Star coloring of sparse graphs

Similar documents
Algorithms and Data Structures 2011/12 Week 9 Solutions (Tues 15th - Fri 18th Nov)

Maximum Flow 5/6/17 21:08. Presentation for use with the textbook, Algorithm Design and Applications, by M. T. Goodrich and R. Tamassia, Wiley, 2015

Algorithmic Discrete Mathematics 6. Exercise Sheet

Main Reference: Sections in CLRS.

Network Flow. Data Structures and Algorithms Andrei Bulatov

Warm Up. Correct order: s,u,v,y,x,w,t

CMPS 6610/4610 Fall Flow Networks. Carola Wenk Slides adapted from slides by Charles Leiserson

Flow networks. Flow Networks. A flow on a network. Flow networks. The maximum-flow problem. Introduction to Algorithms, Lecture 22 December 5, 2001

Introduction to SLE Lecture Notes

3-FACIAL COLOURING OF PLANE GRAPHS

Randomized Perfect Bipartite Matching

The Residual Graph. 12 Augmenting Path Algorithms. Augmenting Path Algorithm. Augmenting Path Algorithm

The Residual Graph. 11 Augmenting Path Algorithms. Augmenting Path Algorithm. Augmenting Path Algorithm

Network flows. The problem. c : V V! R + 0 [ f+1g. flow network G = (V, E, c), a source s and a sink t uv not in E implies c(u, v) = 0

COMPETITIVE LOCAL ROUTING WITH CONSTRAINTS

1 Motivation and Basic Definitions

EECE 301 Signals & Systems Prof. Mark Fowler

On Line Supplement to Strategic Customers in a Transportation Station When is it Optimal to Wait? A. Manou, A. Economou, and F.

Maximum Flow 3/3 4/6 1/1 4/7 3/3. s 3/5 1/9 1/1 3/5 2/2. 1/18/2005 4:03 AM Maximum Flow 1

CHAPTER 7: SECOND-ORDER CIRCUITS

2. VECTORS. R Vectors are denoted by bold-face characters such as R, V, etc. The magnitude of a vector, such as R, is denoted as R, R, V

Competitive Routing in the Half-θ 6 -Graph

Graphs III - Network Flow

Introduction to Congestion Games

Notes for Lecture 17-18

Network Flow Applications

4.2 Continuous-Time Systems and Processes Problem Definition Let the state variable representation of a linear system be

The multisubset sum problem for finite abelian groups

u(t) Figure 1. Open loop control system

FLAT CYCLOTOMIC POLYNOMIALS OF ORDER FOUR AND HIGHER

CS4445/9544 Analysis of Algorithms II Solution for Assignment 1

CONTROL SYSTEMS. Chapter 10 : State Space Response

CSC 364S Notes University of Toronto, Spring, The networks we will consider are directed graphs, where each edge has associated with it

Problem Set If all directed edges in a network have distinct capacities, then there is a unique maximum flow.

Sequences Arising From Prudent Self-Avoiding Walks Shanzhen Gao, Heinrich Niederhausen Florida Atlantic University, Boca Raton, Florida 33431

Network Flows: Introduction & Maximum Flow

LAB # 2 - Equilibrium (static)

We just finished the Erdős-Stone Theorem, and ex(n, F ) (1 1/(χ(F ) 1)) ( n

Some Ramsey results for the n-cube

Today: Max Flow Proofs

T L. t=1. Proof of Lemma 1. Using the marginal cost accounting in Equation(4) and standard arguments. t )+Π RB. t )+K 1(Q RB

To become more mathematically correct, Circuit equations are Algebraic Differential equations. from KVL, KCL from the constitutive relationship

18 Extensions of Maximum Flow

Extremal colorings and independent sets

arxiv: v2 [cs.cg] 30 Apr 2015

Admin MAX FLOW APPLICATIONS. Flow graph/networks. Flow constraints 4/30/13. CS lunch today Grading. in-flow = out-flow for every vertex (except s, t)

Inventory Analysis and Management. Multi-Period Stochastic Models: Optimality of (s, S) Policy for K-Convex Objective Functions

INSTANTANEOUS VELOCITY

Entropy compression method applied to graph colorings

Tight bounds for eternal dominating sets in graphs

Sections 2.2 & 2.3 Limit of a Function and Limit Laws

, the. L and the L. x x. max. i n. It is easy to show that these two norms satisfy the following relation: x x n x = (17.3) max

Lecture 2-1 Kinematics in One Dimension Displacement, Velocity and Acceleration Everything in the world is moving. Nothing stays still.

Notes on cointegration of real interest rates and real exchange rates. ρ (2)

Max Flow, Min Cut COS 521. Kevin Wayne Fall Soviet Rail Network, Cuts. Minimum Cut Problem. Flow network.

PHYSICS 151 Notes for Online Lecture #4

Let. x y. denote a bivariate time series with zero mean.

Practice Problems: Improper Integrals

INDEPENDENT SETS IN GRAPHS WITH GIVEN MINIMUM DEGREE

arxiv: v1 [cs.cg] 21 Mar 2013

Chapter 7: Solving Trig Equations

Math 2142 Exam 1 Review Problems. x 2 + f (0) 3! for the 3rd Taylor polynomial at x = 0. To calculate the various quantities:

Geometric Path Problems with Violations

Longest Common Prefixes

Average Case Lower Bounds for Monotone Switching Networks

The Implied Volatility of Forward Starting Options: ATM Short-Time Level, Skew and Curvature

Price of Stability and Introduction to Mechanism Design

Physics 235 Chapter 2. Chapter 2 Newtonian Mechanics Single Particle

Matching. Slides designed by Kevin Wayne.

