ii , Ij , Im , I1 ,
k ̸=1;k ℓ̸=ij
∪
[ I1′ , I ], [ I1′′ , J ], [ Iij , I(ij)c ]} form K7 . Therefore, | Ukℓ ∪ U234 | = 1.
k ̸=1;k ℓ̸=ij
∪ ∪
Let U(ij)c = ∅. Then, by Theorem 4iii[a], AG(L) is projective if | Uij ∪ Uijk | ≤ 1.
i ̸ =1 i ̸ =1
∪
If | Uij ∪ Uijk | ≥ 4, then K3,7 is a subgraph of AG(L) with the partite sets X = U1
i ̸ =1
∪
and Y = V (AG(L)) \ U1 . So, in the case of U(ij)c = ∅, γ̃(AG(L)) = 2 whenever
i ̸ =1
∪
2≤| Uij ∪ Uijk | ≤ 3 with |Uij | ≤ 1 (refer to Figure 12b).
i ̸ =1

∪4
Figure 12. | n=1 Un | = 6 with |U1 | = 3.

Case 4.2 Suppose |U1 | = 2. Then, |U2 | = 2 and |U3 | = |U4 | = 1. If |U34 | ≥ 3, then the
partite sets X = U1 ∪ U2 and Y = U3 ∪ U4 ∪ U34 form K4,5 as a subgraph in AG(L), which
is a contradiction.
Case 4.2.1 Assume |U34 | = 2. Then, U( pq)c = ∅ for all U pq ̸= ∅; otherwise, the sets
X = U1 ∪ U2 and Y = U3 ∪ U4 ∪ U34 ∪ {[ I pq , I( pq)c ]} form K4,5 in AG(L). In particular,
U12 = ∅.
′ , ( I , I ),
If |Uij | ≥ 2 for some ij ̸= 12, 34 and i < j, then the subgraph AG(L) − { I34 , I34 i j
( Ii , Ij )} contains K3,5 with the partite sets X = Ui ∪ Uj ∪ Uij and Y = Ui′ ∪ Uj′ where i′ ∈
′
Mathematics 2023, 11, 1553 19 of 26

{1, 2} \ {i } and j′ ∈ {3, 4} \ { j}. Note that any N2 -embedding of K3,5 has one hexagonal
and six rectangular faces, and the vertices I34 and I34 ′ are adjacent to I , I ′ , I ′ and I ′ . So, to
i i i i′
′
insert I34 and I34 into an N2 -embedding of K3,5 , we require two faces, say F1 and F2 , which
contains Ii , Ii′ , Ii′ , and Ii′′ . If either F1 or F2 is hexagonal, then the corresponding face may
adopt one of the edges ( Ii , Ij ) or ( Ii′ , Ij ). Let us take that the edge ( Ii , Ij ) is embedded. Now,
to insert an edge ( Ii′ , Ij ), a rectangular face containing Ii′ and Ij as diagonals is required.
However, no such rectangular face exists because the edges ( Ii′ , Ii′ ) and ( Ii′ , Ii′′ ) have been
used twice in F1 and F2 , which is a contradiction.
∪
For all of the remaining cases, that is | Uij | ≤ 2 with |Uij | ≤ 1 and U( pq)c = ∅
ij̸=12,34
when U pq ̸= ∅ for 1 ≤ p ̸= q ≤ 4, we have γ̃(AG(L)) = 2 (refer to Figure 13a).

∪4
Figure 13. | n=1 Un | = 6 with |U1 | = 2.

Case 4.2.2 Assume that |U34 | = 1. Let us take ij ̸= 12, 34.

Let |U12 | ≥ 3, then the subgraph of AG(L) induced by the sets X = U3 ∪ U4 ∪ U34
and Y = U1 ∪ U2 ∪ U12 contains K3,7 so that γ̃(AG(L)) ≥ 3. Thus, |U12 | ≤ 2.
∪
Let |U12 | = 2. If I ∈ Uij ∪ U134 ∪ U234 , then AG(L) contains K3,6 ∪ (K4 − e),
ij̸=12,34
∪
so that, by Remark 1, γ̃(AG(L)) ≥ 3. Therefore, Uij ∪ U134 ∪ U234 = ∅, and in this
ij̸=34,12
case, by Figure 13b, we obtain γ̃(AG(L)) = 2.
Let |U12 | = 1. If |Uij | ≥ 2, then the partite sets X = Ui′ ∪ Uj′ and Y = { Ii , Ii′ , Ij , Iij , Iij′ ,
[ I34 , I12 ]} where i′ ∈ {1, 2} \ {i } and j′ ∈ {3, 4} \ { j} form a minor subgraph K3,6 ∪ (K4 − e)
in AG(L) so that, by Remark 1, γ̃(AG(L)) ≥ 3. If Uij , Ukℓ ̸= ∅ for ij, kℓ ̸= 12, 34
where {i, j} ∩ {k, ℓ} = j = ℓ, then the partite sets X = { Ii , [ Ii′ , Ikℓ ], Iℓ , Iij } and Y =
{ Ik , Ik′ , Im , [ I34 , I12 ]} where m ∈
/ {i, j, k} form ( H4 ∪ (u2 , u3 )) − (u1 , u4 ). A slight modifi-
cation of the proof for H4 in Lemma 3 yields γ̃(AG(L)) ≥ 3. Further, minor changes to
∪
the labels in Figure 13a give γ̃(AG(L)) = 2 whenever | Uij | ≤ 1.
ij̸=12,34
Let U12 = ∅. Then U( pq)c = ∅ for all U pq ̸= ∅; otherwise, AG(L) contains K8 − 4e,
which is isomorphic to ( H4 ∪ (u1 , u3 )) − (v1 , v2 ), so Lemma 3 gives us γ̃(AG(L)) ≥ 3. If
|Uij | ≥ 4, then the partite sets X = Ui ∪ Uj ∪ Uij and Y = Ui′ ∪ Uj′ where i′ ∈ {1, 2} \ {i }
and j′ ∈ {3, 4} \ { j} contain K7,3 in AG(L), which is a contradiction. Suppose |Uij | ∈
{2, 3}. If |Ukℓ | ≥ 2 for some kℓ ̸= ij where {k, ℓ} ∩ {i, j} = k = i, then the subgraph
G15 = AG(L) − { I34 , Ikℓ , Ik′ ℓ , ( Ii , Ij ), ( Ii′ , Ij )} contains K5,3 with the partite sets X = Ui ∪
Uj ∪ Uij and Y = Ui′ ∪ Uj′ where i′ , j′ ∈ / {i, j}. Note that any N2 -embedding of K5,3 has
one hexagonal and six rectangular faces. Further, in AG(L), I34 is adjacent to Ii , Ii′ , Ii′ , Ii′′ ,
and, also, Ikℓ , Ik′ ℓ are adjacent to Ii′ , Ii′′ , Ij . So, to embed the vertices I34 , Ikℓ , and Ik′ ℓ , one
hexagonal and two rectangular faces containing both Ii′ and Ii′′ are required. In such a
 Mathematics 2023, 11, 1553 20 of 26

