Packing Graphs: The packing problem solved

Similar documents
1 Lecture 13: The derivative as a function.

232 Calculus and Structures

1 The concept of limits (p.217 p.229, p.242 p.249, p.255 p.256) 1.1 Limits Consider the function determined by the formula 3. x since at this point

Rules of Differentiation

List Decomposition of Graphs

Function Composition and Chain Rules

Packing and decomposition of graphs with trees

4. The slope of the line 2x 7y = 8 is (a) 2/7 (b) 7/2 (c) 2 (d) 2/7 (e) None of these.

Math 242: Principles of Analysis Fall 2016 Homework 7 Part B Solutions

Derivatives of trigonometric functions

2.4 Exponential Functions and Derivatives (Sct of text)

0.1 Differentiation Rules

160 Chapter 3: Differentiation

f(x + h) f(x) f (x) = lim

(a 1 m. a n m = < a 1/N n

Polynomial Interpolation

Function Composition and Chain Rules

2.11 That s So Derivative

In Leibniz notation, we write this rule as follows. DERIVATIVE OF A CONSTANT FUNCTION. For n 4 we find the derivative of f x x 4 as follows: lim

Tree Decomposition of Graphs

Continuity and Differentiability of the Trigonometric Functions

Differential Calculus Definitions, Rules and Theorems

On convexity of polynomial paths and generalized majorizations

Derivatives. if such a limit exists. In this case when such a limit exists, we say that the function f is differentiable.

Generic maximum nullity of a graph

Continuous measurements: partial selection

Combining functions: algebraic methods

1 ode.mcd. Find solution to ODE dy/dx=f(x,y). Instructor: Nam Sun Wang

arxiv: v3 [math.gr] 7 Dec Introduction

Packing and Covering Dense Graphs

DIGRAPHS FROM POWERS MODULO p

SECTION 2.1 BASIC CALCULUS REVIEW

University Mathematics 2

HOMEWORK HELP 2 FOR MATH 151

Bob Brown Math 251 Calculus 1 Chapter 3, Section 1 Completed 1 CCBC Dundalk

The Complexity of Computing the MCD-Estimator

A = h w (1) Error Analysis Physics 141

Polynomial Interpolation

A Combinatorial Interpretation of the Generalized Fibonacci Numbers

Cubic Functions: Local Analysis

5.1 We will begin this section with the definition of a rational expression. We

Ramsey numbers of some bipartite graphs versus complete graphs

Volume 29, Issue 3. Existence of competitive equilibrium in economies with multi-member households

Decomposing oriented graphs into transitive tournaments

THE IDEA OF DIFFERENTIABILITY FOR FUNCTIONS OF SEVERAL VARIABLES Math 225

The derivative function

Chapter 2 Limits and Continuity

2.1 THE DEFINITION OF DERIVATIVE

2.8 The Derivative as a Function

INTRODUCTION AND MATHEMATICAL CONCEPTS

Complexity of Decoding Positive-Rate Primitive Reed-Solomon Codes

Hy Ex Ez. Ez i+1/2,j+1. Ex i,j+1/2. γ z. Hy i+1/2,j+1/2. Ex i+1,j+1/2. Ez i+1/2,j

1. Questions (a) through (e) refer to the graph of the function f given below. (A) 0 (B) 1 (C) 2 (D) 4 (E) does not exist

Mathematics 5 Worksheet 11 Geometry, Tangency, and the Derivative

does NOT exist. WHAT IF THE NUMBER X APPROACHES CANNOT BE PLUGGED INTO F(X)??????

Efficient algorithms for for clone items detection

The Cardinality of Sumsets: Different Summands

Some Review Problems for First Midterm Mathematics 1300, Calculus 1

Relations between Clar structures, Clar covers, and the sextet-rotation tree of a hexagonal system

. If lim. x 2 x 1. f(x+h) f(x)

Solve exponential equations in one variable using a variety of strategies. LEARN ABOUT the Math. What is the half-life of radon?

LATTICE EXIT MODELS S. GILL WILLIAMSON

Preconditioning in H(div) and Applications

Characterization of reducible hexagons and fast decomposition of elementary benzenoid graphs

lecture 26: Richardson extrapolation

Chapter Primer on Differentiation

This file is /conf/snippets/setheader.pg you can use it as a model for creating files which introduce each problem set.