Physics 240: Worksheet 16 Name

Unit 1 Test Review Physics Basics, Movement, and Vectors Chapters 1-3

Chords in Graphs. Department of Mathematics Texas State University-San Marcos San Marcos, TX Haidong Wu

Lecture Notes 2. The Hilbert Space Approach to Time Series

On a problem of Graham By E. ERDŐS and E. SZEMERÉDI (Budapest) GRAHAM stated the following conjecture : Let p be a prime and a 1,..., ap p non-zero re

CSE 521: Design & Analysis of Algorithms I

Math 333 Problem Set #2 Solution 14 February 2003

Chapter 2. First Order Scalar Equations

Chapter 12: Velocity, acceleration, and forces

5.2 GRAPHICAL VELOCITY ANALYSIS Polygon Method

FIXED POINTS AND STABILITY IN NEUTRAL DIFFERENTIAL EQUATIONS WITH VARIABLE DELAYS

Greedy. I Divide and Conquer. I Dynamic Programming. I Network Flows. Network Flow. I Previous topics: design techniques

Flow Networks Alon Efrat Slides courtesy of Charles Leiserson with small changes by Carola Wenk. Flow networks. Flow networks CS 445

Flow Networks. Ma/CS 6a. Class 14: Flow Exercises

ON THE OSCILLATION OF THIRD ORDER FUNCTIONAL DIFFERENTIAL EQUATIONS. Cairo University, Orman, Giza 12221, Egypt

Classification of 3-Dimensional Complex Diassociative Algebras

Two Coupled Oscillators / Normal Modes

CHAPTER 12 DIRECT CURRENT CIRCUITS

Maximum Flow in Planar Graphs

Languages That Are and Are Not Context-Free

0.1 MAXIMUM LIKELIHOOD ESTIMATION EXPLAINED

NECESSARY AND SUFFICIENT CONDITIONS FOR LATENT SEPARABILITY

7.5 Bipartite Matching. Chapter 7. Network Flow. Matching. Bipartite Matching

6.302 Feedback Systems Recitation : Phase-locked Loops Prof. Joel L. Dawson

20/20 20/20 0/5 0/5 20/20 20/20 5/5 0/5 0/5 5/5 0/20 25/30 20/20 30/30 20/20 0/5 5/5 20/20 0/5 0/5 15/20 15/25 20/20 10/10

CS 473G Lecture 15: Max-Flow Algorithms and Applications Fall 2005

The Asymptotic Behavior of Nonoscillatory Solutions of Some Nonlinear Dynamic Equations on Time Scales

Algorithm Design and Analysis

Rainbow saturation and graph capacities

Computer-Aided Analysis of Electronic Circuits Course Notes 3

Transcription:

Sar coloring of pare graph Yeha B Daniel W. Cranon Mickaël Monaier André Rapad Weifan Wang Abrac A proper coloring of he erice of a graph i called a ar coloring if he nion of eery o color clae indce a ar fore. The ar chromaic nmber χ (G) i he malle nmber of color reqired o obain a ar coloring of G. In hi paper, e dy he relaionhip beeen he ar chromaic nmber χ (G) and he maimm aerage degree Mad(G) of a graph G. We proe ha: 1. If G i a graph ih Mad(G) < 26, hen χ(g) 4. 2. If G i a graph ih Mad(G) < 18 and girh a lea 6, hen χ(g) 5. 3. If G i a graph ih Mad(G) < 8 and girh a lea 6, hen χ(g) 6. 3 Thee rel are obained by proing ha ch graph admi a pariclar decompoiion ino a fore and ome independen e. 1 Inrodcion Le G be a graph. A proper ere coloring of G i an aignmen c of ineger (or label) o he erice of G ch ha c() c() if he erice and are adjacen in G. A k-coloring i a proper ere coloring ing k color. The minimm k ch ha G ha a k-coloring i called he chromaic nmber of G, denoed by χ(g). A proper ere coloring of a graph i acyclic if here i no bicolored cycle in G. In oher ord, he graph indced by he nion of eery eery o color clae i a fore. The acyclic chromaic nmber, denoed by χ a (G), of a graph G i he malle ineger k ch ha G ha an acyclic k-coloring. The acyclic coloring of graph a inrodced by Grünbam in [3]. Finally, a proper ere coloring of a graph i called a ar coloring if he nion of eery o color clae indce a ar fore. The ar chromaic nmber, denoed by χ (G), i he malle ineger k ch ha G ha a ar k-coloring. Deparemen of Mahemaic, Zhejiang Normal Unieriy, Jinha 321004, P. R. China, yhb@zjn.cn DIMACS, Rger Unieriy, Picaaay, Ne Jerey 08854, dcrano@dimac.rger.ed LaBRI UMR CNRS 5800, Unierié Bordea I, 33405 Talence Cede, France, monai@labri.fr LaBRI UMR CNRS 5800, Unierié Bordea I, 33405 Talence Cede, France, rapad@labri.fr Deparemen of Mahemaic, Zhejiang Normal Unieriy, Jinha 321004, P. R. China, f@zjn.cn 1

