Independence Number and Cut-Vertices

Similar documents
On total domination and support vertices of a tree

Graffiti.pc on the 2-domination number of a graph

Some Elementary Lower Bounds on the Matching Number of Bipartite Graphs

Bounds on the k-domination Number of a Graph

ON SOME CONJECTURES OF GRIGGS AND GRAFFITI

LOWER BOUNDS FOR THE DOMINATION NUMBER

Module 7. Independent sets, coverings. and matchings. Contents

DO NOT RE-DISTRIBUTE THIS SOLUTION FILE

Matching Theory. Figure 1: Is this graph bipartite?

On the Relationships between Zero Forcing Numbers and Certain Graph Coverings

The strong chromatic number of a graph

Number Theory and Graph Theory

Abstract. A graph G is perfect if for every induced subgraph H of G, the chromatic number of H is equal to the size of the largest clique of H.

Matching Algorithms. Proof. If a bipartite graph has a perfect matching, then it is easy to see that the right hand side is a necessary condition.

Subdivided graphs have linear Ramsey numbers

HW Graph Theory SOLUTIONS (hbovik) - Q

Theorem 3.1 (Berge) A matching M in G is maximum if and only if there is no M- augmenting path.

Some Upper Bounds for Signed Star Domination Number of Graphs. S. Akbari, A. Norouzi-Fard, A. Rezaei, R. Rotabi, S. Sabour.

Spanning Trees with Many Leaves and Average Distance

A Reduction of Conway s Thrackle Conjecture

CPS 102: Discrete Mathematics. Quiz 3 Date: Wednesday November 30, Instructor: Bruce Maggs NAME: Prob # Score. Total 60

Rigidity, connectivity and graph decompositions

GEODETIC DOMINATION IN GRAPHS

Results on the min-sum vertex cover problem

VIZING S THEOREM AND EDGE-CHROMATIC GRAPH THEORY. Contents

EDGE MAXIMAL GRAPHS CONTAINING NO SPECIFIC WHEELS. Jordan Journal of Mathematics and Statistics (JJMS) 8(2), 2015, pp I.

An Improved Upper Bound for the Sum-free Subset Constant

Fundamental Properties of Graphs

Modules. 6 Hamilton Graphs (4-8 lectures) Introduction Necessary conditions and sufficient conditions Exercises...

Disjoint directed cycles

Definition For vertices u, v V (G), the distance from u to v, denoted d(u, v), in G is the length of a shortest u, v-path. 1

Discrete Wiskunde II. Lecture 6: Planar Graphs

Automated Conjecturing for Proof Discovery

Ma/CS 6b Class 26: Art Galleries and Politicians

Graph Theory S 1 I 2 I 1 S 2 I 1 I 2

Complete Cototal Domination

Bar k-visibility Graphs

Potential Bisections of Large Degree

Discrete mathematics , Fall Instructor: prof. János Pach

Bipartite Roots of Graphs

Math 777 Graph Theory, Spring, 2006 Lecture Note 1 Planar graphs Week 1 Weak 2

The Edge Fixing Edge-To-Vertex Monophonic Number Of A Graph

A taste of perfect graphs (continued)

MC 302 GRAPH THEORY 10/1/13 Solutions to HW #2 50 points + 6 XC points

Section 3.1: Nonseparable Graphs Cut vertex of a connected graph G: A vertex x G such that G x is not connected. Theorem 3.1, p. 57: Every connected

Advanced Combinatorial Optimization September 17, Lecture 3. Sketch some results regarding ear-decompositions and factor-critical graphs.