11.6 DIRECTIONAL DERIVATIVES AND THE GRADIENT VECTOR

MVT and Rolle s Theorem

Subdifferentials of convex functions

A: Derivatives of Circular Functions. ( x) The central angle measures one radian. Arc Length of r

ch (for some fixed positive number c) reaching c

Introduction to Derivatives

7.1 Using Antiderivatives to find Area

Differentiation Rules c 2002 Donald Kreider and Dwight Lahr

Combinatorica 9(1)(1989) A New Lower Bound for Snake-in-the-Box Codes. Jerzy Wojciechowski. AMS subject classification 1980: 05 C 35, 94 B 25

ACCESS TO SCIENCE, ENGINEERING AND AGRICULTURE: MATHEMATICS 1 MATH00030 SEMESTER /2019

A SHORT INTRODUCTION TO BANACH LATTICES AND

The Verlet Algorithm for Molecular Dynamics Simulations

Numerical Differentiation

= 0 and states ''hence there is a stationary point'' All aspects of the proof dx must be correct (c)

Math 102 TEST CHAPTERS 3 & 4 Solutions & Comments Fall 2006

WHEN GENERALIZED SUMSETS ARE DIFFERENCE DOMINATED

1. Consider the trigonometric function f(t) whose graph is shown below. Write down a possible formula for f(t).

MAT 145. Type of Calculator Used TI-89 Titanium 100 points Score 100 possible points

Packing polynomials on multidimensional integer sectors

Chapter 4 Derivatives [ ] = ( ) ( )= + ( ) + + = ()= + ()+ Exercise 4.1. Review of Prerequisite Skills. 1. f. 6. d. 4. b. lim. x x. = lim = c.

INJECTIVE AND PROJECTIVE PROPERTIES OF REPRESENTATIONS OF QUIVERS WITH n EDGES. Sangwon Park

INTRODUCTION AND MATHEMATICAL CONCEPTS

Derivation Of The Schwarzschild Radius Without General Relativity

The Krewe of Caesar Problem. David Gurney. Southeastern Louisiana University. SLU 10541, 500 Western Avenue. Hammond, LA

STUDY PACKAGE. Subject : Mathematics Topic : DIFFRENTIATION Available Online :

Exam 1 Review Solutions

GELFAND S PROOF OF WIENER S THEOREM

(a) At what number x = a does f have a removable discontinuity? What value f(a) should be assigned to f at x = a in order to make f continuous at a?

Differentiation Rules and Formulas

New families of estimators and test statistics in log-linear models

More on generalized inverses of partitioned matrices with Banachiewicz-Schur forms

Exercises for numerical differentiation. Øyvind Ryan

158 Calculus and Structures

Transcription:

Packing Graps: Te packing problem solve Yair Caro an Rapael Yuster Department of Matematics University of Haifa-ORANIM, Tivon 36006, Israel. AMS Subject Classification: 05B05,05B40 (primary), 05B30,51E05,94C30,6K05,6K10 (seconary). Submitte: November 8, 1996; Accepte: December, 1996 Deicate to te memory of Paul Erős Abstract For every fixe grap H, we etermine te H-packing number of K n, for all n>n 0 (H). We prove tat if is te number of eges of H, an gc(h) =is te greatest common ivisor of te egrees of H, ten tere exists n 0 = n 0 (H), suc tat for all n>n 0, P(H, K n )= n n 1, unless n =1moan n(n 1)/ = b mo (/) were 1 b, in wic case P (H, K n )= n n 1 1. Our main tool in proving tis result is te eep ecomposition result of Gustavsson. 1 Introuction All graps consiere ere are finite, unirecte an simple. For te stanar grap-teoretic terminology te reaer is referre to [Bo]. Let H be a grap witout isolate vertices. An H- packing of a grap G is a set L = {G 1,...,G s }of ege-isjoint subgraps of G, were eac subgrap e-mail: zeac603@uvm.aifa.ac.il e-mail: rapy@mat.tau.ac.il 1