Sar coloring hae recenly been ineigaed by Ferin, Rapad, and Reed [2], Nešeřil and Oana de Mendez [4], and Alberon e al. [1]. Alberon e al. proed an inereing relaionhip beeen χ, χ, and χ a of G and ome rcre of G (called F -redcion), hich gie he folloing corollary. Corollar [1] Le G be a planar graph. 1. If g(g) 3, hen χ (G) 20. 2. If g(g) 5, hen χ (G) 16. 3. If g(g), hen χ (G) 9. Here g(g) i he girh of G, i.e. he lengh of a hore cycle in G. Moreoer, Alberon e al. gae a planar graph G ih χ (G) = 10 and hey obered ha for any girh here ei a planar graph G ih χ (G) > 3. In hi paper, e dy he relaionhip beeen ar coloring and maimm aerage degree. The maimm aerage degree of G, denoed by Mad(G), i defined a: { } 2 E(H) Mad(G) = ma, H G. V (H) Or main rel i he folloing heorem. Theorem 2 Le G be a graph. 1. If Mad(G) < 26, hen χ (G) 4. 2. If Mad(G) < 18 and g(g) 6, hen χ (G) 5. 3. If Mad(G) < 8 3 and g(g) 6, hen χ (G) 6. I i ell-knon ha if G i a planar graph, hen Mad(G) < 2 g(g) g(g) 2. Hence, e obain he folloing corollary. Corollary 3 Le G be a planar graph. 1. If g(g) 13, hen χ (G) 4. 2. If g(g) 9, hen χ (G) 5. 3. If g(g) 8, hen χ (G) 6. We hold noe ha par 2 and 3 of Corollary 3 ere obained independenly by Timmon [5] and ha he proof of Theorem 2 in he preen paper e echniqe imilar o [5]. Or proof are baed on an eenion of an oberaion by Alberon e al. [1] (Lemma 5.1). They noiced ha if here ei a pariion of he erice of a graph G ch ha: V (G) = F I and he folloing o condiion hold: 2

P1. G[F] i a fore P2. G[I] i an independen e ch ha for all, y I, d G (, y) > 2, hen G ha a ar 4-coloring; e can conrc hi coloring by ar 3-coloring G[F] ih color 1, 2, and 3, (ee Oberaion 5) and coloring G[I] ih color 4. We eend hi idea by ing many independen e aifying P2 in a cerain ene. In Secion 2 e proe Theroem 2.1, in Secion 3 e proe Theorem 2.2, and in Secion 4 e proe Theorem 2.3. In Secion 5, e offer conclding remark. Some noaion: We e V (G) and E(G) o denoe repeciely he e of erice and edge of a graph G; e e δ(g) o denoe he minimm degree of G. For a ere, e denoe by d() i degree. A ere of degree k (rep. a lea k, a mo k) i called a k-ere (rep. k-ere, k-ere). A k-pah i a pah of lengh k (nmber of edge). A k-hread i a (k + 1)-pah hoe k inernal erice are of degree 2. A k i1,,i k -ere ih i 1 i k i a k-ere ha i he iniial ere of k hread of lengh i 1,, i k. Le S V (G); e denoe by G[S] he bgraph of G indced by he erice of S. Le and y be o erice of G; e denoe by d G (, y) he diance in G beeen and y, i.e. he lengh (nmber of edge) of a hore pah beeen and y in G. In or diagram, a ere i colored black if all of i inciden edge are dran; a ere i colored hie if i may hae more inciden edge han hoe dran. We rie S = S 1 S 2 S 3 o denoe ha he S i pariion he e S. 2 Graph ih Mad(G) < 26 In hi ecion, e proe ha eery graph G ih Mad(G) < 26 i ar 4-colorable. Lemma 4 refer o an independen e I ch ha for all, y I, d G (, y) > 2; e call ch a e 2-independen. Lemma 4 If Mad(G) < 26, hen here ei a ere pariion V (G) = I F ch ha F indce a fore and I i a 2-independen e. Proof. Le G be a conereample ih he fee erice. We may ame ha G i conneced and ha no ere of degree 1, ince ch a ere cold be added o F. We begin by proing for claim abo rcre ha m no appear in G. In each cae, le H be he forbidden rcre, le G = G H, and le V (G ) = I F be he deired pariion of V (G ), hich i garaneed by he minimaliy of G. We conclde ih a dicharging argmen, hich ho ha G m conain a lea one of hee forbidden rcre. Thi conradicion implie ha no conereample ei, o he lemma i re. Claim 1: G ha no 2 1,1 -ere. Sppoe i a 2 1,1 -ere in G. Conider he pariion for he graph obained by remoing and i neighbor from G. We eend hi pariion o G. If any of he 3

econd neighbor of are in I, hen add and i neighbor o F. Oherie, add o I and add he neighbor of o F. Claim 2: G ha no 3 1,1,1 -ere adjacen o a 2 0,1 -ere. Sppoe i a 3 1,1,1 -ere adjacen o a 2 0,1 -ere y. Conider he pariion for he graph obained by remoing, y, and heir neighbor from G. If boh of he econd neighbor of ha are no adjacen o y are in I, hen add all he remoed erice o F. If boh of hee econd neighbor are in F, hen add o I and add all oher erice o F. Th, e may ame ha one econd neighbor of i in F, and he oher i in I. If he econd neighbor of y no adjacen o i in I, hen add all erice o F. Oherie, add y o I and all oher erice o F. Claim 3: G ha no 4 1,1,1,1 -ere adjacen o for 2 0,1 -erice. Sppoe i ch a 4 1,1,1,1 -ere in G. Conider he pariion for he graph obained by remoing, he neighbor of, and he econd neighbor of. To eend he pariion o G, add o I and add he neighbor and econd neighbor of o F. We no ho ha G canno hae cerain ype of cycle. We an J o be he maimm bgraph of G ch ha all i erice eiher are 2 0,1 -erice or are 3 0,1,1 - erice ha are adjacen o a lea one 2 0,1 -ere; frher, if d G () = 3, hen e an d J () 2. To form J, le A be he e of 2 0,1 -erice and of 3 0,1,1 -erice ha are adjacen o o 2 0,1 -erice. Add o A all 3 0,1,1 -erice ha are adjacen o one 2 0,1 -ere and anoher 3 0,1,1 -ere, hich i ielf adjacen o a lea one 2 0,1 -ere. Le J be he bgraph indced by A. Claim 4: Eery componen of J i a ree or a cycle. Sppoe J conain a cycle ih ere e C hich conain a ere r ih d J (r) = 3. Chooe ch a cycle o ha C i minimal. By minimaliy C indce a chordle cycle in G. Le V (G C) = F I be a ere pariion of G C, here e may ame ha a 1-ere in G C or a 2-ere adjacen o a 1-ere i in F. We eend hi pariion o all of G in o phae. In phae 1, e ill form I and F by cceiely adding he erice of C o I and F o ha I i a 2-independen e in G, and he only poible cycle indced by F i C ielf. In phae 2, e modify I and F, if neceary, o ha I i a 2-independen e and F i a acyclic. Before beginning phae 1, e alo need o definiion. For conenience, e aign a cyclic orienaion o C. If erice and are in C F, hen i a predeceor of if he direced pah along C from o i enirely in F. We ay ha F i ellbehaed, if for all erice, ih (F C), (F C), and E(G), e hae eiher ha i a leaf in F or ha ha a mo o predeceor, each of hich i a 2-ere in G. We no ho hy i i efl o hae F ell-behaed. Aiming for a conradicion, ame B i a cycle in F in hich B C. Le and be he endpoin of a maimal componen of B C. Wiho lo of generaliy, ame i a predeceor of. Le be he neighbor of no on C. Since lie on B, i no a leaf in F. Since F i ell-behaed, he o predeceor of hae degree 2 in G. Thi conradic he fac ha d G () = 3. Obere ha a he ar of phae 1 e 4

