This action might not be possible to undo. Are you sure you want to continue?

GRAPH THEORY

Tero Harju

Department of Mathematics University of Turku FIN-20014 Turku, Finland e-mail: harju@utu.ﬁ

1994 – 2011

Contents

1

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 1.1 Graphs and their plane ﬁgures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.2 Subgraphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 1.3 Paths and cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Connectivity of Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.1 Bipartite graphs and trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 2.2 Connectivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Tours and Matchings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1 Eulerian graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Hamiltonian graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 Matchings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Colourings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1 Edge colourings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Ramsey Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Vertex colourings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Graphs on Surfaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 Planar graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Colouring planar graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Genus of a graph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 29 31 35 43 43 47 53 61 61 68 76

2

3

4

5

6

Directed Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 6.1 Digraphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 6.2 Network Flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97

1 Introduction

Graph theory may be said to have its beginning in 1736 when E ULER considered the (general case of the) Königsberg bridge problem: Does there exist a walk crossing each of the seven bridges of Königsberg exactly once? (Solutio Problematis ad geometriam situs pertinentis, Commentarii Academiae Scientiarum Imperialis Petropolitanae 8 (1736), pp. 128-140.) It took 200 years before the ﬁrst book on graph theory was written. This was “Theorie der endlichen und unendlichen Graphen” ( Teubner, Leipzig, 1936) by K ÖNIG in 1936. Since then graph theory has developed into an extensive and popular branch of mathematics, which has been applied to many problems in mathematics, computer science, and other scientiﬁc and not-so-scientiﬁc areas. For the history of early graph theory, see N.L. B IGGS , R.J. L LOYD Press, 1986.

AND

R.J. W ILSON, “Graph Theory 1736 – 1936”, Clarendon

There are no standard notations for graph theoretical objects. This is natural, because the names one uses for the objects reﬂect the applications. Thus, for instance, if we consider a communications network (say, for email) as a graph, then the computers taking part in this network, are called nodes rather than vertices or points. On the other hand, other names are used for molecular structures in chemistry, ﬂow charts in programming, human relations in social sciences, and so on. These lectures study ﬁnite graphs and majority of the topics is included in J.A. B ONDY, U.S.R. M URTY, “Graph Theory with Applications”, Macmillan, 1978. R. D IESTEL, “Graph Theory”, Springer-Verlag, 1997. F. H ARARY, “Graph Theory”, Addison-Wesley, 1969. D.B. W EST, “Introduction to Graph Theory”, Prentice Hall, 1996. R.J. W ILSON, “Introduction to Graph Theory”, Longman, (3rd ed.) 1985. In these lectures we study combinatorial aspects of graphs. For more algebraic topics and methods, see N. B IGGS, “Algebraic Graph Theory”, Cambridge University Press, (2nd ed.) 1993. C. G ODSIL , G.F. R OYLE, “Algebraic Graph Theory”, Springer, 2001. and for computational aspects, see S. E VEN, “Graph Algorithms”, Computer Science Press, 1979.

|X | denotes its size (cardinality. .k] Xi and Xi ∩ X j = ∅ for all different i and j . Here X \ Y = { x | x ∈ X. . Otherwise. and X △Y = ( X \ Y ) ∪ (Y \ X ) is their symmetric difference. • For a real number x. . if it is in P. x ∈ Y }. if both are even. . . . or both are odd. X × Y = {( x. On the other hand. n] = {i. X2 . if n ≡ k (mod 2). y ∈ Y } is their Cartesian product. if X= i∈[1. they have opposite parity. This is one of the great problems in modern mathematics and theoretical computer science. . that is. n}. . y) | x ∈ X. • A family {X1 . n] = {1. If the guessing in NP-problems can be replaced by an efﬁcient systematic search for a solution. the ﬂoor and the ceiling of x are the integers ⌊ x⌋ = max{k ∈ Z | k ≤ x} and ⌈ x⌉ = min{k ∈ Z | x ≤ k}. Intuitively. then P=NP. 2. .3 In these lecture notes we mention several open problems that have gained respect among the researchers. n} for integers i ≤ n. . a problem is in NP 2 . Sections with a star (∗) in their heading are optional. Xk } of subsets Xi ⊆ X of a set X is a partition of X. if it is ﬁrst efﬁcient to guess a solution and then efﬁcient to check that this solution is correct. Graph theory has abundant examples of NP-complete problems. then necessarily P=NP. a problem is in P 1 if there is an efﬁcient (practical) algorithm to ﬁnd a solution to it. and in general. 1 2 Solvable – by an algorithm – in polynomially many steps on the size of the problem instances. . the number of its elements). . It is conjectured (and not known) that P = NP. Finding a solution to any one of these problems is another matter. / • Two integers n. [i. . Indeed. i + 1. For any one NP-complete problem. Notations and notions • For a ﬁnite set X. • For two sets X and Y. k ∈ N (often n = |X | and k = |Y | for sets X and Y) have the same parity. Solvable nondeterministically in polynomially many steps on the size of the problem instances. graph theory has the advantage that it contains easily formulated open problems that can be stated early in the theory. • Let [1.

v2 . c} and edges ψ(e1 ) = aa. . v2 v3 . the 2-sets of V. Note that we can have ψ(e1 ) = ψ(e2 ). Two edges e1 = uv and e2 = uw having a common end. E. v5 . em } is a set (of symbols). In literature. In order to simplify notations. A pair G = (V. The vertex set of a graph G is denoted by VG and its edge set by EG . This is drawn in the ﬁgure of G by placing two (parallel) edges that connect the common ends. The ﬁgure on the right is a geometric representation of the graph G with VG = {v1 . A pair {u. For an edge e = uv ∈ G. The number νG of the vertices is called the order of G. EG ). v5 v6 }. v} | u. v6 } and E G = { v1 v2 . Vertices u and v are adjacent or neighbours. and ε G is the size of G. v} is usually written simply as uv. Graphs can be generalized by allowing loops vv and parallel (or multiple) edges between vertices to obtain a multigraph G = (V. Therefore G = (VG . graphs are also called simple graphs. edges are called lines or links. A graph G can be represented as a plane ﬁgure by drawing a line (or a curve) between the points u and v (representing vertices) if e = uv is an edge of G. we denote νG = |VG | and ε G = | EG | . e2 . v3 . The elements of V are the vertices of G. D EFINITION . b.1 Graphs and their plane ﬁgures Let V be a ﬁnite set. v ∈ V. On the right there is (a drawing of) a multigraph G with vertices V = {a. Notice that then uv = vu. The list of alternatives is long (but still ﬁnite). we also write v ∈ G and e ∈ G instead of v ∈ VG and e ∈ EG . the vertices u and v are its ends. For a graph G. and ψ : E → E(V ) ∪ {vv | v ∈ V } is a function that attaches an unordered pair of vertices to each e ∈ E: ψ(e) = uv. . in which case the vertices are drawn as anonymous circles.1 Graphs and their plane ﬁgures 4 1. E) with E ⊆ E(V ) is called a graph (on V). where E = {e1 . v1 v3 . vertices are called nodes or points. and those of E the edges of G. are adjacent with each other.1. u = v} . v1 v3 v6 v2 v4 v5 Often we shall omit the identities (names v) of the vertices in our ﬁgures. ψ). . subsets of two distinct elements. ψ(e3 ) = bc..e. D EFINITION . if uv ∈ G. . ψ(e2 ) = ab. b a c . and denote by E(V ) = {{u. and ψ(e4 ) = bc. v4 . i. v2 v4 .

1. the required isov5 1 morphism is given by v1 → 1. v4 → 2. In particular. Two isomorphic graphs enjoy the same graph theoretical properties. where the edges are drawn as arrows. In this case. A digraph can contain edges uv and vu of opposite directions. in fact. v5 → 5. Hence G and H are isomorphic if the vertices of H are renamings of those of G. v2 → 3. K = [1. v1 v4 3 5 Isomorphism Problem. and weighted: D EFINITION . and talk of ‘the graph’ although there are. It tells that at least for computational purposes an efﬁcient algorithm for checking whether two graphs are isomorphic or not would be greatly appreciated. Usually. the edges are ordered: E ⊆ V × V. Graphs and digraphs can also be coloured. v ∈ G. v3 → 4. A function α : EG → K is an edge colouring of G. This shows in the ﬁgures. The following graphs are isomorphic. where we tend to replace the vertices by small circles. Two graphs G and H are isomorphic. all isomorphic graphs have the same plane ﬁgures (excepting the identities of the vertices). We also study directed graphs or digraphs D = (V. inﬁnitely many such graphs. Does there exist an efﬁcient algorithm to check whether any two given graphs are isomorphic or not? The following table lists the number 2( 2 ) of all graphs on a given set of n vertices. v2 v3 2 4 Example 1. and they are often identiﬁed. denoted by G ∼ H. If K ⊆ R (often K ⊆ N). then α is a weight function or a distance function. labelled. Isomorphism of graphs D EFINITION . uv = vu. The directed graphs have representations. 5 D EFINITION . k] for some k ≥ 1. A function α : VG → K is a vertex colouring of G by a set K of colours. where the edges have a direction. E).1. if there exists = a bijection α : VG → VH such that uv ∈ EG ⇐⇒ α(u)α(v) ∈ E H for all u. that is. and the number of all nonisomorphic graphs on n vertices. n 1 2 3 4 5 6 7 8 236 9 n graphs 1 2 8 64 1024 32 768 2 097 152 268 435 456 156 1044 12 346 nonisomorphic 1 2 4 11 34 > 6 · 1010 274 668 . Indeed.1 Graphs and their plane ﬁgures Later we concentrate on (simple) graphs.

⊔ ⊓ Let s( G ) be the smallest size of a base set X such that G can be represented as an intersection graph of a family of subsets of X. . but if a problem on graphs is to be programmed. 0 1 1 0 1 1 0 0 1 1 1 0 0 1 0 0 1 1 0 0 1 1 0 0 0 A graph has usually many different adjacency matrices. . let A ⊆ N be a ﬁnite set of natural numbers. the graph in Example 1. Theorem 1.1. and edges Xi X j for all i and j (i = j) with Xi ∩ X j = ∅. . The adjacency matrix of G is the n × n-matrix M with entries Mij = 1 or Mij = 0 according to whether vi v j ∈ G or vi v j ∈ G. and deﬁne. since every respectable programming language has array structures for these. Example 1. Let G be a graph. one for each ordering of its set VG of vertices. For this. and let G A = ( A. let X = {X1 .1 Graphs and their plane ﬁgures 6 Other representations Plane ﬁgures catch graphs for our eyes. Theorem 1. . Every graph is an intersection graph of some family of subsets. Graphs can also be represented by sets.1 has an adjacency matrix on the right. Let VG = {v1 . Then Xu ∩ Xv = ∅ if and only if uv ∈ G. . = How small can s( G ) be compared to the order νG (or the size ε G ) of the graph? It was shown by K OU . As an exercise. Proof. to say the least. . Moreover. Two graphs G and H are isomorphic if and only if they have a common adjacency matrix. . As yet another example. a set Xv = {{v.2. . . unsuitable. that is. and computers are good in crunching numbers. we state: All graphs can be represented in the form G A for some set A of natural numbers. E) be the graph with rs ∈ E if and only if r and s (for r = s) have a common divisor > 1. .2. s( G ) = min{| X | | G ∼ GX for some X ⊆ 2X } . . vn } be ordered. S TOCKMEYER AND W ONG (1976) that it is algorithmically difﬁcult to determine the number s( G ) – the problem is NP-complete. two isomorphic graphs have exactly the same set of adjacency matrices. The following result is obvious from the deﬁnitions. . for all v ∈ G. then these ﬁgures are. X2 . / For instance. . Integer matrices are ideal for computers. Notice that the adjacency matrix is always symmetric (with respect to its diagonal consisting of zeros). Xn . u} | vu ∈ G }.1. Xn } be a family of subsets of a set X. and deﬁne the intersection graph GX as the graph with vertices X1 .

and a solution can be found to the problem by combining the information determined by the parts. when dG (v) is deﬁned as the number of edges that have v as an end.2 Subgraphs Ideally. Indeed. Let v ∈ G be a vertex a graph G. then v is said to be isolated in G. E H ) on the same set of vertices that are not isomorphic. ∑ d G ( v) = 2 · ε G . Degrees of vertices D EFINITION . it depends only on the number of the neighbours of the vertices.1. If dG (v) = 0. EG ) and H = (V.1 (Handshaking lemma). then the number of hands shaken is even. tells that if several people shake hands. then v is a leaf of the graph. The neighbourhood of v is the set NG (v) = {u ∈ G | vu ∈ G } . the number of vertices of odd degree is even.1 holds equally well for multigraphs. there are graphs G = (V.2 Subgraphs 7 1. The following lemma. The second claim follows immediately from the ﬁrst one. For instance. v∈ G Moreover. the existence of an Euler tour is very local. Proof. Note that the degrees of a graph G do not determine G. Every edge e ∈ EG has two ends. The minimum degree and the maximum degree of G are deﬁned as δ( G ) = min{dG (v) | v ∈ G } and ∆( G ) = max{dG (v) | v ∈ G } . The degree of v is the number of its neighbours: dG (v) = | NG (v)| . ⊔ ⊓ Lemma 1. and if dG (v) = 1. In these situations we deal with (small) parts of the graph (subgraphs). but for which dG (v) = d H (v) for all v ∈ V. given a nice problem the local properties of a graph determine a solution. and when each loop vv is counted twice. For each graph G. Lemma 1. due to E ULER (1736). as we shall later see. .

a subgraph H ⊆ G is an induced subgraph. we let G − A ⊆ G be the subgraph induced by VG \ A. = = . In this case. To each subset F ⊆ EG of edges there corresponds a unique spanning subgraph of G. if VH ⊆ VG and E H ⊆ EG . A subgraph H ⊆ G spans G (and H is a spanning subgraph of G). let G − F = G [ EG \ F ] be the subgraph of G obtained by removing (only) the edges e ∈ F from G. Similarly. we write G + F. and. if E H = EG ∩ E(VH ). if each e ∈ F (for F ⊆ E(VG )) is added to G. Kelly-Ulam problem or the Reconstruction Conjecture.g. then G ∼ H. Reconstruction Problem. A graph H is a subgraph of a graph G. The famous open problem.1. e. the set E H of edges consists of all e ∈ EG such that e ∈ E(VH ). G subgraph spanning induced For a set F ⊆ EG of edges. i. In an induced subgraph H ⊆ G. states that a graph of order at least 3 is determined up to isomorphism by its vertex deleted subgraphs G −v (v ∈ G): if there exists a bijection α : VG → VH such that G −v ∼ H −α(v) for all v. EG ∩ E( A)) . F ) . H is induced by its set VH of vertices. G − A = G [VG \ A] . that is. To each nonempty subset A ⊆ VG . denoted by H ⊆ G.. VH = VG .. G −e is obtained from G by removing e ∈ G.e. G −v is obtained from G by removing the vertex v together with the edges that have v as their end. if every vertex of G is in H. G [ F ] = (VG . there corresponds a unique induced subgraph G [ A] = ( A. For a subset A ⊆ VG of vertices.2 Subgraphs 8 Subgraphs D EFINITION . Also. In particular.

Let d = ∆( G ) (= d1 ). → Proof. We can now perform a 2-switch / with respect to the vertices v1 . vy ∈ G replaces the / edges uv and xy by ux and vy.2. Since dG (v1 ) = d. Before proving Berge’s switching theorem we need the following tool. but v j vt ∈ G. . since j > i. ⊔ ⊓ D EFINITION . For a graph G. we have a vertex v and graphs G ′ and H ′ such that G − G ′ and → H − H ′ with NG′ (v) = NH ′ (v). that is. if there exists a graph G = (V. we use induction on the order νG .2. then clearly H has the same degrees as G. and thus also G ′ − H ′ .3 (B ERGE (1973)). Two graphs G and H on a common vertex set V satisfy dG (v) = d H (v) for all v ∈ V if and only if H can be obtained from G by a sequence of 2-switches. dn be a descending sequence of nonnegative integers. there exists a v j with / j ≥ d + 2 such that v1 v j ∈ G. we / ′ as required. → In converse. This gives a new vt graph H. When we repeat this process for all indices i with v1 vi ∈ G for 2 ≤ i ≤ d + 1. obtain a graph G ⊔ ⊓ Theorem 1. . and this proves the → claim. Here di ≥ d j . E) with V = {v1 . . → → Finally. 2s 2s Note that if G − H then also H − G since we can apply the sequence of 2→ → switches in reverse order. G ′ −v − H ′ −v. d1 ≥ d2 ≥ · · · ≥ dn . vt .1. where dG (vi ) = di . we observe that H ′ − H by the ‘reverse 2-switches’. v2 . Suppose that there is a vertex vi with 2 ≤ i ≤ d + 1 such that v1 vi ∈ G. vn } such that di = dG (vi ) for all i. Let G and H have the same degrees. . . . xy ∈ G with ux. . 2s 2s 2s 2s 2s 2s 2s . Let d1 . Now the graphs G ′ −v and H ′ −v have the same → degrees. . there exists a vt (2 ≤ t ≤ n) such that vi vt ∈ G. vd1 +1 }. v j .2 Subgraphs 9 2-switches D EFINITION . . . . Then there is a graph G ′ such that G − G ′ with NG′ (v1 ) = {v2 . By Lemma 1. vj v1 vi Since v1 v j ∈ G. Lemma 1. where v1 vi ∈ H and v1 v j ∈ H. Let G be a graph of order n with a degree sequence d1 ≥ d2 ≥ · · · ≥ dn . If G − H. Proof. Denote G− H → u x u x 2s v y v y if there exists a ﬁnite sequence of 2-switches that carries G to H. and the other / neighbours of v1 remain to be its neighbours. d2 . a 2-switch with respect to the edges uv. By the induction hypothesis. . vi . Such a sequence is said to be graphical.

. 1 is graphical 2. Theorem 1. .1). v2 v6 v4 (1. A sequence d1 . . (⇐) Consider G of order n − 1 with vertices (and degrees) dG (v2 ) = d2 − 1. all discrete graphs of order n are isomorphic with each other.3. . Then in the new graph H. Add a new vertex v1 and the edges v1 vi for all i ∈ [2. ⊔ ⊓ Example 1. The graph G = KV is the complete graph on V. s is graphical ⇐⇒ 3. 0. 3. H AKIMI (1962)).. . dd1 +1 − 1. the graph G on the right has this degree sequence. (⇒) Assume dG (vi ) = di .e. . . .2 and Theorem 1. . The complements G = KV of the complete graphs are called discrete graphs. 1. νG = 1. Clearly. and hence also our original sequence s is graphical. 4. 4.1) v1 v3 v5 Special graphs D EFINITION . . otherwise G is nontrivial. 2. . and they will be denoted by Kn . i. . By Theorem 1. 0 is graphical.1. If this degree is equal to r. By Lemma 1. if every vertex of G has the same degree. . d2 . vd1 +1 }. dd1 +3 . .2 Subgraphs 10 Using the next result recursively one can decide whether a sequence of integers is graphical or not. . if it has only one vertex. 1. d G ( v d1 + 2 ) = d d1 + 2 . 1. dn is graphical (when put into nonincreasing order). But now the degree sequence of G −v1 is in (1. The last sequence corresponds to a graph with no edges. d G ( v n ) = d n as in (1. 0 is graphical 0. then G is r-regular or regular of degree r. 3. . dG (vd1 +1 ) = dd1 +1 − 1. A graph G is said to be regular. d H (v1 ) = d1 . Proof. if every two vertices are adjacent: E = E(V ). dn (with d1 ≥ 1 and n ≥ 2) is graphical if and only if d2 − 1. Consider the sequence s = 4. .3. .4. Indeed. In a discrete graph EG = ∅. . All complete graphs of order n are isomorphic with each other. / The complement of G is the graph G on VG . . . and d H (vi ) = di for all i. 1. d3 − 1. dd1 +1 ]. where EG = {e ∈ E(V ) | e ∈ EG }. . E) is trivial. 2.4 (H AVEL (1955). dd1 +2 . we can suppose that NG (v1 ) = {v2 .1). A graph G = (V. . . .

1.3 Paths and cycles

11

A discrete graph is 0-regular, and a complete graph Kn is (n − 1)-regular. In particular, ε Kn = n(n − 1)/2, and therefore ε G ≤ n(n − 1)/2 for all graphs G that have order n. Many problems concerning (induced) subgraphs are algorithmically difﬁcult. For instance, to ﬁnd a maximal complete subgraph (a subgraph Km of maximum order) of a graph is unlikely to be even in NP. Example 1.4. The graph on the right is the Petersen graph that we will meet several times (drawn differently). It is a 3-regular graph of order 10. Example 1.5. Let k ≥ 1 be an integer, and consider the set B k of all binary strings of length k. For instance, B3 = {000, 001, 010, 100, 011, 101, 110, 111}. Let Qk be the graph, called the k-cube, with VQk = B k , where uv ∈ Qk if and only if the strings u and v differ in exactly one place. 110

111

The order of Qk is νQk = the number of binary strings of length k. Also, Qk is k-regular, and so, by the handshaking lemma, ε Qk = k · 2k−1 . On the right we have the 3-cube, or simply the cube.

2k ,

100

101 010 011 001

000

Example 1.6. Let n ≥ 4 be any even number. We show by induction that there exists a 3-regular graph G with νG = n. Notice that all 3-regular graphs have even order by the handshaking lemma. If n = 4, then K4 is 3-regular. Let G be a 3-regular graph of order 2m − 2, and suppose that uv, uw ∈ EG . Let VH = VG ∪ { x, y}, and E H = ( EG \ {uv, uw}) ∪ {ux, xv, uy, yw, xy}. Then H is 3-regular of order 2m.

x y

w u

v

**1.3 Paths and cycles
**

The most fundamental notions in graph theory are practically oriented. Indeed, many graph theoretical questions ask for optimal solutions to problems such as: ﬁnd a shortest path (in a complex network) from a given point to another. This kind of problems can be difﬁcult, or at least nontrivial, because there are usually choices what branch to choose when leaving an intermediate point.

Walks

D EFINITION . Let ei = ui ui+1 ∈ G be edges of G for i ∈ [1, k]. The sequence W = e1 e2 . . . ek is a walk of length k from u1 to uk+1 . Here ei and ei+1 are compatible in the sense that ei is adjacent to ei+1 for all i ∈ [1, k − 1].

**1.3 Paths and cycles We write, more informally, W : u1 − u2 − . . . − u k − u k+1 → → → → or W : u1 − u k+1 . →
**

k

12

⋆ Write u − v to say that there is a walk of some length from u to v. Here we under→ ⋆ stand that W : u − v is always a speciﬁc walk, W = e1 e2 . . . ek , although we sometimes → do not care to mention the edges ei on it. The length of a walk W is denoted by |W |.

D EFINITION . Let W = e1 e2 . . . ek (ei = ui ui+1 ) be a walk. W is closed, if u1 = uk+1 . W is a path, if ui = u j for all i = j. W is a cycle, if it is closed, and ui = u j for i = j except that u1 = uk+1 . W is a trivial path, if its length is 0. A trivial path has no edges. For a walk W : u = u1 − . . . − uk+1 = v, also → → W −1 : v = u k+1 − . . . − u1 = u → → is a walk in G, called the inverse walk of W. A vertex u is an end of a path P, if P starts or ends in u. ⋆ ⋆ ⋆ The join of two walks W1 : u − v and W2 : v − w is the walk W1 W2 : u − w. → → → (Here the end v must be common to the walks.) Paths P and Q are disjoint, if they have no vertices in common, and they are independent, if they can share only their ends. Clearly, the inverse walk P−1 of a path P is a path (the inverse path of P). The join of two paths need not be a path. A (sub)graph, which is a path (cycle) of length k − 1 (k, resp.) having k vertices is denoted by Pk (Ck , resp.). If k is even (odd), we say that the path or cycle is even (odd). Clearly, all paths of length k are isomorphic. The same holds for cycles of ﬁxed length.

P5

C6

⋆ ⋆ Lemma 1.3. Each walk W : u − v with u = v contains a path P : u − v, that is, there is a → → ⋆ path P : u − v that is obtained from W by removing edges and vertices. →

Proof. Let W : u = u1 − . . . − uk+1 = v. Let i < j be indices such that ui = u j . → → ⋆ If no such i and j exist, then W, itself, is a path. Otherwise, in W = W1 W2 W3 : u − → ⋆ ⋆ ⋆ ⋆ ui − u j − v the portion U1 = W1 W3 : u − ui = u j − v is a shorter walk. By → → → → ⋆ repeating this argument, we obtain a sequence U1 , U2 , . . . , Um of walks u − v with → |W | > |U1 | > · · · > |Um |. When the procedure stops, we have a path as required. (Notice that in the above it may very well be that W1 or W3 is a trivial walk.) ⊔ ⊓

**1.3 Paths and cycles D EFINITION . If there exists a walk (and hence a path) from u to v in G, let dG (u, v) = min{k | u − v} →
**

k

13

⋆ be the distance between u and v. If there are no walks u − v, let dG (u, v) = ∞ by → convention. A graph G is connected, if dG (u, v) < ∞ for all u, v ∈ G; otherwise, it is disconnected. The maximal connected subgraphs of G are its connected components. Denote

c( G ) = the number of connected components of G . If c( G ) = 1, then G is, of course, connected. The maximality condition means that a subgraph H ⊆ G is a connected component if and only if H is connected and there are no edges leaving H, i.e., for every vertex v ∈ H, the subgraph G [VH ∪ {v}] is disconnected. Apparently, every connected / component is an induced subgraph, and

∗ NG (v) = {u | dG (v, u) < ∞}

is the connected component of G that contains v ∈ G. In particular, the connected components form a partition of G.

Shortest paths

D EFINITION . Let G α be an edge weighted graph, that is, G α is a graph G together with a weight function α : EG → R on its edges. For H ⊆ G, let α( H ) =

∑ α( e)

e∈ H

be the (total) weight of H. In particular, if P = e1 e2 . . . ek is a path, then its weight is α( P) = ∑k=1 α(ei ). The minimum weighted distance between two vertices is i

⋆ dα (u, v) = min{α( P) | P : u − v} . → G

In extremal problems we seek for optimal subgraphs H ⊆ G satisfying speciﬁc conditions. In practice we encounter situations where G might represent • a distribution or transportation network (say, for mail), where the weights on edges are distances, travel expenses, or rates of ﬂow in the network; • a system of channels in (tele)communication or computer architecture, where the weights present the rate of unreliability or frequency of action of the connections; • a model of chemical bonds, where the weights measure molecular attraction.

