Technische Universität Ilmenau Institut für Mathematik

Similar documents
Technische Universität Ilmenau Institut für Mathematik

Technische Universität Ilmenau Institut für Mathematik

Technische Universität Ilmenau Institut für Mathematik

Technische Universität Ilmenau Institut für Mathematik

1 Introduction and Results

Cycle Length Parities and the Chromatic Number

arxiv: v1 [math.co] 28 Oct 2016

PGSS Discrete Math Solutions to Problem Set #4. Note: signifies the end of a problem, and signifies the end of a proof.

Subdivisions of Large Complete Bipartite Graphs and Long Induced Paths in k-connected Graphs Thomas Bohme Institut fur Mathematik Technische Universit

Technische Universität Ilmenau Institut für Mathematik

Dominating a family of graphs with small connected subgraphs

AALBORG UNIVERSITY. Total domination in partitioned graphs. Allan Frendrup, Preben Dahl Vestergaard and Anders Yeo

The k-path Vertex Cover in Product Graphs of Stars and Complete Graphs

Partial cubes: structures, characterizations, and constructions

Decomposing oriented graphs into transitive tournaments

Graphs with few total dominating sets

arxiv: v1 [math.co] 20 Oct 2018

Edge Isoperimetric Theorems for Integer Point Arrays

Technische Universität Ilmenau Institut für Mathematik

On the maximum number of maximum independent sets in connected graphs

Choosability and fractional chromatic numbers

The Algorithmic Aspects of the Regularity Lemma

DISTINGUISHING PARTITIONS AND ASYMMETRIC UNIFORM HYPERGRAPHS

Technische Universität Ilmenau Institut für Mathematik

arxiv: v1 [math.co] 7 Aug 2017

Independent Transversals in r-partite Graphs

Testing Equality in Communication Graphs

Technische Universität Ilmenau Institut für Mathematik

Institut für Mathematik

The concentration of the chromatic number of random graphs

Induced subgraphs of prescribed size

Induced subgraphs with many repeated degrees

Bichain graphs: geometric model and universal graphs

A Separator Theorem for Graphs with an Excluded Minor and its Applications

CHVÁTAL-ERDŐS CONDITION AND PANCYCLISM

Longest paths in strong spanning oriented subgraphs of strong semicomplete multipartite digraphs

Packing and Covering Dense Graphs

DECOMPOSITIONS OF MULTIGRAPHS INTO PARTS WITH THE SAME SIZE

Complexity of determining the most vital elements for the 1-median and 1-center location problems

Dirac s Map-Color Theorem for Choosability

HAMBURGER BEITRÄGE ZUR MATHEMATIK

Bipartite decomposition of random graphs

Out-colourings of Digraphs

The 123 Theorem and its extensions

On a Polynomial Fractional Formulation for Independence Number of a Graph

PERIODS IMPLYING ALMOST ALL PERIODS FOR TREE MAPS. A. M. Blokh. Department of Mathematics, Wesleyan University Middletown, CT , USA

Size and degree anti-ramsey numbers

Extremal Graphs Having No Stable Cutsets

Graphs with Integer Matching Polynomial Roots

Linear Algebra and its Applications

A note on network reliability

Note on p-competition Graphs and Paths

Relating 2-rainbow domination to weak Roman domination

Set-orderedness as a generalization of k-orderedness and cyclability

The integers. Chapter 3

Unmixed Graphs that are Domains

Inverse and Disjoint Restrained Domination in Graphs

The chromatic covering number of a graph

Properties and Classification of the Wheels of the OLS Polytope.

Independence in Function Graphs

New lower bounds for hypergraph Ramsey numbers

The Turán number of sparse spanning graphs

Eigenvalues and edge-connectivity of regular graphs

Mutually embeddable graphs and the Tree Alternative conjecture

A lower bound for the Laplacian eigenvalues of a graph proof of a conjecture by Guo

MAX for k-independence in multigraphs

Graph homomorphism into an odd cycle

Maximum k-regular induced subgraphs

On shredders and vertex connectivity augmentation

Turán numbers of vertex-disjoint cliques in r-partite graphs

Durham Research Online

arxiv: v1 [math.co] 4 Jan 2018

Math 104: Homework 7 solutions

MULTIPLICITIES OF MONOMIAL IDEALS

The number of edge colorings with no monochromatic cliques

HAMILTONICITY AND FORBIDDEN SUBGRAPHS IN 4-CONNECTED GRAPHS

Extension of Strongly Regular Graphs

On the Maximum Number of Hamiltonian Paths in Tournaments

arxiv: v3 [math.co] 25 Feb 2019

The edge-density for K 2,t minors

Tournaments and colouring

Maximum Alliance-Free and Minimum Alliance-Cover Sets

Aalborg Universitet. All P3-equipackable graphs Randerath, Bert; Vestergaard, Preben Dahl. Publication date: 2008