te electronic journal of combinatorics 4 (1997), #R1 is isomorpic to H. Te H-packing number of G, enote by P (H, G), is te maximum carinality of an H-packing of G. An H-covering of a grap G is a set L = {G 1,...,G s } of subgraps of G, were eac subgrap is isomorpic to H, were every ege of G appears in at least one member of L. Te H-covering number of G, enote by C(H, G), is te minimum carinality of an H-covering of G. G as an H-ecomposition if it as an H-packing wic is also an H-covering. Te H-packing an H-covering problems are, in general, NP-Complete as sown by Dor an Tarsi [DoTa]. In case G = K n, te H-covering an H-packing problems ave attracte muc attention in te last forty years, an numerous papers were written on tese subjects (cf. [Br95,Ha,MiMu,CoDi,StKaMu] for various surveys). Special cases of tese problems gaine particular interest. 1. P (K k,k n ) wic as been linke to te various Jonson-Sconeim bouns in Coing Teory [BiEt,BrSSlSm,Sc,Jo]. It is known tat P (K k,k n ) is te maximum size of te binary coes of wor-lengt n, constant weigt k, an istance k ork 3. Despite of muc effort only te cases k =3[an k = 4 [] Sc an k =4[]] are solve. Te case k =5is still open [MuYi].. P (C k,k n ) wic is te cycle-system packing problem, solve completely only for k =3,k=4 [an k = 5 [17] ScBi an k =5[17]]. 3. Te packing-covering conjecture for trees saying tat P (T,K n )= ) / an C(T,Kn )= ) / ( is te number of eges of T ) provie n is sufficiently large. Tis conjecture as been prove for all trees on at most 7 vertices [Ro83,Ro93]. A central result concerning H-ecompositions of K n is te teorem of Wilson [Wi] stating tat for sufficiently large n, K n as an H-ecomposition if an only if e(h) ) an gc(h) n 1 were gc(h) is te greatest common ivisor of te egrees of H. Clearly, if te conitions in Wilson s Teorem ol, ten te packing an covering numbers are known. In tis paper we solve all of te conjectures above, for large n, as special consequences of a muc more general result. In fact, for every H, we etermine P (H, K n ), for all n n 0 (H). Teorem 1.1 Let H be a grap wit eges, an let gc(h)=. Ten tere exists n 0 = n 0 (H), suc tat for all n>n 0, P(H, K n )= n n 1, unless n =1moan n(n 1)/ = b mo (/) were 1 b, in wic case P (H, K n )= n n 1 1.

te electronic journal of combinatorics 4 (1997), #R1 3 Proof of te main result As mentione in te abstract, our main tool is te following result of Gustavsson [Gu]: Lemma.1 (Gustavsson s Teorem [Gu) ]LetHbe a grap wit eges. Tere exists N = N(H), an ɛ = ɛ(h) > 0, suc tat for all n>n,ifgis a grap on n vertices an m eges, wit δ(g) n(1 ɛ), gc(h) gc(g), an m, ten G as an H-ecomposition. It is wort mentioning tat N(H) in Gustavsson s Teorem is a rater uge constant; in fact, it is a igly exponential function of. A sequence of n positive integers 1... n is calle grapic if tere exists an n-vertex grap wose egree sequence is { 1,..., n }. We sall nee te following teorem of Erős an Gallai [ErGa], wic gives a necessary an sufficient conition for a sequence to be grapic. Lemma. (Erős an Gallai [ErGa) ] Te sequence 1... n of positive integers is grapic if an only if its sum is even an for every t =1,...,n t i t(t 1) + min{t, i }. (1) Proof of Teorem 1.1: Given H, we coose n 0 = n 0 (H) = max{n(h), ɛ(h), 8}, were N(H) an ɛ(h) are as in Lemma.1. Now let n>n 0. Let n 1=amo, were 0 a 1. Let n(n 1 a)/ = b mo (/), were 0 b / 1. Note tat since = gc(h) an is te sum of te egrees of H, / must be an integer. Also note tat (n 1 a)/ is an integer, an so b is well-efine. We sall use te obvious fact tat ( +1)/, since δ(h). Tis means tat n>n 0 8>4 >(a+). Anoter useful fact is tat b + na is even since if is even ten a an n ave ifferent parity, an if is o ten / is even an so if b is o ten a an n are bot o, an if b is even ten eiter n is even or a is even. In te first part of te proof we sall give a lower boun for P (H, K n ), an in te secon part we sall give an upper boun for P (H, K n ), an notice tat te lower an upper bouns coincie. Proving a lower boun for P (H, K n ): We sall first assume tat a 0. Our first goal is to sow te existence of an n-vertex grap wic as b vertices wit egree + a, an n b vertices wit egree a. For tis purpose we sall use Lemma., wit i = a + for i =1,...,b an i = a