1. others are ∞. The shortest path problem: Given a connected graph G with a weight function α : EG → N. t(v5 ) = min{4. 3 2 Thus choose u2 = v3 . . ui } be any vertex with the least value t(ui+1 ). or all vertices (if we want to travel around). For this. / replace t(v) by min{t(v). ui }. 3 + 1} = 4. t(v4 ) = 2 + 3 = 5. / Dijkstra’s algorithm: (i) Set u0 = u. t(u1 ) + α(u1 v2 )} = min{3. Apply Dijkstra’s algorithm to the vertex v0 . t(v2 ) = 3. t(v4 ) = 4. t(v5 ) = min{4. 1 v1 v3 others are ∞. if the graph represents a network of pipelines. and then one wants to ﬁnd a subgraph with the maximum weight. Let us make the convention that α(uv) = ∞. 3 + 1} = 4. . The algorithm stops. t(v3 ) = 3. 3} = 3. G Example 1. • u0 = v0 . let G be a graph with an integer weight function α : EG → N. v) = t(v). 1 v2 v4 • t(v2 ) = min{3. the weights are volumes or capacities. ∞} = 3. Thus u1 = v1 . 4} = 3. where u is a ﬁxed starting point and v ∈ G. and which connects two given vertices. 3 + 2} = 5. . v5 . 2 • t(v1 ) = min{∞.7. . 1 2 • t(v2 ) = min{3. • t(v4 ) = min{5. • t(v5 ) = min{4. . Dijkstra’s algorithm solves the problem for every pair u. . On the other hand. These are the minimal weights from v0 to each vi . t(v5 ) = 2 + 2 = 4. t(u0 ) = 0 and t(v) = ∞ for all v = u0 .3 Paths and cycles 14 In these examples we look for a subgraph with the smallest weight. Let ui+1 ∈ {u1 . 3 v0 2 2 t(v3 ) = 2 + 1 = 3. / (iii) Conclusion: dα (u. if uv ∈ G. t(ui ) + α(ui v)} . . We consider the minimum problem. t(v2 ) = min{∞. Thus set u3 = v2 . Consider the following weighted graph G. ∞} = 4. 4 + 1} = 4. v. In this case. . v ∈ G. t(v4 ) = min{5. ﬁnd dα (u. 2} = 2. t(u0 ) = 0. call α(uv) the length of uv. Thus choose u4 = v4 . v) for given u. (ii) For i ∈ [0. We have obtained: t(v1 ) = 2. t(v5 ) = 4 . G Assume that G is a connected graph. νG − 1]: for each v ∈ {u1 .

and let P : u0 − u − v be a shortest path from u0 to v. clearly.v2 3 3 3 v3 ∞ 3 v4 ∞ 5 5 v5 ∞ 4 4 4 4 4 15 The correctness of Dijkstra’s algorithm can veriﬁed be as follows. Then. → → where u is any vertex u = v on such a path. u0 − u. Dijkstra’s algorithm makes use of this observation → → iteratively. .3 Paths and cycles The steps of the algorithm can also be rewritten as a table: v1 2 . the length of the path P equals the sum of ⋆ ⋆ the weights of u0 − u and u − v. possibly u = u0 . is a shortest path from u0 to u. and the latter part u − v → → is a shortest path from u to v.1. the ﬁrst ⋆ ⋆ part of the path. ⋆ ⋆ Let v ∈ V be any vertex. Therefore.

For concrete applications in this respect. L EISERSON AND R. ( X.E. if G [ X ] is a discrete graph.1. Example 2.2 Connectivity of Graphs 2. k)bipartite. Bipartite graphs D EFINITION . if VG has a partition to two subsets X and Y such that each edge uv ∈ G connects a vertex of X and a vertex of Y. k)-bipartite graphs are isomorphic. The following result is clear from the deﬁnitions. A subset X ⊆ VG is stable. The solutions in these cases are usually subgraphs without cycles. C.k denote such a graph. A graph G is bipartite if and only if VG has a partition to two stable subsets.5 is bipartite for all k. In this case. Theorem 2.3 . Y ) is a bipartition of G. factorization trees or grammatical trees. Indeed.1 Bipartite graphs and trees In problems such as the shortest path problem we look for minimum solutions that satisfy the given requirements. Y )-bipartite. On the right there is a tree of operations for the arithy z metic formula x · (y + z) + y. Such connected graphs will be called trees. 1993. in search algorithms for databases. Grammatical trees occur especially in linguistics.1. C ORMEN . Let Km. These trees are sometimes called construction y · trees. The k-cube Qk of Example 1. A bipartite graph G (as in the above) is complete (m. and uv ∈ G for all u ∈ X and v ∈ Y. these sets partition B k . and they are used. MIT Press. + Certain structures with operations are representable as trees. consider A = {u | u has an even number of 1′ s} and B = {u | u has an odd number of 1′ s}. decomposition trees. A graph G is called bipartite. and G is ( X. if | X | = m. Clearly. R IVEST. K2.H..L. and they are stable in Qk . x + where syntactic structures of sentences are analyzed. see T.g. e. “Introduction to Algorithms”. All complete (m. |Y | = k.

Since G is connected. Assume that x is the last common ⋆ ⋆ vertex of P and Q: P = P1 P2 .2 (E RDÖS (1965)). We show that each graph G has a bipartite subgraph H ⊆ G such that ε H ≥ 1 ε G . Example 2. then the pair X ′ = X \ {v}.) Now 1 1 1 1 ε H = ∑ d H ( v) ≥ ∑ d G ( v) = ε G . Let v ∈ G be a chosen vertex. and deﬁne X = { x | dG (v. u Thus | P2 | and | Q2 | have the same parity and hence the P2 P1 −1 sum | P2 | + | Q2 | is even. where X consists of those vertices with colour 1. . (⇒) Observe that if G is ( X. Yi )bipartite. → Consequently. v x uw and so uw ∈ EG by assumption. at some point one of the vertices gets both colours. by the deﬁnition of distance. P1 and Q1 are shortest paths v − x. since. 17 Proof. then so are all its subgraphs. i.1 Bipartite graphs and trees Theorem 2. Also.2. say 1 and 2. Y )-bipartite. Y1 ∪ Y2 ∪ · · · ∪ Yp ). . and it is bipartite. If the whole graph can be coloured without contradiction. d H (v) ≥ dG (v)/2. Denote F = EG ∩ {uv | u ∈ X. . if G is disconnected. otherwise. if v ∈ X.. G p of G. let VG = X ∪ Y be a partition such that the 2 number of edges between X and Y is maximum. Indeed.e. and Y of those vertices with colour 2. | P1 | = | Q1 |. Therefore X and Y are / Q1 stable subsets. 2 1 this can be done by using two ‘opposite’ colours. w ∈ G be both in X or both in Y. then each cycle of G is contained in one of the connected components G1 . Q = Q1 Q2 . Since P and Q are shortest paths. we note that it sufﬁces to show the claim for connected graphs. . By the maximum condition. Start from any vertex v1 . Indeed. and G is bipartite as claimed. (⇐) Suppose that all cycles in G are even. and colour it by 1. Assume thus that G is connected. v is on the wrong side. otherwise. Obviously H is a spanning subgraph. x) is even} and Y = {y | dG (v. Y )-bipartite. and let P : v − u and Q : v − w → → be (among the) shortest paths from v to u and w. If Gi is ( Xi . First. 2 v∈ H 2 v∈ G 2 2 . and in this case. then G is ( X. then G has the bipartition ( X1 ∪ X2 ∪ · · · ∪ X p . Indeed. G is not bipartite. the path P2 Q2 is even. y) is odd} . A graph G is bipartite if and only if G it has no odd cycles (as subgraph). v ∈ Y } . Y ′ = Y ∪ {v} does better that the pair X. Y. and let H = G [ F ]. However. an odd cycle C2k+1 is not bipartite. where P2 : x − u and Q2 : x − w are → → ⋆ independent. (That is. ⊔ ⊓ Q2 w 1 Checking whether a graph is bipartite is easy. 2 2 2 Then colour the neighbours of v1 by 2. X ∩ Y = ∅. ⋆ ⋆ Let then u. and proceed by 1 1 colouring all neighbours of an already coloured vertex 2 1 by the opposite colour. VG = X ∪ Y.2.

The following enumeration result for trees has many different proofs.1. ⋆ Proof. and there is a path P : v − u in G −e. there exists a path P : w − v in G. For (ii). if c( G −e) > c( G ). Let w ∈ G. (i) Then c( G −e) = 2.3. Example 2. then f is a bridge of G.2. Let e be a bridge in a connected graph G. We omit the proof. A graph is called acyclic. say C = PeQ. and therefore C is inside H. An edge e ∈ G is a bridge of the graph G. An edge e ∈ G is a bridge if and only if e is not in any cycle of G. ⊔ ⊓ Lemma 2. if G −e has more connected components than G. v in different connected components of G −e . and f is not a bridge of H. This is a → ⋆ ⋆ path of G −e. unless P : w − u → v contains e = uv. An acyclic graph is also called a forest. We note that. e = uv is a bridge ⇐⇒ u. A connected graph is a tree if and only if all its edges are bridges. (ii) Let H be a connected component of G −e.1. if it has no cycles. ⊔ ⊓ Trees D EFINITION . (⇒) If there is a cycle in G containing e. let e = uv. then QP : v − u is a → path in G −e. in which case the part w − u is → → a path in G −e. and so e is not a bridge.3 and the deﬁnition of a tree. On the right (only) the two horizontal lines are bridges. an edge e in a connected G is a bridge if and only if G −e is disconnected. By Theorem 2. Since e is a bridge. then u and v are in the same connected component ⋆ ⋆ of G −e. Theorem 2. A tree is a connected acyclic graph. eP : u − v − u is a cycle in G → → → containing e. that is. In particular. we have Corollary 2.3. . Proof. then C does not contain e (since e is in no cycle). (⇐) If e = uv is not a bridge. for each edge e ∈ G.1 Bipartite graphs and trees 18 Bridges D EFINITION . and most importantly. If f ∈ H is a bridge of H. if f ∈ H belongs to a cycle C of G. For (i). Now. Since G is connected. the ends u and v are not connected in ⋆ G −e. the ﬁrst of which was given by C AYLEY in 1889: There are nn−2 trees on a vertex set V of n elements.

T is connected.4. For acyclic graphs.1 Bipartite graphs and trees On the other hand. In particular.2. Theorem 2. if wv ∈ G. Proof. (iii) T is acyclic and ε T = νT − 1. ⊔ ⊓ (ii)⇒(iii) We prove the claim by induction on n. Let νT = n. Each tree T with νT ≥ 2 has at least two leaves. if → G is acyclic. Since T is acyclic. If n = 1. Suppose that | P| ≥ | Q|.2. ⊔ ⊓ Corollary 2. Assume the claim does not hold.2. then also Pe is a path. then dG (v) = 1. Clearly. ⋆ (i)⇒(ii) Let T be a tree. Hence e must also belong to Q. ⋆ Lemma 2. which contradicts the / maximality assumption for P. Moreover. If e = vw ∈ EG with w ∈ P. Let T be any graph of order n satisfying (ii). Proof. Q : u − v → be two different paths between the same vertices u and v. then w belongs to P. (ii) Any two vertices are connected in T by a unique path. Since P = Q. and therefore u and v belong to different connected components of T −e. . and let P. there exists an edge e which belongs to P but not to Q. the claim holds for n = 2. and suppose it holds for graphs of order less than n. there are only a few trees up to isomorphism: n trees 1 1 2 1 3 1 4 2 5 3 6 6 7 8 11 23 19 n 9 10 11 12 13 14 15 16 trees 47 106 235 551 1301 3159 7741 19 320 The nonisomorphic trees of order 6 are: ⋆ We say that a path P : u − v is maximal in a graph G. and wv is necessarily the last edge of P in order to avoid cycles. Each edge of T is a bridge. and it is clearly acyclic. (i) T is a tree. both ends of a maximal path have degree one. Proof. because νG is ﬁnite. if there are no edges e ∈ G → for which Pe or eP is a path. Suppose thus that n ≥ 2. then the claim is trivial. Then NG (v) ⊆ P. Let P : u − v be a maximal path in a graph G. a contradiction. The following are equivalent for a graph T. Hence NG (v) ⊆ P. Such paths exist.

So 13 plays are needed in this example.2. Then ε G ≥ ε T = νT − 1 = νG − 1. subgraph that is a tree). By Lemma 2. Consider a cup tournament of n teams. If VT = VG . Moreover. and N tries to block all paths s − r (that is. k]. then these are divided into ⌊k⌋ pairs. to designate → at least one edge in each such path). By induction hypothesis. and it satisﬁes the condition (ii). where a play corresponds to an edge of the tree. Let T ⊆ G be a maximum order subtree of G (i. Ei ). Indeed. then 2. and so ε T = ε T −v + 1 = n − 1. It is P’s purpose to designate a path s − r (that is. How many plays are needed to determine the winner? So if there are 14 teams. that is. The answer to our problem is n − 1. ⊔ ⊓ Example 2. (iii)⇒(i) Assume (iii) holds for T. after the ﬁrst round 7 teams continue. that is.5.3. let the connected components of T be Ti = (Vi . If during a round there are k teams left in the tournament. and from each pair only the winner continues.. since the cup tournament is a tree. i i n − 1 = εT = ∑ (|Vi | − 1) = ∑ |Vi | − k = n − k .e. T is connected. Spanning trees Theorem 2. If k is odd. P and N alternate turns so that P designates an edge by +. Each connected graph has a spanning tree. for i ∈ [1.2. Let T be a spanning tree of G. Since T is acyclic. Now. and hence they are trees. and ε T = ∑k=1 |Ei |. For each connected graph G. then one of the teams goes to the next round without having to play. P : u − w − v. for which we have proved that |Ei | = |Vi | − 1. But then T is not / / maximal. to designate → ⋆ all edges in one such path). a connected graph G is a tree if and only if ε G = νG − 1.4.1 Bipartite graphs and trees 20 ⋆ Let P : u − v be a maximal path in T. a spanning graph that is a tree. In Shannon’s switching game a positive player P and a negative player N play on a graph G with two special vertices: a source s and a sink r.5. and after the second round 4 teams continue. We say that a game ( G. We need to show that T is connected. so are the connected graphs Ti . i=1 i=1 k k which gives that k = 1. In this → ⋆ case. Each edge can be des⋆ ignated at most once. Therefore. ⊔ ⊓ Corollary 2. The subgraph → → T −v is connected. The second claim is also clear. there exists an edge uv ∈ EG such that u ∈ T and v ∈ T. Proof. ε T −v = n − 2. Proof. and N by −. ⊔ ⊓ Example 2. we have dT (v) = 1. s. a contradiction. νT = ∑k=1 |Vi |. where vw is the unique edge having an end v. and the claim follows. ε G ≥ νG − 1. r) is .

s 21 L EHMAN proved in 1964 that Shannon’s switching game ( G. choose an edge ei ∈ Ei−1 of smallest / possible weight such that ei does not produce a cycle when added to G [ Ei−1 ]. then P marks by + an edge in the other tree such that this edge reconnects the broken tree. . s. s. • negative. Then P plays as follows. if P has a winning strategy no matter who begins the game.2. In the other direction the claim can be proved along the following lines.1 Bipartite graphs and trees • positive. In graph theoretical terms we wish to ﬁnd an optimal spanning subgraph of a weighted graph. respectively). . P always has two spanning trees for the subgraph H with only edges marked by + in common. . In this way. The ﬁnal outcome is T = (VG . Assume that there exists a subgraph H containing s and r and that has two spanning trees with no edges in common. This is the connector problem. En−1 ). • neutral. otherwise a deletion of any nonbridge will reduce the total weight of the subgraph. . (ii) For each i = 2. If N marks by − an edge from one of the two trees. There remains the problem to characterize those Shannon’s switching games ( G. computers. In converse the claim is considerably more difﬁcult to prove. r) is positive if and only if there exists H ⊆ G such that H contains s and r and H has two spanning trees with no edges in common. and let Ei = Ei−1 ∪ {ei }. if the winner depends on who begins the game. and set E1 = {e1 }. Kruskal’s algorithm: For a connected and weighted graph G α of order n: (i) Let e1 be an edge of smallest weight. n − 1 in this order. Kruskal’s algorithm (also known as the greedy algorithm) provides a solution to the connector problem. 3. The connector problem To build a network connecting n nodes (towns. if N has a winning strategy no matter who begins the game. . Such an optimal subgraph is clearly a spanning tree. r The game on the right is neutral. chips in a computer) it is desirable to decrease the cost of construction of the links to the minimum. r) that are neutral (negative. Let then G α be a graph G together with a weight function α : EG → R + (positive reals) on the edges. for.

If we add ek to T1 . their addresses should differ by one symbol. ( n)] is a weight 2 of one path. 01100) = 3 and h(10000. where a(v) and a(u) differ. e3 = v5 v6 . and the edges indicate the direct links. . and each i ∈ [1.6. It would be a good way to address the vertices so that the Hamming distance of two vertices is the same as their distance in G.2. We deduce that α( T ) ≤ α( T1 ). h(00010. C must contain an edge e that is not in T. The total weight of the spanning tree is thus 9. a(u)) of two addresses of the same length is the number of places. . En−1 ) is a spanning tree of G. Also. Let ek be the ﬁrst edge produced by the algorithm that is not in T1 . α(ek ) ≤ α(e). say T2 . Example 2. e3 = v3 v6 . one by one. we still have a spanning tree. Each computer v has an address a(v). e4 = v2 v3 and e5 = v1 v2 . we can transform T1 to T by replacing edges. Suppose T1 is any spanning tree of G. e2 = v4 v5 . because it contains no cycles. e2 = v4 v5 . such that the total weight does not increase. 3 2 1 4 1 2 v4 1 v5 3 2 v6 2 2 v1 v2 v3 Problem. . then necessarily n = k2 or n = k2 + 2 for some k ≥ 1. When applied to the weighted graph on the right. . there may exist several optimal spanning trees (with the same weight.1 Bipartite graphs and trees 22 By the construction. if two vertices were adjacent. (Why is this so?) Does there exist a weighted tree T α of order n such 2 that the (total) weights of its paths are 1. We now show that T has the minimum total weight among the spanning trees of G. Note that T2 has more edges in common with T than T1 . . 1 5 2 8 4 TAYLOR (1977) proved: if T of order n exists.g. and therefore α( T2 ) ≤ α( T1 ). The Hamming distance h( a(v). 2. Repeating the above procedure.7. The length of an address is the number of its bits. then a cycle C containing ek is created.. A computer network can be presented as a graph G. In particular. This would make it easier for a node computer to forward a message. No solutions are known for any n ≥ 7. 00000) = 1. e4 = v3 v6 . by the construction. it is connected and has n − 1 edges. When we replace e by ek in T1 . (n)? 2 In such a weighted tree T α different paths have different weights. Example 2. However. Also. Indeed. e. the algorithm produces the sequence: e1 = v2 v4 . a bit string (of zeros and ones). e5 = v1 v2 gives another optimal solution (of weight 9). the selection e1 = v2 v5 . Each tree T of order n has exactly (n) paths. where the vertices are the node computers. T = (VG . A message that is sent to v is preceded by the address a(v). Also. Consider trees T with weight functions α : ET → N. α must be injective. of course) for a graph. The outcome of Kruskal’s algorithm need not be unique.

a(v4 ) = ∗ ∗ 11 . if it has an addressing a such that dG (u.2 Connectivity A graph G is said to be addressable. then the claim is obvious. “A Course in Combinatorics”. vk+1 }.2. The Hamming distance remains as it was. a(v3 ) = 1 ∗ 01 . and change it to 0ai . in a reliable network we require that a break-down of a node (computer) should not result in the inactivity of the whole network. . We still want to have h(u. v) = h( a(u). v1 v2 v4 v3 These addresses have length 4. Can you design a star addressing with addresses of length 3? W INKLER proved in 1983 a rather unexpected result: The minimum star address length of a graph G is at most νG − 1. The proof goes by induction. Similarly. The triangle K3 is not addressable. that is. a(v)). . So. the addresses of the vertices are simply 0 and 1. and assume that dT (v1 ) = 1 (a leaf) and v1 v2 ∈ T. A star address will be a sequence of these three symbols. . 010 000 100 110 111 23 We prove that every tree T is addressable. v) is the number of places. For the proof of this. These problems occur especially in different aspects of fault tolerance and reliability of networks. 2. We change this addressing: let ai be the address of vi in T −v1 . v). Moreover. where cost is the criterion. we can address the tree T −v1 by addresses of length k − 1. but where two vertices are always connected by at least two independent paths. In order to gain more generality. We star address this graph as follows: a(v1 ) = 0000 . We may also wish to construct graphs that are as simple as possible. we modify the addressing for general graphs by introducing a special symbol ∗ in addition to 0 and 1. Let then VT = {v1 . the addresses of the vertices of T can be chosen to be of length νT − 1. In the case νT = 2. where one has to make sure that a break-down of one connection does not affect the functionality of the network. ∗00 ∗ ∗∗) = 0. By the induction hypothesis. h(u. a(v2 ) = 10 ∗ 0 . It is now easy to see that we have obtained an addressing for T as required. The special symbol ∗ does not affect h(u. . . v) = dG (u. If νT ≤ 2. v).2 Connectivity Spanning trees are often optimal solutions to problems. h(10 ∗ ∗01. Set the address of v1 to 1a2 . see VAN L INT AND W ILSON. 0 ∗ ∗101) = 1 and h(1 ∗ ∗ ∗ ∗∗. where u and v have a different symbol 0 or 1.

if G is disconnected. → Lemma 2. A vertex v ∈ G is a cut vertex. If e = uv is a bridge. A graph G is k-connected. If νG ≤ 2. D EFINITION . Again. κ ( G ) = 2 and κ ′ ( G ) = 2. On the right.8. v)separating set. if u and v belong to different connected components of G −S. Notice that the minimum degree is δ( G ) = 3. Let G be connected. . assume that G is not complete. that is. F ⊆ EG } . We also say that S separates the vertices u and v and it is a (u. if G is a complete graph. if G is disconnected. For trivial graphs. An edge cut F of G consists of edges so that G − F is disconnected. v} is a separating set.2. ⋆ Lemma 2. If S ⊆ VG separates u and v. The (vertex) connectivity number κ ( G ) of G is deﬁned as κ ( G ) = min{k | k = |S|. Lemma 2. {v} is a separating set. • κ ( G ) = νG − 1. ⊔ / ⊓ D EFINITION . then v is a cut vertex if and only if G −v is disconnected. For νG ≥ 3. Now VG \ {u.4. then κ ′ ( G ) = 0. A minimal edge cut F ⊆ EG is a bond (F \ {e} is not an edge cut for any e ∈ F). if G −S is disconnected. and let uv ∈ G. if c( G −v) > c( G ). The claim follows from this. if κ ′ ( G ) ≥ k. if κ ( G ) ≥ k. then it is 1-connected. Proof. then either G = K2 or one of u or v is a cut vertex. and • otherwise κ ( G ) equals the minimum size of a separating set of G. Example 2. let κ ′ ( G ) = 0. If a connected graph G has no separating sets. In other words. A graph G is k-edge connected. Clearly. • κ ( G ) = 0. Let κ ′ ( G ) = min{k | k = | F |.5. The following lemma is immediate. if G is connected.2 Connectivity 24 Separating sets D EFINITION . G −S disconnected or trivial. then the claim is clear. then every path P : u − v visits a vertex of S. S ⊆ VG } . then it is a complete graph. G − F disconnected.3. A subset S ⊆ VG is a separating set. If G is connected.

Assume G is nontrivial. In both of these cases. since H −e = G − F. → Hence |S| is at least the number of independent paths from u to v. and therefore G −S ⊆ H is an induced subgraph of H). and in this case also κ ( G ) = 0. Let F be an edge cut of G with | F | = κ ′ ( G ). κ ( G ) ≤ κ ′ ( G ) ≤ δ( G ) . either νGu ≥ 2 (and u is a cut vertex) or νGv ≥ 2 (and v is a cut vertex). Clearly. the subgraph H = G −( F \ {e}) is connected for given e ∈ F. and F is minimal. Proof. . If κ ′ ( G ) = 1. v)-separating. κ ( G ) ≤ |S| + 1 ≤ κ ′ ( G ). If F be a bond of a connected graph G. ⋆ Proof. . For any graph G. then S is a separating set and so κ ( G ) ≤ |S| ≤ κ ′ ( G ) − 1 and we are done. Consider the connected subgraph H = G −( F \ {e}) = ( G − F ) + e. .5. and the claim follows.2. . whose removal results in a trivial or a disconnected graph. On the other hand.2 Connectivity 25 Proof. F Now for each f ∈ F \ {e} choose an end different from u and v. we disconnect G. Since G − F is disconnected. H . By Lemma 2. since G is connected. .) Note that since f = e. By Lemma 2. Let u. Then F is a bond. . since if we remove all edges with an end v. either G = K2 or G has a cut vertex. e . If a subset S ⊆ VG is (u. either end of f is different from u or v. Assume then that κ ′ ( G ) ≥ 2. (The choices for different edges need not be different. then G is disconnected. v ∈ G be nonadjacent vertices of a connected graph G. If G −S is disconnected. c( H −e) = 2. there is a vertex of G −S. . then c( G − F ) = 2. then every path u − v of G visits S. Menger’s theorem Theorem 2. ⊔ ⊓ G . and G − F has two connected components. . then G is connected and contains a bridge. and thus c( G − F ) = 2. . In both of these cases. Proof. if G −S is connected. If κ ′ ( G ) = 0.7 (M ENGER (1927)).6. Assume that G = K2 and thus that νG ≥ 3. ⊔ ⊓ Theorem 2. Thus |S| ≤ | F | − 1 = κ ′ ( G ) − 1. ⊔ ⊓ Lemma 2. κ ′ ( G ) ≤ δ( G ). In conclusion. then either G −S = K2 (= e). where e is a bridge. Now. Then the minimum number of vertices separating u and v is equal to the maximum number of independent paths from u to v. Let Gu = ∗ ∗ NG−e (u) and Gv = NG−e (v) be the connected components of G −e containing u and v.6 (W HITNEY (1932)). Let S be the collection of these choices. and G −S does not contain edges from F \ {e}. also κ ( G ) = 1.1. and let e = uv ∈ F. Hence e is a bridge in H. or either u or v (or both) is a cut vertex of G −S (since H −S = G −S. .

k]. and deﬁne Gu to be the graph on Hu ∪ S ∪ {v } having the edges of G [ Hu ∪ S] together with vwi for all i ∈ [1. let S′ be a minimum set that separates u and v. This means that S′ separates u and v in G. and this takes care of the small values of m. This shows that ε Gu ≤ ε G . we obtain k independent → ⋆ ⋆ paths u − wi − v in G. → The case for k = 1 is clear. → ⋆ When we combine these with the above paths u − wi . by the induction hypothesis. all wi ∈ S have to be adjacent to some vertex in Hv . | P| ≥ 3. . By the present assumption.2 Connectivity 26 Conversely. Indeed. A symmetric argument applies to the graph Gv (with a new vertex u). which is deﬁned similarly ⋆ to Gu . the assumption (2. and denote by Hu = NG−S (u) and ∗ Hv = NG−S (v) the connected components of G −S for u and v. . and so y = v. → → (2. But these are paths of G. v)-separating set of the smallest size. we use induction on m = νG + ε G to show that if S = {w1 . there are k independent → → → ⋆ paths u − v in G as required.2) There remains the case. k]. f = xy. .2. v)-separating set of Gu . v)separating set. Together with the path u − w − v. by the assumption (2). → → Notice that. that have only the end u in → common. (2. This is possible → ⋆ only if there exist k paths u − wi . We noted that Gu is smaller than G. required for the induction. and.1) Suppose next that S NG (u) and S NG (v). → ∗ (2) Assume then that NG (u) ∩ NG (v) = ∅. then G has at least (and thus exactly) k ⋆ independent paths u − v. wk . . one for each i ∈ [1. The graph Gu is connected and it is smaller than G.. . In the smaller graph G −w the set S \ {w} is a minimum (u. and Q : y − v. and thus by the ⋆ induction hypothesis. In the smaller graph G − f . v)-separating sets S of k elements. Since k is the size of a minimum (u. So | Hv | ≥ 2 and νGu < νG . Then necessarily w ∈ S. w2 u w1 v If S′ is any (u. there are k independent paths u − v in Gu . S ∩ NG (v) = ∅ or S ∩ NG (u) = ∅.) ⋆ ⋆ Let P = e f Q be a shortest path u − v in G.. by (2). and the claim is clear in this case. wk } is a (u. where e = ux. then. This yields that there are k paths wi − v that have only the end v in common. ⋆ If |S′ | ≥ k. we have |S′ | ≥ k. there are k independent paths u − v → in G − f . also u is nonadjacent to some vertex of S. moreover. where for all (u. Let v be a new vertex. in order for S to be a minimum separating set. v)separating set. either S ⊆ NG (u) or S ⊆ NG (v). and the induction hypothesis yields that there are k − 1 independent ⋆ paths u − v in G −w. w2 . (Note that then. then S′ will separate u from all wi ∈ S \ S′ in G. (1) Assume ﬁrst that u and v have a common neighbour w ∈ NG (u) ∩ NG (v).1) rules out the case Hv = {v}.

