On the Critical Difference of Almost Bipartite Graphs
Vadim E. Levit, Eugen Mandrescu

TL;DR
This paper investigates the properties of almost bipartite graphs, establishing bounds on their independence and matching numbers, characterizing their structure, and relating the critical difference to the core of the graph.
Contribution
It proves that for almost bipartite graphs, the sum of independence and matching numbers is either n-1 or n, and characterizes these cases, extending understanding of their critical difference.
Findings
For almost bipartite graphs, lpha(G)+\u00b5(G) is either n(G)-1 or n(G).
The critical difference d(G) equals | ext{core}(G)| - |N( ext{core}(G))| for these graphs.
Characterizations of graphs achieving each sum value are provided.
Abstract
A set is \textit{independent} in a graph if no two vertices from are adjacent. The \textit{independence number} is the cardinality of a maximum independent set, while is the size of a maximum matching in . If equals the order of , then is called a \textit{K\"{o}nig-Egerv\'{a}ry graph }\cite{dem,ster}. The number is called the \textit{critical difference} of \cite{Zhang} (where ). It is known that holds for every graph \cite{Levman2011a,Lorentzen1966,Schrijver2003}. In \cite{LevMan5} it was shown that is true for every K\"{o}nig-Egerv\'{a}ry…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
On the Critical Difference of Almost Bipartite Graphs
Vadim E. Levit
Department of Computer Science
Ariel University, Israel
Eugen Mandrescu
Department of Computer Science
Holon Institute of Technology, Israel
Abstract
A set is independent in a graph if no two vertices from are adjacent. The independence number is the cardinality of a maximum independent set, while is the size of a maximum matching in . If equals the order of , then is called a König-Egerváry graph [5, 25]. The number is called the critical difference of [27] (where ). It is known that holds for every graph [17, 23, 24]. In [16] it was shown that is true for every König-Egerváry graph.
A graph is (i) unicyclic if it has a unique cycle, (ii) almost bipartite if it has only one odd cycle. It was conjectured in [15, 19] and validated in [1] that holds for every unicyclic non-König-Egerváry graph .
In this paper we prove that if is an almost bipartite graph of order , then . Moreover, for each of these two values, we characterize the corresponding graphs. Further, using these findings, we show that the critical difference of an almost bipartite graph satisfies
[TABLE]
where by core we mean the intersection of all maximum independent sets.
Keywords: independent set, core, matching, critical set, critical difference, bipartite graph, König-Egerváry graph.
1 Introduction
Throughout this paper is a finite, undirected, loopless graph without multiple edges, with vertex set of cardinality , and edge set of size . If , then is the subgraph of spanned by . By we mean the subgraph , if . For , by we denote the partial subgraph of obtained by deleting the edges of , and we use , if . If and , then stands for the set . The neighborhood of a vertex is the set and , and , for . By we mean the chordless cycle on vertices, and respectively the complete graph on vertices.
Let us define the trace of a family of sets on the set as .
A set of vertices is independent if no two vertices from are adjacent, and an independent set of maximum size will be referred to as a maximum independent set. The *independence number of , denoted by , is the cardinality of a maximum independent set *of .
Let is a maximum independent set of , core [12], and corona [4]. An edge is -critical whenever . Notice that holds for each edge .
The number , , is called the difference of the set . The number is called the critical difference of , and a set is critical if [27]. The number is called the critical independence difference of . If is independent and , then is called critical independent [27]. Clearly, is true for every graph .
Theorem 1.1
[27]** The equality holds for every graph .
For a graph , let denote . It is known that is true for every graph [17], while the equality holds for bipartite graphs [21].
A matching (i.e., a set of non-incident edges of ) of maximum cardinality is a maximum matching, and a perfect matching is one covering all vertices of . An edge is -*critical *provided .
Theorem 1.2
For any graph , the following assertions are true:
(i)* [14] no -critical edge has an endpoint in ;*
(ii)* [4] there is a matching from into , for each ;*
(iii) **[12]** if is a connected bipartite graph with , then \alpha(G)>n\left(G\right)/2\if and only if.
It is well-known that hold for every graph . If , then is called a König-Egerváry graph [5, 25]. Various properties of König-Egerváry graphs are presented in [2, 3, 9, 10, 13, 20]. It is known that every bipartite graph is a König-Egerváry* *graph [8, 7]. This class includes also non-bipartite graphs (see, for instance, the graph in Figure 1).
Theorem 1.3
If is of a König-Egerváry graph, then
(i)* [13] every maximum matching matches core into core;*
(ii) **[16]** .
The graph is unicyclic if it has a unique cycle. We call a graph (edge) almost bipartite if it has a unique odd cycle, denoted by . Since is unique, there is no other cycle of sharing vertices with . Let
[TABLE]
and be the bipartite connected* *subgraph of containing , where . Clearly, every unicyclic graph with an odd cycle is almost bipartite.
The smallest number of edges that have to be deleted from a graph to obtain a bipartite graph is called the bipartite edge frustration of and denoted by [6, 26]. Thus, is an almost bipartite graph whenever .