On improving matchings in trees, via bounded-length augmentations 1

RECAP: Extremal problems Examples

Decomposition of Riesz frames and wavelets into a finite union of linearly independent sets

Applied Mathematics Letters

On the Maximum Number of Hamiltonian Paths in Tournaments

Can a Graph Have Distinct Regular Partitions?

New Bounds for Partial Spreads of H(2d 1, q 2 ) and Partial Ovoids of the Ree-Tits Octagon

On a list-coloring problem

CONSTRUCTION OF DIRECTED STRONGLY REGULAR GRAPHS USING FINITE INCIDENCE STRUCTURES

Preprint Christian Garske, Martin Sonntag, Hans-Martin Teichert Niche hypergraphs ISSN

Probabilistic Proofs of Existence of Rare Events. Noga Alon

Lebesgue Measure on R n

MONOCHROMATIC SCHUR TRIPLES IN RANDOMLY PERTURBED DENSE SETS OF INTEGERS

Graphs with Large Variance

Math 421, Homework #7 Solutions. We can then us the triangle inequality to find for k N that (x k + y k ) (L + M) = (x k L) + (y k M) x k L + y k M

Complexity Results for Equistable Graphs and Related Classes

arxiv: v1 [math.co] 29 Jul 2010

Transcription:

Technische Universität Ilmenau Institut für Mathematik Preprint No. M 0/08 A lower bound on independence in terms of degrees Harant, Jochen August 00 Impressum: Hrsg.: Leiter des Instituts für Mathematik Weimarer Straße 5 98693 Ilmenau Tel.: +49 3677 69-36 Fax: +49 3677 69-370 http://www.tu-ilmenau.de/math/

A Lower Bound on Independence in Terms of Degrees Jochen Harant Institut für Mathematik TU Ilmenau, Postfach 00565 D-98684 Ilmenau, Germany email: jochen.harant@tu-ilmenau.de Abstract We prove a new lower bound on the independence number of a simple connected graph in terms of its degrees. Keywords: Independence; stability; connected graph AMS subject classification: 05C69 Introduction We consider finite, simple, and undirected graphs G with vertex set V. For a graph G, we denote its order by n and its size by m, respectively. The degree of u in G is denoted by du) and is the maximum degree of G. A set of vertices I V in a graph G is independent, if no two vertices in I are adjacent. The independence number α of G is the maximum cardinality of an independent set of G. The independence number is one of the most fundamental and well-studied graph parameters [6]. In view of its computational hardness [5] various bounds on the independence number have been proposed. The following classical bound holds for every graph G and is due to Caro and Wei [, 7] α du) +. ) Since the only graphs for which ) is best-possible are the disjoint unions of cliques, additional structural assumptions excluding these graphs allow improvements of ). A natural candidate for such assumptions is connectivity. For connected graphs, Harant and Rautenbach proved [] cf. also [3] and [4]) Theorem If G is a connected graph, then there exist a positive integer k N and a function φ : V N 0 with non-negative integer values such that φu) du) for all u V, α k du) + φu), ) and φu) k ). 3)