Edge Colorings of Complete Multipartite Graphs Forbidding Rainbow Cycles

Progress Towards the Total Domination Game 3 4 -Conjecture

Matching and Factor-Critical Property in 3-Dominating-Critical Graphs

Graph Connectivity G G G

CLAW-FREE 3-CONNECTED P 11 -FREE GRAPHS ARE HAMILTONIAN

Two Characterizations of Hypercubes

DO NOT RE-DISTRIBUTE THIS SOLUTION FILE

1 Matchings in Graphs

CHAPTER 2. Graphs. 1. Introduction to Graphs and Graph Isomorphism

On Covering a Graph Optimally with Induced Subgraphs

AMS /672: Graph Theory Homework Problems - Week V. Problems to be handed in on Wednesday, March 2: 6, 8, 9, 11, 12.

Math 170- Graph Theory Notes

Sharp lower bound for the total number of matchings of graphs with given number of cut edges

Exercise set 2 Solutions

DOMINATION GAME: EXTREMAL FAMILIES FOR THE 3/5-CONJECTURE FOR FORESTS

Domination Cover Pebbling: Structural Results

WALL, NICOLE TURPIN, M.A. On the Reconstruction Conjecture. (2008) Directed by Dr. Paul Duvall. 56 pp.

Ma/CS 6b Class 5: Graph Connectivity

2. CONNECTIVITY Connectivity

1. Chapter 1, # 1: Prove that for all sets A, B, C, the formula

1 Matchings with Tutte s Theorem

Acyclic Edge Colorings of Graphs

Hamiltonian cycles in bipartite quadrangulations on the torus

Solution to Graded Problem Set 4

The Six Color Theorem

Adjacent: Two distinct vertices u, v are adjacent if there is an edge with ends u, v. In this case we let uv denote such an edge.

Math 778S Spectral Graph Theory Handout #2: Basic graph theory

Assignment 4 Solutions of graph problems

Characterizations of Trees

Component Connectivity of Generalized Petersen Graphs

Graph theory - solutions to problem set 1

Problem Set 3. MATH 776, Fall 2009, Mohr. November 30, 2009

Small Survey on Perfect Graphs

Forced orientation of graphs

Lecture 6: Graph Properties

PAIRED-DOMINATION. S. Fitzpatrick. Dalhousie University, Halifax, Canada, B3H 3J5. and B. Hartnell. Saint Mary s University, Halifax, Canada, B3H 3C3

Extremal Graph Theory: Turán s Theorem

[Ramalingam, 4(12): December 2017] ISSN DOI /zenodo Impact Factor

Abstract. 1. Introduction

A note on isolate domination

(Received Judy 13, 1971) (devised Nov. 12, 1971)

Assignment # 4 Selected Solutions

Chapter 2. Splitting Operation and n-connected Matroids. 2.1 Introduction

Math 443/543 Graph Theory Notes 5: Planar graphs and coloring

THE INSULATION SEQUENCE OF A GRAPH

Multiple Vertex Coverings by Cliques

TWO CONTRIBUTIONS OF EULER

K 4 C 5. Figure 4.5: Some well known family of graphs

On Galvin orientations of line graphs and list-edge-colouring

Chapter 4. square sum graphs. 4.1 Introduction

Graph Theory Mini-course

Rainbow game domination subdivision number of a graph

GRAPHS WITH 1-FACTORS

Math 443/543 Graph Theory Notes 2: Transportation problems

Transcription:

Independence Number and Cut-Vertices Ryan Pepper University of Houston Downtown, Houston, Texas 7700 pepperr@uhd.edu Abstract We show that for any connected graph G, α(g) C(G) +1, where α(g) is the independence number of G and C(G) is the number of cut-vertices of G. The bound is sharp and the case of equality is characterized. This inequality was conjectured by the computer program Graffiti during classroom use. 1 Introduction The independence number α = α(g) of a graph G is the cardinality of a largest set of mutually non-adjacent vertices. A vertex v is a cut-vertex of G if the graph G v, formed by deleting v and all edges incident to v, has more components than G does. Similarly, an edge e is a cut-edge if G e has more components than G. In what follows, n = n(g) is the number of vertices and C = C(G) is the number of cut-vertices of the graph G. All graphs are considered to be simple and finite. The results in this paper were inspired by a conjecture of the computer program Graffiti, created by S. Fajtlowicz. This conjecture was made during classroom use in 003. (The use of Graffiti in the classroom is discussed in [1] and [].) Graffiti conjectured that, for any graph with k components, α C + k. This conjecture, which appeared in [4], is correct and follows as a corollary to the lower bound on independence number proven herein. A similar result, also conjectured by Graffiti at around the same time, concerning a lower bound on the independence number in terms of the number 1