In this paper we analyze the relationship between several parameters of a almost bipartite graph , namely, , , , and .
2 Results
Lemma 2.1
If is a almost bipartite graph, then there is an edge , such that .
Proof. For every pair of edges, consecutive on , only one of them may belong to every maximum matching of . In other words, at most one of the edges could be -critical.
Notice that holds for each edge . Every edge of the unique odd cycle could be -critical; e.g., the graph from Figure 2.
Lemma 2.2
[18]** For every bipartite graph , a vertex if and only if there exists a maximum matching that does not saturate .
Lemma 2.2 fails for non-bipartite König-Egerváry graphs; e.g., every maximum matching of the graph from Figure 1 saturates core.
Lemma 2.3
If is a almost bipartite graph, then .
Proof. If , then is bipartite, and hence, . Clearly, , while . Consequently, we get that
[TABLE]
which leads to . The inequality is true for every graph .
Lemma 2.4
Let be a almost bipartite graph. Then if and only if each edge of its unique odd cycle is -critical.
Proof. Assume that . For each , is bipartite, and then we have
[TABLE]
which implies and , since
[TABLE]
In other words, every is -critical.
Conversely, let us choose satisfying . By Lemma 2.1 such an edge exists. Since is -critical, and is bipartite, we infer that
[TABLE]
and this completes the proof.
Lemma 2.5
Let be a almost bipartite graph. If there is some , such that , then is a König-Egerváry graph.
Proof. Let , , and . Suppose, to the contrary, that is not a König-Egerváry graph. By* *Lemma 2.3 and Lemma 2.4, the edge is -critical. Since , it follows that . By Lemma 2.2 there exists a maximum matching of not saturating . Combining with a maximum matching of we get a maximum matching of . Hence is a matching of , which results in . Consequently, using Lemma 2.4 and having in mind that is a bipartite graph of order , we get the following contradiction
[TABLE]
that completes the proof.
Theorem 2.6
If is a almost bipartite non-König-Egerváry graph, then .
Proof. First, one has to prove that every maximum independent set of may be enlarged to some maximum independent set of .
Let , , and . According to Lemma 2.4, the edge is -critical. Hence, there exist , , such that and .
Case 1. Assume that .
If and , then is independent in that causes the contradiction
[TABLE]
Therefore, we have . Then , otherwise we get the following contradiction
[TABLE]
Case 2. Assume now that .
Then . Hence
[TABLE]
Since the set is independent and its size is at least, it is also maximum independent, i.e., .
Second, it is left to prove that for every . Let , and suppose, to the contrary, that . Since, by Lemma 2.5, we have , we can change for some not containing . The set is independent, and . This contradiction completes the proof.
Corollary 2.7
If is a connected almost bipartite non-König-Egerváry graph, then
(i)* ;*
(ii)* core.*
Proof. *(i) *By Theorem 2.6, we infer that:
[TABLE]
which clearly implies .
*(ii) Let *. By Lemma 2.4, the edge is -critical. Hence there exist , such that and . Since , it follows that core, and because , we infer that core. Consequently, we obtain that core.
The assertion in Corollary 2.7(i) may fail for connected unicyclic König-Egerváry graphs; for instance, , while , where and are from Figure 3.
Proposition 2.8
Let be a almost bipartite. Then the following assertions are equivalent:
(i)* , for every ;*
(ii)* there exists some , such that ;*
(iii)* , i.e., is not a König-Egerváry graph.*
Proof. *(i) * *(ii) *Let and assume that there is , such that . Since , there exists some , such that . Hence we infer that , is independent in , and then
[TABLE]
Therefore , and .
In this way, adding more vertices belonging to , one can build some , such that .
*(ii) * *(iii) *We have that , because .
Let . Since is a chordless odd cycle, say , the edge is -critical in , i.e., there is , such that and .
Then, is an independent set in , with
[TABLE]
which implies that the edge is -critical in . Since was an arbitrary edge on , it follows that every edge of is -critical in . By Lemma* *2.4, it follows that .
*(iii) * *(i) *It follows by Lemma 2.5.
Combining Lemma 2.5 and Proposition 2.8, we get the following.
Corollary 2.9
A almost bipartite graph is a König-Egerváry graph if and only if there is some such that .
Theorem 2.10
Let be a connected almost bipartite graph. Then the following assertions are true:
(i)* ;*
(ii)* there exists a matching from into ;*
(iii)* there is a maximum matching of that matches into .*
Proof. If is a König-Egerváry graph, then (i) follows from the definition and the fact that , while (ii), (ii) are true, by Theorem 1.3(i).
For the rest of the proof, we suppose that is not a König-Egerváry graph.
(i) By Lemma 2.3, we have . According to Lemma 2.4, holds for each edge . Consequently, we get that . Since is bipartite, Theorem 1.2(iii) ensures that
[TABLE]
which results in .
*(ii) *If , then the conclusion is clear.
Assume that . By Theorem 1.3(i), in each there is a matching from into . By Theorem 1.2(i), it follows that . Taking into account Corollary 2.7(i), we see that the union of all these matchings gives a matching from into .