hae F = F and I = I o ha F i riially ell-behaed. We ill enre ha F remain ell-behaed hrogho phae 1, o ha a he end of he phae he only poible cycle indced by F i C. (1) 1 1 (2) y (3) Figre 1: The hree key configraion of Claim 4. To eece phae 1, obere ha he immediae neighborhood (in G) of each 3 0,1,1 - ere on C m appear a in Figre 1.1 or Figre 1.2, here he arro indicae C; recall ha he black erice in he figre hae all neighbor hon, b hie erice may hae addiional neighbor (in G). One by one, e eamine he neighborhood of each 3-ere on C proceeding in he order a e enconer hem on C, and aign each ere eiher o I or F. If C conain a ere ih a 3-neighborhood a in Figre 1.2, hen e begin a ch a. Oherie e chooe a 3-ere a in Figre 1.1 for hich, if poible, I. Sppoe fir ha he 3-neighborhood of look like in Figre 1.1. If or i in I, hen add,, o F. If and are no in I, hen add o I and, o F. Thi keep F ell-behaed: If i aigned o I, hen ha no predeceor. If i no aigned o I, hen i m be he cae ha or i in I. If I, hen ha o predeceor, boh of hich are 2-erice in G, herea if I, hen F i riially 5

ell-behaed. Sppoe hen ha he neighborhood of look like in Figre 1.2. If F, hen add o I and,, 1 o F. If I, and 1 F, hen add 1 o I, and,, o F. If, 1 I, hen add,,, 1 o F. To ee ha F i ell-behaed, obere ha hen F, hen i aigned o I, o / F and 1 ha no predeceor. If I, hen he neighbor of no in C i a leaf of F. Frhermore, if alo 1 I, hen he neighbor of 1 no in C i a leaf in F, herea if 1 F, hen 1 / F. Thi complee phae 1. If F i acyclic, hen e hae eended he pariion o G o e may ame ha F conain a cycle. Since F i ell-behaed, hi cycle m be C. We correc hi problem in phae 2. All of C lie in F preciely hen for each 3-ere on C a in Figre 1.1 e hae I, and for each 3-ere a in Figre 1.2 e hae, 1 I. To form I and F, recall ha C conain a ere r ih d J (r) = 3; ha i r i adjacen o a 3 0,1,1 -ere and o o 2 0,1 -erice. If C rn hrogh = r a in Figre 1.1, hen he immediae neighborhood of m look like Figre 1.3. In hi cae e m hae I. If alo y I, hen le I = I +, herea if y F, hen le I = I + +. In he remaining cae C rn hrogh r a in Figre 1.2. If r =, hen i a 2-ere ih I, hich conradic or ampion ha degree 2 erice adjacen o leae in G C are in F, and h F. If r = 1, hen e ge a imilar conradicion for 1. In any cae, I i 2-independen and F = V (G) I i no acyclic, finihing he proof of Claim 4. No e proe ha eery graph G ih minimm degree 2 and Mad(G) < 26 conain one of he configraion forbidden by Claim 1 4. Le G be a conereample, i.e. a graph ih δ(g) 2, ih Mad(G) < 26, and conaining none of he configraion forbidden by Claim 1 4. We e a dicharging argmen ih iniial charge ω() = d() a each ere and ih he folloing for dicharging rle (R1) (R4), here A and J refer o he ere e and bgraph from Claim 4. We rie ω () o denoe he change a each ere afer e apply he dicharging rle. If i a 3-ere, hen a 2-ere i nearby if i adjacen o, or if and hae a common neighbor of degree 2 (in G). Noe ha he dicharging rle do no change he m of he charge. To complee he proof, e ho ha ω () 26 for all V (G); hi lead o he folloing obio conradicion: 26 26 V (G) V (G) V (G) ω () = V (G) Hence no conereample can ei. The dicharging rle are defined a follo. (R1) Each 3-ere gie 2 o each nearby 2-ere. V (G) ω() = 2 E(G) Mad(G) < 26 V (G) V (G). (R2) Each 3 0,1,1 -ere receie 1 from i neighbor of degree a lea 3, nle hi neighbor i in A in hich cae no charge i ranferred. 6