of cut-edges, was also recently published [3]. In this paper we prove the inequality α(g) C(G) + 1, and give a characterization of the case of equality. Main Result First, we define the family of graphs where equality holds. Definition.1. A graph G is called a complete-tree if it is a complete graph or if it can be partitioned into m complete subgraphs, each of order at least, such that every edge with an endpoint in two different members of the partition is a cut-edge and no vertex is incident to more than two parts. The name comes from the fact that the graph formed by replacing each complete subgraph in the partition by a vertex, where two vertices of the new graph are adjacent if and only if there was a cut-edge of the original graph joining the parts, is a tree. Moreover, the structure of these graphs yields the following remarks, which we state without proof since each can easily be proven by induction and/or appeal to the definition. After the remark there is an example of a complete-tree for illustrative purposes. Remark.. 1. A complete-tree has a unique complete-subgraph partition which satisfies the definition.. If G is a complete-tree, then every vertex of G is a member of some maximum independent set and no vertex of G is a member of every maximum independent set. 3. If G is a complete-tree and {A 1, A,..., A m } is its unique complete subgraph partition, then α = m and C = (m 1).

Figure 1: An example of a complete tree with m = 5, C = 8, and α = 5. Theorem.3. Let G be a connected graph with independence number α and C cut-vertices. Then, α C + 1 (1) and equality holds if and only if G is a complete-tree. Proof. Proceeding by induction, notice that the theorem is true for all connected graphs with at most three vertices. Assume that the theorem is true for all connected graphs with less than n vertices and let G be a graph of order n. If G has no cut-vertices, then the theorem follows immediately, so we may assume otherwise. Not every vertex of G can be a cut-vertex, so there must be a cut-vertex with a non-cut-vertex neighbor. Let v be a cutvertex with a non-cut-vertex neighbor w. Moreover, let {G 1, G,..., G k } be the components of G the graph obtained by deletion of v where, without loss of generality, w G 1. 3

From inductive hypothesis we have, α(g) α(g i ) ( C(G i) + 1) = k + 1 C(G i ). () Next we will show the number of cut-vertices in each component does not decrease by more than one with the deletion of v. First observe that, for each i, any vertex of G i which is a cut-vertex of G but not of G i, must be adjacent to v. Now, if there were two such vertices in one component of G, then there must be a cycle in G containing both of them and v contradicting the fact that they were cut-vertices of G but not of the component of G containing them. Thus the number of cut-vertices decreases by at most one in each component, as claimed. Evidently, the number of cut-vertices in the component G 1 does not decrease at all. To see this, suppose p G 1 is a cut-vertex of G but not of G 1. Hence p w is adjacent to v. Now there is a cycle in G containing p, w, and v a contradiction. To find a lower bound on the number of cut-vertices of G, we may count the loss of the cut-vertex v in the component G 1 componentwise to obtain; C(G ) = and then sum C(G i ) C(G) k. (3) The lower bound is proven by combining Inequalities and 3 and then observing that k, as shown below. α(g) α(g i ) ( C(G i) + 1) C(G) + k C(G) + 1. (4) Concerning the case of equality, assume α(g) = C(G) + 1. Then Inequality 4 collapses everywhere to equality. This means that k = and, α(g 1 ) + α(g ) = ( C(G 1) + 1) + ( C(G ) + 1). 4