case, one cannot find two rectangular faces with the diagonal vertices Ii , Ij and Ii′ , Ij . So,
either the edge ( Ii , Ij ) or ( Ii′ , Ij ) cannot be drawn without crossing, which is a contradiction.
Thus, we obtain the result as in the statement-(iv)[b2].
Case 4.2.3 Suppose U34 = ∅.
If |Uij ∪ U(ij)c | ≥ 4 for some ij ∈ / {12, 34}, then the sets X = Ui ∪ Uj ∪ Uij and Y =
Ui′ ∪ Uj′ ∪ U(ij)c where i , j ∈′ ′ / {i, j} form a complete bipartite graph whose crosscap is
more than two.
Let |Uij | ∈ {2, 3} for some ij ∈ / {12, 34}. Then, clearly, U(ij)c must be empty. Let k ℓ ∈
/
{12, 34, ij, (ij)c }. If |Uij ∪ Ukℓ ∪ U(kℓ)c | ≥ 5, then the sets X = Ui ∪ Uj ∪ Uij ∪ {[ Ikℓ , Ikℓc ]}
and Y = Ui′ ∪ Uj′ where i′ ∈ {1, 2} \ {i } and j′ ∈ {3, 4} \ { j} form K6,3 ∪ (K4 − e) and, by
Remark 1, γ̃(AG(L)) ≥ 3. Therefore, 2 ≤ |Uij ∪ Ukℓ ∪ U(kℓ)c | ≤ 4. Now, there are at most
three possibilities:
(i). |Uij | = 3 and |Ukℓ | = 1; this case is pictured in Figure 14.
(ii). |Uij | = 2 and |Ukℓ | = |U(kℓ)c | = 1; this case is pictured in Figure 15a.
atics 2023, 1, 0 (iii). |Uij | = |Ukℓ | = 2; this case is pictured in Figure 15b. 20 of 25
Thus, in all these cases, we have γ̃(AG(L)) = 2.

I3 I2′ I2 I3
b b b b

b
I134
I1 b b
′′ I23 b b
I1′
I23
b

I234 I4
I1′ b b
′
b b I1
I23 b I13
b b
b b

I3 I2′ I2 I3

U34 = ∅ with |Uij | = 3 and |Ukℓ | = 1 for some ij, kℓ ̸= 12, 34

∪4
Figure 13. | Un | = 6 with
∪ |U | = |U | = 2.
n=1Figure 14. | 4n=11Un | = 62 with |U1 | = |U2 | = 2.

I2 I3 I24 I2 I3 I24
b b b b b b

b b
I24 ′
I24
I1 b b
I2′ b b
I1′ I1 b b
I2′ b b I1′
I23 I23

I1′ b
′
b b I4 b
I1 I1′ b
′
b b I4 b
I1
I23 I23
b
I13 b
b b b b b

I2 I3 I24 I2 I3 I24

(a). U34 = ∅ with |Uij | = 2 and |Ukℓ | = |U(kℓ)c | = 1 for some ij, kℓ ̸= 12, 34

(b). U34 = ∅ with |Uij | = |Ukℓ | = 2 for some ij, k ℓ ̸= 12, 34

∪4 ∪
Figure 14. | Un | = 15.
n=1Figure | 4n|=U11U
6 with | n=| |U62 |with
= = 2. |U1 | = |U2 | = 2.

Let |Uij | ≤ 1 for

Letall|Uijij |∈
/≤{112,for 34all}. ij
Then ∈ at least
/ {12, 34}. one Then, Uijat= ∅ for
least oneijU∈ /ij {=12,
∅34 }. ijOth-
for ∈
/ {12, 34}. Other-
erwise the graph induced by { I , I ′ , I , I ′ , I , I , [ ′I , I ′ ], [ I , I ]} form K − 3e in AG(L).
wise, the graph induced 1 1 2 2by 3{ I14, I 13 , I2 , I
24 , I3 ,14I 4 , [ I
2313 , I24 ] , [ I148, I23 ]} forms K8 − 3e in AG(L).
1 2
Clearly γ̃(K8 −Clearly,
3e) ≥ 3γ̃(because
K8 − 3e)the ≥ 3number
becauseof thefaces
number in Nof 2 -embedding
faces in the of K8 − 3e is 17of K8 − 3e is 17,
N2 -embedding
2| E | 2 | E |
which contradicts thecontradicts
which well-known thefact that | F| must
well-known fact thatbe greater must thanbe the girththan
greater valuethe(re-
girth value (refer
∪ ∪ | F|
fer Observationto1). Therefore 1).
Observation | Therefore, Uij | ≤ |3. Thus Uijby| ≤ [20, 3. Theorem
Thus, by2.6iib)], we have
[20, Theorem 2.6iib)], we have
ij̸=12,34 ij̸=12,34
∪ ∪
(AG(L)) |= 2 whenever
γ̃(AG(L)) = 2 γ̃whenever Uij | = 3.| Uij | = 3.
ij̸=12,34 ij̸=12,34
∪4
Case 5 Let | n=1 Un | = 5. Then |U1 | = 2. If Uij = ∅ for all 1 ≤ i < j ≤ 4, then
γ̃(AG(L)) ≤ 1. Observe that we do not want to consider the sets Uij for i ̸= 1 whenever
U(ij)c = ∅ because every vertex in Uij is adjacent to Ii , Ij and ( Ii , Ij ) ∈ E(AG(L)). If
|Uij | ≥ 5 for some i ̸= 1, then the sets X = Ui ∪ Uj ∪ Uij and Y = Ui′ ∪ Uj′ where
′ ′
Mathematics 2023, 11, 1553 21 of 26