(R3) If d J () = 3, hen ge 1 from he bank. (R4) If / A and i adjacen o k 2 0,1 -erice, hen gie k o he bank. We no erify ha ω () 26 for each V (G). Noe ha each leaf in J i a 2 0,1 -ere. Hence, by (R4), he bank receie 1 for each leaf of J. By (R3), he bank gie aa for each 3-ere of J. Claim 4 implie ha J ha aerage degree a mo 2, o J conain a lea a many leae a 3-erice; hence, he bank ha nonnegaie charge. Cae d() = 2 By (R1), receie 2 from each nearby 3-ere, o ω () = 2+2 2 = 26. Cae d() = 3 If i a 3-ere, hen e conider eeral poibiliie: If i a 3 1,1,1 -ere, hen by Claim 2, i no adjacen o a 2 0,1 -ere. Hence, doe no end any charge o he bank. In hi cae, ω () = 3 3 2 = 2. If i a 3 0,1,1 -ere adjacen o o 2 0,1 -erice, hen A. Le be he neighbor of ha i no a 2-ere. If E(J), hen d J () = 3. By (R3), ill ge 1 2 from he bank. By (R1), end a oal of 4 o nearby 2-erice. Th ω () = 3 + 1 4 2 = 26. If / E(J), hen / A, o ha by (R2), receie 1 from. Therefore he ame calclaion applie o. If i a 3 0,1,1 -ere adjacen o a mo one 2 0,1 -ere, hen gie a mo 1 o he bank. Therefore ω () 3 3 2 1 = 26. If i a 3-ere adjacen o a mo one 2-ere, hen ω () 3 2 2 2 1 1 = 26. Cae d() = 4 If i a 4-ere, hen by Claim 3, i adjacen o a mo hree 2 0,1 -erice, o ω () 4 2 3 1 = 2. Cae d() 5 If i a 5-ere, hen ω () d() 4 d() 1 d() = 6 30 d(). Thi implie ha Mad(G) 26, hich proide he needed conradicion. Oberaion 5 If F i a fore, hen χ (G) 3. In each componen T of F, arbirarily chooe a roo r; color each ere of T ih color (1 + d T (r, )) (mod 3). Proof of Theorem 2.1 Le G be a graph ih Mad(G) < 26. By Lemma 4, here ei a ere pariion V (G) = F I. No e color: F ih color 1, 2, and 3 (a in Oberaion 5), and I ih color 4. Thi prodce a ar 4-coloring of G.

3 Graph ih Mad(G) < 18 and g(g) 6 In hi ecion, e proe ha eery graph G ih Mad(G) < 18 and g(g) 6 i ar 5-colorable. The proof in hi ecion and he ne ecion cloely reemble hoe in Secion 2. The main difference i ha in boh hi ecion and he ne, e pli he argmen ino o lemma, raher han one. In hi ecion, e fir proe ha each graph of inere m conain one of 14 configraion; e e hi lemma o proe a rcral decompoiion rel, hich in rn lead o he ar coloring rel. The oline of Secion 4 i imilar. Lemma 6 A graph G ih Mad(G) < 18 and g(g) 6 conain one of he folloing 14 configraion (ee Figre 2 and 3): 1. A 1-ere. 2. A 2 1,1 -ere. 3. A 3 1,1,1 -ere. 4. A 3 0,0,2 -ere. 5. A 3 0,1,1 -ere adjacen o a 3 0,1,1 -ere. 6. A 3 0,0,1 -ere adjacen o o 3 0,1,1 -ere.. A 4 1,1,1,2 -ere. 8. A 4 0,2,2,2 -ere adjacen o a 3-ere. 9. A 4 0,1,1,1 -ere adjacen o a 3 0,1,1 -ere. 10. A 4 0,1,1,1 -ere adjacen o a 4 0,2,2,2 -ere.. A 4 0,0,2,2 -ere adjacen o o 4 0,2,2,2 -ere. 12. A 4 0,0,2,2 -ere adjacen o a 4 0,2,2,2 -ere and a 3 0,1,1 -ere. 13. A 5 1,2,2,2,2 -ere. 14. A 5 0,2,2,2,2 -ere adjacen o a 4 0,2,2,2 -ere. Proof. Le G be a conereample, i.e. a graph ih Mad(G) < 18, ih g(g) 6, and conaining none of he configraion of Lemma 6. We e a dicharging argmen ih iniial charge ω() = d() a each ere ere, and ih he folloing for dicharging rle (R1) (R4), hich decribe ho o rediribe he charge. We rie ω () o denoe he charge a each ere afer e apply he dicharging rle. Noe ha he dicharging rle do no change he m of he charge. To complee he proof, e ho ha ω () 18 for all V (G); hi lead o he folloing obio conradicion: 18 18 V (G) V (G) ω () V (G) = ω() = 2 E(G) Mad(G) < 18 V (G) V (G) V (G) V (G). Hence no conereample can ei. 8

y z (4) (1) y z p q r y p q r (5) r y z p () z y r q p o n (8) y z o n (6) (3) y (2) Figre 2: (1 of 2) The fir 8 of he 14 naoidable configraion in Lemma 6. 9

o p q r z 2 z 3 y 3 3 y 4 4 y 2 2 5 y 5 (9) y z 1 6 z 1 y 6 (10) 1 1 1 1 r1 1 1 r 1 1 2 3 2 3 y 2 y 3 1 2 3 2 3 y 2 y 3 2 z 4 5 4 5 y 4 y 5 2 z p 1 q 1 r 2 6 6 y 6 r 2 p 2 2 2 q 2 () (12) 4 4 3 y 3 3 y 4 4 5 y 5 3 r 3 r 4 4 3 1 1 2 2 y 2 2 y 2 2 1 2 r 2 2 1 r 1 3 3 (13) 1 1 (14) y 3 Figre 3: (2 of 2) The la 6 of he 14 naoidable configraion in Lemma 6. 10