te electronic journal of combinatorics 4 (1997), #R1 4 for i = b +1,...,n. Notice first tat te sum of te sequence is b + na an tis number is even as mentione above. Let 1 t a +. In tis case, (1) ols since t i t(a + ) =t(t 1) + t(a + t +1) t(t 1)+(a+)(a + 1) = t(t 1)+(a+) (a+)<t(t 1) + n (a + ) t(t 1)+(n t) t(t 1) + min{t, i }. For a + t n we sall prove tat (1) ols by inuction on t, were te base case t = a + was prove above. If t>a+ we use te inuction ypotesis to erive tat t t 1 i = t + i t +(t 1)(t ) + min{t, i } = i=t t + min{t, t } (t 1) + t(t 1) + (a + )+(a+) (a + )+t(t 1) + min{t, i } min{t, i } = t(t 1) + min{t, i }. Tus, tere exists a grap G aving b vertices wit egree + a an n b vertices wit egree a. Consier G = K n \ G. Clearly, gc(g), an G as m eges were m = ( ) n b + na = 1 a) (n(n b))=0mo. Also note tat δ(g) n 1 a = n(1 1+a+ n ) n(1 ɛ(h)), since n>n 0 ɛ(h). Tus, G satisfies te conitions of Lemma.1, an terefore G as an H-ecomposition. Tis means tat P (H, K n ) P(H, G) = m = (n(n 1 a) b)) = n n 1 We now eal wit te case a =0. Ifb= 0 ten K n as an H-ecomposition accoring to Wilson s Teorem [Wi], (or accoring to Lemma.1), so, trivially, P (H, K n )= ) =n n 1 = n n 1 If b>we may elete from K n a subgrap G on b vertices wic is regular (tis is oable since b + na = b is even). As in te case were a 0, te remaining grap G = K n \ G satisfies te conitions of Lemma.1 an terefore ) b n ) P (H, K n ) P(H, G) = = ( = n n 1 = n n 1

te electronic journal of combinatorics 4 (1997), #R1 5 Finally, if 1 b ten we can elete from K n a subgrap G on b + vertices wic is regular. Note tat tis can be one since (+1)/ wic implies < +b. Also, if is o ten b an are bot even, so b + is even. Once again, te remaining grap G = K n \ G satisfies te conitions of Lemma.1 an we get P (H, K n ) P(H, G) = ) (b+(/)) ) b n ) = 1= ( 1= n n 1 1. Proving an upper boun for P (H, K n ): Let L be an arbitrary H-packing of K n. Let s enote te carinality of L. Let G enote te ege-union of all te members of L. G contains s eges. Tus G = K n \ G contains ) s eges. Te egree of every vertex in G is 0 mo an so te egree of every vertex in G is a mo. Terefore, te number of eges in G satisfies ( ) n an + c s = for some non-negative integer c. In particular, ) = an+c mo. Tis, in turn, implies tat n(n 1 a)/ = c mo (/). Tus, we must ave c b. Terefore, ) ( an+c n ) an+b s = = n n 1 Since L was an arbitrary H-packing, we ave P (H, K n ) n n 1 Te only remaining case is a = 0 an 1 b. In tis case, we cannot ave c = b. Tis is because every non-isolate vertex of G as egree at least, an terefore tere are at least ( +1)/ eges in G, i.e c/ ( +1)/, wic implies c + 1, but b. We must, terefore ave c b +/. Terefore, s = ) an+c ) an+(b+/) = n n 1 1. 3 Concluing remarks 1. Teorem 1.1, applie to H = K k yiels, for n n 0 (k), tat P (K k,k n )= n k n 1 k 1, unless k 1 n 1 an n(n 1)/(k 1) mo k is less tan k an greater tan 0, in wic case te above formula soul be reuce by 1. Tis solves, in particular, te relate problem in Coing Teory mentione in te introuction.

