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.

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

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

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

. . Proof. and edges Xi X j for all i and j (i = j) with Xi ∩ X j = ∅. one for each ordering of its set VG of vertices. s( G ) = min{| X | | G ∼ GX for some X ⊆ 2X } . E) be the graph with rs ∈ E if and only if r and s (for r = s) have a common divisor > 1.2. S TOCKMEYER AND W ONG (1976) that it is algorithmically difﬁcult to determine the number s( G ) – the problem is NP-complete.1 has an adjacency matrix on the right. . then these ﬁgures are. . Moreover.2. that is. and computers are good in crunching numbers. since every respectable programming language has array structures for these. Let VG = {v1 . Notice that the adjacency matrix is always symmetric (with respect to its diagonal consisting of zeros). and deﬁne. .1.1 Graphs and their plane ﬁgures 6 Other representations Plane ﬁgures catch graphs for our eyes. unsuitable. Every graph is an intersection graph of some family of subsets. and let G A = ( A. The following result is obvious from the deﬁnitions. to say the least. Theorem 1. 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. Xn } be a family of subsets of a set X. . For this. . . for all v ∈ G. 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. vn } be ordered. Integer matrices are ideal for computers. .1. ⊔ ⊓ 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. Then Xu ∩ Xv = ∅ if and only if uv ∈ G. Xn . Example 1. and deﬁne the intersection graph GX as the graph with vertices X1 . . two isomorphic graphs have exactly the same set of adjacency matrices. u} | vu ∈ G }. Let G be a graph. the graph in Example 1. / For instance. As yet another example. but if a problem on graphs is to be programmed. let X = {X1 . . X2 . . = 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 . . Two graphs G and H are isomorphic if and only if they have a common adjacency matrix. Graphs can also be represented by sets. we state: All graphs can be represented in the form G A for some set A of natural numbers. As an exercise. Theorem 1. a set Xv = {{v. let A ⊆ N be a ﬁnite set of natural numbers.

2 Subgraphs Ideally.1 (Handshaking lemma). Lemma 1. v∈ G Moreover. 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 } . and a solution can be found to the problem by combining the information determined by the parts. If dG (v) = 0. E H ) on the same set of vertices that are not isomorphic. then v is a leaf of the graph. . Degrees of vertices D EFINITION . and if dG (v) = 1.1. Proof. Let v ∈ G be a vertex a graph G. The second claim follows immediately from the ﬁrst one. For instance. the existence of an Euler tour is very local. due to E ULER (1736). Every edge e ∈ EG has two ends. tells that if several people shake hands. then v is said to be isolated in G. Note that the degrees of a graph G do not determine G. EG ) and H = (V. The following lemma. The neighbourhood of v is the set NG (v) = {u ∈ G | vu ∈ G } .2 Subgraphs 7 1. the number of vertices of odd degree is even. ∑ d G ( v) = 2 · ε G . but for which dG (v) = d H (v) for all v ∈ V. it depends only on the number of the neighbours of the vertices. ⊔ ⊓ Lemma 1. For each graph G. when dG (v) is deﬁned as the number of edges that have v as an end. and when each loop vv is counted twice. then the number of hands shaken is even. as we shall later see. given a nice problem the local properties of a graph determine a solution. In these situations we deal with (small) parts of the graph (subgraphs). Indeed. there are graphs G = (V.1 holds equally well for multigraphs. The degree of v is the number of its neighbours: dG (v) = | NG (v)| .

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

By Lemma 1. . Let d1 . → In converse.2. Lemma 1. Let G be a graph of order n with a degree sequence d1 ≥ d2 ≥ · · · ≥ dn . . v j . → → Finally. Let G and H have the same degrees. since j > i. .1. Let d = ∆( G ) (= d1 ).2 Subgraphs 9 2-switches D EFINITION . 2s 2s Note that if G − H then also H − G since we can apply the sequence of 2→ → switches in reverse order.3 (B ERGE (1973)). there exists a v j with / j ≥ d + 2 such that v1 v j ∈ G. we use induction on the order νG . then clearly H has the same degrees as G. Since dG (v1 ) = d. . dn be a descending sequence of nonnegative integers. that is. Here di ≥ d j . and the other / neighbours of v1 remain to be its neighbours. Proof. where dG (vi ) = di . Such a sequence is said to be graphical. xy ∈ G with ux. v2 . . . d2 . vy ∈ G replaces the / edges uv and xy by ux and vy. By the induction hypothesis. if there exists a graph G = (V. Then there is a graph G ′ such that G − G ′ with NG′ (v1 ) = {v2 . → Proof. and this proves the → claim. Before proving Berge’s switching theorem we need the following tool. E) with V = {v1 . where v1 vi ∈ H and v1 v j ∈ H. Suppose that there is a vertex vi with 2 ≤ i ≤ d + 1 such that v1 vi ∈ G. We can now perform a 2-switch / with respect to the vertices v1 . we have a vertex v and graphs G ′ and H ′ such that G − G ′ and → H − H ′ with NG′ (v) = NH ′ (v). G ′ −v − H ′ −v. obtain a graph G ⊔ ⊓ Theorem 1. a 2-switch with respect to the edges uv. For a graph G. vd1 +1 }. and thus also G ′ − H ′ . . . This gives a new vt graph H. Now the graphs G ′ −v and H ′ −v have the same → degrees. . 2s 2s 2s 2s 2s 2s 2s . ⊔ ⊓ D EFINITION . . we / ′ as required. d1 ≥ d2 ≥ · · · ≥ dn . When we repeat this process for all indices i with v1 vi ∈ G for 2 ≤ i ≤ d + 1.2. but v j vt ∈ G. . 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. vj v1 vi Since v1 v j ∈ G. there exists a vt (2 ≤ t ≤ n) such that vi vt ∈ G. vi . we observe that H ′ − H by the ‘reverse 2-switches’. 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. vn } such that di = dG (vi ) for all i. vt . If G − H. .