∪
Case 5 Let | 4n=1 Un | = 5. Then, |U1 | = 2. If Uij = ∅ for all 1 ≤ i < j ≤ 4,
then γ̃(AG(L)) ≤ 1. Observe that we do not want to consider the sets Uij for i ̸= 1
whenever U(ij)c = ∅ because every vertex in Uij is adjacent to Ii , Ij and ( Ii , Ij ) ∈ E(AG(L)).
If |Uij | ≥ 5 for some i ̸= 1, then the sets X = Ui ∪ Uj ∪ Uij and Y = Ui′ ∪ Uj′ where
i ′ , j′ ∈
/ {i, j} form K3,7 in AG(L), which is a contradiction.
Case 5.1 Assume |Uij | = 4 for some i ̸= 1. Then, U(mn)c = ∅ whenever Umn ̸= ∅;
otherwise, the sets X = Ui ∪ Uj ∪ Uij ∪ {[ Imn , I(mn)c ]} and Y = Ui′ ∪ Uj′ where i′ , j′ ∈ / {i, j}
form K7,3 as a minor of AG(L). Similarly, U(ij)c = ∅; otherwise K6,4 is a minor of AG(L).
If |Ukℓ | ≥ 2 for some k ̸= 1 and kℓ ̸= ij, then the subgraph G16 = AG(L) − { Ikℓ , Ik′ ℓ , ( Ii , Ij )}
contains K6,3 with the partition sets X = Ui ∪ Uj ∪ Uij and Y = U1 ∪ Ui′ where i′ ∈ / {1, i, j}.
Since {i, j} ∩ {k, ℓ} ̸= ∅, let {i, j} ∩ {k, ℓ} = i = k. Clearly, j ∈ {2, 3, 4} \ {k, ℓ}. Note that
each face in any N2 -embedding of K6,3 is rectangular, and the vertices Ikℓ , Ik′ ℓ are adjacent
to I1 , I1′ and Ij . Therefore, to insert Ikℓ and Ik′ ℓ , two rectangular faces that contain I1 , I1′ and
Ij are required. Next, to insert the edge ( Ii , Ij ), a rectangular face with the diagonals Ii and
Ij is required. However, the edges ( I1 , Ij ) and ( I1′ , Ij ) have been used twice to form the first
two rectangular faces. So, one cannot construct another rectangular face that contains Ii
and Ij with a single left-out vertex of Y, which is a contradiction.
Therefore, for the remaining case, that is, |Ukℓ | ≤ 1 for all k ̸= 1 and k ℓ ̸= ij with
U(mn)c = ∅ whenever Umn ̸= ∅, by using Figure 16a, one can have γ̃(AG(L)) = 2.

∪4
Figure 16. | n=1 Un | = 5 with |U1 | = 2.

Case 5.2 Assume |Uij | = 3 for some i ̸= 1. Let p ∈ / {1, i, j}. Clearly, U(ij)c = ∅;
otherwise, the sets X = Ui ∪ Uj ∪ Uij and Y = U1 ∪ U p ∪ U(ij)c form K5,4 .
If |Ukℓ | = 3 for some k ̸= 1 and kℓ ̸= ij, then the subgraph G15 ′ = AG(L) −
′ ′′ ′
{ Ikℓ , Ikℓ , Ikℓ , ( Ii , Ij ), ( I1 , I p ), ( I1 , I p )} has a similar structure of G15 with the partite sets X =
Ui ∪ Uj ∪ Uij and Y = U1 ∪ U p , and so γ̃(AG(L)) ≥ 3. Suppose |Ukℓ |, |Umn | = 2
for k, m ̸= 1 and kℓ, mn ̸= ij. Let {i, j} ∩ {k, ℓ} = i = k. Then, G17 = AG(L) −
{ Ikℓ , Ik′ ℓ , Imn , Imn
′ , ( I , I )} has K
i j 5,3 with the partite sets X = Ui ∪ U j ∪ Uij and Y = U1 ∪ Uℓ .
Any N2 -embedding of K5,3 has one hexagonal and six rectangular faces. Notice that Ikℓ , Ik′ ℓ
are adjacent to I1 , I1′ , Ij , and Imn , Imn ′ are adjacent to I , I ′ , I . So, to embed I , I ′ , I , and
1 1 i k ℓ k ℓ mn
Imn , one hexagonal and two rectangular faces containing both I1 and I1′ are required. How-
′

ever, the edge ( Ii , Ij ) cannot be drawn without crossing, which is a contradiction. There-
∪
fore, | Ukℓ | ≤ 3 and |Ukℓ | ̸= 3.
k ̸=1;k ℓ̸=ij
∪
Suppose | Ukℓ | = 3. Since |Ukℓ | ̸= 3 for all k ̸= 1 and k ℓ ̸= ij, we have |Ukℓ | = 2
k ̸=1;k ℓ̸=ij
and |Umn | = 1 for some m ̸= 1 and mn ̸= ij, kℓ. Next, we claim that U(kℓ)c = U(mn)c = ∅.
If U(kℓ)c ̸= ∅, then by letting {i, j} ∩ {k, ℓ} = i = k, K7,3 can be formed by the sets X =
Ui ∪ Uj ∪ Uij ∪ Ukℓ and Y = U1 ∪ {[ Iℓ , I(kℓ)c ]}. If U(mn)c ̸= ∅, then AG(L) has a similar
structure to G15 , so that γ̃(AG(L)) ≥ 3.
Mathematics 2023, 11, 1553 22 of 26