2.2 Connectivity

27

If, on the other hand, |S′ | < k, then u and v are still connected in G −S′ . Every path u − v in G −S′ necessarily travels along the edge f = xy, and so x, y ∈ S′ . → / Let Sx = S′ ∪ { x} and Sy = S ′ ∪ {y} .

⋆

These sets separate u and v in G (by the above fact), and they have size k. By our current assumption, the vertices of Sy are adjacent to v, since the path P is shortest and so uy ∈ G (meaning that u is not adjacent to all of Sy ). The assumption (2) yields / that u is adjacent to all of Sx , since ux ∈ G. But now both u and v are adjacent to the vertices of S′ , which contradicts the assumption (2). ⊔ ⊓ Theorem 2.8 (M ENGER (1927)). A graph G is k-connected if and only if every two vertices are connected by at least k independent paths. Proof. If any two vertices are connected by k independent paths, then it is clear that κ ( G ) ≥ k. In converse, suppose that κ ( G ) = k, but that G has vertices u and v connected by at most k − 1 independent paths. By Theorem 2.7, it must be that e = uv ∈ G. Consider the graph G −e. Now u and v are connected by at most k − 2 independent paths in G −e, and by Theorem 2.7, u and v can be separated in G −e by a set S with |S| = k − 2. Since νG > k (because κ ( G ) = k), there exists a w ∈ G that is not in S ∪ {u, v}. The vertex w is separated in G −e by S from u or from v; otherwise there would be a path ⋆ u − v in ( G −e)−S. Say, this vertex is u. The set S ∪ {v} has k − 1 elements, and it → separates u from w in G, which contradicts the assumption that κ ( G ) = k. This proves the claim. ⊔ ⊓ We state without a proof the corresponding separation property for edge connectivity. D EFINITION . Let G be a graph. A uv-disconnecting set is a set F ⊆ EG such that ⋆ every path u − v contains an edge from F. → Theorem 2.9. Let u, v ∈ G with u = v in a graph G. Then the maximum number of edge⋆ disjoint paths u − v equals the minimum number k of edges in a uv-disconnecting set. → Corollary 2.4. A graph G is k-edge connected if and only if every two vertices are connected by at least k edge disjoint paths. Example 2.9. Recall the deﬁnition of the cube Qk from Example 1.5. We show that κ ( Qk ) = k. First of all, κ ( Qk ) ≤ δ( Qk ) = k. In converse, we show the claim by induction. Extract from Qk the disjoint subgraphs: G0 induced by {0u | u ∈ B k−1 } and G1 induced by {1u | u ∈ B k−1 }. These are (isomorphic to) Qk−1 , and Qk is obtained from the union of G0 and G1 by adding the 2k−1 edges (0u, 1u) for all u ∈ B k−1 .

2.2 Connectivity

28

Let S be a separating set of Qk with |S| ≤ k. If both G0 −S and G1 −S were connected, also Qk −S would be connected, since one pair (0u, 1u) necessarily remains in Qk −S. So we can assume that G0 −S is disconnected. (The case for G1 −S is symmetric.) By the induction hypothesis, κ ( G0 ) = k − 1, and hence S contains at least k − 1 vertices of G0 (and so |S| ≥ k − 1). If there were no vertices from G1 in S, then, of course, G1 −S is connected, and the edges (0u, 1u) of Qk would guarantee that Qk −S is connected; a contradiction. Hence |S| ≥ k. Example 2.10. We have κ ′ ( Qk ) = k for the k-cube. Indeed, by Whitney’s theorem, κ ( G ) ≤ κ ′ ( G ) ≤ δ( G ). Since κ ( Qk ) = k = δ( Qk ), also κ ′ ( Qk ) = k. Algorithmic Problem. The connectivity problems tend to be algorithmically difﬁcult. In the disjoint paths problem we are given a set (ui , vi ) of pairs of vertices for i = ⋆ 1, 2, . . . , k, and it is asked whether there exist paths Pi : ui − vi that have no vertices in → common. This problem was shown to be NP-complete by K NUTH in 1975. (However, for ﬁxed k, the problem has a fast algorithm due to R OBERTSON and S EYMOUR (1986).)

Dirac’s fans

D EFINITION . Let v ∈ G and S ⊆ VG such that v ∈ S / in a graph G. A set of paths from v to a vertex in S is called a (v, S)-fan, if they have only v in common. Theorem 2.10 (D IRAC (1960)). A graph G is k-connected if and only if νG > k and for every v ∈ G and S ⊆ VG with |S| ≥ k and v ∈ S, there exists a (v, S)-fan of k paths. / Proof. Exercise. v

∗ ∗ ∗

...

S

⊔ ⊓

Theorem 2.11 (D IRAC (1960)). Let G be a k-connected graph for k ≥ 2. Then for any k vertices, there exists a cycle of G containing them. Proof. First of all, since κ ( G ) ≥ 2, G has no cut vertices, and thus no bridges. It follows that every edge, and thus every vertex of G belongs to a cycle. Let S ⊆ VG be such that |S| = k, and let C be a cycle of G that contains the maximum number of vertices of S. Let the vertices of S ∩ VC be v1 , . . . , vr listed in order around C so that each pair (vi , vi+1 ) (with indices modulo r) deﬁnes a path along C (except in the special case where r = 1). Such a path is referred to as a segment of C. If C contains all vertices of S, then we are done; otherwise, suppose v ∈ S is not on C. It follows from Theorem 2.10 that there is a (v, VC )-fan of at least min{k, |VC |} ⋆ ⋆ paths. Therefore there are two paths P : v − u and Q : v − w in such a fan that end → → ⋆ ⋆ in the same segment (vi , vi+1 ) of C. Then the path W : u − w (or w − u) along C → → contains all vertices of S ∩ VC . But now PWQ−1 is a cycle of G that contains v and all vi for i ∈ [1, r]. This contradicts the choice of C, and proves the claim. ⊔ ⊓

**3 Tours and Matchings
**

3.1 Eulerian graphs

The ﬁrst proper problem in graph theory was the Königsberg bridge problem. In general, this problem concerns of travelling in a graph such that one tries to avoid using any edge twice. In practice these eulerian problems occur, for instance, in optimizing distribution networks – such as delivering mail, where in order to save time each street should be travelled only once. The same problem occurs in mechanical graph plotting, where one avoids lifting the pen off the paper while drawing the lines.

Euler tours

D EFINITION . A walk W = e1 e2 . . . en is a trail, if ei = e j for all i = j. An Euler trail of a graph G is a trail that visits every edge once. A connected graph G is eulerian, if it has a closed trail containing every edge of G. Such a trail is called an Euler tour. Notice that if W = e1 e2 . . . en is an Euler tour (and so EG = {e1 , e2 , . . . , en }), also ei ei+1 . . . en e1 . . . ei−1 is an Euler tour for all i ∈ [1, n]. A complete proof of the following Euler’s Theorem was ﬁrst given by H IERHOLZER in 1873. Theorem 3.1 (E ULER (1736), H IERHOLZER (1873)). A connected graph G is eulerian if and only if every vertex has an even degree.

⋆ Proof. (⇒) Suppose W : u − u is an Euler tour. Let v (= u) be a vertex that occurs → k times in W. Every time an edge arrives at v, another edge departs from v, and therefore dG (v) = 2k. Also, dG (u) is even, since W starts and ends at u. (⇐) Assume G is a nontrivial connected graph such that dG (v) is even for all v ∈ G. Let ⋆ W = e1 e2 . . . en : v0 − vn with ei = vi−1 vi →

be a longest trail in G. It follows that all e = vn w ∈ G are among the edges of W, for, otherwise, W could be prolonged to We. In particular, v0 = vn , that is, W is a closed trail. (Indeed, if it were vn = v0 and vn occurs k times in W, then dG (vn ) = 2(k − 1) + 1 and that would be odd.) If W is not an Euler tour, then, since G is connected, there exists an edge f = vi u ∈ G for some i, which is not in W. However, now e i + 1 . . . e n e1 . . . e i f is a trail in G, and it is longer than W. This contradiction to the choice of W proves the claim. ⊔ ⊓

every graph has an even number of vertices with odd degree. where a postman wishes to plan his route to save the legs. Let G be a graph with a weight function α : EG → R + .1. because such a tour traverses each edge exactly once and this is the best one can do. . ⊔ ⊓ . Theorem 3. then any Euler tour will do as a solution. A connected graph has an Euler trail if and only if it has at most two vertices of odd degree. If G has an Euler trail u − v. v} has an even degree. Here the → → → ⋆ beginning part u − v is an Euler trail of G. then. say u and v. → ⊔ ⊓ The Chinese postman The following problem is due to G UAN M EIGU (1962). If G is eulerian. Consider a village. but still every street has to be walked through. i ]) so that (i) ei+1 has an end vi . / Assume then that G is connected and has at most two vertices of odd degree. . say u − v − w − u. Choose an edge ei+1 (= e j for j ∈ [1. as long as possible: suppose a trail Wi = e1 e2 . Notice that. . In H every vertex ⋆ has an even degree. .2. This problem is akin to Euler’s problem and to the shortest path problem.1. If G is eulerian. The k-cube Qk is eulerian for even integers k. and (ii) ei+1 is not a bridge of Gi = G −{e1 . G has an Euler trail. . Otherwise. .1. and let W0 be the trivial path on v0 . where e j = v j−1 v j . then any trail of G constructed by Fleury’s algorithm is an Euler tour of G. the post ofﬁce). each vertex → w ∈ {u. • Repeat the following procedure for i = 1. as in the proof of Theorem 3. If G has no vertices of odd degree then. 2. and there is a good algorithm for ﬁnding such a tour: Fleury’s algorithm: • Let v0 ∈ G be a chosen vertex. ei has been constructed. . by the handshaking lemma. . because Qk is k-regular. In this case the weight of the optimal tour is the total weight of the graph G. Exercise. Let H be a graph obtained from G by adding a vertex w.3. ei }. The Chinese postman problem is to ﬁnd a minimum weighted tour in G (starting from a given vertex. and hence it has an Euler tour. by Theorem 3. . the weights α(e) play no role in the eulerian case. and therefore G has exactly two such vertices. Theorem 3. unless there is no alternative. and the edges uw and vw. as is natural. Proof. ⋆ Proof.3.1 Eulerian graphs 30 Example 3.

2 Hamiltonian graphs In the connector problem we reduced the cost of a spanning graph to its minimum. e. Kn. . There is a good algorithm by E DMONDS AND J OHNSON (1973) for the construction of an optimal eulerian supergraph by duplications. Unlike for the shortest path and the connector problems no efﬁcient reliable algorithm is known for the travelling salesman problem. It is obvious that each Kn is hamiltonian whenever n ≥ 3. The rightmost multigraph is eulerian. if it visits each vertex once. A path P of a graph G is a Hamilton path. the poor postman has to walk at least one street twice. n]. Unfortunately. he is looking for a minimum weighted Hamilton cycle of a graph. the vertices of which are the towns and the weights on the edges are the ﬂight times. We can attack this case by reducing it to the eulerian case as follows. and in general if there is a street corner of an odd number of streets. .3. Indeed. 3. Example 3. this algorithm is somewhat complicated. if it is added to G parallel to an existing edge e′ = uv with the same weight. and we shall skip it. then so is ui → .m is hamiltonian if and only if n = m ≥ 2. In terms of graphs. if one of the streets is a dead end. α(e′ ) = α(e).g. and so that the overall ﬂight time is as short as possible. if it has a Hamilton cycle. where the cost is measured by an active user of the graph. Hamilton cycles D EFINITION . and thus we can choose where to start the cycle. A graph is hamiltonian.m have a . This happens. Indeed. if P visits every vertex of G once. .2. un → u1 → . For instance.2 Hamiltonian graphs 31 If G is not eulerian. Obviously. in the travelling salesman problem a person is supposed to visit each town in his district. ui−1 for each i ∈ [1. 4 3 2 1 2 2 2 3 4 3 1 2 2 2 3 4 3 3 1 3 2 Above we have duplicated two edges.. it is widely believed that no practical algorithm exists for this problem. Similarly. a cycle C is a Hamilton cycle. as is easily seen. There are different problems. An edge e = uv will be duplicated. Note that if C : u1 → u2 → · · · → un is a Hamilton cycle. he should plan the travel so as to visit each town once. . and this he should do in such a way that saves time and money. Also. let Kn.

This graph is not hamiltonian. each cycle in Kn. because C is a cycle and ui vi ∈ G for all i. c( G −S) = 3 > 2 = |S| for the set S of black vertices.m has even length as the graph is bipartite.4.3. Y )-bipartite of even order with |X | = |Y |. Theorem 3. due to O RE. u ∈ S. k]. and let C : u − u be a Hamilton cycle of G. Therefore the conclusion of Lemma 3. some interesting general conditions. ⋆ Proof. then there is nothing to prove. Therefore G does not satisfy the condition of Lemma 3. If C does not use the edge e. Proof. then for every nonempty subset S ⊆ VG . The following theorem. Lemma 3. Interestingly this graph is ( X. Let ∅ = S ⊆ VG . Gi . But then necessarily | X | = |Y |.1) no good characterization is known for hamiltonian graphs. and v j = vt for all j = t. and thus the cycle visits the sets X. which appears in many places in graph theory as a counter example for various conditions. In G.1 is not sufﬁcient to ensure that a graph is hamiltonian. Then G is hamiltonian if and only if G + uv is hamiltonian. i ∈ [1. − vn = v is a Hamilton → → → → → . Now vi ∈ S for each i by the choice of ui . Let G be a graph of order νG ≥ 3. and hence suppose that k > 1. If uv ∈ G. Thus |S| ≥ k as required. (⇐) Denote e = uv and suppose that G + e has a Hamilton cycle C. Assume thus that uv ∈ G. Let u. Y equally many times.3. / (⇒) This is trivial since if G has a Hamilton cycle C. where | X | = n and |Y | = m. Consider the Petersen graph on the right. Assume → G −S has k connected components. Unlike for eulerian graphs (Theorem 3.1. It is also 3-regular. Example 3. Let ui be the last vertex of C that belongs to Gi . We may ⋆ then assume that C : u − v − u. . If G is hamiltonian. the problem to determine if G is hamiltonian is NPcomplete. Now u = v1 − v2 − . The case k = 1 is trivial. v ∈ G be such that dG (u) + dG (v) ≥ νG . however. but it does satisfy the condition c( G −S) ≤ |S| for all S = ∅. and hence it is not hamiltonian. and let vi be the vertex that follows ui in C. .2 Hamiltonian graphs 32 bipartition ( X. v ∈ G be such that dG (u) + dG (v) ≥ n. Indeed. then it is a Hamilton cycle of G. since X and Y are stable subsets. Suppose thus that e is on C. Now. then C is also a Hamilton cycle of G + uv.1. There are. generalizes an earlier result by D IRAC (1952). Y ). c( G − S ) ≤ |S | . ⊔ ⊓ Example 3. and let u.4 (O RE (1962)). Denote n = νG . Consider the graph on the right.

then so are Gk−1 . a contradiction. and it is denoted by cl ( G ) (= Gk ) . ⊔ ⊓ . Proof. where the edges are added in the given orders. . First.2 Hamiltonian graphs 33 path of G. Gk of graphs such that G0 = G and Gi+1 = Gi + uv . . . There exists an i with 1 < i < n such that uvi ∈ G and vi−1 v ∈ G. . However. In each step of the construction of cl ( G ) there are usually alternatives which edge uv is to be added to the graph. since each complete graph is hamiltonian. In the other direction. . deﬁne inductively a sequence G0 . f i }. . . G1 . we have Hk−1 ⊆ H ′ . . G1 and G0 by Theorem 3. er } and H ′ = G + { f 1 . v1 v2 ◦ ◦ v i −1 vi ◦ ◦ vn ⋆ ⋆ But now u = v1 − vi−1 − vn − vn−1 − vi+1 − vi − v1 = u is a Hamilton cycle → → → → → → in G. . By the choice of ek . where u and v are any vertices such that uv ∈ Gi and dGi (u) + dGi (v) ≥ νG . . Therefore H ⊆ H ′ . . but ek ∈ Hk−1 . ⊔ ⊓ Theorem 3. ⊔ ⊓ Closure D EFINITION . and thus if G is hamiltonian. If cl ( G ) is hamiltonian. and hence H ′ = H. The Claim (ii) follows from (i). . (ii) If cl ( G ) is a complete graph. . . . Let G be a graph of order νG ≥ 3. we deduce that H ′ ⊆ H. This / procedure stops when no new edges can be added to Gk for some k. Then d Hk −1 (u) + d Hk −1 (v) ≥ n. say H = G + {e1 . let G = G0 .4. For the initial values. For a graph G. the ﬁnal result cl ( G ) is independent of the choices. then G is hamiltonian. otherwise. . ei } and Hi′ = ′ G + { f 1 . G ⊆ cl ( G ) and G spans cl ( G ). in Gk . Lemma 3.2. . The closure cl ( G ) is uniquely deﬁned for all graphs G of order νG ≥ 3. and thus also d H ′ (u) + / d H ′ (v) ≥ n. so is cl ( G ). . Suppose there are two ways to close G. . since ek ∈ Hk . . v ∈ G either uv ∈ Gk or dGk (u) + dGk (v) < νG . . and therefore the above procedure is not deterministic. Proof. Gk = cl ( G ) be a construction sequence of the closure of G. G1 . .3. f s } . . Let Hi = G + {e1 . Let ek = uv be the ﬁrst edge such that ek = f i for all i. . dG (v) < n − dG (u) would contradict the assumption. Denote n = νG . . we have G = H0 = H0 . The result of this procedure is the closure of G. (i) G is hamiltonian if and only if its closure cl ( G ) is hamiltonian. Symmetrically. that is. for all u. which means that e = uv must be in H ′ . .5. . For. .

in this case. and therefore dG (u) = i < n/2. and adding edges to G does not decrease any of its degrees. v2 . Suppose that for all nonadjacent vertices u and v. First of all. Consequently. for di = dG (vi ). Let A = {w | vw ∈ G. . Assume on the contrary that G = Kn . Theorem 3. the ﬁrst example where DNA computing was applied. / By our choice. ordered so that d1 ≤ d2 ≤ · · · ≤ dn . we may suppose that G is closed. then G is hamiltonian. ⊔ ⊓ Note that the condition (3. for n ≥ 3. dG (u) + dG (v) ≥ νG . because G is hamiltonian if and only if cl ( G ) is hamiltonian. for each vertex from B = {w | uw ∈ G. and let uv ∈ G with dG (u) ≤ dG (v) be / such that dG (u) + dG (v) is as large as possible. In particular.3. then G is 2 hamiltonian. Proof. We become satisﬁed by two general results. Consequently. there are at least i vertices w with dG (w) ≤ i. at least partly because the problem has practical signiﬁcance. and | B | = ( n − 1) − d G ( u ) = ( n − 1) − i . . Let G be a graph of order νG ≥ 3. (Indeed. since now dG (u) + dG (v) ≥ νG for all u. The second claim is immediate. and thus G is hamiltonian. w = u}. dn−i < n − i. if δ( G ) ≥ 1 νG .) There are some general improvements of the previous results of this chapter. di ≤ i =⇒ dn−i ≥ n − i . if G satisﬁes (3. we must have dG (u) + dG (v) < n. We show that.1). we have cl ( G ) = Kn for n = νG . ⊔ ⊓ Chvátal’s condition The hamiltonian problem of graphs has attracted much attention. G = cl ( G ). v ∈ G whether adjacent or not. was the hamiltonian problem. If for every i < n/2. and thus there are at least n − i vertices w with dG (w) < n − i. so does G + e for every e. . dG (w) ≤ i for all w ∈ A. Then G is hamiltonian.7 (C HVÁTAL (1972)). This contradicts the obtained bound di ≤ i and the condition (3. vn }.6. Proof. dG (w) ≤ dG (v) < / n − dG (u) = n − i. Similarly. Also dG (u) < n − i. where the graphs are somehow restricted. that is.1) is easily checkable for any given graph. and quite many improvements in various special cases. and. . moreover. and so di ≤ dG (u) = i.1) | A | = ( n − 1) − d G ( v ) ≥ d G ( u ) = i . Because G is closed. G = Kn . Let G be a graph with VG = {v1 .2 Hamiltonian graphs 34 Theorem 3. . and thus G is hamiltonian.1). w = v}. Since dG (u) + dG (v) ≥ νG for all nonadjacent vertices. (3.

although you cannot add any edges to M to form a larger matching. Maximum matchings D EFINITION . and a job may be suited for several applicants. then M is a perfect matching.3. if for no matching M ′ .3. | M | < | M ′ |. One of them is the job assignment problem. The two vertical edges on the right constitute a matching M that is not a maximum matching. D EFINITION . An odd path P = e1 e2 . if it saturates every v ∈ A. where we are given n applicants and m jobs. if M contains no adjacent edges.3 Matchings 35 3. Let M be a matching of G. then G is a path or a cycle. A matching M saturates v ∈ G. a subset M ⊆ EG is a matching of G. Exercise. Given a set X of boys and a set Y of girls. if • P alternates between EG \ M and M (that is. e2k+1 is M-augmented. if v is an end of an edge in M. and we should assign each applicant to a job he is qualiﬁed. under what condition can each boy marry a girl who cares to marry him? This problem has many variations. e2i+1 ∈ G − M and e2i ∈ M). . It is clear that every perfect matching is maximum. D EFINITION . . The problem is then to ﬁnd a pairing of the elements so that each element is paired (matched) uniquely with an available companion. The two ends of an edge e ∈ M are matched under M. which is stated as follows. This matching is not maximum because the graph has a matching of three edges. If M saturates VG . For a graph G. A special case of the matching problem is the marriage problem. and • the ends of P are not saturated. Lemma 3. On the right the horizontal edges form a perfect matching. Proof. M saturates A ⊆ VG . ⊔ ⊓ We start with a result that gives a necessary and sufﬁcient condition for a matching to be maximum. A matching M is a maximum matching. One can use the ﬁrst part of the proof to construct a maximum . The problem is that an applicant may be qualiﬁed for several jobs. Also.3 Matchings In matching problems we are given an availability relation between the elements of a set. If G is connected with ∆( G ) ≤ 2.