2 and Theorem 1. . . 2. the graph G on the right has this degree sequence. and d H (vi ) = di for all i. The graph G = KV is the complete graph on V. otherwise G is nontrivial. . But now the degree sequence of G −v1 is in (1. s is graphical ⇐⇒ 3. 3. dn is graphical (when put into nonincreasing order). Theorem 1. Proof. . 3. . 1. d2 .4. 4. 1. i. H AKIMI (1962)). . . A graph G = (V. . . all discrete graphs of order n are isomorphic with each other. . dd1 +3 . if every two vertices are adjacent: E = E(V ). . 0 is graphical. The complements G = KV of the complete graphs are called discrete graphs. v2 v6 v4 (1. 0. dn (with d1 ≥ 1 and n ≥ 2) is graphical if and only if d2 − 1. .3. . and they will be denoted by Kn . . In a discrete graph EG = ∅. . . A sequence d1 . we can suppose that NG (v1 ) = {v2 . νG = 1. d G ( v d1 + 2 ) = d d1 + 2 . By Theorem 1. dd1 +1 ]. (⇐) Consider G of order n − 1 with vertices (and degrees) dG (v2 ) = d2 − 1.1).e. Indeed. if every vertex of G has the same degree. if it has only one vertex. All complete graphs of order n are isomorphic with each other. then G is r-regular or regular of degree r. vd1 +1 }. d G ( v n ) = d n as in (1. dd1 +1 − 1. . 1.4 (H AVEL (1955).2 Subgraphs 10 Using the next result recursively one can decide whether a sequence of integers is graphical or not. If this degree is equal to r. 1 is graphical 2. Add a new vertex v1 and the edges v1 vi for all i ∈ [2. where EG = {e ∈ E(V ) | e ∈ EG }. . By Lemma 1. dG (vd1 +1 ) = dd1 +1 − 1. . d H (v1 ) = d1 .3. / The complement of G is the graph G on VG .1. Clearly. Then in the new graph H. dd1 +2 . The last sequence corresponds to a graph with no edges. E) is trivial. . . . Consider the sequence s = 4. (⇒) Assume dG (vi ) = di . d3 − 1. ⊔ ⊓ Example 1. 2. . 0 is graphical 0.. A graph G is said to be regular. . 4. . 1.1).1) v1 v3 v5 Special graphs D EFINITION . and hence also our original sequence s is graphical.

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.

. t(v4 ) = 4. On the other hand. G Example 1. t(v5 ) = 4 . For this. • t(v4 ) = min{5. ui }. G Assume that G is a connected graph. v) = t(v). The algorithm stops. Thus u1 = v1 . We consider the minimum problem. v. . v5 . if the graph represents a network of pipelines. 3 + 2} = 5. (ii) For i ∈ [0. / Dijkstra’s algorithm: (i) Set u0 = u. 3} = 3. Apply Dijkstra’s algorithm to the vertex v0 . ∞} = 4. 3 v0 2 2 t(v3 ) = 2 + 1 = 3. and which connects two given vertices. Let ui+1 ∈ {u1 . . t(u0 ) = 0 and t(v) = ∞ for all v = u0 .7. 1 2 • t(v2 ) = min{3. call α(uv) the length of uv. t(v5 ) = min{4. νG − 1]: for each v ∈ {u1 . The shortest path problem: Given a connected graph G with a weight function α : EG → N. v ∈ G. . v) for given u. t(v3 ) = 3. and then one wants to ﬁnd a subgraph with the maximum weight. • t(v5 ) = min{4. ∞} = 3. 2} = 2. if uv ∈ G. / (iii) Conclusion: dα (u. 3 + 1} = 4. or all vertices (if we want to travel around). t(v5 ) = 2 + 2 = 4. t(v2 ) = min{∞. t(u1 ) + α(u1 v2 )} = min{3. 3 + 1} = 4. We have obtained: t(v1 ) = 2.3 Paths and cycles 14 In these examples we look for a subgraph with the smallest weight. t(ui ) + α(ui v)} . Consider the following weighted graph G. where u is a ﬁxed starting point and v ∈ G. 2 • t(v1 ) = min{∞. the weights are volumes or capacities. These are the minimal weights from v0 to each vi . . 1 v1 v3 others are ∞. • u0 = v0 . t(v2 ) = 3. / replace t(v) by min{t(v). Let us make the convention that α(uv) = ∞. . ﬁnd dα (u. Dijkstra’s algorithm solves the problem for every pair u. . ui } be any vertex with the least value t(ui+1 ). In this case. others are ∞.1. 4} = 3. t(v4 ) = min{5. t(v4 ) = 2 + 3 = 5. let G be a graph with an integer weight function α : EG → N. t(u0 ) = 0. 3 2 Thus choose u2 = v3 . Thus choose u4 = v4 . 4 + 1} = 4. t(v5 ) = min{4. 1 v2 v4 • t(v2 ) = min{3. Thus set u3 = v2 . .