(iii) Let be a maximum matching of and be a matching from into , that exists by Part (ii). The matching must saturate , because otherwise it can be enlarged with edges from . Hence, all the edges of saturating can be replaced by the edges of , and the resulting matching is a maximum matching of that matches into .
The almost bipartite graph from Figure 2 has and as maximum matchings, but only matches core into core. Notice that is not a König-Egerváry graph.
Proposition 2.11
If there is a matching from into , then
[TABLE]
Proof. Let be a matching from into . According to Theorem 1.2(ii), there is a matching, say , from into . Consequently, we get that
[TABLE]
and this completes the proof.
Theorem 2.12
If is a connected almost bipartite graph, then
[TABLE]
Proof. If is a König-Egerváry graph, the result is true by Theorem 1.3(ii).
Otherwise, let . Then is a bipartite graph, and by Lemma 2.4, we get that and . For every , it follows that , which implies
[TABLE]
Hence, using Theorem 2.10, Proposition 2.11, and Theorem 1.3(ii), we obtain
[TABLE]
Let be some critical independent set of . By Theorem 1.3(ii), we have
[TABLE]
for every . It is clear that
[TABLE]
Since, by Proposition 2.8(i), for every , we have
[TABLE]
Thus, by Theorem 1.3(ii), it follows
[TABLE]
Consequently, we infer that , where
[TABLE]
Using Corollary 2.7(i), we deduce that
[TABLE]
which completes the proof.
Theorem 2.13
[1]** If is unicyclic and non-König-Egerváry, then .
Lemma 2.14
[11]** Every connected bipartite graph has a spanning tree with the same independence number.
Theorem 2.15
If is an almost bipartite non-König-Egerváry graph, then
[TABLE]
Proof. Case 1. is connected.
By Lemma 2.14, every bipartite subgraph of has a spanning tree , having the same independence number, and hence, the same matching number, i.e., and .
Consequently, , which gives . By Theorem 2.6, we have that .
Let be the graph obtained from by substituting every with an appropriate . Thus is a connected unicyclic graph, having as its unique cycle.
Since is a non-König-Egerváry graph, Proposition 2.8(i) implies , for every . Therefore, , for every .
Claim 1. .
Every independent set of is independent in as well, while . Hence,
[TABLE]
Thus, .
Claim 2. .
Since and have the same vertex sets and , we get that .
By Proposition 2.8(ii), there exists some , such that . Hence,
[TABLE]
Clearly, is an independent set in as well.
Let . Then, , and also
[TABLE]
Thus
[TABLE]
In conclusion, we get that .
Claim 3. .
Along the lines of the proof of Claim 2, we know that there exists a set , such that . Therefore, Proposition 2.8 implies that is a non-König-Egerváry graph. Hence,
[TABLE]
By Claim 2, it means that .
Claim 4. .
By Claim 2, Claim 3, Theorem 2.12, Claim 1, and Theorem 2.13, we finally obtain the following:
[TABLE]
which completes the proof.
Case 2. is disconnected.
Clearly, , where is the connected component of containing the unique odd cycle, and is a nonempty bipartite graph. By Case 1,
[TABLE]
while Theorem 1.3*(ii) *implies
[TABLE]
Since
[TABLE]
we conclude with
[TABLE]
as required.
3 Conclusions
It is known that for every graph [17, 23, 24], while by definition of .
By Theorems 1.3, 2.15 for both König-Egerváry graphs and almost bipartite graphs. Otherwise, every relation between and is possible. For instance, the non-König-Egerváry graphs from Figure 4 satisfy
[TABLE]
The opposite direction of the displayed inequality may be found in , where
[TABLE]
Another example reads as follows:
[TABLE]
where is from Figure 5.
Problem 3.1
Characterize graphs enjoying .
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] A. Bhattacharya, A. Mondal, T. S. Murthy, Problems on matchings and independent sets of a graph , Discrete Mathematics 341 (2018) 1561–1572.
- 2[2] J. M. Bourjolly, P. L. Hammer, B. Simeone, Node weighted graphs having König-Egervary property , Math. Programming Study 22 (1984) 44–63.
- 3[3] J. M. Bourjolly, W. R. Pulleyblank, König-Egerváry graphs, 2-bicritical graphs and fractional matchings , Discrete Applied Mathematics 24 (1989) 63–82.
- 4[4] E. Boros, M.C. Golumbic, V. E. Levit, On the number of vertices belonging to all maximum stable sets of a graph , Discrete Applied Mathematics 124 (2002) 17–25.
- 5[5] R. W. Deming, Independence numbers of graphs - an extension of the König-Egerváry theorem , Discrete Mathematics 27 (1979) 23–33.
- 6[6] T. Došlić,, D. Vukičević, Computing the bipartite edge frustration of fullerene graphs, Discrete Applied Mathematics 155 (2007) 1294–1301.
- 7[7] E. Egerváry, On combinatorial properties of matrices , Matematikai Lapok 38 (1931) 16–28.
- 8[8] D. König, Graphen und Matrizen , Matematikai Lapok 38 (1931) 116–119.