If G is ( X. Deﬁne N ⊆ EG by / N = ( M \ {e2i | i ∈ [1. e2k ∈ M. e2k+1 ∈ M. ⊔ ⊓ Example 3. since | N | > | M |. and S ⊆ X. Proof.3. . Indeed. . Hence A : u − v is a path (of → odd length). e4 . We have d H (v) ≤ 2 for each v ∈ H. (⇐) Assume N is a maximum matching. Hence | N | > | M |. This proves the theorem. . because v is an end of at most one edge in M and N. which starts and ends with an edge from N. there is a connected component A of H. because an alternating cycle has even length. the matching M = {(0u. e2k+1 in G.3 Matchings 36 matching in an iterative manner starting from any matching M and from any Maugmented path. k]}) ∪ {e2i+1 | i ∈ [0. Y )-bipartite. Because A is a connected component of H. This A cannot be a cycle. each connected component A of H is either a path or a cycle.9 (H ALL (1935)). Let G be a ( X. denote NG (S) = {v | uv ∈ G for some u ∈ S} . consequently. Consider the k-cube Qk for k ≥ 1. . Then G contains a matching M saturating X if and only if |S| ≤ | NG (S)| for all S ⊆ X. . Y )-bipartite graph. and it is clearly perfect. known as the Theorem 3. .5.8 (B ERGE (1957)). the ends u and v are not saturated by M. 1u) | u ∈ B k−1 }. . . Theorem 3. Now. each connected component (path or a cycle) A alternates between N and M. Hall’s theorem For a subset S ⊆ VG of a graph G. Consider the subgraph H = G [ M △ N ] for the symmetric difference M △ N. and | N | = | M | + 1. (3. has 2k−1 edges. e3 . . Now. N is a matching of G. (⇒) Let a matching M have an M-augmented path P = e1 e2 . A is an M-augmented path. which has more edges from N than from M.2) . and it ⋆ thus contains equally many edges from N and M. k]} . e1 . Since no v ∈ A can be an end of two edges from N or from M. and. Each maximum matching of Qk has 2k−1 edges. A matching M of G is a maximum matching if and only if there are no M-augmented paths in G.3. Here e2 . but M is not. By Lemma 3. Therefore M is not a maximum matching. . then NG (S) ⊆ Y. The following result.

k · | NG (S)| = | E2 | ≥ | E1 | = k · |S|. the induced subgraph H2 = G [VG \ A]. H contains a matching M saturating X \ {u}. then choose an edge uv ∈ G with u ∈ X. and hence | X | = |Y |. Clearly.1 (F ROBENIUS (1917)). by Menger’s → theorem. H1 contains a matching M1 that saturates R (with the other ends in NG ( R)). then not all x ∈ S can be matched with different y ∈ NG (S). There exists a matching saturating X if (and ⋆ only if) the number of independent paths x − y is equal to | X |. ⊔ ⊓ . By regularity. then the claim is clear. satisﬁes (3. and assume (3.2). and hence we have NG ( X \ A) ⊆ B. This proof of the direction (⇐) uses Menger’s theorem. and hence. We conclude that |S| = | A| + | B| ≥ | X |. and consider the induced subgraph H = G −{u. where A ⊆ X and B ⊆ Y. by the induction hypothesis. we get a matching M1 ∪ M2 saturating X in G. For all S ⊆ X \ {u}. G has a matching that saturates X.2) for G. (⇒) Let M be a matching that saturates X. then we would have | NG (S ∪ R)| = | NH2 (S)| + | NH1 ( R)| < |S| + | NG ( R)| = |S| + | R| = |S ∪ R| (since S ∩ R = ∅). Also. If G is a k-regular bipartite graph with k > 0. Since | X | = |Y |. v}. Now. k · | X | = ε G = k · |Y |. If | NG (S)| ≥ |S| + 1 for every nonempty S ⊆ X with S = X. If | X | = 1. Let S ⊆ X. If |S| > | NG (S)| for some S ⊆ X. Therefore. (⇐) Let G satisfy Hall’s condition (3. We prove the claim by induction on | X |. and thus that |X \ A| ≤ | NG ( X \ A)| ≤ | B| using the condition (3. Now M ∪ {uv} is a matching saturating X in G.2) implies the existence of a matching that saturates every proper subset of X. By Theorem 3. Combining the matchings for H1 and H2 . Let H be the graph obtained from G by adding two new vertices x.2) (since G does).3 Matchings 37 Proof. by the induction hypothesis. which contradicts (3. as was required.9. and by E2 the set of the edges with an end in NG (S). H2 has a matching M2 that saturates X \ R (with the other ends in Y \ NG ( R)). for A = R ∪ NG ( R). it sufﬁces to show that every set S that separates x and y in H has at least |X | vertices. Proof.3. Y )-bipartite graph.2). Let then | X | ≥ 2. Let G be k-regular ( X. By the induction hypothesis. vertices in X \ A are not adjacent to vertices of Y \ B. | NH (S)| ≥ | NG (S)| − 1 ≥ |S| . Let S = A ∪ B. x X\A A Y\B B y ⊔ ⊓ Corollary 3. and hence. The induced subgraph H1 = G [ R ∪ NG ( R)] satisﬁes (3. then G has a perfect matching. Indeed.2). y such that x is adjacent to each v ∈ X and y is adjacent to each v ∈ Y. ⊔ ⊓ Second proof. For this. Denote by E1 the set of the edges with an end in S. this matching is necessarily perfect. if there were a subset S ⊆ X \ R such that | NH2 (S)| < |S|. E1 ⊆ E2 . and so | NG (S)| ≥ |S|. Suppose then that there exists a nonempty subset R ⊆ X with R = X such that | NG ( R)| = | R|.

n]. . An m × n latin rectangle is an m × n integer matrix M with entries Mij ∈ [1. A2 . D EFINITION . and therefore each i is missing from exactly n − m columns. 6}. Let S = {S1 . Then S has a transversal if and only if the union of any k of the subsets Si of S contains at least k elements. As an example. Now |∪i∈ I Ai | ≥ | I |. 4. We show the following: Let M be an m × n latin rectangle (with m < n). Sm } be a family of ﬁnite nonempty subsets of a set S. Let S be a family of ﬁnite nonempty sets. (Si need not be distinct. S2 . An } has a transversal. . and let S1 = S2 = {1. For S = {S1 . Y )-bipartite graph G such that there is an edge (i. . S4 }. . Denote by m( G ) be the number of edges in a maximum matching of a graph G. Let Ai ⊆ [1. then M is a latin square. if it has an odd (even) number of vertices. and this transversal can be added as a new row to M. S2 . . | Ai | = n − m for each i. If we add the set S5 = {2. The connection of transversals to the Marriage Theorem is as follows. Form an ( X. By Marriage Theorem. 6].) A transversal (or a system of distinct representatives) of S is a subset T ⊆ S of m distinct elements one from each Si . since otherwise at least one element from the union would be in more than n − m of the sets Ai with i ∈ I. and hence ∑i∈ I | Ai | = | I |(n − m) for all subsets I ⊆ [1.6. This proves the claim. 2}. 3. Denote by codd ( G ) the number of odd connected components in G. s) if and only if s ∈ Si . . m]. A connected component of a graph G is said to be odd (even). S3 . The claim follows when we show that M can be extended to an (m + 1) × n latin rectangle. let S = [1. 5. The possible transversals T of S are then obtained from the matchings M saturating X in G by taking the ends in Y of the edges of M. 3} to S . Example 3. However. the set T = {1. Clearly. then it is impossible to ﬁnd a transversal for this new family. n] be the set of those elements that do not occur in the i-th column of M. Tutte’s theorem The next theorem is a classic characterization of perfect matchings. Note that in a m × n latin rectangle M. . 4} is a transversal. Then M can be extended to a latin square by the addition of n − m new rows. we always have that m ≤ n. . . if m = n. 2. Moreover.3 Matchings 38 Applications of Hall’s theorem D EFINITION . 3} and S4 = {1. each row has all the n elements.3. Corollary 3.2. Let S = Y and X = [1. n] such that the entries in the same row and in the same column are different. the family { A1 . S3 = {2.

S ⊆VG 39 νG + |S| − codd ( G −S) 2 (3. If νG = 1. the right hand side of (3. Let w be an inter⋆ mediate vertex on a shortest path u − v. Note that the condition in (ii) includes the case. there is a maximum matching that does not saturate v. But now N ′ = N ∪ {e} \ {e′ } is a maximum matching that does not saturate w. ⊔ ⊓ . Suppose then that for each vertex v. since. Let e = xy ∈ M. By the induction hypothesis. v) and dG (w. Therefore. If y is also unsaturated by N. by the beginning of the proof. N saturates both u and v. m ( G ) ≤ |S | + m ( G −S ) ≤ |S | + m( G ) − 1 ≥ 1 (νG − 1) + |S′ | − codd ( G −(S′ ∪ {v})) 2 1 = ((νG + |S| − codd ( G −S))) − 1. N ∩ M ⊂ N ′ ∩ M contradicts the choice of N. then the claim is trivial. 2 2 Indeed. w) < dG (u. The proof proceeds by induction on νG . Now dG (u. By assumption. for all S ⊆ VG . ν + |S| − codd ( G −S) |VG \ S| − codd ( G −S) = G ..10 (Tutte-Berge Formula).e. We can choose N such that the intersection M ∩ N is maximal. this must be the minimum of the right hand side. v) is as small as possible. each odd component of G −S must have at least one unsaturated vertex. contradicting maximality of N. where S = ∅. Assume ﬁrst that there exists a vertex v ∈ G such that v is saturated by all maximum matchings.3). i. We observe ﬁrst that ≤ holds in (3. for all S′ ⊆ G −v. v) < dG (u. v) ≥ 2. In this case. Suppose that νG ≥ 2. then N ∪ {e} is a matching. We prove the result for connected graphs. Since dG (u. We claim that m( G ) = (νG − 1)/2. and hence there exists another vertex x = w saturated by M but which is unsaturated by N. Each maximum matching of a graph G has m( G ) = min elements.3. since otherwise uv ∈ G could be added to M.3) Proof. and hence. 2 The claim follows from this. contradicting the maximality of M. there exists a maximum → matching N that does not saturate w. Therefore there exists an edge e′ = yz in N. and let M be a maximum matching having two different unsaturated vertices u and v. However. The result then follows for disconnected graphs by adding the formulas for the connected components. where z = x. denote S = S′ ∪ {v}. Then m( G −v) = m( G ) − 1. The (maximum) matchings N and M leave equally many vertices unsaturated. and choose M so that the distance dG (u. Suppose to the contrary. every maximum matching leaves exactly one vertex unsaturated.3 Matchings Theorem 3. v). m( G ) = (νG − 1)/2.3) gets value (νG − 1)/2. for S = ∅. For a subset S′ ⊆ G −v. It also follows that y = w.

Its applications are mainly in the proofs of other results that are related to matchings. 3 i=1 v∈S and so. Proof. G has a perfect matching. Theorem 3. because G has no bridges.3 Matchings 40 For perfect matchings we have the following corollary. v∈S The ﬁrst of these implies that mi = ∑ v ∈ Gi d G ( v ) − 2 · ε Gi is odd. There is a good algorithm due to E DMONDS (1965). Tutte’s theorem does not provide a good algorithm for constructing a perfect matching.12 (P ETERSEN (1891)).3. because removing the black vertex results in a graph with three odd connected components. Using Theorem 3. Furthermore.11 we can give a short proof of P ETERSEN’s result for 3-regular graphs (1891). mi = 1. Example 3. but this algorithm is somewhat involved. then it has a perfect matching. Hence the number of odd connected components of G −S satisﬁes t≤ 1 1 t ∑ m i ≤ 3 ∑ d G ( v) = |S | . be the odd connected components of G −S. Theorem 3. Here removing the middle vertex creates two odd components. The next 3-regular graph (known as the Sylvester graph) does not have a perfect matching. codd ( G −S) ≤ |S|. t]. ∑ v ∈ Gi dG (v) = 3 · νGi and ∑ d G ( v) = 3 · |S | . by Theorem 3. Let G be a nontrivial graph. which uses ‘blossom shrinkings’. Let S be a proper subset of VG . because the theorem requires ‘too many cases’. Since G is 3-regular. The following are equivalent. and therefore mi ≥ 3. since for a perfect matching we have m( G ) = (1/2)νG .11. If G is a bridgeless 3-regular graph. (i) G has a perfect matching. Denote by mi the number of edges with one end in Gi and the other in S. This graph is the smallest regular graph with an odd degree that has no perfect matching. and let Gi . (ii) For every proper subset S ⊂ VG . i ∈ [1.7.11 (T UTTE (1947)). ⊔ ⊓ . The simplest connected graph that has no perfect matching is the path P3 .

Example 3. A stable matching of G is the following: M = {28. We omit the proof of the next theorem. The algorithm by G ALE works as follows. The ﬁnal outcome. Theorem 3. Consider a bipartite graph G with a bipartition ( X. is a perfect matching. then set Mi = ( Mi−1 \ {zy}) ∪ { xy}. We write u < x v. each vertex x ∈ G supplies an order of preferences of the vertices of NG ( x). a stable matching exists for all lists of preferences. 35.14. Proof.13.n be a complete bipartite graph.8. For instance. There is a catch. Then iterate the following process until all vertices are saturated: Choose a vertex x ∈ X that is unsaturated in Mi−1 . v ∈ X. if x prefers v to u. it is not the case that x and y prefer each other better than their matched companions: xv ∈ M and y < x v. say M = Mt . / (1) Add y to P( x). if for each unmatched pair xy ∈ M / (with x ∈ X and y ∈ Y).) Theorem 3. or uy ∈ M and x <y u. For bipartite graphs G. (2) If y is not saturated. Let the bipartition be ( X.) A matching M of G is said to be stable. . First of all. AND S HAPLEY (1962) Procedure. the graph on the right does have a perfect matching (of 4 edges). Let y ∈ Y be the most preferred vertex for x such that y ∈ P( x). (Here u. v ∈ Y. That was the good news. of course. Suppose the preferences are the following: 1: 5 5: 4 < 1 < 3 2: 6 < 8 < 7 6: 2 3: 8 < 5 7: 2 < 4 4 3 2 1 8 7 6 5 4: 7 < 5 8: 3 < 2 Then there is no stable matchings of four edges. (You should check that there is no stable matching containing the edges 15 and 26.3 Matchings 41 Stable Marriages D EFINITION . (3) If zy ∈ Mi−1 and z <y x. then set Mi = Mi−1 ∪ { xy}. and P( x) = ∅ for all x ∈ X. the procedure terminates. if x ∈ X. Set M0 = ∅.3. Y ). if x ∈ Y. since the iteration continues until there are no unsaturated vertices x ∈ X. and u. In addition. Let G = Kn. Then G has a perfect and stable matching for all lists of preferences. which leaves 1 and 6 unmatched. A stable matching need not saturate X and Y. 47}. since a vertex x ∈ X takes part in the iteration at most n times (once for each y ∈ Y). Y ) of the vertex set.

⊔ ⊓ . xy ∈ Mi . but xz was not added). Thus. which means that z ∈ P( x) at this step (otherwise x would have ‘proposed’ z). but y < x z for some z ∈ Y. then y <v z for the vertex z ∈ Y such that vz ∈ M. Then xy is added to the matching at some step. Assume the that xy ∈ M. k < i (where z was put to the list P( x). the matching M = Mt is stable. for some u ∈ X. then x <y z. Note ﬁrst that. Similarly. uz ∈ Mk−1 and x <z u. if x <y v for some v ∈ X.3 Matchings 42 Also. by (3).3. Hence x took part in the iteration at an earlier step Mk . if xy ∈ Mi and zy ∈ M j for some x = z and i < j. and so in M the vertex z is matched to some w with x <z w.

Note that it is possible that Ei = ∅ for some i..1. This is clear. E2 . Proof. We adopt a simpliﬁed notation G α [ i 1 . In the general case. and we study the properties of this pair G α = ( G. if α(vu) = i for some vu ∈ G. . then i is available for v. Moreover. where Ei = {e | α(e) = i }. a graph G with a colouring α is given. . or it . . k] of a graph G is an assignment of k colours to its edges. In the chromatic theory. when one is interested in classifying relations between objects. G is ﬁrst given and then we search for a colouring that the satisﬁes required properties. The edge chromatic number χ′ ( G ) of G is deﬁned as χ′ ( G ) = min{k | there exists a proper k-edge colouring of G } . Each colour set Ei in a proper k-edge colouring is a matching. That is. if no two adjacent edges obtain the same colour: α(e1 ) = α(e2 ) for adjacent e1 and e2 . . .1 Edge colourings Colourings of edges and vertices of a graph G are useful. i t ] = G [ Ei1 ∪ Ei2 ∪ · · · ∪ Ei t ] for the subgraph of G consisting of those edges that have a colour i1 . k] are incident with each other. . . . train) of an edge tells the nature of a link. i 2 .4 Colourings 4. ∆( G ) ≤ χ′ ( G ) ≤ ε G . in transportation networks with bus and train links. ⊔ ⊓ . A k-edge colouring α : EG → [1. We write G α to indicate that G has the edge colouring α. Edge chromatic number D EFINITION . where the colour (buss. . for each graph G. α). i2 . If v ∈ G is not incident with a colour i. There are two sides of colourings. One of the important properties of colourings is ‘properness’. e. . the edges having other colours are removed. In a proper colouring adjacent edges or vertices are coloured differently. Ek } of EG . A vertex v ∈ G and a colour i ∈ [1. Lemma 4. .g. This is the situation. A k-edge colouring α can be thought of as a partition {E1 . The colouring α is proper. .

in which both colours are incident with each vertex v with dG (v) ≥ 2. Then there exists a 2-edge colouring (that need not be proper). where ei = vi vi+1 . the claim is trivial. . since each vertex vi with odd degree dG (vi ) ≥ 3 is entered and departed at least once in the tour by an edge of the original graph G: ei−1 ei . A star. Otherwise. and hence G0 is eulerian. t − 1] are incident with both colours. All vertices are among these ends.2. Optimal colourings We show that for bipartite graphs the lower bound is always optimal: χ′ ( G ) = ∆( G ). . α restricted to EG is a colouring of G as required by the claim. Now. Deﬁne 1. et be an Euler tour of G. We deﬁne a new graph G0 by adding a vertex v0 to G and connecting v0 to each v ∈ G of odd degree. otherwise. G has a vertex v1 with dG (v1 ) ≥ 4. et be an eulerian tour of G0 .1 can be equal. k]}| . Proof. Let G be a connected graph that is not an odd cycle. when G = K1. D EFINITION . Hence the ends of the edges ei for i ∈ [2.n is a star. If G is an even cycle. there is a required colouring α of G0 as above. since now dG (v) is even for all v. if i is odd . Let e0 e1 . Lemma 4. Let e1 e2 . if i is even .4. then a 2-edge colouring exists as required.1 Edge colourings 44 Example 4. let cα (v) = |{i | v is incident with i ∈ [1. (1) Suppose ﬁrst that G is eulerian. 2 v0 1 2 1 1 2 ⊔ ⊓ . Assume that G is nontrivial. But often the inequalities are strict. . for instance. The three numbers in Lemma 4. The condition dG (v1 ) ≥ 4 guarantees this for v1 . For a k-edge colouring α of G. (2) Suppose then that G is not eulerian. By the previous case. This happens. and a graph with χ′ ( G ) = 4.1. α( ei ) = 2. where ei = vi vi+1 (and vt+1 = v1 ). In G0 every vertex has even degree including v0 (by the handshaking lemma). . Hence the claim holds in the eulerian case.

4. by the construction of β. ∆( G ) ≤ χ′ ( G ) ≤ ∆( G ) + 1. then χ′ ( G ) = ∆( G ). Furthermore. for all vertices v. cα (v) = dG (v). where ∆ = ∆( G ). if 45 ∑ c β ( v) > ∑ cα ( v) . but it need not have any proper k-edge colourings. Therefore ∑u∈ G c β (u) > ∑u∈ G cα (u). Let ∆ = ∆( G ). Proof. v∈ G v∈ G Also. Lemma 4. α is a proper colouring.4.) We obtain a recolouring β of G as follows: β( e) = γ ( e ). α ( e ). Suppose the connected component H is not an odd cycle. Let α be an optimal ∆-edge colouring of a bipartite G. due to V IZING. we have c β (u) ≥ cα (u) for all u = v. G would contain an odd cycle. The following important theorem.4. Suppose that the colour i is available for v. ⊔ ⊓ Vizing’s theorem In general we can have χ′ ( G ) > ∆( G ) as one of our examples did show. Lemma 4. Suppose on the contrary that χ′ ( G ) > ∆ + 1. / Since d H (v) ≥ 2 (by the assumption on the colour j) and in β both colours i and j are now incident with v.3. Then the connected component H of G α [i. If G is bipartite. Let α be an optimal k-edge colouring of G. j] that contains v.3.1 Edge colourings A k-edge colouring β is an improvement of α. Notice that we always have cα (v) ≤ dG (v). By Lemma 4. if it cannot be improved. if e ∈ H . Note also that a graph G always has an optimal k-edge colouring.2. which contradicts the optimality of α. Proof. then by Lemma 4. H has a 2-edge colouring γ : E H → {i. if e ∈ H . An edge colouring α of G is proper if and only if cα (v) = dG (v) for all vertices v ∈ G. (We have renamed the colours 1 and 2 to i and j. j}. For any graph G. then cα (v) = dG (v). We need only to show that χ′ ( G ) ≤ ∆ + 1. in which both i and j are incident with each vertex x with d H ( x) ≥ 2. By Lemma 4.2 (V IZING (1964)). shows that the edge chromatic number of a graph G misses ∆( G ) by at most one colour. c β (v) = cα (v) + 1. α is optimal.1 (K ÖNIG (1916)). Thus an improvement of a colouring is a change towards a proper colouring. and the colour j is incident with v at least twice. If there were a v ∈ G with cα (v) < dG (v). and in this case α is optimal. Proof. Hence H is an odd cycle. But a bipartite graph does not contain such cycles. Therefore. ⊔ ⊓ Theorem 4. . and let v ∈ G. is an odd cycle. and ∆ = χ′ ( G ) as required. and let α be an optimal (∆ + 1)-edge colouring of G. Theorem 4. The next lemma is obvious. and if α is proper.

β is an optimal (∆ + 1)-edge colouring of G. j]. Claim 4. (4.1). it+1 = ir . Let t be the smallest index such that for some r < t. and so Claim 1. . u j . such that the claim holds for these. . Let then the colouring γ be obtained from β by recolouring the edges vu j by i j+1 for r ≤ j ≤ t. . u2 u r −1 . there exists an available colour b(u) for u. Claim 3. that v is not incident with b(u j ) = i j+1 . by the counter hypothesis. Indeed. and obtain in this way an improvement of α. . β(vu j ) = i j+1 . .. We can recolour the edges vuℓ by iℓ+1 for ℓ ∈ [1. cγ (u) ≥ c β (u) follows as for β. Such an index t exists. Here v gains a new colour i j+1 . . 46 Moreover. c β (v) = cα (v) and c β (u) ≥ cα (u) for all u. Assume we have already found the vertices u1 . α is not a proper colouring. because dG (v) is ﬁnite. say α(vu1 ) = i1 = α(vx) . . β(e) = α(e).4. . the fact ir = it+1 ensures that ir is a new colour incident with ut . each uℓ gains a new colour iℓ+1 (and may loose the colour iℓ ). . . with j ≥ 1. . and thus that cγ (ut ) ≥ c β (ut ). Also. and hence a colour i1 that is incident with v at least twice. Indeed. . such that α(vu j ) = i j and i j+1 = b(u j ) . γ is an optimal (∆ + 1)-edge colouring of G. i r = i t +1 v i3 i2 i1 x ur . For all other vertices. contrary to the claim. for each uℓ either its number of colours remains the same or it increases by one. . i r = i t +1 v i2 i1 u1 i1 x it ut i r −1 ur . Suppose. . since each u j (1 ≤ j ≤ r − 1) gains a new colour ji+1 although it may loose one of its old colours. Indeed.1) Claim 2. let u1 be as in (4. ut . u2 i3 i2 u1 i1 x u r −1 . Now. . Let β be a recolouring of G such that for 1 ≤ j ≤ r − 1. There is a sequence of vertices u1 .1 Edge colourings We have (trivially) dG (u) < ∆ + 1 < χ′ ( G ) for all u ∈ G. u2 . u2 ur . This contradicts the optimality of α. it ut ir u1 u r −1 . ir i r +1 v ir . vut is recoloured by ir = it+1 . For each u ∈ G. and proves Claim 2. Therefore.. and hence there exists a v ∈ G with cα (v) < dG (v).. and for all other edges e.

Since C ′ needs three colours. − u5 − u1 the inner cycle of G such that vi ui ∈ EG → → → for all i. By Lemma 4. by Vizing’ theorem. . In fact. This can be done uniquely (up to permutations): v1 − v2 − v3 − v4 − v5 − v1 . → → → → → Hence v1 − u1 . Ramsey theory studies unavoidable patterns in combinatorics. . v4 − u4 . We consider an instance of this theory mainly for edge colourings (that need not be proper). χ′ ( G ) = 3 or 4. v2 − u2 . that is.4.2 Ramsey Theory In general. . Observe that every vertex is adjacent to all colours 1. when the word ‘optimal’ is forgotten: colour ﬁrst the edges as well as you can (if nothing better.2. 2 3 1 1 1 1 2 3 2 3 4. However. . H1 is a cycle (vur −1 ) P1 (ur v) and H2 is a cycle (vur −1 ) P2 (ut v). Now C uses one colour (say 1) once and the other two twice. v3 − u3 . where ⋆ ⋆ both P1 : ur −1 − ur and P2 : ur −1 − ut are paths. and use the proof iteratively to improve the colouring until no improvement is possible – then the proof says that the result is a proper colouring. the edges of P1 and P2 → → have the same colours with respect to β and γ (either i0 or ir ). The answer is known (only) for some special classes of graphs. ir ] and H2 of G γ [i0 . ir ] containing the vertex v are cycles. the claim follows. this means that 1 → → → → → ′ . Indeed. However. Suppose 3 colours sufﬁce. This contradiction proves the theorem. A typical example of a Ramsey property is the following: given 6 persons each pair of whom are either friends or enemies. then arbitrarily in two colours). the connected components H1 of G β [i0 . . the problem whether χ′ ( G ) is ∆( G ) or ∆( G ) + 1 is NP-complete. cannot be a colour of any edge in C Edge Colouring Problem. This is not possible.2 Ramsey Theory 47 By Claim 1. v5 − u5 . 2. there are then 3 persons who are mutual friends or mutual enemies. By H OLYER (1981). ⊔ ⊓ Example 4. there is a colour i0 = b(v) that is available for v. for which χ′ ( G ) = ∆( G ) + 1. 3. We show that χ′ ( G ) = 4 for the Petersen graph. it is one of the famous open problems of graph theory to ﬁnd such a characterization. − v5 − v1 be → → → the outer cycle and C ′ : u1 − . In graph theoretic terms this means that each colouring of the edges of K6 with 2 colours results in a monochromatic triangle. Let C : v1 − . Vizing’s theorem (nor its present proof) does not offer any characterization for the graphs. since P1 ends in ur while P2 ends in a different vertex ut .4. The proof of Vizing’s theorem can be used to obtain a proper colouring of G with at most ∆( G ) + 1 colours.

where r ∈ [1.2 Ramsey Theory 48 Turan’s theorem for complete graphs We shall ﬁrst consider the problem of ﬁnding a general condition for K p to appear in a graph. H p−1 of order t (when t > 0). then G contains a complete subgraph K p . Suppose then that t ≥ 1. p) edges. p) = 2( p − 1) 2 p−1 The next result shows that the above bound T (n. If t = 0. . K p G − A. we can apply the induction hypothesis to obtain ε G− A ≤ T (n − p + 1. One can compute that the number ε H of edges of H is equal to r p−2 2 r n − 1− . A complete p-partite graph G consists of p discrete and disjoint induced subgraphs G1 . Let n = ( p − 1)t + r for 1 ≤ r ≤ p − 1 and t ≥ 0. We prove the claim by induction on t. If a graph G has ε G > triangle K3 . and let H = Hn. . i ∈ [1. Now ε G ≤ T (n − p + 1. K p H.4. Theorem 4. Gi and Gj with i = j. Proof. . (Here r is the positive residue of n modulo ( p − 1). . G2 . we have the following interesting case. then T (n. Note that a complete p-partite graph is completely determined by its discrete parts Gi .) By its deﬁnition. 2 ⊔ ⊓ When Theorem 4. Each v ∈ A is adjacent to at most p − 2 vertices of A. / Also.2) T (n. p) = n(n − 1)/2. and there is nothing to prove. otherwise G [ A ∪ {v}] = K p . . and that every nondiscrete graph contains K2 . G p ⊆ G. D EFINITION . ( p − 1)( p − 2) = T (n. (4. . p − 1] and t ≥ 0. p) is optimal. such that there are r parts H1 .1 ( M ANTEL (1907)). p) . . Corollary 4. Let p ≥ 3. and let G be a graph of order n such that ε G is maximum subject to the condition K p G. . . and is thus determined by n and p.3 is applied to triangles K3 .3 (T URÁN (1941)). and p − 1 vertices of this K p induce a subgraph K p−1 ⊆ G. Hr of order t + 1 and p − 1 − r parts Hr +1 . . Because n − p + 1 = (t − 1)( p − 1) + r. 1 2 4 νG edges. .p be the complete ( p − 1)-partite graph of order n = t( p − 1) + r. . It is clear that every graph contains K1 . then G contains a . since adding any one edge to G results in a K p . p). p) + (n − p + 1)( p − 2) + which proves the claim. where uv ∈ G if and only if u and v belong to different parts. If a graph G of order n has ε G > T (n. p]. and νG− A = n − p + 1. Now G contains a complete subgraph G [ A] = K p−1.