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

MIT Press. Y ) is a bipartition of G. see T. Such connected graphs will be called trees. In this case. in search algorithms for databases. k)-bipartite graphs are isomorphic. 1993. These trees are sometimes called construction y · trees. Theorem 2. A graph G is bipartite if and only if VG has a partition to two stable subsets. Indeed. A subset X ⊆ VG is stable. “Introduction to Algorithms”. A bipartite graph G (as in the above) is complete (m.5 is bipartite for all k. C..L. K2. and they are stable in Qk . + Certain structures with operations are representable as trees. Example 2. if | X | = m. For concrete applications in this respect. The k-cube Qk of Example 1. consider A = {u | u has an even number of 1′ s} and B = {u | u has an odd number of 1′ s}. Let Km. A graph G is called bipartite. if G [ X ] is a discrete graph.g. 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. L EISERSON AND R. On the right there is a tree of operations for the arithy z metic formula x · (y + z) + y. All complete (m. Grammatical trees occur especially in linguistics.E. Y )-bipartite. these sets partition B k .2 Connectivity of Graphs 2.1. factorization trees or grammatical trees. and G is ( X.3 .H. R IVEST. x + where syntactic structures of sentences are analyzed. The solutions in these cases are usually subgraphs without cycles. e. and uv ∈ G for all u ∈ X and v ∈ Y. Clearly. |Y | = k.k denote such a graph. Bipartite graphs D EFINITION . and they are used. k)bipartite. C ORMEN . ( X. decomposition trees. The following result is clear from the deﬁnitions.1.1 Bipartite graphs and trees In problems such as the shortest path problem we look for minimum solutions that satisfy the given requirements.

By the maximum condition. and in this case. VG = X ∪ Y. Indeed. we note that it sufﬁces to show the claim for connected graphs. otherwise. y) is odd} . Since P and Q are shortest paths. let VG = X ∪ Y be a partition such that the 2 number of edges between X and Y is maximum. However. by the deﬁnition of distance. Denote F = EG ∩ {uv | u ∈ X. and G is bipartite as claimed. and colour it by 1. 2 2 2 Then colour the neighbours of v1 by 2. (⇐) Suppose that all cycles in G are even. d H (v) ≥ dG (v)/2. and proceed by 1 1 colouring all neighbours of an already coloured vertex 2 1 by the opposite colour. Y ′ = Y ∪ {v} does better that the pair X. if v ∈ X. Let v ∈ G be a chosen vertex. say 1 and 2. Assume thus that G is connected. where X consists of those vertices with colour 1.2 (E RDÖS (1965)). where P2 : x − u and Q2 : x − w are → → ⋆ independent. 2 1 this can be done by using two ‘opposite’ colours. i. G is not bipartite. Indeed.2. → Consequently. x) is even} and Y = {y | dG (v. then G is ( X. if G is disconnected. ⋆ ⋆ Let then u. G p of G. X ∩ Y = ∅. (That is. v is on the wrong side. and let P : v − u and Q : v − w → → be (among the) shortest paths from v to u and w. Y1 ∪ Y2 ∪ · · · ∪ Yp ). Therefore X and Y are / Q1 stable subsets. an odd cycle C2k+1 is not bipartite. . v x uw and so uw ∈ EG by assumption. and it is bipartite. Y.) Now 1 1 1 1 ε H = ∑ d H ( v) ≥ ∑ d G ( v) = ε G . (⇒) Observe that if G is ( X. | P1 | = | Q1 |. then G has the bipartition ( X1 ∪ X2 ∪ · · · ∪ X p . . otherwise. ⊔ ⊓ Q2 w 1 Checking whether a graph is bipartite is easy. Obviously H is a spanning subgraph. and deﬁne X = { x | dG (v. P1 and Q1 are shortest paths v − x. We show that each graph G has a bipartite subgraph H ⊆ G such that ε H ≥ 1 ε G . Assume that x is the last common ⋆ ⋆ vertex of P and Q: P = P1 P2 . Indeed. then the pair X ′ = X \ {v}. . 17 Proof.1 Bipartite graphs and trees Theorem 2. then so are all its subgraphs. then each cycle of G is contained in one of the connected components G1 . Q = Q1 Q2 . Y )-bipartite.. u Thus | P2 | and | Q2 | have the same parity and hence the P2 P1 −1 sum | P2 | + | Q2 | is even. 2 v∈ H 2 v∈ G 2 2 . v ∈ Y } . Start from any vertex v1 . Yi )bipartite. If the whole graph can be coloured without contradiction. . Since G is connected. the path P2 Q2 is even. since. If Gi is ( Xi . and let H = G [ F ]. Y )-bipartite. and Y of those vertices with colour 2. First.2. Example 2. at some point one of the vertices gets both colours. A graph G is bipartite if and only if G it has no odd cycles (as subgraph).e. Also. w ∈ G be both in X or both in Y.