∪
Suppose | Ukℓ | ≤ 2. As mentioned, U(kℓ)c = ∅ when |Ukℓ | = 2 for k ̸= 1 and
k̸=1;k ℓ̸=ij
kℓ ̸= ij. Suppose |Ukℓ | = 1 and |U(kℓ)c | ≥ 2. Then, AG(L) − { Ikℓ , I(kℓ)c , I(′ kℓ)c , ( Ii , Ij ), ( I1 , Iℓ ),
and
( I1′ , Iℓ )} has K5,3 with the partite sets X = Ui ∪ Uj ∪ Uij and Y = U1 ∪ Uℓ . Note that
any N2 -embedding of K5,3 has one hexagonal and six rectangular faces, Ikℓ is adjacent to
I1 , I1′ , Ij , I(kℓ)c , I(′ kℓ)c , and I(kℓ)c , I(′ kℓ)c are adjacent to Ik , Iℓ , Ikℓ . So, the three vertices
Ikℓ , I(kℓ)c , I(′ kℓ)c together with the edges ( Ii , Ij ), ( I1 , Iℓ ), ( I1′ , Iℓ ) cannot be embedded, and,
also
, γ̃(AG(L)) ≥ 3. Therefore, |Ukℓ ∪ U(kℓ)c | ≤ 2. Further, if |Ukℓ ∪ U(kℓ)c | = |Uℓm ∪ U(ℓm)c | =
2 for kℓ ̸= ij and ℓm ̸= ij, kℓ, then AG(L) contains K3,7 , which is a contradiction.
∪
Thus, an N2 -embedding of AG(L) can be retrieved from Figure 16a for | U pq | ≤ 3
pq̸=ij
with U( pq)c = ∅ if |U pq | = 2.
Case 5.3 Assume |Uij | = 2 for some i ̸= 1. Clearly, |U(ij)c | ≤ 1; otherwise, the sets
X = Ui ∪ Uj ∪ Uij and Y = U1 ∪ U p ∪ U(ij)c where p ∈ / {1, i, j} form K5,4 .
If |Ukℓ |, |Umn | = 2 for k, m ̸= 1 and kℓ, mn ̸= ij, then U(ij)c , U(kℓ)c , U(mn)c = ∅. Further,
an N2 -embedding of AG(L) in the case of |Uij | = |Ukℓ | = |Umn | = 2 is given in Figure 16b
so that γ̃(AG(L)) = 2.
Suppose |Ukℓ | = 2, |Umn | ≤ 1 for k, m ̸= 1 and kℓ, mn ̸= ij. If U(ij)c , U(kℓ)c ̸= ∅,
then the sets X = U1 ∪ U p ∪ U(ij)c and Y = Ui ∪ Uj ∪ Uij ∪ {[ I(kℓ) , I(kℓ)c ]} where p ∈ /
{1, i, j} form K5,4 − e in AG(L) so that, by Remark 1, we have γ̃(AG(L)) ≥ 3. Fur-
ther, since |Ukℓ | = 2, we have |U(kℓ)c | ≤ 1. Therefore, |U(ij)c ∪ U(kℓ)c | ≤ 1. Suppose
|U(ij)c ∪ U(kℓ)c | = 1, say U(ij)c ̸= ∅. Then, U(mn)c = ∅; otherwise, X = U1 ∪ U p ∪ U(ij)c
and Y = Ui ∪ Uj ∪ Uij ∪ {[ I(mn) , I(mn)c ]} where p ∈ / {1, i, j} form K4,5 − e in AG(L). So,
|Umn ∪ U(mn)c | ≤ 1. Suppose not, that is, U(ij)c , U(kℓ)c = ∅, then |U(mn)c | ≤ 1; other-
wise, AG(L) − { Iij , Iij′ , Ikℓ , Ik′ ℓ , ( Im , In ), ( I1 , Im′ ), ( I1′ , Im′ )} ∼
= K5,3 with the partite sets X =
U1 ∪ Um′ ∪ U(mn)c and Y = Um ∪ Un ∪ Umn where m ∈ ′ / {1, m, n} is a similar structure to
G17 which has a crosscap of at least three. So, |Umn ∪ U(mn)c | ≤ 2.
Suppose |Ukℓ |, |Umn | ≤ 1 for k, m ̸= 1 and kℓ, mn ̸= ij. Then, by Theorem 4(ii),
∪ ∪ ∪
γ̃(AG(L)) = 2 provided | Ukℓ | = 2 with | U( pq)c | = 1 or | Ukℓ | = 1 with
k ̸=1;k ℓ̸=ij p ̸ =1 k ̸=1;k ℓ̸=ij
∪
|U(ij)c | = 1, U(kℓ)c = ∅ or U(ij)c = ∅, |U(kℓ)c | ≤ 2 or Ukℓ = ∅ with |U(ij)c | = 1.
k ̸=1;k ℓ̸=ij
∪ ∪
Hence, γ̃(AG(L)) = 2 whenever 4 ≤ | Uij ∪ U(ij)c | ≤ 6 with | U(ij)c | ≤ 1 or
i ̸ =1 i ̸ =1
∪ ∪
| Uij | = 3 with |Uij ∪ U(ij)c | ≤ 3 and a unique U(ij)c ̸= ∅ or Uij = 2 with |U(ij)c | = 1.
i ̸ =1 i ̸ =1
Case 5.4 Assume |Uij | = 1 for all i ̸= 1. Then, |U(ij)c | ≤ 3; otherwise, the sets
X = Ui ∪ Uj ∪ Uij and Y = U1 ∪ Ui′ ∪ U(ij)c where i′ ∈ / {1, i, j} form K3,7 .
Suppose |Ukℓ | = |Umn | = 1 for k, m ̸= 1 and kℓ, mn ̸= ij. If U(ij)c , U(kℓ)c , U(mn)c ̸= ∅,
then the sets X = U1 ∪ U2 ∪ U3 and Y = { I4 , [ Iij , I(ij)c ], [ Ikℓ , I(kℓ)c ], [ Imn , I(mn)c ]} form
H4 as a minor of AG(L), which is a contradiction. Assume that |U(ij)c | = 3. If I ∈
U(kℓ)c ∪ U(mn)c , then G18 = AG(L) − { I, Ikℓ , Imn , ( Ii , Ij ), ( I1 , Ii′ ), ( I1′ , Ii′ )} contains K6,3 with
the partite sets X = U1 ∪ Ui′ ∪ U(ij)c and Y = Ui ∪ Uj ∪ Uij and any N2 -embedding of
K3,6 has nine rectangular faces. Here, it is not hard to verify that all the left-out vertices
and edges cannot be embedded into the nine rectangular faces so that γ̃(AG(L)) ≥ 3.
Therefore, U(kℓ)c ∪ U(mn)c = ∅. Here, the graph AG(L) − { Ikℓ , Imn } is a subgraph of
the graph in Figure 2a, and the suitable labels in Figure 2b give two different faces in
the N2 -embedding of AG(L) − { Ikℓ , Imn } that contains the vertices N ( Ikℓ ) and N ( Imn ) so
that γ̃(AG(L)) = 2. Assume |U(ij)c | ≤ 2. If |U(ij)c ∪ U(kℓ)c | ≥ 4, then the subgraph
AG(L) − { I(kℓ)c , I(′ kℓ)c , Imn , ( Ii , Ij ), ( I1 , Ii′ ), ( I1′ , Ii′ )} has a similar structure to G15 so that we
∪
have γ̃(AG(L)) ≥ 3. Additionally, by Theorem 4ii, AG(L) is projective when | U(ij)c | ≤
i ̸ =1
1. For all of the remaining cases, γ̃(AG(L)) = 2 can be verified by drawing the N2 -
Mathematics 2023, 11, 1553 23 of 26