Since each summand on the right hand side of the above equation is also a lower bound for its counterpart on the left, it follows that, α(g 1 ) = C(G 1) α(g ) = C(G ) + 1, + 1. Therefore G 1 and G are complete-trees by inductive hypothesis. Moreover, we also find from the collapsed Inequality 4, C(G) = C(G 1 ) + C(G ) = C(G ). (5) Hence, no new cut-vertices were introduced by deletion of v and the total number of cut-vertices decreases by exactly two. Since v itself was a cutvertex and the number of cut-vertices in G 1 does not decrease, let z G be the unique cut-vertex of G which is not a cut-vertex of G. It turns out that z is also the unique neighbor of v in G since, echoing previous reasoning, any other neighbor of v in G would introduce a cycle and contradict the nature of z just described. Thus, it necessarily follows that vz is a cut-edge of G. Now, since G 1 is a complete-tree, partition it into complete subgraphs satisfying the definition such that H is the part containing the vertex w. If there was a vertex p H which was not adjacent to v, then form a maximum independent set of G 1 containing p and a maximum independent set of G not containing z (that such sets exist in complete-trees is the second part of Remark.). Now take the union of these two sets together with v and we have a maximum independent set of G with α(g 1 ) + α(g ) + 1 = α(g) + 1 vertices a contradiction. Therefore, v is adjacent to every vertex of H. Moreover, suppose there was a vertex x G 1 H which was adjacent to v. Then, since G 1 is a complete-tree and x / H, any path in G 1 from x to w contains a cut-edge. Let rs be a cut-edge on a path in G 1 from x to w. Hence, r and s are cut-vertices of the component G 1. Now there is a cycle in G containing v, w, r, s, and x (or only four of these vertices if it turned out that r = x or s = x). However, by the structure of complete-trees, r is only adjacent to s and the all the vertices in the complete subgraph it is a 5

member of, and similarly for s. Thus it must be the case that r and s were not cut-vertices of G and only became cut-vertices after deletion of v. This contradicts Equation 5, which informs us that the the cut-vertices of G and of G are the same with the exception of the loss of v and z. Therefore no such vertex x could exist and the only neighbors v has in G 1 are the vertices of the complete subgraph H which contains w. To conclude the implication, we see that G can be partitioned into complete subgraphs satisfying the definition of a complete-tree by including v in the complete subgraph H of the partitioning of G 1 and then joining this to G by the cut-edge vz. The converse is settled quite easily by appealing to the structure of complete-trees laid out in the third part of Remark.. 3 Concluding Remarks The author would like to thank Siemion Fajtlowicz, Greg Henry, Craig Larson, and Dillon Sexton for helpful discussions about this theorem. Moreover, after the presented proof was discovered, both Henry and Larson found independent proofs of their own, though neither of these made the characterization of equality quite as simple. It should also be pointed out here that a nearly symmetric upper bound is known. Theorem 3.1. Let G be an n > 1 vertex connected graph with independence number α and C cut-vertices. Then α n C 1. (6) This theorem is much simpler in the sense that it is basically a statement about trees. It was proven independently by Fajtlowicz, Pepper, and Larson. Fajtlowicz s proof of an equivalent statement to this can be found in Ermelinda DeLaViña s list of conjectures of Graffiti and Graffiti.pc, Written on the Wall II, conjecture #1. This list can be found at: http://cms.dt.uh.edu/faculty/delavinae/research/wowii/. 6

References [1] S. Fajtlowicz, Toward Fully Automated Fragments of Graph Theory, Graph Theory Notes of New York XLII (00) 18 5. [] R. Pepper, On New Didactics of Mathematics: Learning Graph Theory via Graffiti, S. Fajtlowicz, P. Fowler, P. Hansen, M. Janowitz, F. Roberts, (Eds.), Graphs and Discovery, DIMACS Series (69) in Discrete Mathematics and Theoretical Computer Science, 341-351, 005. [3] R. Pepper, G. Henry, D. Sexton, Cut-Edges and the Independence Number, MATCH Commun. Math. Comput. Chem. 56 (006) 403-408. [4] R. Pepper, Binding Independence, Ph.D. Dissertation, University of Houston, 004. 7