The dicharging rle are defined a follo. Each 3-ere gie: (R1) 2 o each 2-ere ha i no adjacen o a 2-ere. (R2) 4 o each 2-ere ha i adjacen o a 2-ere ( i a 2 0,1-ere). (R3) 1 o each 3 0,1,1-ere. (R4) 2 o each 4 0,2,2,2-ere. To complee he proof, i ffice o erify ha ω () 18 for all V (G). Cae d() = 2 If i a 2 0,1 -ere, hen receie 4 from i neighbor of degree a lea 3. Hence, ω () = 2+ 4 = 18. Oherie, i adjacen o o 3-erice, hich each gie 2 o. Hence ω () = 2 + 2 2 = 18. Cae d() = 3 By Lemma 6.4, i no inciden o a 2-hread. If i a 3 0,1,1 -ere, hen gie 2 o each adjacen 2-ere and receie 1 from i hird neighbor (ince hi hird neighbor i neiher a 2-ere, nor a 3 0,1,1 -ere, nor a 4 0,2,2,2 -ere). Hence, ω () = 3 2 2 + 1 = 18. If i a 3 0,0,1 -ere, hen i adjacen o a mo one 3 0,1,1 -ere and o no 4 0,2,2,2 -ere. Hence, ω () 3 2 1 = 18. Finally, ame ha i no adjacen o a 2-ere. Since i no adjacen o a 4 0,2,2,2 -ere, ω () 3 3 1 = 18. Cae d() = 4 By Lemma 6.8, i inciden o a mo hree 2-hread. If i inciden o eacly hree 2-hread, hen he forh neighbor of i neiher a 2-ere, nor a 3 0,1,1 -ere, nor a 4 0,2,2,2 -ere; hence, hi forh neighbor gie 2 o. Th ω () = 4 3 4 + 2 = 18. Ame ha i inciden o eacly o 2-hread. If he hird (or forh) neighbor of i a 2-ere or a 4 0,2,2,2 -ere, hen he final neighbor of doe no receie charge from (by Lemma 6. and 6.9 6.12). In hi cae, ω () = 4 2 4 2 = 18. I i alo poible ha one or boh of he remaining neighbor of are 3 0,1,1 -erice. In hi cae, ω () 4 2 4 2 1 = 18. If i inciden o a mo one 2-hread, hen ω () 4 4 3 2 = 18. Cae d() = 5 By Lemma 6.13, i inciden o a mo for 2-hread. If i inciden o eacly for 2-hread, hen i fifh neighbor i neiher a 2- ere, nor a 4 0,2,2,2 -ere. Hence, ω () = 5 4 4 1 = 18. If i inciden o a mo hree 2-hread, hen ω () 5 3 4 2 2 = 19. Cae d() 6 Triially, ω () d() d() 4 = 3 d() 18. Thi implie ha Mad(G) 18, hich proide he needed conradicion. Lemma If G i a graph ih Mad(G) < 18 and g(g) 6, hen here ei a ere pariion V (G) = F I 1 I 2 ch ha:

P1. F indce a fore, P2. I 1 i an independen e ch ha for all, y I 1, d G[F I1](, y) > 2, and P3. I 2 i an independen e ch ha for all, y I 2, d G (, y) > 2. Thi decompoiion a fir inrodced by Timmon in [5]. Proof. Le G be a conereample ih he fee erice. By Lemma 6, G conain one of he 14 configraion hon in Figre 2 and 3. In each configraion, le N denoe he e of black erice. In each cae, e delee a bgraph H and pariion V (G H) ino e F, I 1, and I 2 ha aify P 1, P 2, and P 3. We no ho ho o add he deleed erice o F, I 1, and I 2 o ha he reling e ill aify P 1, P 2, and P 3. We conider each of he 14 configraion in Figre 2 and 3. (1) Le G = G. By he minimaliy of G, he ere e of G i he dijoin nion of e F, I 1, and I 2 ha aify P1, P2, and P3. No add o F. (2) Le G = G\N. If, y F, hen add and o F and add o I 1. Oherie, add,, o F. (3) Le G = G\N. If, y, F, hen add,, z o F and add o I 1. If eacly o of he erice, y, are in F, hen.l.o.g., y F and I 1 (rep. I 2 ); add,, z o F and add o I 2 (rep. I 1). If a mo one of he erice, y, i in F, hen add,,, z o F. (4) Le G = G \ {, }. Conider he cae, F. If y, z I 1 I 2 hen add and o F. Oherie: if y, z I 2 F, hen add o F and o I 1 ; imilarly, if y, z I 1 F, hen add o F and add o I 2. Finally, if / F or / F, hen add and o F. (5) Le G = G \ N. If, p, y, F, hen add, q,, o F, add o I 1, and add r o I 2. If eacly hree of, p, y, are in F, hen W.l.o.g.,, p, y F and I 1; add o I 1 and, p, r,, o F. The remaining cae are eaier and are lef o he reader. (6)-(9) Thee eay cae are lef o he reader. (10) Le G = G \ N. We conider o cae: (i) If a lea o of y 4, y 5, y 6 are in independen e, hen add o an independen e and add he oher erice o he fore; (ii) If a mo one of y 4, y 5, y 6 i in an independen e, hen add o he oher independen e and add all he oher erice o he fore. () Le G = G \ N. Add and o I 1, add z o I 2, and add all he oher erice o he fore. (12) Le G = G \ N. We j conider he cae here q 1 and q 2 are in F ; he remaining cae are eaier. Add and o I 1, add z o I 2, and add he oher erice o he fore. 12