Note that Theorem is best-possible for the connected graphs which arise by adding bridges to disjoint unions of cliques, i.e. it is best-possible for the intuitively most natural candidate of a connected graph with small independence number. In [3], a weaker version of Theorem is proved. This result is obtained from Theorem by replacing the inequality 3) by φu) k. For an integer l with 0 l m let fl) = min du)+ φu), where the minimum is taken over all integers 0 φu) du) with φu) = l. Obviously, f is strictly increasing. With this function f, it follows the existence of positive integers k and k such that α k + fk ) put k = k and use the result in [3]) and α k + fk ) with k = k ) and Theorem ). After extending f to real arguments, in [4], it is proved that the function l + fl) is continuous and strictly increasing and that k is at least the unique zero k 0 of this function. Finally, α k 0 + is the main result in [4]. Here we will show that the continuous function l + fl) is also strictly increasing. If we assume that for the graph in question then f) > f0) =. It will be proved du)+ du)+ that there is a unique solution l 0 of the equation l + = fl) and, because l + fl) is strictly increasing and + f) < 0, it follows l 0 >. Consequently, l 0 + = fl 0 ) > f l 0 + ), since f is strictly increasing, hence, l 0 > k 0. The inequality α l 0 + is the content of the following Theorem. In case, Remark gives a lower bound on the improvement du)+ l 0 + ) k 0 + ) = fl 0 ) fk 0 ). Theorem Let G be a finite, simple, connected, and non-complete graph on n 3 vertices of size m n. Moreover, let α n be the independence number, be the maximum degree of G, n j be the number of vertices of degree j in G, and jj + ) xj) = jj + ) j + j))n +... + j + )n j+ + n j j + +... + n ) for j {,..., }. Then i) there is a unique j 0 {,..., } such that 0 xj 0 ) < n +... + n j0 and ii) α du) + ) + n + ) + n + n ) +... + n +... + n j0 + j 0 + )j 0 + ) + xj 0) j 0 + )j 0 = xj 0) + n j0 + + n j0 + +... + j 0 )n +. Proof of Theorem In the sequel let k be the lower bound on α of Theorem. By Theorem, it follows Lemma k fk )).

For a finite family F of integers let maxf ) be a maximum member of F. Note that a member of a family may occur more than once. If for instance F = {,, } then F \{maxf )}) {maxf ) } = {,, }. The following Lemma, Lemma 3, and Lemma 4 are proved in [3] and [4]. Lemma Given an integer l with 0 l m, the following algorithm calculates fl): Input: The family F = {du) u V }. j := 0, while j < l do begin F := F \ {maxf )}) {maxf ) }; j := j + end Output: fl) = m F m+. Lemma 3 is a consequence of Lemma. Lemma 3 Given an integer 0 l m, i) there are unique integers j and x with j {,..., } and x {0,..., n +... + n j } such that l = n + n + n ) +... + n + n +... + n j+ ) + x = x + n j+ + n j+ +... + j)n and ii) fl) = n +... + n j x) j+ + x j + n j j +... + n = n +... + n j ) j+ + x jj+) + n j j +... + n. By Lemma 3, it follows Lemma 4 If l = x + n j+ + n j+ +... + j)n with j {,..., } and x {0,..., n +... + n j } than fl + ) fl) = jj+). Using Lemma 3, the calculation of fl) is possible without taking a minimum and without using the algorithm above. We will now define the function f for real l [, m). For given j {,..., } and a real number x with 0 x < n +... + n j let the real numbers l and fl) implicitly) be defined as l = x + n j+ + n j+ +... + j)n and fl) = n +... + n j ) j+ + x jj+) + n j j +... + n. We will prove Lemma 5. Lemma 5 The function g with gl) = l + fl) is continuous and strictly increasing on [, n). Proof. Consider l [, n). Then there are j {,..., } and x with 0 x < n +... + n j such that l = x + n j+ + n j+ +... + j)n. If j = then n > l n + n 3 +... + )n = m n, a contradiction to n m. Hence, j, and l belongs to the interval Ij) = [n j+ + n j+ +... + j)n, n j + n j+ +... + j + )n ). By Lemma 3, gl + ɛ) gl) = ɛ jj+) ) and, consequently, gl) is continuous and, because j, strictly increasing on Ij). Note that Ij) Ij ) = if j j and that I)... I ) = [, m n) [, n). It is easy to see that g is also continuous in l = n j+ + n j+ +... + j)n for j {3,..., }. Since the classical bound due to Caro and Wei is tight only for complete graphs, it follows g0) = < 0, and, by Lemma, gk )) 0. Using Lemma 5, there is a unique du)+ zero l 0 = xj 0 ) + n j0 + + n j0 + +... + j 0 )n of g with < l 0 k ) α ) < n and 0 xj 0 ) < n +... + n j0. It follows Lemma 6. Lemma 6 α k l 0 +, where l 0 0, n] is the unique solution of l + = fl). 3