and in the second case. then it has a complete subgraph of order p or a stable subset of order q. In the ﬁrst case. Let p.5. q) such that for all n ≥ R( p. 2] contains a 1-monochromatic K p or a 2-monochromatic Kq . q). Theorem 4. any 2-edge colouring of Kn → [1. Before proving this. By the deﬁnition of Ramsey numbers.4 and 4. Be patient.4 (R AMSEY (1930)). This proves the claim.6 (E RDÖS and S ZEKERES (1935)). G [ NG (v)] contains a complete subgraph B of order p − 1 or a stable subset S of order q. we give an equivalent statement. if all edges of H have the same colour i. Theorem 4. q) is known as the Ramsey number for p and q.6. q) such that for all n ≥ R( p. and denote by A = VG \ ( NG (v) ∪ {v}) the set of vertices that are not adjacent to v. It is clear that R( p. Theorem 4. q) or | A| ≥ R( p.2 Ramsey Theory 49 Ramsey’s theorem D EFINITION . q − 1) + R( p − 1. q − 1) + R( p − 1. The number R( p. this will follow from Theorem 4. and R( p. Let p. the same complete subgraph of order p is good for G. if G [ X ] is a discrete graph. The following theorem is one of the jewels of combinatorics. and in the second case the same stable set of order q is good for G. If | A| ≥ R( p. In the ﬁrst case. q) − 1 vertices different from v. q) . q).4. 2) = p and R(2. Then there exists a (smallest) integer R( p. q). q) = q. Let v ∈ G. q ≥ 2 be any integers. q − 1). Let α be an edge colouring of G. Theorems 4. B ∪ {v} induces a complete subgraph K p in G. q) exists for all p. Then there exists a (smallest) integer R( p. Recall that a subset X ⊆ VG is stable. q) ≤ R( p. It is clear that R( p. q ≥ 2 be any integers. Since G has R( p.5 follow from the next result which shows (inductively) that an upper bound exists for the Ramsey numbers R( p. q). It is now sufﬁcient to show that if G is a graph of order R( p. q − 1) (or both). A subgraph H ⊆ G is said to be (i-) monochromatic. q) exists for p = 2 or q = 2. Proof. q − 1) + R( p − 1. ⊔ ⊓ . then G [ A] contains a complete subgraph of order p or a stable subset S of order q − 1. The Ramsey number R( p. any graph G of order n contains a complete subgraph of order p or a stable set of order q. Assume ﬁrst that | NG (v)| ≥ R( p − 1. and it is thus exists for p + q ≤ 5. q). either | NG (v)| ≥ R( p − 1. We use induction on p + q. S ∪ {v} is a stable subset of G of q vertices. q ≥ 2.

q). Theorem 4. Small Ramsey numbers. p) (where p = q) is for p = 5. k − 1] be obtained from α by identifying the colours k − 1 and k: β( e) = 1 α(e) if α(e) < k − 1 . .6 and the induction hypothesis. qk−2 . R ADZISZOWSKI. By Theorem 4. R( p. q2 . . p−1 50 Proof. not so much is known about these numbers. Generalizations Theorem 4. Proof. It has been veriﬁed that 43 ≤ R(5.2 Ramsey Theory A concrete upper bound is given in the following result. Let n = R(q1 . any k-edge colouring of Kn has an i-monochromatic Kqi for some i. . qk−2 . . . For p = 2 or q = 2. q − 1) + R( p − 1. . q) ≤ R( p. 2000 on the Web . we show that R(q1 . qk ) such that for all n ≥ R. 5) ≤ 49. Then there exists an integer R = R(q1 . k] with k ≥ 2. the claim is clear.4. of Combin. . . .4.. We use induction on p + q for the general statement. qk ) ≤ R(q1 . Let β : EKn → [1. q ≥ 2. where p = R(qk−1 . p\q 3 4 5 6 7 8 9 10 3 6 9 14 18 23 28 36 40-43 4 9 18 25 35-41 49-61 55-84 69-115 80-149 5 14 25 43-49 58-87 80-143 95-216 121-316 141-442 The ﬁrst unknown R( p. . Electronic J. Theorem 4. . p). S. For k > 2. .8. q) p+q−3 + p−1 p+q−3 p−2 p+q−2 . The case k = 2 is treated in Theorem 4. = ⊔ ⊓ In the table below we give some known values and estimates for the Ramsey numbers R( p.P. k] be an edge colouring. k − 1 if α(e) = k − 1 or k . .7 ( E RDÖS and S ZEKERES (1935)).4 can be generalized as follows. p). but to determine the exact value is an open problem. . q) ≤ p+q−2 . Let qi ≥ 2 be integers for i ∈ [1. qk ). Assume that p. As can be read from the table1 . The proof is by induction on k. For all p. . q ≥ 3. and let α : EKn → [1. p−1 ≤ which is what we wanted. . R( p.

n ) = 2n − 1 2n if n is even. To this end. . .4. Example 4. or the complement of G contains a complete subgraph Kn . even in more general cases. . . then R( T. . the generalized Ramsey numbers R( H1 . Hk ) α and for all k-edge colourings α of Kn . H2 . . Examples of Ramsey numbers∗ Some exact values are known in Corollary 4. Hk ) can be much smaller than their counter parts (for complete graphs) in Theorem 4. R2 ( G ) ≥ 4νG − 1 3 ( G connected). G ) (k times G ).3 ) = 18. However. for some dear graphs (see R ADZISZOWSKI’s survey). H2 . Kn ) contains a subgraph isomorphic to T. Below we list some of these results for cases. R2 (Cn ) = 2n − 1 if n ≥ 5 and n odd . where the graphs are equal. .8. Here is a list of some special cases: n R2 ( Pn ) = n + − 1. if n is odd. .3. R2 (K2. R2 (K3. Then there exists an integer R( H1 .3 ) = 10. H α and thus monochromatic subgraph H p α Kn has a (k − 1)-monochromatic or a k-monochromatic subgraph.4. 2 6 if n = 3 or n = 4. let Rk ( G ) = R( G. Hk ) such that for all complete graphs Kn with n ≥ R( H1 . that is. ⊔ ⊓ β Since for each graph H. by Theorem 4. In the latter case. . . .2. . . Kn ) = (m − 1)(n − 1) + 1 . R2 (K1. . we have Corollary 4. . . . Hk be arbitrary graphs. since this subgraph is monochromatic in Kn ) or Kn has a (k − 1)β = K . This generalization is trivial from Theorem 4. Kn has an i-monochromatic Kqi for some 1 ≤ i ≤ k − 2 α (and we are done. The best known lower bound of R2 ( G ) for connected graphs was obtained by B URR AND E RDÖS (1976).2 Ramsey Theory β 51 By the induction hypothesis. . . . H2 . any graph G of order at least R( T. Let k ≥ 2 and H1 . H2 . 3n/2 − 1 if n ≥ 6 and n even. H ⊆ Km for m = νH . We leave the following statement as an exercise: If T is a tree of order m.2. and this proves the claim. Kn contains an i-monochromatic subgraph Hi for some i.8. G. .

where 3 occurs n times. we shortly describe Schur’s theorem. . H AJNAL. The following gives a short proof using Ramsey’s theorem. But (t − j) + ( j − i) = t − i proves the claim.9 (D EUBER. 13]. consider the partition {1. the only nontrivial result for complete graphs is: R3 (K3 ) = 17. In fact. Note. . For this. . Then R2 (W5 ) = 15 and R2 (W6 ) = 17. then you are bound to ﬁnd a class. 7. if |i − j| ∈ Sk . that is. The following impressive theorem improves the results we have mentioned in this chapter and it has a difﬁcult proof.n ) are known for n ≤ 16.n ) ≤ 4n − 2. {2. K α has a monochromatic triangle. “Ramsey Theory”.2 Ramsey Theory 52 The values R2 (K2. . .L. Sn of N S(n) . B. and R ÖDL (around 1973)).g. deﬁne an edge colouring α of K by α(ij) = k. j − i. Needless to say that no exact values are known for Rk (Kn ) for k ≥ 4 and n ≥ 3. 12}. Indeed. but also..L. S CHUR (1916) solved this problem in a general setting. 3). 13}. . 3. R2 (K2. there are three vertices i. and let K be a complete on N S(n) . P ÓSA. Sn of N S(n) has a class Si containing two integers x. 11. there exists an integer S(n) such that any partition S1 . that in Corollary 4. . 10. 9} of the set N13 = [1. However. 4. As an application of Ramsey’s theorem. Let Wn denote the wheel on n vertices. We observe that in no partition class there are three integers such that x + y = z. t such that 1 ≤ i < j < t ≤ S(n) with t − j. It is a cycle Cn−1 . Example 4. For a partition S1 . much less is known. There are quite many interesting corollaries to Ramsey’s theorem in various parts of mathematics including not only graph theory. For the square C4 . see R. Theorem 4.17 ) is either 65 or 66. .L. 8.4. 6. . By Theorem 4. and in general. we know that R3 (C4 ) = 11. any sufﬁciently large complete graph) such that any 2-edge colouring of G has a monochromatic (induced) subgraph Kn . Wiley. t − i ∈ Sk for some k. but no nontrivial upper bound is known for R3 (K5 ).4. where x + y = z has a solution. R OTHSCHILD AND J. let S(n) = R(3. {5.8. It follows from Theorem 4. j. For three colours.2 the monochromatic subgraph Hi is not required to be induced. . (2nd ed. . y such that x + y ∈ Si . . Also.) 1990. Let H be any graph. and 385 ≤ R3 (K5 ). The value R2 (K2. geometry and algebra. E RDÖS. if you try to partition N14 into three classes. G RAHAM . S PENCER. there exists a graph G (well. e. Note that W4 = K4 . For each n ≥ 1. 128 ≤ R3 (K4 ) ≤ 235. . however. 3. Then there exists a graph G such that any 2-edge colouring of G has an monochromatic induced subgraph H. where a vertex v with degree n − 1 is attached.4 that for any complete Kn .

Let G t be a new graph obtained by adding n + 1 new vertices v and u1 . Again. and let π be any permutation of the colours. n]. if it has no subgraphs isomorphic to K3 . un such that G t has all the edges of G plus the edges ui v and ui x for all x ∈ N (vi ) and for all i ∈ [1. we have α(u) = α(v). ⊔ ⊓ Example 4.3 Vertex colourings 53 4. if there is a proper k-colouring for G. . k] → [1. . A k-colouring (or a k-vertex colouring) of a graph G is a mapping α : VG → [1. . The (vertex) chromatic number χ( G ) of G is deﬁnes as χ( G ) = min{k | there exists a proper k-colouring of G } . . Each proper vertex colouring α : VG → [1.3 Vertex colourings The vertices of a graph G can also be classiﬁed using colourings. . . and if π : [1. A colour i ∈ [1. if adjacent vertices obtain a different colour: for all uv ∈ G. then uv ∈ G implies that α(u) = α(v). k]. Proof. The following construction is due to G RÖTZEL: Let G be any triangle-free graph with VG = {v1 .5. V2 . If χ( G ) = k. In this chapter. k] is a bijection. k] is proper. . . v2 . we shall concentrate on proper vertex colourings. These colourings tell that certain vertices have a common property (or that they are similar in some respect). . It follows that πα is a proper colouring. . and hence also that πα(u) = πα(v). is 3-chromatic. . The graph on the right. Then the colouring β = πα is a proper k-colouring of G. which is often called a wheel (of order 7). u2 .6. Vk } of the vertex set VG .5.4. . Example 4. The chromatic number D EFINITION . the ‘names’ of the colours are immaterial: Lemma 4. if α : VG → [1. if they share the same colour. . then G is k-chromatic. We show that there are triangle-free graphs with arbitrarily large chromatic numbers. a graph G is 2-colourable if and only if it is bipartite. A graph G is k-colourable. The colouring α is proper. Indeed. if no neighbour of v is coloured by i. where adjacent vertices get different colours. k] is said to be available for a vertex v. vn }. where Vi = {v | α(v) = i }. A graph is triangle-free. Let α be a proper k-colouring of G. k] provides a partition {V1 . By the deﬁnitions.

v j . we obtain larger triangle -free graphs with high chromatic numbers. Now using inductively the above construction starting from the triangle-free graph K2 . . χ( G t ) > k. G t is triangle-free and it is k + 1-chromatic Indeed. and hence there is an available colour i for v. Proof. If G is k-critical for k ≥ 2. so does {vi . . say with α(v) = k. . extend it by setting α(ui ) = α(vi ) and α(v) = k + 1.3 Vertex colourings 54 Claim. Let then G be k-critical. Since G is critical. but the latter triangle is already in G. and so a triangle contains at most (and thus exactly) one vertex ui ∈ U. But there are k colours. Connectivity claim follows from this observation. Therefore χ( G t ) = k + 1. AND W ILF (1968). Note that for any graph G with the connected components G1 . vk } induces a triangle. Secondly. but δ( G ) = dG (v) ≤ k − 2 for v ∈ G. Each Kn is n-critical. and thus must have a cycle of odd length. let U = {u1 .7. . a contradiction. v j . G2 . since a 3-chromatic G is not bipartite. U is stable. m]} . Example 4. If {ui . vk } by the deﬁnition of G t . .4. If α is a proper k-colouring of G. since in Kn −(uv) the vertices u and v can gain the same colour. a contradiction. and δ( G ) ≥ k − 1. If we recolour v by i. . We show ﬁrst that G t is triangle-free. Recolour each vi by α(ui ). (iii) k ≤ 1 + max H ⊆ G δ( H ). Then α(ui ) = k. ⊔ ⊓ The case (iii) of the next theorem is due to S ZEKERES Theorem 4. a contradiction. Critical graphs D EFINITION . χ( G ) = max{χ( Gi ) | i ∈ [1. For the chromatic number we notice ﬁrst that χ( G t ) ≤ (k + 1). The 3-critical graphs are exactly the odd cycles C2n+1 for n ≥ 1. Let G be any graph with k = χ( G ). In a critical graph an elimination of any edge and of any vertex will reduce the chromatic number: χ( G −e) < χ( G ) and χ( G −v) < χ( G ) for e ∈ G and v ∈ G. This gives a proper (k − 1)-colouring to G. (ii) G has at least k vertices of degree ≥ k − 1. Now v is adjacent to only δ( G ) < k − 1 vertices. there is a proper (k − 1)-colouring of G −v. (i) G has a k-critical subgraph H. . then a proper (k − 1)-colouring is obtained for G. Now. un }.11.10. Assume that α is a proper k-colouring of G t . if χ( H ) < k for all H ⊆ G with H = G. . Theorem 4. A k-chromatic graph G is said to be k-critical. Gm . The graph K2 = P2 is the only 2-critical graph. then it is connected. .

v} is a separating set.10. . if uv ∈ G.3 Vertex colourings 55 Proof. and α(vi ) ≤ dG (vi ) + 1 for all i. v ∈ G. m]}. the chromatic number χ( G ) usually takes much lower values – as seen in the bipartite case. For (ii). The claim follows. v ∈ G. Denote Gi = G [ Ai ∪ {v}]. Moreover. vn } be ordered in some way. Lemma 4. ⊔ ⊓ Although. By Theorem 4. Let v be a cut vertex of a connected graph G. Then χ( G ) = max{χ( Gi ) | i ∈ [1. (ii) For all u. For all graphs G. Lemma 4. and let Ai . In fact. For (iii). there exists a proper colouring α : VG → [1. v} is a separating set. Of course. every k-critical graph H must have at least k vertices. Proof. a critical graph does not have cut vertices.8. Suppose each Gi has a proper k-colouring αi . We use greedy colouring to prove the claim. χ( G ) ≤ ∆( G ) + 1. then {u. m]. For (i). Proof. d H (v) ≥ k − 1 for every v ∈ H. χ( G ) − 1 ≤ δ( H ). we observe that a k-critical subgraph H ⊆ G is obtained by removing vertices and edges from G as long as the chromatic number remains k. and deﬁne α : VG → N inductively as follows: α(v1 ) = 1. By Lemma 4. for i ∈ [1. The next proof of Brook’s theorem is by L OVÁSZ (1975) as modiﬁed by B RYANT (1996). also dG (v) ≥ k − 1 for every v ∈ H. . By Theorem 4. clearly. be the connected components of G −v.6. we may take αi (v) = 1 for all i. v) = 2. Then α is proper. Let G be a 2-connected graph. In particular. and α(vi ) = min{ j | α(vt ) = j for all t < i with vi vt ∈ G } .4.5. let H ⊆ G be k-critical. ⊔ ⊓ Lemma 4. ⊔ ⊓ Brooks’ theorem For edge colourings we have Vizing’s theorem. then {u. The claim follows from this.10. Then the following are equivalent: (i) G is a complete graph or a cycle. ∆( G ) + 1] such that α(v) ≤ dG (v) + 1 for all vertices v ∈ G. . let H ⊆ G be k-critical. . we always have χ( G ) ≤ ∆( G ) + 1. but no such strong results are known for vertex colouring. because. These k-colourings give a k-colouring of G. / (iii) For all u. which proves this claim. Let VG = {v1 . if dG (u.7. . the maximum value ∆( G ) + 1 is obtained only in two special cases as was shown by B ROOKS in 1941.

. First of all. and thus that ∆( G ) = 2 as required. there exists a vertex u ∈ NG (w) ∪ {w} such / that u is adjacent to a vertex v ∈ NG (w). v ∈ NG (w) such that uv ∈ G. v ∈ W1 . vw ∈ VG −{ x. Let then G be any k-critical graph for k ≥ 2. Note that U1 ⊆ W ∪ U. Then χ( G ) = ∆( G ) + 1 if and only if either G is an odd cycle or a complete graph. assume the claim holds for k-critical graphs. Suppose now that G is neither complete nor a cycle (odd or even). then all paths from z to u must pass through x (or y. ∆(Kn ) = n − 1. This contradiction. We claim that W = {w}. v} is a → → separating set of G. Thus VG = W1 ∪ { x. Assume then that (iii) holds. and by (iii).6. Hence dG ( x. u w x v y There exists a vertex z ∈ U1 . (Since uw. then G is complete. and x (or y. But then dG (u) > d H (u). v) = 2 (the shortest path is u − w − v). Indeed. (=⇒) Assume that k = χ( G ). and χ(Kn ) = n. ⊔ ⊓ Theorem 4. where w ∈ W. there is a partition VG = W ∪ {u. and that (ii) implies (iii). Since v is not a cut vertex. Now G is connected. v} ∪ U.12 ( B ROOKS (1941)). and by (iii). It is clear that (i) implies (ii). it is 2-connected. {u. respectively). If z ∈ W (or z ∈ U. from which the theorem follows. since G is connected. This / means that dG (u. We shall show that either G is a complete graph or dG (v) = 2 for all v ∈ G. otherwise. Since w is not a cut vertex (since G has no cut vertices). y} ∪ U1 . then k = 2 = ∆( G ). and this contradicts the choice of w. χ(C2k+1 ) = 3. since G is 2-connected. We show that χ( G ) ≤ ∆( G ). Assume that w ∈ W1 . Consequently. Assume then that there are different vertices u. where all paths from W1 to U1 pass through x or y. y) = 2. and therefore there exists an edge uv ∈ G with u ∈ H and v ∈ H. dG (w) = ∆( G ). and let H ⊂ G be a k-critical proper subgraph. Since χ( H ) = k = ∆( G ) + 1 > ∆( H ). say xu ∈ G. ∆(C2k+1 ) = 2. But then dG (v) > dG (w). since H = Kn or H = Cn . Let G be connected.3 Vertex colourings 56 Proof. and hence that also u. Suppose on the contrary that |W | ≥ 2. We may suppose that G is k-critical. we must have χ( H ) = ∆( H ) + 1. and thus H is a complete graph or an odd cycle. respectively). and all paths from a vertex of W to a vertex of U go through either u or v. y}). dG (v) ≥ 2 for all v. Indeed. there exists an x ∈ W with x = w such that xu ∈ G or xv ∈ G. proves the claim. { x. (⇐=) Indeed. y} is a separating set. respectively) would be a cut vertex of G. there exists a y ∈ U such that uy ∈ G. Let w be a vertex of maximum degree. Let k = ∆( G ) + 1. If G is an even cycle. / and ∆( G ) > ∆( H ).4. Proof. If the neighbourhood NG (w) induces a complete subgraph. By Lemma 4. We need only to show that (iii) implies (i).

and. | NG (vi ) ∩ {v1 . where χ G (k) = |{α | α : VG → [1. the chromatic number problem is NP-complete. n − 1]. . . and so α(vn ) ≤ ∆( G ). This problem requires that we ﬁnd χ( G ). In graph theoretical terminology we consider the graph G = (V. In the storage problem we would like to ﬁnd a partition of the set V with as few classes as possible such that no class contains two incompatible elements. or worse. Indeed. . D EFINITION . (4.4. The chromatic polynomial of G is the function χ G : N → N. (However. and for all i ≥ 3. . . Therefore for each i ∈ [1. w ) ≥ d H ( vi+1 . vn } such / that vn = w. ⊔ ⊓ Example 4. By (4.) The chromatic polynomial A given graph G has many different proper vertex colourings α : VG → [1.3 Vertex colourings 57 By Lemma 4. and we would like to colour the vertices of G properly using as few colours as possible. . consequently. v2 ) = 2. graph theorists who will ﬁght during a conference). we ﬁnd at least one j > i such that vi v j ∈ G (possibly v j = w).5 to be certain on this point. In particular. Unfortunately. of the same colour 1. Already the problem if χ( G ) = 3 is NP-complete. Order VH = {v3 . vi−1 }| < dG (vi ) ≤ ∆( G ) . such that H = G −{v1 . v2 . v2 ∈ G with dG (v1 . . Also. This shows that G has a proper ∆( G )colouring. where vi v j ∈ E just in case vi and v j are incompatible. . and. there is an available colour for vn . α(vi ) ≤ ∆( G ) for all i ∈ [1. vn in this order as follows: α(v1 ) = 1 = α(v2 ) and α(vi ) = min{r | r = α(v j ) for all v j ∈ NG (vi ) with j < i } .3) The colouring α is proper. as we have seen. . χ( G ) ≤ ∆( G ). vn }. there exist v1 .8. v4 . wv2 ∈ G with v1 v2 ∈ G. . . see Lemma 4. for all 1 ≤ i < n. . indeed. n − 1]. some of which are not compatible (like chemicals that react with each other.8. no good algorithms are known for determining χ( G ). say v1 w. and since vn has at most ∆( G ) neighbours. Then colour v1 . w ) . . . . w = vn has two neighbours. v1 and v2 . E). v2 } is connected. . k] for sufﬁciently large natural numbers k.3). k] a proper colouring}| . Suppose we have n objects V = {v1 . . d H ( vi . the problem whether χ( G ) = 2 has a fast algorithm.

Remark. and for more nonregular graphs it should be avoided. f has no end u or v} ∪ {wx | wu ∈ G or wv ∈ G } . v2 . Let k ≥ χ(K4 ) = 4. u e v x . Therefore. in the discrete graph K 4 has no edges. χ( G ) = min{k | χ G (k) = 0} . Then χ G ( k ) = χ G − e ( k ) − χ G ∗ e ( k ). Let G be a graph. but its applications have turned out to be elsewhere. Theorem 4. Consider the complete graph K4 on {v1 . to attack the famous 4-Colour Theorem. v}) ∪ { x} is obtained from G by contracting the edge e. after which k − 1 colours are available for v2 . The vertex v1 can be ﬁrst given any of the k colours. if we can ﬁnd the chromatic polynomial of G. Therefore there are k(k − 1)(k − 2)(k − 3) different ways to properly colour K4 with k colours. and the vertices u and v are deleted. 2. when EG∗e = { f | f ∈ EG . B IRKHOFF AND L EWIS (1946). Let G be a graph. and thus any k-colouring is a proper colouring. then we easily compute the chromatic number χ( G ) just by evaluating χ G (k) for k = 1. (Of course.13 will give the tools for constructing χ G . D EFINITION . If k < χ( G ). and let e ∈ G. Then v3 has k − 2 and ﬁnally v4 has k − 3 available colours. Therefore χ K 4 ( k ) = k4 . and so χK4 (k) = k(k − 1)(k − 2)(k − 3) . and let x = x(uv) be a new contracted vertex. Example 4.13. . no loops or parallel edges are allowed in the new graph G ∗ e. The considered method for checking the number of possibilities to colour a ‘next vertex’ is exceptional. and. . until we hit a nonzero value. and by replacing all edges wu and wv by wx. indeed. The graph G ∗ e on VG∗e = (VG \ {u. v3 .4.) Theorem 4. On the other hand. v4 }.3 Vertex colourings 58 This notion was introduced by B IRKHOFF (1912). . then clearly χ G (k) = 0.9. e = uv ∈ G. Hence G ∗ e is obtained by introducing a new vertex x.

χ T (k) = χ T −e (k) − χ T ∗e (k). when we set α( x) = α(u) for the contracted vertex x = x(uv). ⊔ ⊓ Example 4. then α corresponds to a unique proper k-colouring of G. . ⊔ ⊓ Theorem 4. By Lemma 4. Suppose that n ≥ 3. We use induction on n. Indeed.9. χ T −e (k) = k · k(k − 1)n−2 . Then we have the following reductions. e G = G−e − G∗e f = G − {e. G2 . . by the induction hypothesis. Then χ T (k) = k(k − 1)n−1 . the claim is obvious. Proof. Let e = uv. Here T ∗ e is a tree of order n − 1. since there G −e and G ∗ e have less edges than G. and thus. and so Lemma 4. . Proof. namely α. then α corresponds to a unique proper k-colouring of G ∗ e. Therefore χ T ( k ) = k ( k − 1) n − 1 . For n ≤ 2. Hence the number of such colourings is χ G∗e (k). Let T be a tree of order n. χKn (k) = kn for the discrete graph. ⊔ ⊓ The connected components of a graph can be coloured independently.13. The proper k-colourings α : VG → [1.14. f } − ( G − e) ∗ f G−e . also P1 − P2 is a polynomial. (2) If α(u) = α(v). . .13.4. By Theorem 4. and for two polynomials P1 and P2 . where v is a leaf. Theorem 4.15. Gm . which together show that χ G−e (k) = χ G (k) + χ G∗e (k): (1) If α(u) = α(v).3 Vertex colourings 59 Proof. Hence the number of such colourings is χ G (k). The proof is by induction on ε G .9. Consider the graph G of order 4 from the above. Let the graph G have the connected components G1 .10. and the induction hypothesis. namely α. k] of G −e can be divided into two disjoint cases. χ Gm (k) . The claim follows from Theorem 4. The chromatic polynomial is a polynomial. . The graph T −e consists of the isolated v and a tree of order n − 1. χ T ∗e (k) = k(k − 1)n−2 . Then χ G (k) = χ G1 (k)χ G2 (k) . and let e = vu ∈ T.

For instance.4. χ G ( x) = 0 for all real numbers 0 < x < 1. The coefﬁcients alternate in sign. For instance. and so χ G (k) = χ G−e (k) − χ G∗e (k) = k(k − 1)2 (k − 2) − k(k − 1)(k − 2) = k(k − 1)(k − 2)2 = k4 − 5k3 + 8k2 − 4k .13 is time consuming. In our example. Chromatic Polynomial Problems. for 3 colours. as an exercise). we have obtained χ G−e (k) = χ G−{e. which would tell from any polynomial P(k) whether it is a chromatic polynomial of some graph. we know the chromatic polynomials for trees (and complete graphs. f } (k) − χ( G−e)∗ f (k) = k ( k − 1) 3 − k ( k − 1) 2 = k ( k − 1) 2 ( k − 2) . when G is connected. χ G (m) ≤ m(m − 1)n − 1 for all positive integers m. The coefﬁcient of kn−1 equals −ε G .3 Vertex colourings 60 Theorem 4. The constant term is 0.13 reduces the computation of χ G to the discrete graphs. The coefﬁcient of kn equals 1. R EED (1968) conjectured that the coefﬁcients of a chromatic polynomial should ﬁrst increase and then decrease in absolute value. Also. . but it seems to satisfy the general properties (that are known or conjectured) of these polynomials. there is known no characterization. It is difﬁcult to determine χ G of a given graph. However. there are 6 proper colourings of the given graph. and so there is no need to prolong the reductions beyond these. since the reduction method provided by Theorem 4. R EED (1968) and T UTTE (1974) proved that for each G of order νG = n: • • • • • • • The degree of χ G (k) equals n. the polynomial k4 − 3k3 + 3k2 is not a chromatic polynomial of any graph.

There are. especially in group theory and topology. where the lines (or continuous curves) corresponding to the edges do not intersect each other except at their ends. the surface graphs. Such graphs are used.. but may be apparent from another. . that rely on the (topological or geometrical) properties of the drawings of graphs.5 Graphs on Surfaces 5. where one tries to (or has to) avoid crossing the wires or laser beams. The geometry of the plane will be treated intuitively. when it is replaced by a path u − x − v of length two by introducing a new vertex x. the planar graphs.5 of S. We restrict ourselves in this chapter to the most natural of these. D EFINITION . if it has a plane ﬁgure P( G ). S KIENA. An edge e = uv ∈ G is subdivided. A subdivision H of a graph → → G is obtained from G by a sequence of subdivisions.4 is a planar graph. however. However. Deﬁnition D EFINITION . Indeed. e. A graph G is a planar graph. called the plane embedding of G. the algorithms are all rather difﬁcult to implement. There are fast algorithms (linear time algorithms) for testing whether a graph is planar or not. 1990.g. Also. the properties of a graph are not necessarily immediate from one representation. The complete bipartite graph K2.1 Planar graphs The plane representations of graphs are by no means unique. Addison-Wesley. “Implementing Discrete Mathematics: Combinatorics and Graph Theory with Mathematica”. in the design of electrical (or similar) circuits. Planar graphs come into use also in some parts of mathematics. A planar graph will be a graph that can be drawn in the plane so that no two edges intersect with each other. important families of graphs. a graph G can be drawn in arbitrarily many different ways. Most of them are based on an algorithm designed by A USLANDER AND PARTER (1961) see Section 6.

). Let P( G ) be a plane embedding of a planar graph G.1 Planar graphs 62 The following result is clear.) that are not on the exterior boundary are interior vertices interior edges. (v) A bridge of G belongs to the boundary of only one face. (ii) P( G ) has a unique exterior face. A face of P( G ) is an interior face.5. Vertices (edges. resp. More importantly. The edges that surround a face F constitute the boundary ∂( F ) of F. resp. The vertices (edges. Regard now each edge e = uv ∈ G as a line from u to v.1.) on the exterior boundary are called exterior vertices exterior edges.). every point x ∈ F has a disk centred at x and contained in F. The set (R × R ) \ EG is open. The exterior boundary is the boundary of the exterior face. Then F is a region. if it is bounded. (iii) Each edge e belongs to the boundary of at most two faces. F2 F1 F0 F3 Embeddings P( G ) satisfy some properties that we accepts at face value. Let F be an open set of the plane R × R. the way we draw a graph G in the plane does not change its maximum degree or its chromatic number. that is. (iv) Each cycle of G surrounds (that is. The (unique) face that is unbounded is called the exterior face of P( G ). (vi) An edge that is not a bridge belongs to the boundary of exactly two faces. We recall ﬁrst some elements of the plane geometry. and it is divided into a ﬁnite number of disjoint regions. resp. (i) Two different faces F1 and F2 are disjoint. and P( G ) one of its plane embeddings. called the faces of P( G ). resp. The boundary ∂( F ) of a region F consists of those points for which every neighbourhood contains points from F and its complement. and their boundaries can intersect only on edges.2. there are – as we shall see – some nontrivial topological (or geometric) properties that are shared by the plane embeddings. Lemma 5. its interior contains) at least one internal face of P( G ). y ∈ F can be joined by a continuous curve the points of which are all in F. D EFINITION . . For instance. Lemma 5. if any two points x. Geometric properties It is clear that the graph theoretical properties of G are inherited by all of its plane embeddings. Let G be a planar graph. A graph is planar if and only if its subdivisions are planar.

This means. there are no cycles in G. we have the following invariant property of planar graphs. and the claim holds. We shall prove the claim by induction on the number of faces ϕ of a plane embedding P( G ). in particular. it is a tree. Let e ∈ G be an edge that is not a bridge. Let P( G ) be a plane embedding of a connected planar graph such that P( G ) has ϕ faces.3. since each P( G ) has an exterior face. if and only if the connected components of G are trees.1 Planar graphs 63 If P( G ) is a plane embedding of a graph G. we have ε G = νG − 1. or inside any geometric triangle. since the two faces of P( G ) that are separated by e are merged into one face of P( G −e). If G is a planar graph of order νG ≥ 3.3. then ε G ≤ 3νG − 6. then so is any drawing P′ ( G ) which is obtained from P( G ) by an injective mapping of the plane that preserves continuous curves. If ϕ = 1.4. Theorem 5. Suppose then that the claim is true for all plane embeddings with less than ϕ faces for ϕ ≥ 2.5. Then νG − ε G + ϕ = 2 .1.2. that every planar graph has a plane embedding inside any geometric circle of arbitrarily small radius. Euler’s formula Lemma 5. where ϕ is the number of faces of P( G ). Moreover. First. ⊔ ⊓ In particular. . The next general form of Euler’s formula was proved by L EGENDRE (1794). A plane embedding P( G ) of a planar graph G has no interior faces if and only if G is acyclic.4. and the claim follows. and hence νG − (ε G − 1) + ( ϕ − 1) = 2. Proof. and let P( G ) be any of its plane embeddings. Now P( G −e) has ϕ − 1 faces. then ε G ≤ 2νG − 4. By the induction hypothesis. Proof. νG− e − ε G−e + ( ϕ − 1) = 2.1 (Euler’s formula). The subgraph G −e is planar with a plane embedding P( G −e) = P( G )−e obtained by simply erasing the edge e. and since G is connected. Then every plane embedding of G has the same number of faces: ϕ G = ε G − νG + 2 ⊔ ⊓ Maximal planar graphs Lemma 5. In this case. Corollary 5. that is. Let G be a connected planar graph. if G has no triangles C3 . Let G be a planar graph. by Lemma 5. This is clear from Lemma 5. notice that ϕ ≥ 1. by Theorem 2. then.

The induced subgraph H [K ∪ { x}] is complete. The second claim is proved similarly except that. Clearly. A planar graph G is maximal. Theorem 5. K5 and K3. each face F of P( G ) contains at least four edges on its boundary (when G is connected and ε G ≥ 4). then there is nothing to prove. then it holds for G.5. Proof. δ( G ) · νG ≤ It follows that δ( G ) ≤ 5. v∈ G ⊔ ⊓ . in this case. Suppose thus that ε G ≥ 3. and connect the vertices of K to x. Proof.1 Planar graphs 64 Proof. the case where ε G ≤ 2 is clear. since εG = i=1 ∑ ε Gi ≤ 3 ∑ νGi − 6k = 3νG − 6k ≤ 3νG − 6 . Also. Assume that the set of the boundary vertices of F induces a complete subgraph K. The edges of K are either on the boundary or they are not inside F (since F is a face. i=1 νG νG It is thus sufﬁcient to prove the claim for connected planar graphs. Then there are two nonadjacent vertices on the boundary of F. but K5 has 5 vertices and 10 edges.3 has 6 vertices and 9 edges. if an edge is removed from K3. If G is disconnected with connected components Gi . we have |K | < 4 as required. ⊔ ⊓ An upper bound for δ( G ) for planar graphs was achieved by H EAWOOD. and since H is planar. ⊔ ⊓ ∑ dG (v) = 2ε G ≤ 6νG − 12 . then δ( G ) ≤ 5.5.3 . if we remove one edge from K5 . Each face F of an embedding P( G ) contains at least three edges on its boundary ∂( F ).3 are not planar graphs. if G + e is nonplanar for every e ∈ G.2 (H EAWOOD (1890)). and if the claim holds for these smaller (necessarily planar) graphs Gi . k]. The result is a plane embedding of a graph H with VH = VG ∪ { x} (that has G as its induced subgraph). However.) Add a new vertex x inside F. By the second claim of Lemma 5.1. for i ∈ [1. Proof. If G is a planar graph. ⊔ ⊓ D EFINITION .4. By the handshaking lemma and the previous lemma.3.4. Hence 3ϕ ≤ 2ε G . The ﬁrst claim follows from Euler’s formula. / Example 5. since each edge lies on at most two faces. a planar graph of order 5 has at most 9 edges. Suppose νG ≥ 3. Theorem 5. the result is not maximal! Lemma 5. By Lemma 5. a triangle-free planar graph of order 6 has at most 8 edges. the result is a maximal planar graph. If νG ≤ 2. but K3. Let F be a face of a plane embedding P( G ) that has at least four edges on its boundary.

5 it is not planar. The theorem on planar graphs.5. Theorem 5. For a maximal planar graph G of order νG ≥ 3. Indeed. Hence 3ϕ = 2ε G . 3. and therefore they are nonplanar.W.2. and. The graphs K5 and K3.3 are the smallest nonplanar graphs. ε G = 3νG − 6 .4. 0000 0100 1010 1110 1101 1001 0001 0011 1000 1100 0010 .5 (K URATOWSKI (1930)). since there are now no bridges. that is. then an edge can be added to the graph (inside the face). see J. Proof. Historia Math. Example 5.V. then G is triangulated. The cube Qk is planar only for k = 1. L. and the graph remains to be planar. ⊔ ⊓ Kuratowski’s theorem Theorem 5. if G contains a subdivision of K5 or K3. each Qk with k ≥ 4 has Q4 as a subgraph. The subgraph of Q4 that is a subdivision of K3. if a face has a boundary of at least four edges. every face of a plane embedding P( G ) has a boundary of exactly three edges. the graph Q4 contains a subdivision of K3.3 as a subgraph. For history of the result.1 Planar graphs 65 By the previous lemma. 356 – 368. K ENNEDY.3 as a subgraph. On the other hand.M. and also independently by F RINK AND S MITH (1930). then G is not planar. AND M.1. The claim follows from Euler’s formula.3 . S YSLO.2. If G is a maximal planar graph with νG ≥ 3.3 is given below. Therefore Theorem 5. the theorem was proven earlier by P ONTRYAGIN (1927-1928). This theorem (due to K URATOWSKI in 1930) is one of the jewels of graph theory. A graph is planar if and only if it contains no subdivision of K5 or K3. Hence we have proved Corollary 5. In fact. Q UINTAS . and thus by Theorem 5. by Lemma 5. We prove this result along the lines of T HOMASSEN (1981) using 3-connectivity. 2. We prove the converse of this result in what follows.5 will give a simple criterion for planarity of graphs. Each face F of an embedding P( G ) is a triangle having three edges on its boundary. 12 (1985).

where the edges of E are exterior edges. This contradiction shows that G is 3-connected. Since G is minimal nonplanar with respect to ε G . Ak be the connected components of G −v.8. say S = { x. Then G is 3-connected. Proof. By the minimality assumption. . and thus G contains a Kuratowski graph. On the contrary. Therefore T = . . This path can be regarded as a subdivision of uv. and both G1 and G2 contain a connected component of G −S. the graph G ∗ e has a separating set S with |S| = 2. Let E ⊆ EG be the set of the boundary edges of a face F in a plane embedding of G. S would separate G already). Proof. Necessarily x ∈ S. This is a geometric proof. and this will contradict the choice of G. there are paths ⋆ u − v in G1 and G2 . where uv is an exterior edge. We can glue these plane embeddings together at v to obtain a plane embedding of G. otherwise. We show then that G is 2connected.7. . where v is an exterior vertex. (Maybe uv ∈ G. assume that v is a cut vertex of G. z} (for. if it is a subdivision of K5 or K3. then. Let G be a 3-connected graph of order νG ≥ 5. they have plane embeddings. Proof. G is connected. This will map the given face as the exterior face of the image plane embedding. by the ⋆ minimality of G. H1 H2 It is now easy to glue H1 and H2 together on the edge uv to obtain a plane embedding of G + uv.1 Planar graphs 66 D EFINITION . and let A1 . v}. On the contrary suppose that for any e ∈ G. Let G be a nonplanar graph without Kuratowski graphs of minimal total size ε G + νG . and let x = x(uv) be the contracted vertex. and so. . Then there exists a plane embedding P( G ). ⊔ ⊓ Lemma 5. ⊔ ⊓ Lemma 5.) If both H1 and H2 are planar. Indeed. We conclude that H1 or H2 is nonplanar. Now ε H1 < ε G .5. say H1 . by Lemma 5. Since G is 2-connected (by the above). both u and v are adjacent to a vertex of each connected → component of G −S. Let G1 and G2 be any subgraphs of G such that EG = EG1 ∪ EG2 . Then use geometric inversion with respect to this circle. S = VG1 ∩ VG2 . since G2 ⊆ H2 . Let e = uv. and thus of G. Assume then that G has a separating set S = {u. H1 contains a Kuratowski graph H.3 . A graph G is called a Kuratowski graph. there is a path u − v → in H2 . Let Hi = Gi + uv. However. Choose a circle that contains every point of the plane embedding (including all points of the edges) such that the centre of the circle is inside the given face.6. Lemma 5. Then there exists an edge e ∈ G such that the contraction G ∗ e is 3-connected.6. the subgraphs Gi = G [ Ai ∪ {v}] have plane embeddings A1 A2 P( Gi ).

By Lemma 5. Consider . There exists a vertex y ∈ A with zy ∈ G. there exists a path P : u − y′ in G −{z. and this contradicts / the choice of A. Assume that e and S are chosen such that G − T has a connected component A with the least possible number of vertices. w}. where H is a subdivision of K5 . The only 3-connected graph of order 4 is the planar graph K4 . If d H ( x) = 2. since y ∈ B. G contains a subdivision of K3.8.9. Lemma 5. We do not consider the details of these cases. and hence G ∗ e has a plane embedding P( G ∗ e) by the induction hypothesis. v ∈ B. If there exists at most one edge xy ∈ H such that uy ∈ G (or vy ∈ G). Suppose then that d H ( x) = 3 or 4. a Kuratowski graph cannot be created by contractions. In this case. and hence → this P goes through y. that is. The proof is by induction on νG . then the claim is obviously true. v} would separate G. Proof.3 .10. and both u and v have 3 neighbours in the subgraph of G corresponding to H. ⊔ ⊓ v2 x v1 v3 v1 v4 v2 u v v3 v4 Lemma 5. since G is 3-connected.9. There remains only one case. y. then so does G. there exists a vertex w such that R = {z. v.1 Planar graphs 67 {u. Let G be a graph. G ∗ e has no Kuratowski subgraphs. The proof consists of several cases depending on the Kuratowski graph. If for some e ∈ G the contraction G ∗ e has a Kuratowski subgraph.) The graph G ∗ (zy) is not 3connected by assumption. Proof. Every 3-connected graph G without Kuratowski subgraphs is planar. B A y u T v z Since uv ∈ G. where x = x(uv) is the contracted vertex for e = uv. Hence | B| < | A|. By Lemma 5. It can be that w ∈ {u. and so B ⊆ A. For each / ⋆ y′ ∈ B. as in the above. then one easily sees that G contains a Kuratowski graph. (Otherwise {u. G − R has a connected component B such that u. and how the subdivision is made. there exists an edge e = uv ∈ G such that G ∗ e (with a contracted vertex x) is 3-connected. and hence. Therefore y′ is connected to y also in G − T. Let H be a Kuratowski graph of G ∗ e. but by symmetry we can suppose that w = u. w} separates G. v}. The inclusion is proper. Therefore we can assume that νG ≥ 5.5. y′ ∈ A. ⊔ ⊓ By the next lemma. z} separates G.

G has a plane embedding (obtained from P( G )−u by putting u inside the triangle (v.2 Colouring planar graphs 68 the part P( G ∗ e)− x. For the equality. by (1). . y} form a subdivision of K3. Add a new vertex at each crossing.7. v1 . z. v j }.5. For this we need to show that ×(K6 ) ≥ 3. This contradiction proves the claim. ⊔ ⊓ Example 5. z ∈ {vi . ⊔ ⊓ u v Proof of Theorem 5. Also. Now. Therefore. Assume.5. y (2) Assume there are y. by symmetry. P( G ∗ e)− x is a plane embedding of G −{u. Then. for instance. (1) If NG (u) \ {v} ∈ Pi. / / {u.10.1. vk } in order along the cycle C. . vi+1 ) and by drawing the edges with an end u inside this triangle).i+1 (i + 1 is taken modulo k) for some i. Any graph G can be drawn in the plane so that three of its edges never intersect at the same point. The crossing number ×( G ) is the minimum number of intersections of its edges in such plane drawings of G. . NG (u) \ {v} = NG (v) \ {u} by the assumption dG (v) ≤ dG (u). vi . z ∈ NG (u) \ {v} such that y ∈ Pij and z ∈ Pij for some i and j.2 Colouring planar graphs The most famous problem in the history of graph theory is that of the chromatic number of planar graphs. On the contrary. Here C is a cycle of G (since G is 3-connected). suppose that G is a nonplanar graph that has a minimal size ε G such that G does not contain a Kuratowski subgraph. v}. The problem was known as the 4-Colour Conjecture for more than 120 years. then. by Lemma 5. vi+1 } ∪ {v. we can assume that NG (u) \ {v} ⊆ NG (v). clearly. since ε G = 2c + 15 ≤ 3(c + 6) − 6 = 3νG − 6. v3 give a subdivision of K5 . Now c ≥ 3. But now u. Now since G −{u. that dG (v) ≤ ⋆ dG (u). one is invited to design a drawing with exactly 3 crossings. we need to show that each nonplanar graph G contains a Kuratowski subgraph. and by Lemma 5. . until it was solved by A PPEL AND H AKEN in 1976: if G is a . We show that ×(K6 ) = 3. it is planar. dG (v) = dG (u) > 3. We obtain a plane embedding of G −u by drawing (straight) edges vvi for 1 ≤ i ≤ k. G is 3-connected.j : vi − v j → be the path along C from vi to v j . By Theorem 5. v} = ( G ∗ e)− x. 5.3. and. and let C be the boundary of the face of P( G ∗ e)− x containing x (in P( G ∗ e)). Therefore G is planar if and only if ×( G ) = 0.3. ×(K5 ) = 1. z u v By (1) and (2). Let Pi.3 and Lemma 5. v2 . v. where y. This results in a planar graph G on c + 6 vertices and 2c + 15 edges. v2 . vi . and NG (u) ⊆ VC ∪ {v} and NG (v) ⊆ VC ∪ {u}. Let X (K6 ) be a drawing of K6 using c crossings so that two edges cross at most once. Let NG (v) \ {u} = {v1 .

7 (4-Colour Theorem). ⊔ ⊓ The ﬁnal word on the chromatic number of planar graphs was proved by A PPEL AND H AKEN in 1976. δ( H ) ≥ k − 1. If G is a planar graph. Proof. The 4-Colour Conjecture has had a deep inﬂuence on the theory of graphs during the last 150 years. then χ( G ) ≤ 5. the claim holds for νG ≤ 6. The proof is by induction on νG . Theorem 5. because G is 6-critical. Suppose the claim does not hold. Recall that for k-critical graphs H.2.2 Colouring planar graphs 69 planar graph. Theorem 5. Proof. then χ( G ) ≤ 4. j]. and thus there exists a vertex v with dG (v) = δ( G ) ≥ 5. then χ( G ) ≤ 4. If G is a planar graph. v4 lies inside the region enclosed by the cycle C. and so χ( G ) ≤ 6. By the following theorem. there is a colour i available for v in the 6-colouring of G −v. and let G be a 6-critical planar graph. then χ( G ) ≤ 6. χ( G −v) ≤ 6. Lemma 5. Now. By Theorem 5. ⊔ ⊓ The proof of the following theorem is partly geometric in nature. the path P24 must meet C at some vertex of C. The 5-colour theorem We prove H EAWOOD’s result (1890) that each planar graph is properly 5-colourable. Let α be a proper 5-colouring of G −v. exactly one of v2 . since the vertices of P24 are coloured by 2 and 4. j] (for. ⋆ Let Pij : vi − v j be a path in G [i. v5 v v4 v1 v2 v3 P13 Consider the subgraph G [i. The vertices vi and v j are in the same connected component of G [i. Since dG (v) ≤ 5. . but C contains no such colours. dG (v) = 5. Such a colouring exists. a planar graph G has a vertex v with dG (v) ≤ 5. there exists a neighbour vi ∈ NG (v) such that α(vi ) = i. By assumption. χ( G ) > 5.6 (H EAWOOD (1890)). and it requires the assistance of a computer. By the geometric → assumption.2. 5]. Clearly. Suppose these neighbours vi of v occur in the plane in the geometric order of the ﬁgure. If G is a planar graph. This is a contradiction.5. otherwise we interchange the colours i and j in the connected component containing v j to obtain a recolouring of G. and let C = (vv1 ) P13 (v3 v). each planar graph can be decomposed into two bipartite graphs. and then recolour v with the remaining colour j). The solution of the 4-Colour Theorem is difﬁcult. and therefore for each i ∈ [1. By Theorem 5. where vi and v j have the same colour i. j] ⊆ G made of colours i and j. By the induction hypothesis.11. since G is planar.

The illustrated map can be 4-coloured. E) be a 4-chromatic graph. . . Using this notion of a dual graph. Fn be the countries of a map M. because a bridge in such a map would be a boundary inside a country. . These curves divide the plane into regions. consider ﬁnitely many simple closed curves in the plane. and it cannot be coloured using only 3 colours.5. they are between the sets V1 and V3 . It is clear that (V. As an example. Let E2 be the rest of the edges. Let G = (V. V2 and V4 . and we wish to colour each country so that no neighbouring countries obtain the same colour. Let F1 . and deﬁne a graph G with VG = {v1 . . vn } such that vi v j ∈ G if and only if the countries Fi and Fj are neighbours. E2 ) are both bipartite. E1 ) and (V. and two neighbouring countries share a common edge (not just a single vertex). that is. Proof.8. since the sets Vi are stable. where the maps are deﬁned by curves in the plane. . Map-colouring can be used in rather generic topological setting. . ⊔ ⊓ Map colouring∗ The 4-Colour Conjecture was originally stated for maps. because every two faces have a common border. we deﬁne a map as a connected planar (embedding of a) graph with no bridges. χ( G ) ≤ 4. How many colours are needed? A border between two countries is assumed to have a positive length – in particular. The deﬁne E1 as the subset of the edges of G that are between the sets V1 and V2 . The regions are 2-colourable. E2 ) are bipartite. In the map-colouring problem we are given several countries with common borders. The edges of this graph represent the boundaries between countries. we can state the map-colouring problem in new form: What is the chromatic number of a planar graph? By the 4-Colour Theorem it is at most four. countries that have only one point in common are not allowed in the map colouring. V3 and V4 . The map-colouring problem is restated as follows: How many colours are needed for the faces of a plane embedding so that no adjacent faces obtain the same colour. V2 and V3 . v2 . . . Formally. E1 ) and (V. Let Vi = α−1 (i ) be the set of vertices coloured by i in a proper 4-colouring α of G.2 Colouring planar graphs 70 Theorem 5. Hence a country is a face of the map. V1 and V4 . We deny bridges. . Then the edges of G can be partitioned into two subsets E1 and E2 such that (V. F2 . It is easy to see that G is a planar graph.

To see this. he proved that if the number of edges around each region is divisible by 3. As we shall see in the next chapter. otherwise. but also some further errors. colour it by 2. In a triangulation. the graph where the vertices correspond to the regions. In 1879 C AYLEY pointed out some difﬁculties that lie in the conjecture. One can triangulate any planar graph G (drawn in the plane). a given set S of conﬁgurations can be proved to be unavoidable. A conﬁguration is said to be reducible. B IRKHOFF introduced one of the basic notions (reducibility) needed in the proof of the 4CC. we have ∑(6 − dG (v)) = 12. The ﬁnal notion for the solution was due to H EESCH. if the region is inside an odd number of curves. . a conﬁguration is a part that is contained inside a cycle. This consists of assigning to a vertex v the charge 6 − dG (v). The same year A LFRED K EMPE published a paper. is bipartite. to 39 by O RE AND S TEMPLE (1970). and the edges correspond to the neighbourhood relation.5.2 Colouring planar graphs That is. TAIT published two papers on the 4CC in the 1880’s that contained clever ideas. a student of D E M ORGAN at University College of London. to 35 by W INN (1940). The basic idea in K EMPE’s argument (known later as Kempe chains) was the same as later used by H EAWOOD to prove the 5-Colour Theorem. and. v Now. Also. colour a region by 1. For more than 10 years K EMPE’s proof was considered to be valid. where he claimed a proof of the 4CC. For instance. (Theorem 5. by adding edges to divide the faces into triangles. if for a triangulation. to 95 by M AYER (1976). then the map is 4-colourable. During the following 120 years many outstanding mathematicians tried to solve the problem. and some of them even thought that they had been successful. and in 1922 F RANKLIN showed that the 4CC holds for maps with at most 25 regions. In 1890 H EAWOOD showed that K EMPE’s proof had serious gaps. H EAWOOD discovered the number of colours needed for all maps on other surfaces than the plane. From Euler’s formula we see that for the sum of the charges. if it is not contained in a triangulation of a minimal counter example to the 4CC. The search for avoidable sets began in 1904 with work of W EINICKE. who in 1969 introduced discharging.6). This number was increased to 27 by R EYNOLDS (1926). that does not contain a conﬁguration from S. 2 1 2 2 2 1 1 2 1 2 1 71 1 History of the 4-Colour Theorem That four colours sufﬁce planar maps was conjectured around 1850 by F RANCIS G UTHRIE. one can ‘redistribute’ the charges so that no v comes up with a positive charge. An unavoidable set is a set of conﬁgurations such that any triangulation must contain one of the conﬁgurations in the set.

x2 . We can assume that the planar graph G is connected. The following book contains the ideas of the proof of the 4-Colour Theorem.L. R UBIN AND TAYLOR (1979) that χ ℓ ( G ) ≤ ∆( G ) + 1 . They based the proof on reducibility using Kempe chains. and that it is given by a near-triangulation. Let Λ consist of lists such that |Λ(v)| = 3 for all v ∈ C. but equality does not hold in general. (If the . it was shown by V OIGT (1993) that there exists a planar graph with χℓ ( G ) = 5. List colouring D EFINITION . it was proved by V IZING (1976) and E RDÖS .10. S ANDERS . y2 . {1.9 (T HOMASSEN (1994)). There were difﬁculties with his approach that were solved in 1976 by A PPEL AND H AKEN. In fact. However. (K OCH assisted with the computer calculations. 3} {2. Y ).2 Colouring planar graphs 72 According to H EESCH one might be satisﬁed with a set of 8900 conﬁgurations to prove the 4CC. since proper colourings are special cases of list colourings. |Λ(v)| = k}. if each vertex v gets a colour from its list. At the moment. 1986. an embedding. The lists for the vertices shown in the ﬁgure show that χℓ (K3.3 is not 2choosable. Then χℓ ( G ) ≤ 5. let the bipartition of K3. Theorem 5. Example 5. where X = { x1 . if χℓ ( G ) ≤ k. and |Λ(v)| = 5 for all v ∈ C.3 ) > 2. 2} {1. Let G be a graph so that each of its vertices v is given a list (set) Λ(v) of colours.4. / Proof.5. G is k-choosable. Dover. 3} Obviously χ( G ) ≤ χℓ ( G ). S AATY AND P. where the interior faces are triangles. T HOMASSEN proved a stronger statement: Theorem 5. For planar graphs we do not have a ‘4-list colour theorem’.C. A proper colouring α : VG → [1. Then G has a Λ-list colouring α. Indeed. Because of these simpliﬁcations also the computer time is much less than in the original proof. and ended up with an unavoidable set with over 1900 conﬁgurations and some 300 discharging rules. m] of G is a (Λ-)list colouring. “The Four-Color Problem”. Let G be a planar graph. 3} y3 x1 x2 x3 {1. The list chromatic number χℓ ( G ) is the smallest integer k such that G has a Λ-list colouring for all lists of size k. x3 } and Y = {y1 . K AINEN. 3} y2 {2. α(v) ∈ Λ(v).) A simpliﬁed proof by R OBERTSON . Also. the smallest such a graph was produced by M IRZAKHANI (1996). T. S EYMOUR AND T HOMAS (1997) uses 633 conﬁgurations and 32 discharging rules.3 be ( X. Let G be a planar graph and let C be the cycle that is the boundary of the exterior face. y3 }. Indeed. and it is of order 63. The bipartite graph K3. 2} y1 {1. The proof used 1200 hours of computer time.

(Thus we prove a stronger statement than claimed. vk . We deﬁne new lists for H as follows: Λ′ (vi ) ⊆ Λ(vi ) \ {r. The result is known as Fáry’s Theorem. Let x ∈ C be a vertex. and it was rediscovered independently by A NDREEV (1970) and T HURSTON (1985). → Since |Λ(v)| = 3. . y}.2 Colouring planar graphs 73 boundary of a face has more than 3 edges. Note that the exterior boundary is unchanged by a triangulation of the interior faces. For the vertex v. . then we can add an edge inside the face. Theorem 5.) For νG ≤ 3.) This is because adding edges to a graph can only make the list colouring more difﬁcult. This gives a Λ′ -list colouring for G. where the edges are straight lines. Let v ∈ C be a vertex ⋆ adjacent to x. and otherwise Λ′ (z) = Λ(z). ⋆ ary of H is the cycle x → v1 → · · · → vk → y − x.12 (K OEBE (1936)). . s ∈ Λ(v) that differ from the colour c of x.11 (WAGNER (1936)). Can all planar graphs be drawn in the plane such that the edges are straight lines of integer lengths? We say that two circles kiss in the plane. ⊔ ⊓ v1 Straight lines and kissing circles∗ We state an interesting result of WAGNER.5.. . we choose α(v) = r or s such that α(v) = α(y). . the proof of which can be deduced from the above proof of Kuratowski’s theorem. · · · · · · · · Theorem 5. H has a Λ′ -list colouring α. if they intersect in one point and their interiors do not intersect. The exterior bound. vk v2 Consider the subgraph H = G −v. For a set of circles. we draw a graph by putting an edge between two midpoints of kissing circles. The following improvement of the above theorem is due to K OEBE (1936). s} such that |Λ′ (vi )| = 3 for each i ∈ [1. Since Λ′ (z) ⊆ Λ(z) for all z. there are two colours r. Now νH = νG − 1. in which case xy ∈ G. C : v → x − v. k]. This raises a difﬁcult problem: Integer Length Problem. y}. A planar graph G has a plane embedding. x It can be that NG (v) = { x. A graph is planar if and only if it is a kissing graph of circles. for which we ﬁx a colour c ∈ Λ( x). → Let NG (v) = { x. and by the induction hypothesis (with c still ﬁxed for x). where y ∈ C. The proof is by induction on νG under the additional constraint that one of the vertices of C has a ﬁxed colour. Suppose then that νG ≥ 4. we have that α is a Λ-list colouring of G. v1 .. that is. and vi are v y ordered such that the faces are triangles as in the ﬁgure. the claim is obvious.

It turns out to be a well-quasi-order. A recent result of R OBERTSON AND S EYMOUR (1983-2000) on graph minors is (one of) the deepest results of graph theory. but this is not the case. A graph H is a minor of G. −1. +1 A positive answer to this 4-slope problem for planar graphs would prove the 4-Colour Theorem. E). consider a set S of curves of the plane (that are continuous between their end points). It is known that all planar graphs are string graphs (this is a trivial result). At ﬁrst it might seem that every graph is a string graph. = G imply H G. that is. that is. G and G L and L H imply G ∼ H. . The string graph of S is the graph G = (S. −1 0 ∞ The Minor Theorem∗ D EFINITION . where uv ∈ E if and only if the curves u and v intersect.13 is around 500 pages long. .2 Colouring planar graphs 74 Graphs can be represented as plane ﬁgures in many different ways.5. Indeed. the proof of Theorem 5. G2 . For this. H G. every inﬁnite sequence G1 . The proof goes beyond these lectures. Is every planar graph a line segment graph for some set L of lines? Note that there are also nonplanar graphs that are line segment graphs. The following properties of the minor relation are easily established: (i) G (ii) H (iii) H G. e G a subgraph a contraction Note that every subgraph H ⊆ G is a minor. The above question remains open even in the case when the slopes of the lines are +1. The conditions (i) and (iii) ensure that the relation is a quasi-order. all complete graphs are such graphs. A graph is a line segment graph if it is a string graph for a set L of straight line segments in the plane. it is reﬂexive and transitive. of graphs has two graphs Gi and Gj with i < j such that Gi Gj . 0 and ∞. Line Segment Problem. if H is isomorphic to a graph obtained from a subgraph of G by successively contracting edges. Indeed. . denoted by H G. .

15 (Minor Theorem 2). Each property P of graphs deﬁnes a family of graphs. D EFINITION . this algorithm is not known! Hadwiger’s Problem. More importantly. the family of those graphs that satisfy this property. A graph G is nonplanar if and only if K5 G. (3) acyclic graphs. . namely. H ADWIGER conjectured in 1943 that for every graph G. the family of trees is not closed under taking subgraphs. if all minors of a graph G satisfy P whenever G does. This problem is solved by the above results. An old problem asks if there exists an algorithm that would determine whether a graph can be drawn so that its cycles do not form (nontrivial) knots. However. R OBERTSON AND G or K3.14 (WAGNER (1937)). The following families of graphs are minor closed: the family of (1) all graphs. since the property ‘knotless’ is inherited by minors: there exists a fast algorithm to do the job.3 ⊔ ⊓ S EYMOUR (1998) proved the Wagner’s conjecture: Theorem 5. if χ( G ) ≥ r.15. (2) planar graphs (and their generalizations to other surfaces). Let P be a property of graphs inherited by minors. The conjecture is trivial for r = 2. and it is known to hold for all r ≤ 6. if every minor H of a graph G ∈ F is also in F . The cases for r = 5 and 6 follow from the 4-Colour Theorem. Every graph can be drawn in the 3-dimensional space without crossing edges. In particular. A family F of graphs is said to be minor closed. one of the graphs is a (proper) minor of another. Kχ( G) G. Then there exists a ﬁnite set F of graphs such that G satisﬁes P if and only if G does not have a minor from F . then G has a complete graph Kr as its minor. One of the impressive application of Theorem 5.13 (Minor Theorem). However. A property P of graphs is said to be inherited by minors. WAGNER proved a minor version of Kuratowski’s theorem: Theorem 5. the subgraph order of trees (T1 ⊆ T2 ) is not a well-quasi-order. By Theorem 5. that is. Proof.15 concerns embeddings of graphs on surfaces. see the next chapters. in any inﬁnite family F of graphs.2 Colouring planar graphs 75 Theorem 5. The minor order is a well-quasi-order on graphs. The acyclic graphs include all trees.5. and thus it is not minor closed. Exercise. one can test (with a fast algorithm) whether a graph can be embedded onto a surface.

Theorem 5. Every compact surface has a triangulation. We shall study the ‘easy surfaces’ – those that are compact and orientable. y) ∈ R of its points. sets of points) A and B are topologically equivalent (or homeomorphic) if there exists a bijection f : A → B such that f and its inverse f −1 : B → A are continuous. if X is connected (there is a continuous line inside X between any two given points) and every point x ∈ X has a neighbourhood that is topologically equivalent to an open planar disk D ( a) = { x | dist ( a. two ﬁgures are topologically equivalent if one can be deformed to the other by bending.5. m. if X is closed and bounded. There are quite many interesting surfaces many of which are rather difﬁcult to draw. i = 1. . and shrinking without tearing it apart or gluing any of its parts together. and can be entirely characterized by the number of holes in them. All these deformations should be such that they can be undone. Two ﬁgures (i. is a triangulation of X if X = ∪m 1 Ti and any nonempty intersection Ti ∩ Tj with i = j is either a vertex or i= an edge. A compact surface X is orientable if and only if it has no subsets that are topologically equivalent to the Möbius band. The following is due to R ADÓ (1925). Background on surfaces We shall ﬁrst have a quick look at the general surfaces and their classiﬁcation without going into the details. A triangulation is said to be oriented if the triangles are assigned orientations such that common edges of two triangles are always oriented in reverse directions.16. . since it it not bounded. There are also non-orientable compact surfaces such as the Klein bottle and the projective plane. .. squeezing.3 Genus of a graph A graph is planar. Theorem 5. Consider the space R3 . Equivalently. A ﬁnite set of triangles Ti .17. orientability can be described as follows. We deal with surfaces of the real space.3 Genus of a graph 76 5. . if it can be drawn in the plane without crossing edges. A set of points X is a surface.e. A plane is an important special case of a surface. In this section we study shortly drawing graphs in other surfaces. . Such a permutation induces a direction for the edges of the triangle. which has its (usual) distance function d( x. stretching. A surface is orientable if it admits an oriented triangulation. These are surfaces that have both an inside and an outside. A subset of a compact surface X is a triangle if it is topologically equivalent to a triangle in the plane. x) < 1}. and in this case a surface X is compact. In particular. Each triangle of a surface can be oriented by choosing an order for its vertices up to cyclic permutations. 2. This number is the genus of the surface. Note that the plane is not compact.

. It is often difﬁcult to imagine how a ﬁgure (say. (Such a deformation is not allowed by topological equivalence. a compact surface is obtained by gluing together the edges having the same label in the direction that they have. then it is topologically equivalent to a sphere S = S0 or a connected sum of tori: Sn = S1 #S1 # . (ii) if X is nonorientable. . result due to R ADÓ (1920). where S1 is a torus. a graph) can be drawn in a surface. .18 (D EHN AND H EEGAARD (1907)).) The next result is known as the classiﬁcation theorem of compact surfaces. and • each edge is given an orientation (clockwise or counter clockwise).) A plane model. which consists of a polygon in the plane such that • each edge of the polygon is labelled by a letter. Given a plane model M. . #P for some n ≥ 1. (This corresponds to the Möbius band. It is nonoriented if and only if. and thus a compact surface. and concatenating the labels with the convention that a−1 is chosen if the direction of the edge is counter clockwise.3 Genus of a graph In the Möbius band (which itself is not a surface according the above deﬁnition) one can travel around and return to the starting point with left and right reversed. for some label a. where P is a projective plane. and difﬁcult to prove. the edges labelled by a have the same direction when read clockwise. There is a helpful. the sphere is represented by the word abb−1 a−1 . Hence. can also be represented by a (circular) word by reading the model clockwise. #S1 for some n ≥ 1. stating that every compact surface (orientable or not) has a description by a plane model. then X is topologically equivalent to a connected sum of projective planes: Pn = P#P# . Let X be a compact surface. • each letter is a label of exactly two edges of the polygon. the Klein bottle by aba−1 b and the projective plane by abb−1 a. . Theorem 5. Then (i) if X is orientable. a a b Sphere b b a Torus a b b a b a a Klein bottle a b b Projective plane From a plane model one can easily determine if the surface is oriented or not. 77 A connected sum X#Y of two compact surfaces is obtained by cutting an open disk off from both surfaces and then gluing the surfaces together along the boundary of the disks.5. the torus by aba−1 b−1 .

5. have many other plane models and representing words as well. A word representing a connected sum of two surfaces. Another way to see this is to use a projection of the sphere to a plane: ¡ ¢£ . if G can be drawn in S without crossing edges. According to the next theorem a sphere has exactly the same embeddings as do the plane.5. as do the other surfaces. 78 Klein bottle Drawing a graph (or any ﬁgure) in a surface can be elaborated compared to drawing in a plane model. if a graph can be embedded in one sphere. In the one direction the claim is obvious: if G is a planar graph. then it can be drawn in a bounded area of the plane (without crossing edges). e3 b e2 e1 e1 e2 b e3 e4 a e5 Sphere D EFINITION .3 Genus of a graph These surfaces. if S is a surface. In this process no crossings of edges can be created. represented by words W1 and W2 . then a graph G has an S-embedding. then there is a small area of the sphere. where a line that enters an edge of the polygon must continue by the corresponding point of the other edge with the same label (since these points are identiﬁed when we glue the edges together). then it can be embedded in any sphere – the size of the sphere is of no importance. By studying the relations of the representing words. Theorem 5. where there are no points of the edges. On the other hand. and this bounded area can be ironed on the surface of a large enough sphere. Let S0 be (the surface of) a sphere. where the lines enter and leave the edges of the plane model. Recall that in the plane model for the Klein bottle the vertical edges of the square have the same direction. is obtained by concatenating these words to W1 W2 . In general. and hence G is planar. if G is embeddable in a sphere S0 . e4 a e5 Example 5. and stretch it open until it looks like a region of the plane. The black dots indicate. We then puncture the sphere at this area.18 can be proved. On the right we have drawn K6 in the Klein bottle. Clearly.

6. ¡ ¢£ ¡ ¡ ¡ ¢£ ¢£ ¢£ ¡ ¡ ¢£ ¢£ 1 1 5 2 6 3 7 4 1 1 Example 5. A graph G has an S0 -embedding if and only if it is planar. K7 has genus 1 as can be seen from the plane model (of the torus) on the right. 79 Therefore instead of planar embeddings we can equally well study embeddings of graphs in a sphere. Torus Consider next a surface which is obtained from the sphere S0 by pressing a hole in it. Also. then it can be drawn in any one of the above surfaces without crossing edges. a planar graph drawn in a sphere has no exterior face – all faces are bounded (by edges).19. In topological terms the surface has been distorted by a continuous transformation. its embeddability properties will remain the same. This is a torus S1 (or an orientable surface of genus 1). The S1 -embeddable graphs are said to have genus equal to 1. If a sphere is deformed by pressing or stretching. Most importantly. . If a graph G is S1 -embeddable. since the sphere is closed and it has no boundaries.5.3 have genus 1. This is sometimes convenient. The smallest nonplanar graphs K5 and K3.3 Genus of a graph Theorem 5. Sometimes it is easier to consider handles than holes: a torus S1 can be deformed (by a continuous transformation) into a sphere with a handle.

Theorem 5. For K5 . consider a graph G and any of its plane (or sphere) drawing (possibly with many crossing edges) such that no three edges cross each other in the same point (such a drawing can be obtained). The same handle can be utilized by several edges. clearly g( G ) ≤ n. in particular. This result has an easy intuitive veriﬁcation. We deﬁne the genus g( G ) of a graph G as the smallest integer n. ¡¢ ¡¢ £¤¥ £¤¥ D EFINITION . The drawing of an S4 can already be quite complicated. and thus the genus g( G ) of G exists. ¡¢ £¤¥ ¡¢ £¤¥ . We should note that the above argument does not determine g( G ).5. For planar graphs. However. Indeed. we have g( G ) = 0.3 Genus of a graph 80 Genus Let Sn (n ≥ 0) be a sphere with n holes in it. g(K3. The next theorem states that any graph G can be embedded in some surface Sn with n ≥ 0. g(K4 ) = 0.20. for which G is Sn -embeddable. because we do not put any restrictions on the places of the holes (except that we must not tear the surface into disjoint parts). However. once again an Sn can be transformed (topologically) into a sphere with n handles. Also. and. since K5 is nonplanar. but is embeddable in a torus.3 ) = 1. we have g(K5 ) = 1. At each of these crossing points create a handle so that one of the edges goes below the handle and the other uses the handle to cross over the ﬁrst one. only that G can be embedded in some Sn . Every graph has a genus.

1 1 ε G − (νG − 2) . ⊔ ⊓ Theorem 5.22 (Y OUNGS (1963)).12. as before.3 Genus of a graph 81 Euler’s formula with genus∗ The drawing of a planar graph G in a sphere has the advantage that the faces of the embedding are not divided into internal and external. then clearly g( H ) ≤ g( G ). The faces of an embedding of a connected graph G in a surface of genus g( G ) are 2-cells. The complete graph K4 can be embedded in a torus such that it has a face that is not a 2-cell. g( G ) ≥ 2. and the genus of the torus is 1. Proof. Lemma 5. By the previous lemma. If G is a planar graph.23.5. Now every face has a border of at least three edges. 3ϕ G ≤ 2ε G . then g( G ) = 0.21. . Theorem 5. In particular. For a connected G with νG ≥ 3 we have 3ϕ G ≤ 2ε G . ϕ G = ε G − νG + 2 − 2g( G ). We deduce that K8 cannot be embedded in the surface S1 of the torus. then νG − ε G + ϕ G = 2 − 2g( G ) . let G = K8 . a face of an embedding of G in Sn (with g( G ) = n) is a region of Sn surrounded by edges of G. If νG = 3. If H ⊆ G. and by the generalized Euler’s formula. every nonbridge is on the boundary of exactly two faces. In this case νG = 8. Let again ϕ G denote the number of faces of an embedding of G in Sn . if every simple closed curve (that does not intersect with itself) can be continuously deformed to a single point. We omit the proof of the next generalization of Euler’s formula. In this case we need the knowledge that ϕ G is counted in a surface that determines the genus of G (and in no surface with a larger genus). and. Combining these we obtain that 3ε G − 3νG + 6 − 6g( G ) ≤ 2ε G . ⊔ ⊓ g( G ) ≥ By this theorem. and the above formula is the Euler’s formula for planar graphs. and the claim follows. since H is obtained from G by omitting vertices and edges. The external face of G becomes an ‘ordinary face’ after G has been drawn in S0 . We omit the proof of the general condition discovered by Y OUNGS: Theorem 5. As an example. But this is because g(K4 ) = 0. In general. If G is a connected graph. A face of an embedding P( G ) in a surface is a 2-cell. ε G = 28. D EFINITION . Assume thus that νG ≥ 4. we can compute lower bounds for the genus g( G ) without drawing any embeddings. and so 5 g( G ) ≥ 3 . For a connected G with νG ≥ 3. 6 2 Proof. then the claim is trivial. Since the genus is always an integer.

The number of edges in Kn is equal to ε G = 1 n(n − 1). g( G ) ≤ g(Kn ). 12 Therefore g(K6 ) = ⌈3 · 2/12⌉ = ⌈1/2⌉ = 1. then χ( G ) ≤ 7+ 1 + 48g . For the complete bipartite graphs.3 Genus of a graph Lemma 5.5.23. we 2 obtain g(Kn ) ≥ (1/6)ε G − (1/2)(n − 2) = (1/12)(n − 3)(n − 4) . g(K7 ) = 1. Theorem 5. 12 Proof.25. χ( G ) ≤ 4. g(Km. If n ≥ 3. By Theorem 5.28 (H EAWOOD ). but g(K8 ) = 2. H EAWOOD proved it ‘only’ for g ≥ 1. . If g( G ) = g ≥ 1. If n ≥ 3. For all graphs G of order n ≥ 3. we know the exact genus for the complete bipartite graphs: Theorem 5.26. Using the result of R INGEL AND Y OUNGS and some elementary computations we can prove that the above theorem is the best possible. By Theorem 5. g( G ) ≤ (n − 3)(n − 4) . 12 Also. then g( K n ) ≥ 82 (n − 3)(n − 4) .n ) = (m − 2)(n − 2) .13. then g( K n ) = (n − 3)(n − 4) .27 ( R INGEL (1965)).24 (H EAWOOD (1890)). it is surprising that the generalization of the 4-Colour Theorem for genus ≥ 1 is much easier. the proof of the 4-Colour Theorem. Also.25 (R INGEL AND Y OUNGS (1968)). 2 Notice that for g = 0 this theorem would be the 4-colour theorem. For a graph G of order n. For the complete graphs Kn a good lower bound was found early. ⊔ ⊓ This result was dramatically improved to obtain Theorem 5. H EAWOOD proved a hundred years ago: Theorem 5. is extremely long and difﬁcult. 4 Chromatic numbers∗ For the planar graphs G. Theorem 5. This in mind.

By C ONWAY and G ORDON in 1983 every spatial embedding of K6 contains linked cycles. . Moreover. A spatial embedding of a graph G is said to have linked cycles. and χ( G ) ≤ ⌊(7 + 1 + 48g)/2⌋ = 7. and it contains K6 and the Petersen graph. Johns Hopkins.29. “Graphs on Surfaces”. then g( G ) = 1.5. T HOMASSEN. for G = K7 we have that χ(K7 ) = 7 and g(K7 ) = 1. S EYMOUR AND T HOMAS (1993) that there is a set of 7 graphs such that a graph G has a spatial embedding without linked cycles if and only if G does not have a minor belonging to this set. and then drawing the edges in different planes that contain the line. Such a drawing is called spatial embedding of the graph. Every graph in the set has 15 edges. see B. This family of forbidden graphs was originally found by S ACHS (without proof). For further results and proofs concerning graphs in surfaces. Three dimensions∗ Every graph can be drawn without crossing edges in the 3-dimensional space. there exists a graph G with genus g( G ) = g so that χ( G ) = 7+ 1 + 48g 2 . the vertices of G can be put in a sphere. 83 If a nonplanar graph G can be embedded in a torus. if two cycles of G form a link (they cannot be separated in the space). which is curious.3 Genus of a graph Theorem 5. Alternatively. Indeed. For each g ≥ 1. It was shown by R OBERTSON . M OHAR AND C. such an embedding can be achieved by putting all vertices of G on a line. and drawing the edges as straight lines crossing the sphere inside. 2001.

Note that e ∈ D does not imply e−1 ∈ D. Then A is its • subdigraph. v). then the undirected edge with the same ends is in U ( D ).1 Digraphs In some problems the relation between the objects is not symmetric.6 Directed Graphs 6. The maximal induced di-connected subdigraphs are the di-components → of D. Let D be a digraph. . A walk W = e1 e2 . if VA = X and E A = ED ∩ ( X × X ). A = D [ X ]. k]. there exist directed paths u − v → ⋆ and v − u. We still write uv for (u. if G = U ( D ) and e ∈ D implies ∈ D. for all u = v. Let D be a digraph. The underlying graph U ( D ) of a digraph D is the graph on VD such that if e ∈ D. . Similarly. ⋆ The digraph D is di-connected. u)). In this case. / ⋆ D EFINITION . we deﬁne directed paths and directed cycles as directed walks and closed directed walks without repetitions of vertices. For these cases we need directed graphs. ek : u − v of U ( D ) is a directed → walk. if ei ∈ D for all i ∈ [1. A digraph (or a directed graph) D = (VD . if VA ⊆ VD and E A ⊆ ED . ED ) consists of the vertices VD and (directed) edges ED ⊆ VD × VD (without loops vv). if. For each pair e = uv deﬁne the inverse of e as e−1 = vu (= (v. D is said to be an oriented graph. • induced subdigraph. where the edges are oriented from one vertex to another. Can you make the streets one-way. but note that now uv = vu. and still be able to drive from one house to another (or exit the town)? Deﬁnitions D EFINITION . As an example consider a map of a small town. e −1 A digraph D is an orientation of a graph G. D EFINITION . .

. if α(v) = i. although the digraph D is not di-connected. and PQ is a directed path. since D − A does not contain directed cycles. assign a colour α(v) = i. . There is a nice connection between the lengths of directed paths and the chromatic number χ( D ) = χ(U ( D )). ( D − A) + e contains a ⋆ ⋆ directed cycle C : u − v − u. → Consider then an edge e = vu ∈ A. Every graph G has an orientation D. This shows that α is a proper colouring of U ( D ).2. where the part u − v is a directed path in D − A. (You offer and accept a handshake. D We have the following handshaking lemma. where the longest directed paths have lengths χ( G ) − 1.1 (R OY (1967).) Lemma 6. ⊔ ⊓ The bound χ( D ) − 1 is the best possible in the following sense: Theorem 6.6. Here 1 ≤ i ≤ k + 1. Indeed. but its directed paths are of length one. A digraph D has a directed path of length χ( D ) − 1. Then I ∑ dD (v) = |D| = ∑ dO (v). The indegree and the outdegree of a vertex are deﬁned as follows I dD (v) = |{e ∈ D | e = xv}|. This digraph has a path of length ﬁve. Let k be the length of the longest directed path in D − A. and therefore χ( D ) ≤ k + 1. ⋆ First we observe that if P = e1 e2 . By the minimality of A. Let D be a digraph. if e = uv ∈ D − A. D v∈ D v∈ D Directed paths The relationship between paths and directed paths is in general rather complicated. and → → → hence α(u) = α(v). For each vertex v ∈ D. In particular. if a longest directed path from v has length i − 1 in D − A. Let A ⊆ ED be a minimal set of edges such that the subdigraph D − A contains no directed cycles. . then α(u) = α(v). ⋆ Since PQ : u − w. er (r ≥ 1) is any directed path u − v in D − A. that is. Theorem 6. dO (v) = |{e ∈ D | e = vx}| .1 Digraphs 85 Note that a graph G = U ( D ) might be connected. D EFINITION .G ALLAI (1968)). then there exists a directed path Q : v − w of → length i − 1. → ⋆ then α(u) = α(v). Proof. k ≥ χ( D ) − 1.1. α(u) = i = α(v).

⊔ ⊓ . Indeed. On the other hand. Now χ G (k) is a polynomial that satisﬁes the recurrence χ G (k) = χ G−e (k) − χ G∗e (k). where b is the number of acyclic orientations D of G −e that extend in both ways e1 and e2 . we show that a( G ) satisﬁes the same recurrence. Let a( G ) be the number of acyclic orientations of G. An orientation D of an undirected graph G is acyclic. For (6. if G is discrete. As usual the set of colours is [1. if α(u) < α(v). and if D has no directed path v − u. We orient each edge uv ∈ G by setting uv ∈ D. To prove the claim. The proof is by induction on ε G . a( G ) = (−1)n χ G−e (−1) + (−1)n−1 χ G∗e (−1) = (−1)n χ G (−1) . Let G be a graph of order n. Therefore b = a( G ∗ e). First.2) Each acyclic orientation of G ∗ e corresponds in a natural way to an acyclic orientation D of G −e that satisﬁes (6.3 (S TANLEY (1973)).1) (6. and a( G ) = 1 = (−1)n (−1)n = (−1)n χ G (−1) as required. → → (6. ⊔ ⊓ D EFINITION . if D is an acyclic orientation of G −e for e = uv. if it has no directed cycles. Note that since D is acyclic. k]. since χ G (−1) measures the number of proper colourings of G using −1 colours! Theorem 6. Indeed. Clearly.6. where χ G is the chromatic polynomial of G. we choose → → ⋆ ⋆ e1 . then χ G (k) = kn .1 Digraphs 86 Proof. it cannot have both ways u − v and v − u. we observe that every acyclic orientation of G gives an acyclic orientation of G −e. if D has no ⋆ ⋆ directed path u − v. it extends to an acyclic orientation of G by putting e1 : u → v or e2 : v → u.2).1). the so obtained orientation D has no directed paths of length ≥ k − 1. by the induction hypothesis. we choose e2 . and the proof is completed. Proof. → → We conclude that a( G ) = a( G − e) + b. if a( G ) = a( G − e) + a( G ∗ e) then. The next result is charming. Let k = χ( G ) and let α be a proper k-colouring of G. The acyclic orientations D that extend in both ways are exactly those that contain ⋆ ⋆ neither u − v nor v − u as a directed path. Then the number of the acyclic orientations of G is a( G ) = (−1)n χ G (−1) .

. (1) Let D be a digraph such that U ( D ) is connected.1 Digraphs 87 One-way trafﬁc Every graph can be oriented. that is. Example 6. but the result may not be di-connected. A connected graph G is di-orientable if and only if G has no bridges. In the oneway trafﬁc problem the resulting orientation should be di-connected. R OBBINS’ theorem solves this problem. where w is the last vertex inside H. In conclusion. Now. Theorem 6. Proof. an } of n letters. If H = G. Let D be a digraph. . ⋆ In the di-orientation D H of H there is a directed path P′ : u − w.4 (R OBBINS (1939)). for otherwise someone is not able to drive home. then any orientation of G has at least two di-components (both sides of the bridge). which eP → → → H C contradicts the maximality assumption on H. The edge e / is not a bridge and thus there exists a cycle ⋆ ⋆ C ′ = ePQ : v − u − w − v → → → u P′ P w e Q v in G. A directed Euler tour of D is a directed closed walk that uses each edge exactly once. ⊔ ⊓ Example 6. y for which D I ( v) − dO ( v)| = 1.2. for directed graphs. if there is a di-connected oriented graph D such that G = U ( D ). . Consider words over an alphabet A = {a1 . then we are done. we orient → ⋆ e : v − u and the edges of Q in the direction Q : w − v to obtain a directed cycle → → ⋆ ⋆ ′ Q : v − u − w − v. . Then D has a directed Euler trail if and I only if dD (v) = dO (v) for all vertices v with possibly excepting two vertices x. there exists an edge e = vu ∈ G such that u ∈ H but v ∈ H (because G is connected). This proves the claim. D (2) Let D be a digraph such that U ( D ) is connected. Hence G has a cycle C. The following two results are left as exercises. each word w is a sequence of letters. Otherwise. a2 . |d D D The above results hold equally well for multidigraphs. Then D has a directed Euler tour if and I only if dD (v) = dO (v) for all vertices v. and a cycle is always di-orientable. In the case .1. where we allow parallel directed edges between the vertices. A directed Euler trail of D is a directed walk that uses each edge exactly once. G [V ∪ V ] has a di-orientation. If G has a bridge e. Let then H ⊆ G be maximal such that it has a di-orientation D H . D EFINITION . Suppose then that G has no bridges.6. The following problem was ﬁrst studied by H UTCHINSON AND W ILF (1975) with a motivation from DNA sequencing. that is. A graph G is di-orientable.

and the question is: does there exist a word w in which each letter ai occurs exactly si times. Let T be an orientation of Kn . The chromatic number of Kn is χ(Kn ) = n. Proof. if n = 2. s1 = 3. there is a directed path v − u of length at most two. In a problem instance. Suppose that there exists an x. . T. Tournaments D EFINITION . r12 = 2. which contradicts the maximality assumption made for v. There are four tournaments of four vertices that are not isomorphic with each other.6 (L AUDAU (1953)). ⊔ ⊓ The vertices of a tournament can be easily reached from one vertex (sometimes called the king). and hence by Theorem 6. then the word a1 a2 a1 a1 a2 is a solution to the problem. Theorem 6. j ≤ n. Example 6. But there are d vertices in A = {y | vy ∈ T }. This is then a directed Hamilton path visiting each vertex once. a tournament T of order n has a directed path of length n − 1. and let dO (v) = d be the maximum outdegree in T T.1. → Proof. It follows that xv ∈ T and xu ∈ T for all u with vu ∈ T. It is rather obvious that a directed Euler trail of D gives a solution to the sequencing problem. C. For instance. r21 = 1. G. r22 = 0. we are given nonnegative integers si and rij for 1 ≤ i.1 Digraphs 88 of DNA. for which the directed distance from v to x is at least three. Let v be a vertex of a tournament T of maximum outdegree. Consider a multidigraph D with VD = A for which there are rij edges ai a j .) Theorem 6. and thus d + 1 vertices in {y | xy ∈ T } = A ∪ {v}.6. Here the new digraph has the edge vu instead of uv. ⋆ Then for all u. Every tournament has a directed Hamilton path. A tournament T is an orientation of a complete graph. ⊔ ⊓ Problem. (Isomorphism of directed graphs is deﬁned in the obvious way. the letters are A.3. and ai is followed by a j exactly rij times. Ádám’s conjecture states that in every digraph D with a directed cycle there exists an edge uv the reversal of which decreases the number of directed cycles. and r11 = 1. It follows that the outdegree of x is d + 1.5 (R ÉDEI (1934)).

s2 (4) = 3. . the level scorings will eventually stabilize in a ranking of the tournament: there exits an m for which the mth-level scoring gives the same ordering as do the . We continue by deﬁning inductively the mth-level scoring by sm ( j) = ∑ s m −1 (i ) . and in another. and let T be the scoring digraph as in the ﬁgure.6 states that a winner v either defeated a team u or v defeated a team that defeated u. .1 Digraphs 89 Example 6. s2 (6) = 3 . Example 6. D EFINITION . s1 (1) = 4. . A ranking of a tournament is a linear ordering of the teams vi1 > vi2 > · · · > vin that should reﬂect the scoring of the teams. . s1 (6) = 1 . ji∈ T This tells us how good teams j beat. s1 (5) = 2. However. 2. ji∈ T It can be proved (using matrix methods) that for a di-connected tournament with at least four teams. The situation can be presented as a tournament. So. s2 (2) = 5.6. − vin . s1 (4) = 2.4. In the above tournament. if vi won v j in their mutual game. In our example. it seems that 3 is the winner. it is a looser. and there is an edge vi v j . .5. Consider a tournament of six teams 1. and suppose that each game has a winner. 6. Consider a tournament of n teams that play once against each other. is team 1 the winner? If so. where it is a winner. Theorem 6. s1 (3) = 3. s2 (5) = 4. Now. s2 (3) = 9. if v comes out with the most victories in the tournament. A team v is a winner (there may be more than one winner). but this extends to a directed Hamilton cycle (by adding 3 − 1)! So for every team → there is a Hamilton path. we have s2 (1) = 8. is 2 or 3 next? Deﬁne the second-level scoring for each team by s2 ( j ) = ∑ s1 ( i ) . and some of these may do unjust for the ‘real’ winner. 1 6 5 2 3 4 Let s1 ( j) = dO ( j) be the winning number of the team j (the number of teams T beaten by j). Here 1 − 2 − 4 − 5 − 6 − 3 is a di→ → → → → rected Hamilton path.but 4 and 6 have the same score. s1 (2) = 3. . One way of ranking a tournament could be by a Hamilton path: the ordering can be obtained from a directed Hamilton → → → path P : vi1 − vi2 − . where the vertices correspond to the teams vi . a tournament may have several directed Hamilton paths.

A ] f ( e) . A] = {e ∈ D | e = uv. if e ∈ E. In modelling a transportation network by a digraph. in every electrical network the amount of current ﬂowing in a vertex equals the amount ﬂowing out that vertex.A] f ( e) and f − ( A) = ∑ e ∈[ A. if e ∈ N. In our example the level scoring gives 1 − 3 − 2 − 5 − 4 − 6 as the ranking → → → → → of the tournament. f + (u) = ∑ v∈ N f (uv) and f − (u) = ∑ v∈ N f (vu) . These networks usually possess also some additional requirements. and f − (v) = f + (v) for all v ∈ {s.E. Harris in the 1950s. / Denote VN = V and EN = E. then the level scoring should be carried out with respect to the di-components. v ∈ A} . If T is not di-connected. • two distinct vertices s and r. r} . Flows D EFINITION . called the source and the sink of N. . for which α(e) = 0. E). A network N consists of • an underlying digraph D = (V. The connection to Kirchhoff’s Current Law (1847) is immediate. we must make sure that the number of goods remains the same at each crossing of the roads. According to this law.6. Let A ⊆ VN be a set of vertices.6. D EFINITION . and • a capacity function α : V × V → R + (nonnegative real numbers). The second condition states that there should be no loss. We adopt the following notations: / 5 s 6 4 4 5 4 2 5 4 2 r [ A.2 Network Flows 90 (m + k)th-level scorings for all k ≥ 1. The value f (e) can be taught of as the rate at which transportation actually happens along the channel e which has the maximum capacity α(e). / Example 6. u ∈ A. and f : VN × VN → R any function such that f (e) = 0. 6. ∑ e ∈[ A. Goods are transported from speciﬁc places (warehouses) to ﬁnal locations (marketing places) through a network of roads. / f + ( A) = In particular. A ﬂow in a network N is a function f : VN × VN → R + such that 0 ≤ f (e) ≤ α(e) for all e. The problem setting for such networks was proposed by T.2 Network Flows Various transportation networks or water pipelines are conveniently represented by weighted directed graphs.

. Then 0= ∑ ( f + (v) − f − (v)) = ∑ v∈ A v∈ A f + ( v) − ∑ v∈ A f − (v) = f + ( A) − f − ( A) = val( f A ).. ei where ι(e) = α( e) − f ( e) f ( e) if e is along P . s. In the above example. if ei = vi+1 vi ∈ N. r}. r. A ﬂow f of a network N is a maximum ﬂow. r. For a ﬂow f and each subset A ⊆ VN . The previous network has a ﬂow that is indicated on the right. Let N = ( D. E. The second claim is also clear. then the value α(e) gives the capacity (x m3 /min) of the pipe e. because the junctions cannot handle this much water. and let P = e1 e2 . Proof. Lemma 6. in the above ﬁgure the source should not try to feed the network the full capacity (11 m3 /min) of its pipes. and against P. D EFINITION . . α) be a network with a ﬂow f . 4 4 5 s 4 4 3 3 0 4 1 r 91 A ﬂow f in N is something that the network can handle. s. (i) If A ⊆ N \ {s. Every network N has a zero ﬂow deﬁned by f (e) = 0 for all e. if e is against P. ⊔ ⊓ Improvable ﬂows Let f be a ﬂow in a network N. r}.6. v ∈ A cancel each out. deﬁne the resultant ﬂow from A and the value of f as the numbers val( f A ) = f + ( A) − f − ( A) and val( f ) = val( f s ) (= f + (s) − f − (s)) . en be an undirected path in N where an edge ei is along P.2 Network Flows If N = ( D. where the third equality holds since the values of the edges uv with u. if ei = vi vi+1 ∈ N. α) is a network of water pipes. if there does not exist any ﬂow f ′ such that val( f ) < val( f ′ ).2. (ii) val( f ) = −val( f r ).g. val( f ) = 9. Let A ⊆ N \ {s. . The value val( f ) of a ﬂow is the overall number of goods that are (to be) transported through the network from the source to the sink. then val( f A ) = 0. We deﬁne a nonnegative number ι( P) for P as follows: ι( P) = min ι(e).

we have ( f ′ )− (v) = ( f ′ )+ (v). Example 6. For a subset S ⊂ VN with s ∈ S and r ∈ S. if ι( P) > 0. S] (= {uv ∈ N | u ∈ S. val( f ) = val( f S ) = f + (S) − f − (S) . 4 4 5 s 5 4 3 2 0 4 2 r Then f ′ is a ﬂow. s. A ⋆ path P : s − r is ( f -)improvable. if ι( P) > 0. Hence. v ∈ S}) . α) be a network. → On the right. Let f be a ﬂow in a network N. A cut [S] is a minimum cut. Now val( f ′ ) = val( f ) + ι( P). In our original network the capacity of the cut for the indicated vertices is equal to 10. then we can improve the ﬂow. let the cut by S be / [S] = [S. 5 s 6 4 4 5 2 4 4 2 5 r . Deﬁne f ( e) + ι( P ) ′ f ( e) = f ( e) − ι( P ) f ( e) if e is along P .2 Network Flows 4 92 5 s 3 4 4 3 4 0 4 1 r D EFINITION . if e is not in P .6. the bold path has value ι( P) = 1. Let N be a network. since at each intermediate vertex v ∈ {s. For a ﬂow f and a cut [S] of N. r}. r. / The capacity of the cut [S] is the sum α[S ] = α+ (S ) = ∑ e ∈[S ] α( e) . if there is no cut [ R] with α[ R] < α[S]. since P has exactly one edge sv ∈ N for the source s. Lemma 6.3. Proof. ⊔ ⊓ Max-Flow Min-Cut Theorem D EFINITION .4. if e is against P . Let N = ( D. If f is a maximum ﬂow of N.7. and / the capacities of the edges are not exceeded. then it has no improvable paths. Lemma 6. and therefore this path is improvable.

as required. and by R OBACKER in 1955 – 56. ι( P) > 0} . (ii) if e = vu ∈ N. The following main result of network ﬂows was proved independently by E LIAS . a maximum ﬂow cannot have improvable paths. then f (e) = 0. and let ⋆ S I = {u ∈ N | for some path P : s − u. The present approach is due to Ford and Fulkerson. then f (e) = α(e). and (2) f (e) = 0 for all e = vu with u ∈ S. → ⊔ ⊓ .8. if f is a maximum ﬂow and [S] a minimum cut. val( f ) ≤ α[S]. ⊔ ⊓ Theorem 6. the equality val( f ) = α[S] holds if and only if (1) f + (S) = α[S] and (2) f − (S) = 0. then f is a maximum ﬂow and [S] a minimum cut.1. Conversely. By Lemma 6. By Lemma 6. A ﬂow f of a network N is maximum if and only if there are no f -improvable paths in N. Furthermore. Proof.7.3. This holds if and only if f (e) = α(e) for all e ∈ [S] (since f (e) ≤ α(e)). val( f ) = val( f S ) = f + (S) − f − (S). Theorem 6. Also. and val( f ) = val( f s ). f + (S) = ∑ e ∈[S ] f ( e) ≤ ∑ e ∈[S ] α( e) = α[S ] . Hence val( f S ) =val( f s ) − ∑ v∈S I f (sv) + ∑ v∈S I f (vs) + val( f S I ) + ∑ v∈S I f (sv) − ∑ v∈S I f (vs) =val( f s ) = val( f ) .4. and hence val( f ) ≤ α[S]. then val( f ) ≤ α[S] . assume that N contains no f -improvable paths. Corollary 6.6. and f − (S) ≥ 0.2 Network Flows 93 Proof. If f is a ﬂow and [S] a cut such that val( f ) = α[S]. F EINSTEIN . equality holds if and only if for each u ∈ S and v ∈ S. S HANNON. By the deﬁnition of a ﬂow. r}). by F ORD AND F ULKERSON. v ∈ S. In particular. / This proves the claim. / (i) if e = uv ∈ N. For a ﬂow f and any cut [S] of N. Proof. Now val(S I ) = 0 (since S I ⊆ N \ {s. Let S I = S \ {s}.

there exists a / ⋆ ⋆ path P : s − u with ι( P) > 0. yr ∈ N (α(yr) = 1) for all x ∈ X. Now Theorem 6. and consider a network N with vertices {s. (i) The value val( f ) of a maximum ﬂow equals the maximum number of edge-disjoint di⋆ rected paths s − r.8 showed also Theorem 6. f (e) = 0. Applications to graphs⋆ The Max-Flow Min-Cut Theorem is a strong result. the value val( f ) of a maximum ﬂow equals the capacity α[S] of a minimum cut. For this.1 implies now that f is a maximum ﬂow (and [S] is a minimum cut). y ∈ Y together with the edges xy ∈ N (α( xy) = | X | + 1). and so f (e) = α(e).5. ⊔ ⊓ Corollary 6. Let f 0 be the zero ﬂow. y ∈ Y.3 by increasing and decreasing the values of the edges by integers only. where the capacities of the edges are equal to one (α(e) = 1 for all e ∈ N). Then G has a matching that saturates X if and only if N has a maximum ﬂow of value | X |.9. Theorem 3. By the same argument. whose removal destroys → ⋆ all the directed connections u − v from D. We mention a connection to the Marriage Theorem. Y ). r} ∪ X ∪ Y. → (ii) The capacity of a minimum cut [S] equals the minimum number of edges whose removal ⋆ destroys the directed connections s − r from s to r. Moreover. Since u ∈ S. for an edge e = vu ∈ N with v ∈ S and u ∈ S. Next we apply the theorem to unit networks. Let N be a network. where u ∈ S and v ∈ S. Let the edges (with their capacities) be sx ∈ N (α(sx) = 1).10 gives Theorem 3. Exercise. since v ∈ S. Let u and v be two vertices of a digraph D. In a network N. → / → Therefore ι(e) = 0. We obtain results for (directed) graphs. ⊔ ⊓ The proof of Theorem 6.9. and many of our previous results follow from it. A maximum ﬂow is constructed using Lemma 6. if xy ∈ G for x ∈ X.6. we have val( f ) = α[S]. Lemma 6. ⊔ ⊓ Theorem 6. → Proof. The maximum number of edge⋆ disjoint directed paths u − v equals the minimum number of edges. Consider an edge e = uv ∈ N.9. Proof. Corollary 6. where the capacity function α : V × V → N has integer values. → . Then N has a maximum ﬂow with integer values. / By Theorem 6.2.7.2 Network Flows 94 Set S = S I ∪ {s}. ι( Pe) = 0 for the path Pe : s − v.10 (Max-Flow Min-Cut). f0 (e) = 0 for all e ∈ V × V. let G be a bipartite graph with a bipartition ( X. Let N be a unit network with source s and sink r.

3) becomes ∑ e = vu ∈ H α( e) = 0 . A graph with a nowhere zero 4-ﬂow.3. A network N with source s and sink r is obtained by setting the capacities equal to 1. 1 3 2 2 The condition (6. if α(e) = 0 for all e ∈ H. α) of an undirected graph G is an orientation H of G together with an edge colouring α : E H → [0. The claim follows then easily from Corollary 6. Corollary 6. → Proof. ⊔ ⊓ The next corollary is Menger’s Theorem for edge connectivity. let α(e−1 ) = −α(e) for all e ∈ H in the orientation H of G so that the condition (6. A k-ﬂow ( H. whose removal destroys all the → ⋆ connections u − v from G. k − 1] such that for all vertices v ∈ V. .4) generalizes to the subsets A ⊆ VG in a natural way. Let u and v be two vertices of a graph G. Proof. For convenience. The claim follows immediately from Corollary 6.2. 1 2 1 (6. In particular.A] α( e) = 0 .3.2 Network Flows 95 Proof. Consider the digraph D that is obtained from G by replacing each (undirected) edge uv ∈ G by two directed edges uv ∈ D and vu ∈ D.4. ⊔ ⊓ Seymour’s 6-ﬂows∗ D EFINITION . the sum of the incoming values equals the sum of the outgoing values. The maximum number of edge⋆ disjoint paths u − v equals the minimum number of edges.10. In the k-ﬂows we do not have any source or sink. The claim follows from Lemma 6. ⊔ ⊓ Corollary 6. (6. ∑ e = vu ∈ H α( e) = ∑ f = uv∈ H α( f ) . ∑ e ∈[ A.4) Example 6.5 and Corollary 6. A k-ﬂow is nowhere zero. A graph G is k-edge connected if and only if any two distinct vertices of G are connected by at least k independent paths.6. (6.8.3) that is.5) since the values of the edges inside A cancel out each other.

Y )-bipartite. we have that / f ( G ) ≤ 3. and set α(e) = 2. In order to fully appreciate Seymour’s result. 96 Tutte’s Problem. and so 5 is the best one can think of. ⊔ ⊓ Example 6. Proof. The ﬂow number f ( G ) of a bridgeless graph G is the least integer k for which G has a nowhere zero k-ﬂow.9. Indeed. Orient the edges e ∈ M from / Y to X. and set α(e) = 1. Since W arrives and leaves each vertex equally many times. then f (Kn ) = Proof. For each 3-regular bipartite graph G. and consider an Euler tour W of G. Theorem 6. we mention that it was proved as late as 1976 (by J AEGER) that every bridgeless G has a nowhere zero k-ﬂow for some integer k. let α(e) = 1. Conversely. A connected graph G has a ﬂow number f ( G ) = 2 if and only if it is eulerian. 2 3 if n is odd .13. Tutte’s conjecture resembles the 4-Colour Theorem. By Corollary 3.11. Theorem 6. and so G is eulerian.1. It was conjectured by T UTTE (1954) that every bridgeless graph has a nowhere zero 5-ﬂow. y3 ∈ Y such that xy2 . Suppose G is eulerian.11 (S EYMOUR’s (1981)). Then necessarily the degrees of the vertices are even. ⊔ ⊓ .6. and two neighbours y2 .12. if n is even . xy3 ∈ M.6. S EYMOUR’s remarkable result reads as follows: Theorem 6. The proof of this uses the 4-Colour Theorem.2 Network Flows Lemma 6. The Petersen graph has a nowhere zero 5-ﬂow but does not have any nowhere 4-ﬂows. Exercise. Theorem 6. Since each x ∈ X has exactly one neighbour y1 ∈ Y such that xy1 ∈ M. We have f (K4 ) = 4. Let D be the orientation of G corresponding to the direction of W. a 3-regular graph has a perfect matching M. ⊔ ⊓ D EFINITION . the function α is a nowhere zero 2-ﬂow. Omitted. Every bridgeless graph has a nowhere zero 6-ﬂow. If an edge uv ∈ D. Orient the edges e ∈ M from X to Y. we have f ( G ) ≤ 3. then G has no bridges. and indeed. Proof. let G be ( X. the conjecture is known to hold for the planar graphs. If G has a nowhere zero k-ﬂow for some k. let α be a nowhere zero 2-ﬂow of an orientation D of G. and if n > 4.

5 edge. 12 distance. 62 face. 4 Euler trail. 78 end (of a path).Index 2-sets. 10 connected. 5 disconnected. path. 24 cycle.edge. 53 chromatic polynomial χ G . cycle. 96 forest. 54 crossing number. 84 di-orientable. 4 graphical sequence. boundary. 16 complete graph Kn . 24 embedding. 24 boundary. 63 eulerian. trail. 92 cut vertex. 62 bridge. 35 available colour.k . 53 complement G. 90 ﬂow number. 87 walk. 10 disjoint walks. 4 augmented path. 13 distance function. 11 cut (in a network). 80 graph. 28 ﬂow. 38 even cycle. 7 di-connected di-component. 12 end (of an edge). 53 bipartite. 27 discrete graph. 12 degree d G (v). 68 cube. 10 complete bipartite Km. 58 critical. 87 digraph. 90 choosable. 24 contracted vertex.vertex. 92 capacity function. 5. 31 . 18 genus. 57 closed walk. 4 edge chromatic number χ′ ( G ). 43 edge cut. 12 exterior: face. 6 adjacent. 31 hamiltonian. 43. 62 fan. 84 directed Euler tour. cycle. 29 even component. 23 adjacency matrix. 29 Euler’s formula. 24 connected (component). 13 disconnecting set. 16 bond. 77 connectivity number κ ( G ). 9 Hamilton path. 43 edge colouring. 4 acyclic. 72 chromatic number χ( G ). 13 connected sum. 18 digraph. 12 colouring. tour. 18 capacity. 84 directed graphs (digraph). 86 addressable.

66 latin rectangle. 77 proper colouring. 41 stable set. 24 Shannon’s switching game. 24 separating set. 35 Petersen graph. 8 interior: face. 84 induced subgraph. 38 latin square. 91 same parity. edge. 91 matching. 84 inverse walk (path). 38 odd cycle. 4 Möbius band. 74 monochromatic. 61 subgraph. 90 size ε G . 89 regular graph. 7 network. 8 surface. 12 induced subdigraph. vertex. 10 resultant ﬂow. 88 kiss (circles). 18 triangle. 62 intersection graph. 76 orientation. 74 linked cycles. 84 oriented. 45 incident colours. 53 triangulation. 7 weighted distance. 49 multigraph. 88 trail. 3 odd component. 3 path. 4 near-triangulation. 92 improvement (colouring). 38 leaf. 13 minor. 79 tournament. 4 neighbourhood NG (v). 76 triangle-free. 83 list chromatic number. 43 indegree. 84 subdivision. 35 maximal planar graph. 22 homeomorphic. 70 matching. 7 line segment graph. 43. 3 topologically equivalent.Index Hamming distance. 11. 79 stable matching. 7 ﬂow. 76 torus. 72 loop. 72 neighbour. 4 orientable surface. 20 sink and source. 10 98 . 12 optimal colouring. 8 spanning tree. 38 tree. 85 parallel edges. 35 minimum cut. 76 improvable (path). 61 plane model. 84 outdegree. 76 oriented graph. 72 list colouring. 5 join of walks. 4 partition. 12 king. 61 plane embedding. 45 order νG . 53 Ramsey number. 76 symmetric difference. 6 inverse pair. 10 NP-complete problems. 35 separates. 90 nontrivial graph. 4 spanning subgraph. 16 subdigraph. 20 spatial embedding. 92 degree δ( G ). 29 transversal. 7 isomorphic. 83 sphere. 85 independent paths. 3 saturate (matching). 32 planar graph. 73 Kuratowski graph. 77 map. 49 ranking. 64 maximum degree ∆ ( G ). 76 trivial graph. 12 isolated vertex. 78 sphere with a handle. 12 perfect matching.

5 walk. 12 2-cell. 91 99 . 94 vertex. 81 2-switch. 89 zero ﬂow. 13 weight function. 84 unit networks.Index trivial path. 52 winning number. 5 wheel. 9 underlying digraph. 11 weight. 4 vertex colouring. 90 underlying graph.

Sign up to vote on this title

UsefulNot useful- Graph Theory
- Graph Theory With Applications
- Intro to Graph Theory
- Graph Theory With Applications - J. Bondy, U. Murty
- GT English
- Introduction to Graph Theory Second Edition Solution Manual (by Douglas B. West)
- Graph Theory & Algorithms - M. Ashraf Iqbal
- Graph Theory
- Introduction to Graph Theory -- D. B. West
- Graph Theory
- C. Vasudev - Combinatorics and Graph Theory - New Age - 577p
- Lecture Notes on Graph Theory
- graph theory by narsing deo
- Graph Theory Bondy Murty
- Graph Theory Book
- Handbook of Graph Theory
- Matveev Mechanics and Theory of Relativity
- Graph Theory
- Graph Theory Douglas B West 2 Ed
- Chartrand - Graphs and Digraphs
- Graph Theory
- Graph Theory Answer Key
- Algebraic Graph Theory
- Graph Theory
- Lowell W. Beineke & Robin J. Wilson & Peter J. Cameron - Topics in Algebraic Graph Theory
- New Frontiers Graph Theory
- MAtrix Computations
- Differential geometry
- Graph Theory

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.