(13) Le G = G \ N. Add o an independen e ha doen conain y 5 and add all he oher erice o he fore. (14) Le G = G \ N. Add o I 1, add o I 2, and add all oher oher erice o he fore. Proof of Theorem 2.2 Le G be a graph ih Mad(G) < 18 and g(g) 6. By Lemma, here ei a ere pariion V (G) = F I 1 I 2. No e color: G[F] ih he color 1, 2, and 3 (a in Oberaion 5), G[I 1 ] ih he color 4, and G[I 2 ] ih he color 5. Thi prodce a ar 5-coloring of G. Oberaion 8 The condiion g(g) 6 can be dropped. Since here are no 3-hread, e j e hi ampion o enre ha hie and black erice are diinc in he configraion in Figre 2 and 3. 4 Graph ih Mad(G) < 8 3 and g(g) 6 In hi ecion, e proe ha eery graph G ih Mad(G) < 8 3 6-colorable. and g(g) 6 i ar Lemma 9 A graph G ih Mad(G) < 8 3 and g(g) 6 conain one of he folloing 8 configraion: 1. A 1-ere. 2. A 2 0,1 -ere adjacen o a 4-ere. 3. A 3 0,1,1 -ere adjacen o a 3-ere. 4. A k 1,1,2,,2 -ere. 5. A k 0,2,,2 -ere adjacen o a 3-ere. 6. A k 0,1,2,,2 -ere adjacen o k 0,1,1,2,,2 -ere.. A k 0,0,2,,2 -ere adjacen o o 3 0,1,1 -erice. 8. A k 0,0,2,,2 -ere adjacen o a k 0,1,1,2,,2-ere and o a k 0,1,2,2,,2 -ere. 13

y 2 z r y 3 (1) (2) y 2 (3) z n 2 y n 2 n 2 1 1 z n 2 y n 2 n 2 1 z 2 y 2 2 z 1 1 (4) 2 2 z 2 y 2 2 z 1 1 (5) 2 2 z n 2 z 2 z 1 y n 2 y 2 n 2 2 1 3 3 1 1 2 2 r m 3 m 3 m 3 r 2 2 2 r 1 1 1 z n 2 z 2 z 1 y n 2 y 2 n 2 2 r 2 1 2 1 1 1 (6) 3 () 2 3 z n 2 y n 2 n 2 z 2 2 y 2 z 1 1 2 2 a l 2 b l 2 c l 2 a 2 b 2 c 2 a 1 b 1 c 1 r m 3 m 3 m 3 r 2 2 2 1 r 1 1 1 1 (8) Figre 4: The 8 naoidable configraion in Lemma 9. 14

Proof. Or proof i imilar o he proof of Lemma 4 and 6. Le G be a conereample. We e a dicharging argmen ih iniial charge ω() = d() a each ere ere, and ih he folloing hree dicharging rle (R1) (R3), hich decribe ho o rediribe he charge. We rie ω () o denoe he charge a each ere afer e apply he dicharging rle. Noe ha he dicharging rle do no change he m of he charge. To complee he proof, e ho ha ω () 8 3 for each V (G); hi lead o he folloing obio conradicion: 8 8 3 3 V (G) V (G) Hence no conereample can ei. V (G) ω () V (G) = ω() = 2 E(G) Mad(G) < 8 V (G) V (G) V (G) 3. The dicharging rle are defined a follo. (R1) Each 5-ere gie 2 3 o each adjacen 2-ere hich i adjacen o a 2- ere ( i a 2 0,1 -ere). (R2) Each 3-ere gie 1 3 o each adjacen 2-ere hich i no adjacen o a 2- ere. (R3) Each 4-ere gie 1 3 o each adjacen 3 0,1,1-ere and 1 3 o each adjacen 5 0,2,2,2,2 -ere. We call 3 0,1,1 -erice and 5 0,2,2,2,2 -erice ligh erice. If a ere i he iniial or final ere of a hread, e call i an end of he hread. By Lemma 9.1., δ(g) 2. Cae d() = 2 If i no adjacen o any 2-erice, hen i receie 1 3 from each neighbor by (R2); o, ω () = 2 + 2 1 3 = 8 3. If i a 2 0,1-ere, hen i i adjacen o a 5-ere. By (R3), gie 2 3 o, o ω () = 2 + 2 3 = 8 3. Cae d() = 3 By Lemma 9.2 and he girh condiion, i no he end of a 2-hread. Moreoer, by Lemma 9.3, i he end of a mo o 1-hread. If i he end of eacly o 1-hread, hen i i adjacen o a 4-ere, by Lemma 9.3. So, ω () = 3 2 1 3 + 1 3 = 8 3 by (R2) and (R3). If i he end of a mo one 1-hread, hen ω () 3 1 3 = 8 3. Cae d() = 4 By (R2) and (R3), ω () 4 4 1 3 = 8 3. Cae d() = 5 By Lemma 9.4, i he end of a mo for 2-hread. If i he end of eacly for 2-hread ( i ligh), hen i i neiher adjacen o a 3-ere, by Lemma 9.4 and 9.5, nor o a 5 0,2,2,2,2 -ere, by Lemma 9.6. Hence, receie 1 3 by (R3). So, ω () = 5 4 2 3 + 1 3 = 8 3. If i he end of eacly hree 2-hread, hen i adjacen o a mo one ligh ere, by Lemma 9. and 9.8. So, ω () 5 3 2 3 1 3 8 3. Finally, if i he end of a mo o 2-hread, hen ω () 5 2 2 3 3 1 3 8 3. 15