Considering the equation l 0 + = fl 0 ), i.e. + xj 0 ) + n j0 + + n j0 + +... + j 0 )n = n +... + n j0 ) it follows x 0 = j 0j 0 +) j 0 j 0 +) We obtain Lemma 7. jj+) jj+) j 0 + + xj 0) j 0 j 0 +) + n j 0 j 0 + j 0))n +... + j 0 + )n j 0 + + n j 0 j 0 + +... + n ). Lemma 7 If j {,..., } and l = x + n j+ + n j+ +... + j)n with 0 x < n +... + n j, then l + = fl) if and only if x = j+ j))n +... + j+ )n j+ + n j j+ +... + n ). Now we complete the proof of Theorem. By Lemma 4 and Lemma 6, α k fl 0 ) = f0) + f) f0)) +... + f l 0 ) f l 0 )) + fl 0 ) f l 0 )) = du)+ ) + n +) + n +n ) +... + n +...+n j0 + j 0 +)j 0 +) + xj 0) j 0 +)j 0 because l 0 = xj 0 ) + n j0 + + n j0 + +... + j 0 )n = n + n + n ) +... + n + n +... + n j0 +) + xj 0 ). j 0 +... + n With fl 0 ) = l 0 + = xj 0)+n j0 ++n j0 ++...+ j 0 )n + Theorem is proved. 3 Remarks The following Remark is proved in the introduction. Remark If du)+ then l 0 > k 0. Remark compares the lower bound l 0 + on α in Theorem to the lower bound k 0 + on α in the main result in [4]. Remark If du)+ and k 0 = n + n + n ) +... + n + n +... + n j+ ) + x with 0 x < n +... + n j then l 0 k 0 k 0 jj+). Proof. Remark implies l 0 k 0 = fl 0 ) fk 0 ) > fk 0 ) fk 0 ). According to Lemma, the family F contains the member, the member,..., and the member exactly n times, n times,..., n times, respectively. Therefore, let the output fl) of the algorithm in Lemma be denoted by f n,...,n l). With this notation, for example f n,...,n ) = f n,...,n +,n 0). Using Lemma 3 ii), it follows f n,...,n k 0 ) = f n,...,n j,n j +x,n +...+n j x0) and f n,...,n k 0 ) = f n,...,n j,n j +x,n +...+n j xk 0 ). Consequently, f n,...,n k 0 )) f n,...,n k 0 ) = f n,...,n j,n j +x,n +...+n j xk 0 ) f n,...,n j,n j +x,n +...+n j x0) = f n,...,n j,n j +x,n +...+n j x)k 0 ) f n,...,n j,n j +x,n +...+n j x)k 0 )) +f n,...,n j,n j +x,n +...+n j x)k 0 ) f n,...,n j,n j +x,n +...+n j x)k 0 )) +... +f n,...,n j,n j +x,n +...+n j x)) f n,...,n j,n j +x,n +...+n j x)0)). Note that the expressions f n,...,n j,n j +x,n +...+n j x)s) f n,...,n j,n j +x,n +...+n j x)s ) equal fractions of type aa+) see Lemma 3 and Lemma 4) with a j for s =,..., k 0. Thus, f n,...,n k 0 ) f n,...,n k 0 ) k 0 jj+). ) 4

For integers r and s, consider the graph G r,s obtained from s copies of the clique K r on r vertices and adding s mutually independent edges between these cliques such that G r,s is connected. It follows = r, n j = 0 for j < r, n r = sr s ), n r = s ), and α = s for G r,s. Using Theorem, we obtain xr ) = r )r r )r r )n r + n r r ) ) = r )r r )r sr 4s ) r )s ) + r = 0. Hence, j 0 = r, xj 0 )+n j0 ++n j0 ++...+ j 0 )n + = nr + = s = α, and Remark 3 follows. Remark 3 There are infinitely many graphs G such that the lower bound on α of Theorem is tight. References [] Y. Caro, New Results on the Independence Number, Technical Report, Tel-Aviv University, 979. [] J. Harant and D. Rautenbach, Independence in Connected graphs, submitted to Discrete Math.. [3] J. Harant and I. Schiermeyer, On the independence number of a graph in terms of order and size, Discrete Math. 3 00), 3-38. [4] J. Harant and I. Schiermeyer, A lower bound on the independence number of a graph in terms of degrees, Discuss. Math., Graph Theory 6 006), 43-437. [5] J. Håstad, Clique is hard to approximate within n ɛ, in: Proceedings of 37th Annual Symposium on Foundations of Computer Science FOCS), 996, 67-636. [6] V. Lozin and D. de Werra, Foreword: Special issue on stability in graphs and related topics, Discrete Appl. Math. 3 003), -. [7] V.K. Wei, A Lower Bound on the Stability Number of a Simple Graph, Technical memorandum, TM 8-7 - 9, Bell laboratories, 98. 5