te electronic journal of combinatorics 4 (1997), #R1 6. Teorem 1.1, applie to H = C k yiels, for n n 0 (k), tat P (C k,k n )= n k n 1 unless n is o an ) =1,mok. 3. If n n 0 (H) an gc(h) = 1, ten P (H, K n )= (n ) e(h). Note tat by first eleting from K n any set of b<e(h) eges were b = ) mo e(h), te remaining grap satisfies te conitions in Gustavsson s Teorem, an since te set of elete eges may be cosen as a subgrap of H we ave C(H, K n )= (n ) e(h), solving, in particular, te packing-covering conjecture for trees. Our approac allows us to solve te covering problem as well. Tis is one in a fortcoming paper [CaYu]. 4 Acknowlegment Te autors wis to tank N. Alon, T. Etzion, R. Mullin, J. Sconeim an Y. Roitty for useful iscussions, elpful information, an sening important references. References [1] S. Bitan an T. Etzion, Te last packing number of quaruples an cyclic SQS, Design, Coes an Cryptograpy 3 (1993), 83-313. [] A.E. Brouwer, Optimal packing of K 4 s into a K n, J. Combin. Teory, Ser. A 6 (1979), 78-97. [3] A.E. Brouwer, Block Designs, in: Capter 14 in Hanbook of Combinatorics, R. Graam, M. Grötscel an L. Lovász Es. Elsevier, 1995. [4] A. Brouwer, J. Searer, N. Sloane an W. Smit, A new table of constant weigt coes, IEEE Trans. Inform. Teory 36 (1990), 1334-1380. [5] B. Bollobás, Extremal Grap Teory, Acaemic Press, 1978. [6] Y. Caro an R. Yuster, Covering graps: Te covering problem solve, submitte. [7] C.J. Colbourn an J.H. Dinitz, CRC Hanbook of Combinatorial Design, CRC press 1996.

te electronic journal of combinatorics 4 (1997), #R1 7 [8] D. Dor an M. Tarsi, Grap ecomposition is NPC -Acomplete proof of Holyer s conjecture, Proc. 0t ACM STOC, ACM Press (199), 5-63. [9] P. Erős an T. Gallai, Graps wit prescribe egrees of vertices (Hungarian), Mat. Lapok 11 (1960), 64-74. [10] T. Gustavsson, Decompositions of large graps an igraps wit ig minimum egree, Doctoral Dissertation, Dept. of Matematics, Univ. of Stockolm, 1991. [11] H. Hanani, Balance incomplete block esigns an relate esigns, Discrete Mat. 11 (1975), 55-369. [1] S.M. Jonson, A new upper boun for error-correcting coes, IEEE Trans. Inform. Teory 8 (196), 03-07. [13] W.H. Mills an R.C. Mullin, Coverings an packings, in: Contemporary Design Teory: A collection of Surveys, 371-399, eite by J. H. Dinitz an D. R. Stinson. Wiley, 199. [14] R.C. Mullin an J. Yin, On packing of pairs by quintuples v =3,9,17(mo0), Ars Combinatoria 35 (1993), 161-171. [15] Y. Roitty, Packing an covering of te complete grap wit a grap G of four vertices or less, J. Combin. Teory, Ser. A 34 (1983), 31-43. [16] Y. Roitty, Packing an covering of te complete grap IV, te trees of orer 7, Ars Combinatoria 35 (1993), 33-64. [17] A. Rosa an S. Znam, Packing pentagons into complete graps: ow clumsy can you get, Discrete Mat. 18 (1994), 305-316 [18] J. Sconeim, On maximal systems of k-tuples, Stuia Sci. Mat. Hungar. (1966), 363-368. [19] J. Sconeim an A. Bialostocki, Packing an covering of te complete grap wit 4-cycles, Canaian Mat. Bull. 18 (1975), 703-708. [0] R.G. Stanton, J.G. Kalbfleisc an R.C. Mullin, Covering an packing esigns, Proc. n Capel Hill Conf. on Combinatorial Matematics an its applications. Univ. Nort Carolina, Capel Hill (1970) 48-450. [1] R. M. Wilson, Decomposition of complete graps into subgraps isomorpic to a given grap, Congressus Numerantium XV (1975), 647-659.