Cae d() 6 By Lemma 9.4, i he end of a mo d() 1 2-hread. If i he end of eacly d() 1 2-hread, hen i i neiher adjacen o a 3- ere, by Lemma 9.4 and 9.5, nor o a 5 0,2,2,2,2 -ere, by Lemma 9.6; hence, ω () = d() (d() 1) 2 3 8 3. If i he end of eacly d() 2 2-hread, hen i adjacen o a mo one ligh ere, by Lemma 9. and 9.8. So, ω () d() (d() 2) 2 3 1 3 3. Finally, if i he end of a mo d() 3 2-hread, hen ω () d() (d() 3) 2 3 3 1 3 3. Thi implie ha Mad(G) 8 3, hich proide he needed conradicion. Lemma 10 Le G be a graph ih Mad(G) < 8 3 ere pariion V (G) = F I 1 I 2 I 3 ch ha: and g(g) 6. Then here ei a P1. F indce a fore. P2. I i i an independen e ch ha for all, y I i, d G (, y) > 2 for i = 1, 2, 3. Proof. Le G be a conereample ih he fee erice. By Lemma 9, G conain one of he 8 configraion in Figre 4. In each configraion, le N denoe he e of black erice. In each cae, e delee a bgraph H and pariion V (G H) ino e F, I 1, I 2, and I 3 ha aify P 1 and P 2. We no ho ho o add he deleed erice o F, I 1, I 2, and I 3 o ha he reling e ill aify P 1 and P 2. We conider each of he 8 configraion in Figre 4: (1) Le G = G. By he minimaliy of G, here ei a ere pariion V (G ) = F I 1 I 2 I 3 aifying P1 and P2. Add o F. (2) Le G = G \ {, } and V (G ) = F I 1 I 2 I 3. If a lea one of he erice and belong o an independen e, hen add and o F. So ame ha, F. If all hree y i are in diinc independen e, hen add and o F. Oherie, add o he fore and add o an independen e ha doe no conain any y i. (3) Le G = G \ {,, z}. If hree or more of he erice, r,, y 2 belong o independen e, hen add,, z o he fore. So ppoe ha a mo o of hee erice belong o independen e. Add o an independen e ha doe no conain any of, r,, y 2 ; add all he oher erice o he fore. I may happen ha and are in he ame independen e; in hi cae, e can eiher add,, z o he fore or e can moe o he fore. (4) Le G = G \N. Add o an independen e ha conain neiher 1 nor 2 and add all he oher erice o he fore. (5) Le G = G \ {black erice ecep }. Add o an independen e ha conain neiher 1 nor 2 and add all he oher erice o he fore (moe o he fore if neceary). 16

(6) Le G = G\N. Add o an independen e ha conain neiher 2 nor 3, add o an independen e ha conain neiher nor 1, and add he all he oher erice o he fore. () Le G = G\N. Add r o an independen e ha conain neiher 1 nor 2, and add o an independen e ha conain neiher 1, nor 2, nor r; if here i no choice for, hen add o he fore. Add o an independen e ha conain neiher r nor, and add all he oher erice o he fore. (8) Le G = G \ N. Add o an independen e ha conain neiher 1 nor 2, add o an independen e ha conain neiher nor 3, and add o an independen e ha conain neiher nor. Finally, add all he oher erice o he fore. Proof of Theorem 2.3 Le G be a graph ih Mad(G) < 8 3 and g(g) 6. By Lemma 10, here ei a ere pariion V (G) = F I 1 I 2 I 3. No e color: F ih he color 1, 2, and 3 (a in Oberaion 5), I 1 ih he color 4, I 2 ih he color 5, and I 3 ih he color 6. Thi prodce a ar 6-coloring of G. Oberaion The condiion g(g) 6 can be dropped. Since here are no 3- hread, e j e hi ampion o enre ha hie and black erice are diinc in he configraion in Figre 4. 5 Conclion In hi paper, e proed ha if G i a graph ih Mad(G) < 26 18 (rep. Mad(G) < and g(g) 6, Mad(G) < 8 3 and g(g) 6), hen χ (G) 4 (rep. 5, 6). Are hee rel opimal? Le define a fncion f a follo. Definiion 12 Le f : N R be defined by Obere ha: f(n) = inf{mad(h) χ (H) > n} f(1) = 1. Acally, if G conain an edge, hen χ (G) > 1; oherie, G i a independen e, o χ (G) = 1 and Mad(G) = 0. Noe alo ha: 1

f(2) = 3 2. If G conain a pah of lengh 3, hen χ (G) > 2 and Mad(G) 3 2. Oherie, G i a ar fore, o χ (G) 2 and Mad(G) < 2. Finally, noe ha: f(3) = 2. If Mad(G) < 2, hen G i a fore and χ (G) = 3. In conra, if G i a 5-cycle, hen Mad(G) = 2 and χ (G) = 4. Problem 13 Wha i he ale of f(n) for n = 4, 5, 6? y z Figre 5: A graph G ih Mad(G) = 18 and χ (G) > 4. From Theorem 2.1, e kno ha: 26 f(4). The graph G, in Figre 5, ha Mad(G) = 18 and χ (G) > 4, o G gie an pper bond on f(4). Hence: 26 = 182 198 f(4) = 18 Finally, obere ha here ei a ar 4-colorable graph G ih Mad(G) = 5 2 ha doe no admi a pariion V = F I ch ha for all, y I, d G (, y) > 2. See Figre 6. 6 Acknoledgemen Thank o Craig Timmon and André Kündgen for helpfl dicion. 18

Figre 6: A ar 4-colorable graph G ih Mad(G) = 5 2 ha ha no pariion F I ch ha for all, y I, d G (, y) > 2. Reference [1] M.O. Alberon, G.G. Chappell, H.A. Kieread, A. Kündgen, and R. Ramamrhi. Coloring ih no 2-colored P 4. The Elecronic Jornal of Combinaoric, (1):R26, 2004. [2] G. Ferin, A. Rapad, and B. Reed. On ar coloring of graph. LNCS, 2204:140 153, 2001. [3] B. Grünbam. Acyclic coloring of planar graph. Irael J. Mah., 14:390 408, 193. [4] J. Nešeřil and P. Oana de Mendez. Coloring and homomorphim of minor cloed clae. Dicree and Compaional Geomery: The Goodman-Pollack Fechrif, 651 664, 2003. [5] C. Timmon. Sar coloring planar graph. Maer hei, California Sae Unieriy San Marco, May 200. 19