embedding.
∪
Thus, γ̃(AG(L)) = 2 when 2 ≤ | U(ij)c | ≤ 3 with at least one of the sets’ U(ij)c = ∅.
i ̸ =1
Suppose |Ukℓ | = 1 and Umn = ∅ for k, m ̸= 1 and kℓ, mn ̸= ij. If |U(ij)c | = 3 and
U(kℓ)c ̸= ∅, then the subgraph AG(L) − { I(kℓ)c , Ikℓ , ( Ii , Ij ), ( I1 , Ii′ ), ( I1′ , Ii′ )} has a similar
structure to G18 , and, if |U(ij)c | = |U(kℓ)c | = 2, then the subgraph
′ ′
AG(L) − { I(kℓ)c , I(kℓ)c , Ikℓ , ( Ii , Ij ), ( I1 , Ii′ ), ( I1 , Ii′ )} has a similar structure to G15
so that γ̃(AG(L)) ≥ 3. Further, AG(L) is projective if |U(ij)c ∪ U(kℓ)c | ≤ 1. Thus,
γ̃(AG(L)) = 2 whenever |U(ij)c ∪ U(kℓ)c | ∈ {2, 3}.
Suppose Ukℓ , Umn = ∅ for k, m ̸= 1 and kℓ, mn ̸= ij. Then, γ̃(AG(L)) = 2 whenever
2 ≤ |U(ij)c | ≤ 3.
∪
Case 6 Let | 4n=1 Un | = 4. Then, by Theorem 4(i), |Uij ∪ U(ij)c | ≥ 3 for
some Uij , U(ij)c ̸= ∅. Further, if |Uij ∪ U(ij)c | ≥ 6 with Uij , U(ij)c ̸= ∅, then the subgraph
∪
induced by the sets X = Ui ∪ Uj ∪ Uij and Y = Uk ∪ U(ij)c contains one of the graph’s
k ̸=i,j
K3,7 , K4,6 , or K5,5 as a subgraph so that γ̃(AG(L)) ≥ 3. Therefore, 3 ≤ |Uij ∪ U(ij)c | ≤ 5 for
some Uij , U(ij)c ̸= ∅.
(i) Suppose |Uij ∪ U(ij)c | = 5 for Uij , U(ij)c ̸= ∅. If either |Uij | = 3 or |U(ij)c | = 3,
∪
then the sets X = Ui ∪ Uj ∪ Uij and Y = Uk ∪ U(ij)c form K4,5 , which is a contradiction.
k ̸=i,j
So, either |Uij | = 4 or |U(ij)c | = 4. With no loss of generality, assume that |Uij | = 4. If
Ukℓ , U(kℓ)c ̸= ∅ for kℓ ̸= ij, (ij)c , then clearly |{i, j} ∩ {k, ℓ}| = 1 and |{m, n} ∩ {k, ℓ}| = 1
where m, n ∈ {1, 2, 3, 4} \ {i, j}. So, let us take {i, j} ∩ {k, ℓ} = { j} and {m, n} ∩ {k, ℓ} =
{m}. This implies that ( Ikℓ , Ii ), ( I(kℓ)c , Im ) ∈ E(AG(L)). Then, the subgraph AG(L) −
{ Ii , Ikℓ , I(kℓ)c } contains K5,3 with the partite sets X = Uj ∪ Uij and Y = Um ∪ Un ∪ U(ij)c .
Now, the path Ii − Ikℓ − I(kℓ)c has to be embedded into a single face of any N2 -embedding
of K5,3 . Further, the vertices Ii and I(kℓ)c are adjacent to Ij and Im . So, after embedding these
four edges, the edge ( Ikℓ , In ) cannot be embeded, which means γ̃(AG(L)) ≥ 3. Therefore,
U(kℓ)c = ∅ when Ukℓ ̸= ∅ for all kℓ ̸= ij, (ij)c , and, in such cases, γ̃(AG(L)) = 2.
(ii) Suppose |Uij ∪ U(ij)c | = 4 for Uij , U(ij)c ̸= ∅. If |Ukℓ ∪ U(kℓ)c | ≥ 3 for kℓ ̸= ij,
then the subgraph AG(L) − {Ukℓ ∪ U(kℓ)c } contains a crosscap two graph K5,3 or K4,4 with
∪
the partite sets X = Ui ∪ Uj ∪ Uij and Y = Um ∪ U(ij)c . Since |Ukℓ ∪ U(kℓ)c | ≥ 3,
m̸=i,j
we can take |Ukℓ | ≥ 2. Notice that the path Ikℓ − I(kℓ)c − Ik′ ℓ together with the edges
( Ikℓ , Im ), ( Ikℓ , Ii ), ( Ik′ ℓ , Im ), and ( Ik′ ℓ , Ii ) should be embedded into a single face of an N2 -
embedding of K5,3 . Thereafter, the face cannot adopt the edges ( I(kℓ)c , Ij ) and ( I(kℓ)c , In )
where n ∈ / {i, j, m}, which implies that γ̃(AG(L)) ≥ 3. Therefore, |Ukℓ ∪ U(kℓ)c | = 2 for all
Ukℓ , U(kℓ)c ̸= ∅ with kℓ ̸= ij and 1 ≤ i, j ≤ 4.
If |Uij | = 3, then, by Figure 17a, we obtain γ̃(AG(L)) = 2. If not, then |Uij | =
2. Suppose |Ukℓ ∪ U(kℓ)c | = |Umn ∪ U(mn)c | = 2 for Ukℓ , U(kℓ)c , Umn , U(mn)c ̸= ∅ with
kℓ, mn ̸= ij. Then, the subgraph AG(L) − {[ Ikℓ , I(kℓ)c ], [ Imn , I(mn)c ]} contains K4,4 with the
partite sets X = Ui ∪ Uj ∪ Uij and Y = Ui′ ∪ Uj′ ∪ U(ij)c , where i′ , j′ ∈ / {i, j}. Note that every
face of any N2 -embedding of K4,4 is rectangular, and the vertices [ Ikℓ , I(kℓ)c ] and [ Imn , I(mn)c ]
are adjacent to the four vertices Ii , Ij , Ii′ , and Ij′ . So, to embed the vertices [ Ikℓ , I(kℓ)c ] and
[ Imn , I(mn)c ], two distinct rectangular faces with boundaries Ii , Ij , Ii′ , and Ij′ are required,
which is a contradiction. Therefore, at least one U(kℓ)c = ∅ when Ukℓ ̸= ∅ for kℓ ̸= ij and
1 ≤ i ̸= j ≤ 4. In this case, an N2 -embedding of AG(L) is given in Figure 17b.
(iii) Suppose 2 ≤ |Uij ∪ U(ij)c | ≤ 3 for all Uij , U(ij)c ̸= ∅ with 1 ≤ i ̸= j ≤ 4. Then,
by Theorem 4i, there exists Ukℓ such that Ukℓ , U(kℓ)c ̸= ∅ with |Ukℓ ∪ U(kℓ)c | = 3 and
∪
U(mn)c ̸= ∅.
mn̸=k ℓ,(k ℓ)c ;Umn ̸=∅
 Then the subgraph AG(L) − {[ Ikℓ , I(kℓ)c ], [ Imn , I(mn)c ]} contains K4,4 with partite sets X =
Ui ∪ Uj ∪ Uij and Y = Ui′ ∪ Uj′ ∪ U(ij)c , where i′ , j′ ∈
/ {i, j}. Note that every face of any N2 -
embedding of K4,4 is rectangular, and the vertices [ Ikℓ , I(kℓ)c ] and [ Imn , I(mn)c ] are adjacent
to the four vertices Ii , Ij , Ii′ and Ij′ . So to embed the vertices [ Ikℓ , I(kℓ)c ] and [ Imn , I(mn)c ],
Mathematics 2023, 11, 1553
it requires two distinct rectangular faces with boundaries Ii , Ij , Ii′ and Ij′ , a contradiction. 24 of 26
Therefore at least one U(kℓ)c = ∅ when Ukℓ ̸= ∅ for kℓ ̸= ij and 1 ≤ i ̸= j ≤ 4. In this case,
an N2 -embedding of AG(L) is given in Figure 16b.

I24 I1 I2 I24 I14 I3 I4 I14

b b b b b b b b

I14 b
I23 b

′ I34
I4 b
′
b
I12 b b I3 I1 b
I12 b b b
I2
I12

b b b I34 b
I4 I2 b b b I12 b I1
I3 ′′ ′
I12 I34
I13 I23 b
b
b b b b b b b b

I24 I1 I2 I24 I14 I3 I4 I14

(a). |U12 | = 3 and |U13 ∪ U24 | = |U14 ∪ U23 | = 2 (b). |U12| = |U34| = 2 and |U14 ∪ U23| = 2
∪4
Figure 17. | n=1 Un | = 4 with |U12 ∪ U34 | = 4.

Suppose |Uij ∪ U(ij)c | = 3 for all 1 ≤ i ̸= j ≤ 4. That is, |U12 ∪ U34 | = |U13 ∪ U24 | =
|U14 ∪U
(iii) 23 | = 3. 2Without
Suppose ≤ |Uij ∪ a loss |≤
U(ij)cof generality,
3 for all we Uij ,letU(ij|U ̸=| =
)c 12 ∅ |with
U13 | 1=≤|Ui14̸= | =j ≤ 2. 4.
Now, Then con-
bysider the bipartite
Theorem 4i, there graphexists G19U= kℓ
AG(L)
such −
that {(
U I2
kℓ
,
, I
U3 ) ,
(kℓ)
( Ic 2 ,̸ =I4 )∅, ( I ,
with
3 I4 ) ,|(UI2 ,
kℓ
I∪34 )
U , ( I
(kℓ)3 ,
c |
I =
24 ) , (
3 I4 ,
andI23 )}
∪
with the partite U(sets
mn) c X̸ = =∅. U 1 ∪ U 12 ∪ U 13 ∪ U 14 and Y = U 2 ∪ U 3 ∪ U 4 ∪ U 34 ∪ U 24 ∪ U 23 .
̸=kℓ,(kℓ) c ;U γ̸̃ = ∅19 ) = 2 and the faces of any N2 -embedding of G19 have one of the following
mnNote that mn ( G
Suppose |Uij ∪ U(ij)c | = 3 for all 1 ≤ i ̸= j ≤ 4. That is |U12 ∪ U34 | = |U13 ∪ U24 | =
possibilities:
|U14 ∪ U23 | = 3. Without loss of generality, we let |U12 | = |U13 | = |U14 | = 2. Now con-
• Nine rectangular and two hexagonal faces;
sider the bipartite graph G19 = AG(L) − {( I2 , I3 ), ( I2 , I4 ), ( I3 , I4 ), ( I2 , I34 ), ( I3 , I24 ), ( I4 , I23 )}
• Ten rectangular faces and one octagonal face.
Mathematics 2023, 1, 0 with partite sets X = U1 ∪ U12 ∪ U13 ∪ U14 and Y = U2 ∪ U3 ∪ U4 ∪ U34 ∪ U24 ∪ U23 . Note 24 of 25
that γ̃(Since,
G19 ) = in 2G19and, the the only common
faces of anyneighbor
N2 -embedding for I2 and of GI34 19 in X
have is I
one 1 , no
of rectangular
the following face
has both I2 and I34 . Therefore, the edge ( I2 , I34 ) should be embedded in a face of a length
possibilities:
of more than four; so the edges of are I3 , either
⟩(in
⟨Yfaces I24 ) and two( Ihexagonal
4 , I23 ). Thus, weorhave one to embed the
• three mutually
9 rectangular disjoint
and twoedges
hexagonal faces octagonal face.
three mutually disjoint edges of ⟨ Y ⟩ in either two hexagonal faces
at most two mutually disjoint edges of ⟨Y ⟩ and so or one octagonal face.
• But 10 in any case, and
rectangular the faces may adopt
one octagonal faces.
However,
γ̃(AG(L))in≥any case,
3. For thethe faces may
remaining adopt
cases, weat havemost two mutually
γ̃(AG(L)) = 2. disjoint edges of ⟨Y ⟩,
and,Since,
so, γ̃in G19 , the≥only
(AG(L)) 3. Forcommon neighbor
the remaining cases, forwe I2 have and Iγ̃ in X is I=
34(AG(L)) 1 , 2.
no rectangular
face has both I2 and I34 . Therefore the edge ( I2 , I34 ) should be embedded in a face of
Remark 2. As an illustration, we consider the case (v)[a] in Theorem 5. Let |U1 | = |U2 | =
length
Remark more than
|U3 | = 2. |U4As
4; so are the edges
| =an1illustration,
and |U23 | =we4.consider
( I3 , I24the
If |U
) and case ( I4(v)[a]
, I23 ).inThus Theorem we have 5. Letto|U 1 | = |U2 | =
embed the
24 | = |U34 | = 1, then the corresponding 5-partite
|U 3 | = |U as4 |in=Figure
1 and 17a, |U23 has
| = crosscap
4. If |U24 |two. = |UAlso,
34 | =if1,|U then the corresponding five-partite
graph, 24 | = 2, then the crosscap of corre-
sponding 5-partite graph, given in Figure 17b, is not equal |to
graph, as in Figure 18a, has a crosscap two. Additionally, if U24 | = 2,
two. then the the
Moreover, crosscap
5-partiteof
the corresponding five-partite graph, given
graph G in Figure 17b is minimal with respect to γ̃( G ) ̸= 2. in Figure 18b, is not equal to two. Moreover,
the five-partite graph G in Figure 18b is minimal with respect to γ̃( G ) ̸= 2.

U2 U1 U3 U4 U2 U1 U3 U4

b b

b b b b b b

b b

b b b b b b b b b b b b
I23 ′ ′′ ′′′ I34 I23 ′ ′′ ′′′ ′
I24
I23 I23 I23 I24 I23 I23 I23 I24

(a) A crosscap two 5-partite graph (b) A minimal 5-partite graph with crosscap ̸= 2
Figure 17. 5-partite graphs.
Figure 18. Five-partite graphs.
6. Conclusions
The forbidden subgraphs for crosscap two surface (Klein bottle) are not known yet.
In this regard, an open problem will be to determine a family of graphs that has crosscap
number two. This paper provides a class of r-partite graphs, where 2 ≤ r ≤ 5, that can
be both embedded and not embedded in crosscap two surface. This was done by using
the classification of all lattices with at most 4 atoms whose annihilating-ideal graph has
crosscap two.
Mathematics 2023, 11, 1553 25 of 26

6. Conclusions
The forbidden subgraphs for a crosscap two surface (a Klein bottle) are not known yet.
In this regard, an open problem will be to determine a family of graphs that has a crosscap
number two. This paper provides a class of r-partite graphs, where 2 ≤ r ≤ 5, that can
be both embedded and not embedded in a crosscap two surface. This was completed
by using the classification of all lattices with at most four atoms whose annihilating-ideal
graph has a crosscap two.

Author Contributions: Conceptualization, T.A.; methodology, T.A.; investigation, K.M.; writing—

original draft preparation, T.A. and K.M.; writing—review and editing, J.A.A.-B. and W.M.F.; project
administration, J.A.A.-B.; funding acquisition, W.M.F. All authors have read and agreed to the pub-
lished version of the manuscript.
Funding: This research project was funded by the Deanship of Scientific Research (DSR) at King Ab-
dulaziz University, under grant no. KEP–44–130–42. The first, third, and fourth authors, therefore,
acknowledge the DSR for its technical and financial support.
Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.
Acknowledgments: The The authors gratefully thank to the referees for the constructive comments
and recommendations which definitely help to improve the readability of the paper.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Robertson, N.; Seymour, P. Graph minors. XX. Wagners conjecture. J. Comb. Theory Ser. B 2004, 92, 325–357. [CrossRef]
2. Kawarabayashi, K.; Mohar, B. Some recent progress and applications in graph minor theory. Graphs Combin. 2007, 23, 1–46.
[CrossRef]
3. Thomassen, C. A simpler proof of the excluded minor theorem for higher surfaces. J. Comb. Theory Ser. B 1997, 70, 306–311.
[CrossRef]
4. Archdeacon, D. A Kuratowski Theorem for the Projective Plane. Ph.D. Thesis, The Ohio State University, Columbus, OH, USA,
1980.
5. Glover, H.; Huneke, J.P.; Wang, C.S. 103 graphs that are irreducible for the projective plane. J. Comb. Theory Ser. B 1979, 27,
332–370. [CrossRef]
6. Beck, I. Coloring of commutative rings. J. Algebra 1988, 116, 208–226. [CrossRef]
7. Halaš, R.; Jukl, M. On Beck’s coloring of posets. Discrete Math. 2009, 309, 4584–4589. [CrossRef]
8. Estaji, E.; Khashyarmanesh, K. The zero-divisor graph of a lattice. Results Math. 2012, 61, 1–11. [CrossRef]
9. Anderson, D.F.; Asir, T.; Tamizh Chelvam, T.; Badawi, A. Graphs from Rings, 1st ed.; Springer International Publishing: New York,
NY, USA, 2021.
10. Behboodi, M.; Rakeei, Z. The annihilating-ideal graph of commutative rings I. J. Algebra Appl. 2011, 10, 727–739. [CrossRef]
11. Afkhami, M.; Bahrami, S.; Khashyarmanesh, K.; Shahsavar, F. The annihilating-ideal graph of a lattice. Georgian Math. J. 2016,
23, 1–7. [CrossRef]
12. Asir, T.; Mano, K. Classification of rings with crosscap two class of graphs. Discrete Appl. Math. 2019, 256, 13–21. [CrossRef]
13. Asir, T. The genus two class of graphs arising from rings. J. Algebra Appl. 2018, 17, 1850193. [CrossRef]
14. Asir, T.; Mano, K. Classification of non-local rings with genus two zero-divisor graphs. Soft Comput. 2020, 24, 237–245. [CrossRef]
15. Chiang-Hsieh, H.-J. Classification of rings with projective zero-divisor graphs. J. Algebra. 2008, 319, 2789–2802. [CrossRef]
16. Chiang-Hsieh, H.-J.; Smith, N.O.; Wang, H.-J. Commutative rings with toroidal zero-divisor graphs. Houston J. Math. 2010, 36,
1–31.
17. Pucanović, Z.S.; Petrović, Z.Z. Toroidality of intersection graphs of ideals of commutative rings. Graphs Combin. 2014, 30, 707–
716. [CrossRef]
18. Smith, N.O. Planar zero-divisor graphs. Int. J. Commut. Rings 2003, 2, 177–188.
19. Shahsavar, F. On the planar and outer planar annihilating-ideal graphs of a lattice. Algebras Groups Geom. 2015, 32, 479–494.
20. Parsapour, A.; Ahmad Javaheri, K. The embedding of annihilating-ideal graphs associated to lattices in the projective plane.
Bull. Malays. Math. Sci. Soc. 2019, 42, 1625–1638. [CrossRef]
21. Parsapour, A.; Ahmad Javaheri, K. When a line graph associated to annihilating-ideal graph of a lattice is planar or projective.
Czech. Math. J. 2018, 68, 19–34. [CrossRef]
22. Parsapour, A.; Ahmad Javaheri, K. Line graphs associated to annihilating-ideal graph attached to lattices of genus one. Trans.
Comb. 2023, 12, 175–190.
Mathematics 2023, 11, 1553 26 of 26

23. Nation, J.B. Notes on Lattice Theory; University of Hawaii: Honolulu, HI, USA, 1998; pp. 1–143.
24. Davey, B.A.; Priestley, H.A. Introduction to Lattices and Order; Cambridge University Press: Cambridge, MA, USA, 2002.
25. White, A.T. Graphs, Groups and Surfaces; Elsevier: Amsterdam, The Netherlands, 1973.
26. Ellingham, M.N.; Stephens, C.; Zhab, X. The nonorientable genus of complete tripartite graphs. J. Comb. Theory Ser. B 2006, 96,
529–559. [CrossRef]
27. Jungerman, M. The non-orientable genus of the symmetric quadripartite graph. J. Comb. Theory Ser. B 1979, 26, 154–158.
[CrossRef]

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury
to people or property resulting from any ideas, methods, instructions or products referred to in the content.