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

4. both ends of a maximal path have degree one. ⋆ (i)⇒(ii) Let T be a tree. then w belongs to P. Let P : u − v be a maximal path in a graph G. because νG is ﬁnite. and let P. The following are equivalent for a graph T. Proof.2. Hence e must also belong to Q. (ii) Any two vertices are connected in T by a unique path. For acyclic graphs.1 Bipartite graphs and trees On the other hand. 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. Proof. Since P = Q. and therefore u and v belong to different connected components of T −e. . which contradicts the / maximality assumption for P. and it is clearly acyclic. Proof. If n = 1.2. if → G is acyclic. Moreover. Such paths exist. T is connected. then dG (v) = 1. Suppose that | P| ≥ | Q|. In particular. then also Pe is a path. ⊔ ⊓ (ii)⇒(iii) We prove the claim by induction on n. there exists an edge e which belongs to P but not to Q. Clearly. (i) T is a tree. Let νT = n. Theorem 2. Hence NG (v) ⊆ P. Q : u − v → be two different paths between the same vertices u and v. Since T is acyclic. Then NG (v) ⊆ P. Suppose thus that n ≥ 2. Each edge of T is a bridge. if there are no edges e ∈ G → for which Pe or eP is a path. Let T be any graph of order n satisfying (ii). If e = vw ∈ EG with w ∈ P. the claim holds for n = 2. (iii) T is acyclic and ε T = νT − 1. then the claim is trivial.2. and suppose it holds for graphs of order less than n. Assume the claim does not hold. Each tree T with νT ≥ 2 has at least two leaves. a contradiction. if wv ∈ G. ⊔ ⊓ Corollary 2. ⋆ Lemma 2. and wv is necessarily the last edge of P in order to avoid cycles.

The second claim is also clear. then one of the teams goes to the next round without having to play. a connected graph G is a tree if and only if ε G = νG − 1. ε T −v = n − 2.5. k]. Each connected graph has a spanning tree. Indeed. It is P’s purpose to designate a path s − r (that is. P : u − w − v. s. Let T ⊆ G be a maximum order subtree of G (i. Spanning trees Theorem 2. that is.2. where a play corresponds to an edge of the tree. i i n − 1 = εT = ∑ (|Vi | − 1) = ∑ |Vi | − k = n − k . In this → ⋆ case. By induction hypothesis. after the ﬁrst round 7 teams continue. Now. and it satisﬁes the condition (ii). Let T be a spanning tree of G. 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. ⊔ ⊓ Corollary 2. The answer to our problem is n − 1. We say that a game ( G. The subgraph → → T −v is connected.5. ε G ≥ νG − 1. so are the connected graphs Ti . where vw is the unique edge having an end v. Since T is acyclic.e. subgraph that is a tree). and the claim follows. But then T is not / / maximal. since the cup tournament is a tree. P and N alternate turns so that P designates an edge by +. By Lemma 2. and hence they are trees. (iii)⇒(i) Assume (iii) holds for T. Each edge can be des⋆ ignated at most once. We need to show that T is connected. let the connected components of T be Ti = (Vi . Then ε G ≥ ε T = νT − 1 = νG − 1.. For each connected graph G. for i ∈ [1. Moreover. and N by −. If VT = VG . If during a round there are k teams left in the tournament.3. then 2. then these are divided into ⌊k⌋ pairs. So 13 plays are needed in this example. and N tries to block all paths s − r (that is. i=1 i=1 k k which gives that k = 1. to designate → at least one edge in each such path). and so ε T = ε T −v + 1 = n − 1. to designate → ⋆ all edges in one such path).1 Bipartite graphs and trees 20 ⋆ Let P : u − v be a maximal path in T. we have dT (v) = 1. r) is . and ε T = ∑k=1 |Ei |. ⊔ ⊓ Example 2. T is connected. and from each pair only the winner continues. and after the second round 4 teams continue. there exists an edge uv ∈ EG such that u ∈ T and v ∈ T. If k is odd. a spanning graph that is a tree. for which we have proved that |Ei | = |Vi | − 1. ⊔ ⊓ Example 2. How many plays are needed to determine the winner? So if there are 14 teams. Therefore. that is. νT = ∑k=1 |Vi |.4. Proof. a contradiction.2. Proof. Ei ). Consider a cup tournament of n teams.

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

Let ek be the ﬁrst edge produced by the algorithm that is not in T1 . When we replace e by ek in T1 . When applied to the weighted graph on the right. The length of an address is the number of its bits. (n)? 2 In such a weighted tree T α different paths have different weights. A message that is sent to v is preceded by the address a(v). This would make it easier for a node computer to forward a message. Also. then a cycle C containing ek is created. e4 = v2 v3 and e5 = v1 v2 . h(00010. e4 = v3 v6 .g. of course) for a graph. e. We now show that T has the minimum total weight among the spanning trees of G. e5 = v1 v2 gives another optimal solution (of weight 9). No solutions are known for any n ≥ 7. Consider trees T with weight functions α : ET → N. it is connected and has n − 1 edges.1 Bipartite graphs and trees 22 By the construction. such that the total weight does not increase. Example 2. Note that T2 has more edges in common with T than T1 . a(u)) of two addresses of the same length is the number of places. the algorithm produces the sequence: e1 = v2 v4 .6. and therefore α( T2 ) ≤ α( T1 ). En−1 ) is a spanning tree of G. . the selection e1 = v2 v5 . Indeed. 3 2 1 4 1 2 v4 1 v5 3 2 v6 2 2 v1 v2 v3 Problem. their addresses should differ by one symbol.2. e3 = v5 v6 . 1 5 2 8 4 TAYLOR (1977) proved: if T of order n exists. where the vertices are the node computers. C must contain an edge e that is not in T. because it contains no cycles. e3 = v3 v6 . if two vertices were adjacent. 01100) = 3 and h(10000. Also. In particular. We deduce that α( T ) ≤ α( T1 ). and the edges indicate the direct links. Also. . α must be injective. by the construction. Each computer v has an address a(v). 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. we still have a spanning tree. Each tree T of order n has exactly (n) paths. . there may exist several optimal spanning trees (with the same weight. and each i ∈ [1. 00000) = 1. However. we can transform T1 to T by replacing edges. α(ek ) ≤ α(e).. If we add ek to T1 . T = (VG . . then necessarily n = k2 or n = k2 + 2 for some k ≥ 1. a bit string (of zeros and ones).7. one by one. A computer network can be presented as a graph G. Repeating the above procedure. where a(v) and a(u) differ. . e2 = v4 v5 . e2 = v4 v5 . Example 2. (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. say T2 . Suppose T1 is any spanning tree of G. The outcome of Kruskal’s algorithm need not be unique. ( n)] is a weight 2 of one path. The Hamming distance h( a(v). 2. The total weight of the spanning tree is thus 9.

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

if u and v belong to different connected components of G −S. then the claim is clear. • κ ( G ) = 0. If νG ≤ 2. and let uv ∈ G. The following lemma is immediate. G − F disconnected. On the right. then either G = K2 or one of u or v is a cut vertex. if G is connected. and • otherwise κ ( G ) equals the minimum size of a separating set of G. ⋆ Lemma 2. Proof.2. The claim follows from this. For νG ≥ 3. {v} is a separating set. if κ ( G ) ≥ k. if κ ′ ( G ) ≥ k. that is. if G is a complete graph. Now VG \ {u. If G is connected. Let κ ′ ( G ) = min{k | k = | F |. if G is disconnected. A minimal edge cut F ⊆ EG is a bond (F \ {e} is not an edge cut for any e ∈ F). → Lemma 2. v)separating set.8. Lemma 2. S ⊆ VG } . Clearly. D EFINITION . The (vertex) connectivity number κ ( G ) of G is deﬁned as κ ( G ) = min{k | k = |S|. then every path P : u − v visits a vertex of S. In other words. If S ⊆ VG separates u and v. Example 2. then κ ′ ( G ) = 0. then v is a cut vertex if and only if G −v is disconnected. A subset S ⊆ VG is a separating set. For trivial graphs. G −S disconnected or trivial. A graph G is k-edge connected. Let G be connected. We also say that S separates the vertices u and v and it is a (u. then it is 1-connected. if G is disconnected.4. F ⊆ EG } . v} is a separating set. . If e = uv is a bridge. If a connected graph G has no separating sets. ⊔ / ⊓ D EFINITION . • κ ( G ) = νG − 1. A vertex v ∈ G is a cut vertex.3. κ ( G ) = 2 and κ ′ ( G ) = 2. assume that G is not complete. A graph G is k-connected. Again.5. An edge cut F of G consists of edges so that G − F is disconnected. Notice that the minimum degree is δ( G ) = 3.2 Connectivity 24 Separating sets D EFINITION . if G −S is disconnected. then it is a complete graph. let κ ′ ( G ) = 0. if c( G −v) > c( G ).

In both of these cases. By Lemma 2. then G is connected and contains a bridge. . Now. F Now for each f ∈ F \ {e} choose an end different from u and v.2 Connectivity 25 Proof. if G −S is connected. since H −e = G − F. then S is a separating set and so κ ( G ) ≤ |S| ≤ κ ′ ( G ) − 1 and we are done. and G − F has two connected components. and F is minimal. On the other hand. . the subgraph H = G −( F \ {e}) is connected for given e ∈ F. ⊔ ⊓ Lemma 2. v)-separating. Let u. either νGu ≥ 2 (and u is a cut vertex) or νGv ≥ 2 (and v is a cut vertex). there is a vertex of G −S. → Hence |S| is at least the number of independent paths from u to v. By Lemma 2.6 (W HITNEY (1932)). Hence e is a bridge in H. κ ( G ) ≤ κ ′ ( G ) ≤ δ( G ) . and the claim follows. (The choices for different edges need not be different. Proof.6. Since G − F is disconnected.2. e . Thus |S| ≤ | F | − 1 = κ ′ ( G ) − 1. Assume then that κ ′ ( G ) ≥ 2. Let Gu = ∗ ∗ NG−e (u) and Gv = NG−e (v) be the connected components of G −e containing u and v.) Note that since f = e. H . and G −S does not contain edges from F \ {e}. Menger’s theorem Theorem 2. v ∈ G be nonadjacent vertices of a connected graph G. and let e = uv ∈ F. ⊔ ⊓ Theorem 2. whose removal results in a trivial or a disconnected graph. since G is connected. where e is a bridge. then either G −S = K2 (= e). κ ′ ( G ) ≤ δ( G ). In conclusion. Assume that G = K2 and thus that νG ≥ 3. If G −S is disconnected. κ ( G ) ≤ |S| + 1 ≤ κ ′ ( G ). . also κ ( G ) = 1. then G is disconnected. ⋆ Proof. Assume G is nontrivial. Consider the connected subgraph H = G −( F \ {e}) = ( G − F ) + e. If F be a bond of a connected graph G. If κ ′ ( G ) = 1. or either u or v (or both) is a cut vertex of G −S (since H −S = G −S. If κ ′ ( G ) = 0. . Let F be an edge cut of G with | F | = κ ′ ( G ). . then every path u − v of G visits S. . either G = K2 or G has a cut vertex. ⊔ ⊓ G .7 (M ENGER (1927)). and therefore G −S ⊆ H is an induced subgraph of H). . either end of f is different from u or v. Proof. Then the minimum number of vertices separating u and v is equal to the maximum number of independent paths from u to v. For any graph G. . then c( G − F ) = 2. and thus c( G − F ) = 2.5.1. Let S be the collection of these choices. we disconnect G. If a subset S ⊆ VG is (u. In both of these cases. c( H −e) = 2. . and in this case also κ ( G ) = 0. Then F is a bond. . since if we remove all edges with an end v. Clearly.

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

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. ⊔ ⊓

Notice that.3. v} has an even degree. The k-cube Qk is eulerian for even integers k. i ]) so that (i) ei+1 has an end vi . then any Euler tour will do as a solution. In H every vertex ⋆ has an even degree. → ⊔ ⊓ The Chinese postman The following problem is due to G UAN M EIGU (1962). Exercise. Let G be a graph with a weight function α : EG → R + . A connected graph has an Euler trail if and only if it has at most two vertices of odd degree.3. because Qk is k-regular. Consider a village. where e j = v j−1 v j . . Otherwise. . the weights α(e) play no role in the eulerian case. by the handshaking lemma. ei has been constructed. and therefore G has exactly two such vertices. as in the proof of Theorem 3. . ei }. If G is eulerian. ⋆ Proof. / Assume then that G is connected and has at most two vertices of odd degree. and let W0 be the trivial path on v0 . the post ofﬁce). If G has an Euler trail u − v. unless there is no alternative. then any trail of G constructed by Fleury’s algorithm is an Euler tour of G.2. and (ii) ei+1 is not a bridge of Gi = G −{e1 .1. ⊔ ⊓ .1. This problem is akin to Euler’s problem and to the shortest path problem. . and hence it has an Euler tour. every graph has an even number of vertices with odd degree. 2. and the edges uw and vw. G has an Euler trail. Here the → → → ⋆ beginning part u − v is an Euler trail of G. and there is a good algorithm for ﬁnding such a tour: Fleury’s algorithm: • Let v0 ∈ G be a chosen vertex. as is natural. . If G is eulerian. then. as long as possible: suppose a trail Wi = e1 e2 . each vertex → w ∈ {u. Proof. The Chinese postman problem is to ﬁnd a minimum weighted tour in G (starting from a given vertex.1. If G has no vertices of odd degree then. Let H be a graph obtained from G by adding a vertex w. . where a postman wishes to plan his route to save the legs. by Theorem 3. Theorem 3. but still every street has to be walked through. . In this case the weight of the optimal tour is the total weight of the graph G. because such a tour traverses each edge exactly once and this is the best one can do. • Repeat the following procedure for i = 1.1 Eulerian graphs 30 Example 3. say u − v − w − u. . say u and v. Theorem 3. . Choose an edge ei+1 (= e j for j ∈ [1.

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

Assume thus that uv ∈ G. and let u. k]. then there is nothing to prove. − vn = v is a Hamilton → → → → → . Then G is hamiltonian if and only if G + uv is hamiltonian. If G is hamiltonian. Unlike for eulerian graphs (Theorem 3. c( G −S) = 3 > 2 = |S| for the set S of black vertices.2 Hamiltonian graphs 32 bipartition ( X. but it does satisfy the condition c( G −S) ≤ |S| for all S = ∅. Gi . (⇐) Denote e = uv and suppose that G + e has a Hamilton cycle C. then C is also a Hamilton cycle of G + uv. Suppose thus that e is on C. Theorem 3.1. Y ). Now. .3.4. Y equally many times. Let u. But then necessarily | X | = |Y |. We may ⋆ then assume that C : u − v − u. however. In G. There are. and hence suppose that k > 1. due to O RE. which appears in many places in graph theory as a counter example for various conditions.m has even length as the graph is bipartite. then it is a Hamilton cycle of G. / (⇒) This is trivial since if G has a Hamilton cycle C. Interestingly this graph is ( X. each cycle in Kn. Example 3. Therefore the conclusion of Lemma 3. Y )-bipartite of even order with |X | = |Y |. Lemma 3.1. since X and Y are stable subsets. and thus the cycle visits the sets X. Consider the graph on the right. the problem to determine if G is hamiltonian is NPcomplete. and hence it is not hamiltonian. ⊔ ⊓ Example 3. and let C : u − u be a Hamilton cycle of G. because C is a cycle and ui vi ∈ G for all i. This graph is not hamiltonian. If C does not use the edge e. The case k = 1 is trivial. where | X | = n and |Y | = m. i ∈ [1. Consider the Petersen graph on the right.3. and let vi be the vertex that follows ui in C. v ∈ G be such that dG (u) + dG (v) ≥ n. Assume → G −S has k connected components. If uv ∈ G. Thus |S| ≥ k as required. Let G be a graph of order νG ≥ 3. The following theorem. u ∈ S. Now vi ∈ S for each i by the choice of ui .1) no good characterization is known for hamiltonian graphs. c( G − S ) ≤ |S | . It is also 3-regular. Indeed. and v j = vt for all j = t. Let ui be the last vertex of C that belongs to Gi . Therefore G does not satisfy the condition of Lemma 3.4 (O RE (1962)). then for every nonempty subset S ⊆ VG . Let ∅ = S ⊆ VG . . generalizes an earlier result by D IRAC (1952). Denote n = νG . some interesting general conditions. v ∈ G be such that dG (u) + dG (v) ≥ νG . Now u = v1 − v2 − . Proof.1 is not sufﬁcient to ensure that a graph is hamiltonian. ⋆ Proof.

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

Consequently. in this case. and thus G is hamiltonian. Let G be a graph of order νG ≥ 3.1) | A | = ( n − 1) − d G ( v ) ≥ d G ( u ) = i . (3. Then G is hamiltonian. . the ﬁrst example where DNA computing was applied. We show that. because G is hamiltonian if and only if cl ( G ) is hamiltonian. Consequently. moreover. and adding edges to G does not decrease any of its degrees. for each vertex from B = {w | uw ∈ G.2 Hamiltonian graphs 34 Theorem 3. and therefore dG (u) = i < n/2. and.) There are some general improvements of the previous results of this chapter. was the hamiltonian problem. . Theorem 3. and let uv ∈ G with dG (u) ≤ dG (v) be / such that dG (u) + dG (v) is as large as possible. The second claim is immediate. Let G be a graph with VG = {v1 . vn }. If for every i < n/2. since now dG (u) + dG (v) ≥ νG for all u. / By our choice.1). and quite many improvements in various special cases. Assume on the contrary that G = Kn . This contradicts the obtained bound di ≤ i and the condition (3. we have cl ( G ) = Kn for n = νG . if δ( G ) ≥ 1 νG . at least partly because the problem has practical signiﬁcance. so does G + e for every e. . Also dG (u) < n − i. . ⊔ ⊓ Chvátal’s condition The hamiltonian problem of graphs has attracted much attention. Proof. Similarly. then G is 2 hamiltonian. v2 . dG (w) ≤ dG (v) < / n − dG (u) = n − i. . w = u}. then G is hamiltonian. where the graphs are somehow restricted. In particular. if G satisﬁes (3.3. Proof. that is. Suppose that for all nonadjacent vertices u and v. we must have dG (u) + dG (v) < n. v ∈ G whether adjacent or not.1).7 (C HVÁTAL (1972)). We become satisﬁed by two general results. G = cl ( G ). ⊔ ⊓ Note that the condition (3. ordered so that d1 ≤ d2 ≤ · · · ≤ dn . Because G is closed. w = v}. and so di ≤ dG (u) = i.6. for di = dG (vi ). di ≤ i =⇒ dn−i ≥ n − i . and | B | = ( n − 1) − d G ( u ) = ( n − 1) − i . (Indeed. G = Kn .1) is easily checkable for any given graph. Let A = {w | vw ∈ G. for n ≥ 3. Since dG (u) + dG (v) ≥ νG for all nonadjacent vertices. dG (u) + dG (v) ≥ νG . and thus G is hamiltonian. First of all. we may suppose that G is closed. dG (w) ≤ i for all w ∈ A. dn−i < n − i. there are at least i vertices w with dG (w) ≤ i. and thus there are at least n − i vertices w with dG (w) < n − i.

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

e1 . the ends u and v are not saturated by M. Theorem 3. . . . each connected component (path or a cycle) A alternates between N and M. Now. Then G contains a matching M saturating X if and only if |S| ≤ | NG (S)| for all S ⊆ X. known as the Theorem 3. Y )-bipartite graph. By Lemma 3. ⊔ ⊓ Example 3. .3. e4 . consequently. (⇒) Let a matching M have an M-augmented path P = e1 e2 .8 (B ERGE (1957)). We have d H (v) ≤ 2 for each v ∈ H. Let G be a ( X. Indeed. . Hence A : u − v is a path (of → odd length). N is a matching of G. which has more edges from N than from M. Each maximum matching of Qk has 2k−1 edges. e2k+1 in G. A matching M of G is a maximum matching if and only if there are no M-augmented paths in G. (3. Deﬁne N ⊆ EG by / N = ( M \ {e2i | i ∈ [1. Now. has 2k−1 edges. and | N | = | M | + 1.5. Y )-bipartite. Consider the k-cube Qk for k ≥ 1. Proof. . because an alternating cycle has even length. denote NG (S) = {v | uv ∈ G for some u ∈ S} . This A cannot be a cycle. then NG (S) ⊆ Y. each connected component A of H is either a path or a cycle. k]}) ∪ {e2i+1 | i ∈ [0. This proves the theorem. and it ⋆ thus contains equally many edges from N and M. but M is not. 1u) | u ∈ B k−1 }. Hall’s theorem For a subset S ⊆ VG of a graph G.2) .3. since | N | > | M |. e2k ∈ M. . . Consider the subgraph H = G [ M △ N ] for the symmetric difference M △ N. Because A is a connected component of H. and it is clearly perfect. because v is an end of at most one edge in M and N. e2k+1 ∈ M. . e3 . k]} . Therefore M is not a maximum matching. Since no v ∈ A can be an end of two edges from N or from M. which starts and ends with an edge from N. the matching M = {(0u. (⇐) Assume N is a maximum matching. Hence | N | > | M |. and S ⊆ X. and. If G is ( X. The following result. . there is a connected component A of H. A is an M-augmented path.9 (H ALL (1935)).3 Matchings 36 matching in an iterative manner starting from any matching M and from any Maugmented path. Here e2 .

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

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

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

Theorem 3. i ∈ [1. be the odd connected components of G −S. There is a good algorithm due to E DMONDS (1965). G has a perfect matching. codd ( G −S) ≤ |S|. If G is a bridgeless 3-regular graph. v∈S The ﬁrst of these implies that mi = ∑ v ∈ Gi d G ( v ) − 2 · ε Gi is odd. mi = 1. The simplest connected graph that has no perfect matching is the path P3 . (i) G has a perfect matching. Theorem 3. Let G be a nontrivial graph. then it has a perfect matching. and let Gi . Here removing the middle vertex creates two odd components. and therefore mi ≥ 3. since for a perfect matching we have m( G ) = (1/2)νG .3. Its applications are mainly in the proofs of other results that are related to matchings.7. The next 3-regular graph (known as the Sylvester graph) does not have a perfect matching.11 (T UTTE (1947)).3 Matchings 40 For perfect matchings we have the following corollary. Denote by mi the number of edges with one end in Gi and the other in S. 3 i=1 v∈S and so. Let S be a proper subset of VG . because removing the black vertex results in a graph with three odd connected components. because G has no bridges. Since G is 3-regular. Using Theorem 3. Example 3. because the theorem requires ‘too many cases’. ∑ v ∈ Gi dG (v) = 3 · νGi and ∑ d G ( v) = 3 · |S | . Proof. ