1. Introduction
All graphs considered in this paper are simple undirected graphs. Let
be a graph, where
and
denote the vertex set and the edge set of G, respectively. We call a graph
is a bipartite graph if
,
and each edge
if and only if e has one vertex in X and the other one in Y. If
,
is called a balanced bipartite graph. For a bipartite graph
, if for any
,
, there is
, then call G a complete bipartite graph. In particular, denote a complete bipartite graph by
when
,
. For a vertex x of G, the degree of x is the number of incidence edges of x in G, and it is denoted by
. A 2-factor of a graph G is a spanning subgraph of G in which all vertices have a degree exactly 2. Call
,
,
and
, the order of G, the number of edges of G, the minimum degree of G and the average degree of G, respectively. When each vertex of G has degree k, call Gk-regular. The missing degree of
in a bipartite graph
is
.
An h-cycle is a graph
with
,
. We often denote an h-cycle as
. A Hamilton cycle of G is a cycle that visits every vertex of G exactly once. If G has a Hamilton cycle, then it is called Hamiltonian. For
,
denotes the subgraph of G induced by U. A component of G is a maximal connected induced subgraph. For any graph G, F is a 2-factor of G if and only if F is a 2-regular subgraph of G and
. Clearly, a 2-factor F of G is a collection of vertex disjoint cycles that cover all vertices of G.
In the next section, a statement of the problem is introduced. Afterwards, the proof of the main results is established.
2. Statement of the Problem
The problem of the existence of 2-factors in a graph has been extensively studied. The special case that a 2-factor has one component, that is, Hamilton cycle problem has received wide attention.
Many sufficient conditions for a graph to be Hamiltonian have been obtained. In 1952, Dirac [1] gave a minimum degree condition for Hamiltonian graphs. The following theorem is a classic result about sufficient conditions on Hamiltonian graphs.
Theorem 1. (Dirac [1]) Let G be a graph of order
. If
, then G contains a Hamilton cycle.
In 1959, Erdős and Gallai [2] gave a condition on the number of edges to guarantee that a graph has a Hamilton cycle.
Theorem 2. [2] Let G be a graph of order
. If
, that is,
, then G contains a Hamilton cycle.
In 1960, Ore [3] obtained a well-known result on Hamiltonian graphs, which is stronger than Theorem 1.
Theorem 3. (Ore [3]) Let G be a graph of order
. If
for every pair of non-adjacent vertices u and v in
, then G contains a Hamilton cycle.
If a graph satisfies the condition of Theorem 2, then it also satisfies the one of Theorem 3. So Theorem 3 is stronger than Theorem 2.
Both Theorem 1 and Theorem 3 imply that G has a 2-factor with exactly one component. The research on the existence of 2-factors in a graph is mostly motivated by results from Hamilton cycles. In 1997, Chen et al. [4] proved that for
, the condition of Theorem 3 for a graph to be Hamiltonian
implies that the graph contains a 2-factor with exactly k components.
Theorem 4. [4] Let k be a positive integer and let G be a graph of order
. If
for every pair of non-adjacent vertices u and v in
, then G contains a 2-factor with exactly k vertex disjoint cycles.
Chen et al. [4] also obtained a corollary of Theorem 4, which generalizes the classic Hamiltonian result of Theorem 1.
Corollary 1. [4] Let k be a positive integer and let G be a graph of order
. If
, then G contains a 2-factor with exactly k vertex disjoint cycles.
When G is a complete bipartite graph
, we can see that the conclusions of Theorem 4 and Corollary 1 are the best possible that any 2-factor can contain at most
components.
Moon and Moser [5] considered a bipartite version of Theorem 3 and obtained the following result for balanced bipartite graphs.
Theorem 5. [5] Let
be a balanced bipartite graph of order
, with
for every pair of non-adjacent vertices
and
, then G contains a Hamilton cycle.
The problem of determining whether a given graph has vertex disjoint cycles or not, is NP-complete. Many scholars have researched such a problem as the existence of a 2-factor with exactly k vertex disjoint cycles in balanced bipartite graphs. They have investigated degree conditions for partitioning in terms of, for example, minimum degree, average degree, degree sum of independent vertices and so on.
In 1999, Wang [6] gave a Dirac-Type degree condition for a balanced bipartite graph to contain a 2-factor with exactly k components and obtained the following result.
Theorem 6. [6] Let k be a positive integer and let
be a bipartite graph with
, if
, then G contains a 2-factor with exactly k vertex disjoint cycles.
In 2000, Chen et al. [7] obtained the following Ore-Type result, which generalizes Theorem 5.
Theorem 7. [7] Let k be a positive integer and let
be a balanced bipartite graph of order 2n where
. If
for every
and
, then G contains a 2-factor with exactly k vertex disjoint cycles.
Later, in 2001, Li et al. [8] improved Theorem 6 with Ore-Type degree condition. They also showed that the degree condition in Theorem 8 is sharp when
.
Theorem 8. [8] Let k be a positive integer and let
be a bipartite graph with
. If
for every pair of non-adjacent vertices
, then G has a 2-factor with exactly k vertex disjoint cycles.
In 2017, Chiba and Yamashita [9] proved that the degree condition in Theorem 5 also guarantees the existence of a 2-factor with exactly k vertex disjoint cycles when
. The following result generalizes Theorem 7 and Theorem 8.
Theorem 9. [9] Let k be a positive integer and let
be a bipartite graph with
. If
for every pair of non-adjacent vertices
and
, then G has a 2-factor with exactly k vertex disjoint cycles.
As early as 1963, Moon and Moser [5] considered a bipartite version of Ore's Theorem (Theorem 3) and they obtained Theorem 5. They also discussed the problem of how many edges are necessary to ensure that a balanced bipartite graph contains a Hamilton cycle. Moon and Moser [5] gave the conditions on the minimum degree and the number of edges for the existence of a Hamilton cycle in balanced bipartite graphs.
Theorem 10. [5] Let G be a balanced bipartite graph of order 2n
and
, where
. If
, then G contains a Hamilton cycle.
Theorem 10 implies that if there is no restriction on the minimum degree (i.e.
), then the sufficient condition for a balanced bipartite graph to be Hamiltonian is
.
Let H and G be two graphs. We call G H-free if G contains no subgraphs isomorphic to H. For a graph H and positive integers m, n, the bipartite Turán number of H, denoted by
, is the maximum number of edges among H-free bipartite graphs with two parts of sizes m and n, respectively. Recently, Zhang et al. [10] determined that the bipartite Turán number of F for any 2-factor F in
.
Theorem 11. [10]
.
Theorem 11 implies the following corollary.
Corollary 2. [10] Let G be a balanced bipartite graph of order
. If
, then G contains all non-isomorphic 2-factors of
.
3. Main Results
In this paper, we consider conditions on the minimum degree and number of edges for a balanced bipartite graph to contain some class of 2-factors and obtain the following result. This result extends Theorem 10 under condition (1). Since Theorem 10 implies that if G is a balanced bipartite graph of order 2n,
and
, then G contains a Hamilton cycle. Theorem 10 show that if G is a balanced bipartite graph of order 2n,
and
, then G contains a 2-factor with 2 components.
Theorem 12. Let G be a balanced bipartite graph of order 2n
and
. Then G contains a 2-factor with k components,
-cycle,
,
-cycle, if one of the following is satisfied:
(1)
,
and
;
(2)
,
and
.
4. The Proof of Theorem 12
Proof. Let
be a balanced complete bipartite graph, where
,
. Let k be a positive integer with
. Let
be the collection of non-isomorphic 2-factors of
with k components,
-cycle,
,
-cycle, when
,
and when
,
.
Let G be a spanning subgraph of
with
. When
, by Theorem 10, G contains a Hamilton cycle.
(1)
,
.
For any
, let two components of F be
-cycle and
-cycle. Note that
. Let
(arranged in clockwise) be a Hamilton cycle of G. There are exactly 2n different ways to take a pair of vertices
with distance
on the Hamilton cycle C in the clockwise direction. We know that the sequence of vertices
is a
-cycle and
is a
-cycle in
. Then these two vertex disjoint cycles form a 2-factor isomorphic to F (when the subscript is greater than 2n, do operation module 2n). Noting that
, the number of two-edge sets
in
is exactly 2n and such 2n two-edge sets are pairwise non-intersecting. Compared with
, G has at most
missing edges. So G contains at least five such two-edge sets. Taking such a two-edge set, we can obtain vertex disjoint
-cycle and
-cycle in G, based on the Hamilton cycle C. I.e., G contains a 2-factor isomorphic to F.
(2)
,
.
For any
, let three components of F be
-cycle,
-cycle,
-cycle. Note that
. Since
and
, for any 2-factor
, G contains a 2-factor isomorphic to
. Let
be a 2-factor with two components: a
-cycle and a
-cycle. Since
, we can see that
. Then G contains a 2-factor isomorphic to
with two components: a
-cycle
and a
-cycle
. Let
For the sake of convenience, we assume
Write
.
To the contrary, we assume that G contains no subgraphs isomorphic to F. Then
does not contain vertex disjoint
-cycle and
-cycle. (Otherwise, combined with
-cycle
, G contains a 2-factor isomorphic to F, a contradiction.) Noting that
is a spanning subgraph of
and
, we know that
by (1). This means that
has at least
missing edges compared with
, which leads to the following claim.
Claim 1.
has at least
missing edges between
and
.
According to the number of edges between two vertex sets
and
in G, there are three cases to be discussed.
Case 1. For any
-set
,
contains a
-cycle.
By Claim 1, G has at least
missing edges between
and
with respect to
. This means that there are at least two missing incidence edges at each vertex in
(except at most four vertices) on average. Since
,
. We can take a
-set
such that there are at least
missing edges between
and
. For the sake of convenience, the partition of X is adjusted as follows:
. By the assumption,
contains a
-cycle. Then
contains no vertex disjoint
-cycle,
-cycle. (Otherwise, G contains a 2-factor isomorphic to
, a contradiction.) Noting that
is a spanning subgraph of
and
, we know that
by (1). So
has at least
missing edges with respect to
. According to above discussion, G has at least
missing edges between
and
, and at least
missing edges between
and
with respect to
. So there are at least
missing edges in total. It follows that
, a contradiction.
Case 2. For any
-set
,
contains a
-cycle.
By symmetry, the proof in Case 2 is same as the one in Case 1.
Case 3. There exist two
-sets
,
such that both
and
do not contain
-cycle.
Let
,
. If
,
,
has at least
missing edges. If
,
,
has at least
missing edges. If
,
,
has at least
missing edges by Theorem 10. If both
and
have at least
missing edges, by Claim 1, G has at least
missing edges, a contraction. So at least one of
and
, say
, has exactly
missing edges. According to the above discussion, the unique possibility is that
has exactly
missing edges and
. This means that the minimum degree vertex, say w, is unique in
, and
is a complete bipartite graph, where
is the subgraph of
by removed of w. Next, we discuss two cases according to the minimum degree vertex belonging to A
or
.
Case 3.1. If the minimum degree vertex w of
in A, then by
, w has at least
neighbors in
. So there are at most
missing edges between
and
. Noting that
, there are at least
missing edges between
and
. By the pigeonhole principle, there exists at least one vertex
such that there are at least two missing edges between
and
. So we can take a
-set
such that there are at least
missing edges between S and
. (We can choose the first
vertices with missing degree as large as possible of
in
) Noting that
is a complete bipartite graph and
,
contains a
-cycle. (See Figure 1)
Figure 1. Choosing a
-set
to replace
in Case 3.1, where
and
.
Case 3.2. If the minimum degree vertex w of
in
, then w has exactly
non-neighbors in
. Let
be a vertex of such
vertices and
has the smallest missing degree in
. Recalling that both
and
have at least
missing edges and
,
has at most
missing edges. By the pigeonhole principle, there are at most
missing edges between
and
. Recalling that
has at least
missing edges and
, the number h of missing edges between
and
satisfies that
This means that there are at least
missing edges between
and
. Therefore, we can pick a
-set
such that the number of missing edges between S and
is at least
. (We can choose the first
vertices with missing degree as large as possible of
in
.) Noting that
and there are exactly
missing edges between
and
, there are at most
missing edges between S and
. By Corollary 2,
contains a
-cycle. (See Figure 2)
Figure 2. Choosing a
-set
to replace
in Case 3.2, where
and
.
Totally, in Case 3.1 or Case 3.2, we can always pick a
-set
such that
contains a
-cycle and there are at least
missing edges between S and
. Now, the partition of X is adjusted as follows:
. Noting that
contains a
-cycle,
contains no vertex disjoint
-cycle,
-cycle. (Otherwise, G contains a 2-factor isomorphic to F, a contradiction.) Noting that
is a subgraph of
and
, we know that
by (1). So
has at least
missing edges with respect to
. According to above discussion, G has at least
missing edges between X and
, at least
missing edges between S and
, and at least
missing edges between
and
with respect to
. In total, there are at least
missing edges with respect to
. It follows that
, a contradiction.
In conclusion, for any balanced bipartite graph G of order
and
, if
, then G contains a 2-factor isomorphic to F for every
; if
, then G contains a 2-factor isomorphic to F for every
.
□
5. Conclusions and Suggestions
Theorem 12 extends one result of Theorem 10 under condition (1). Furthermore, it is possible to relax the condition
into
, but there are more cases to be discussed.
The condition
is necessary, because G contains no 2-factors when
. Figure 3 is a counterexample graph.
Figure 3.
and
, G contains no 2-factors.
It would be interesting to discuss whether the condition
can be replaced with
when
.
There are some limitations to our results. Theorem 12 only considers conditions on the minimum degree and number of edges for a balanced bipartite graph to contain 2-factors with k components, where
. Actually, it is significant to consider the more general problem for the cases
.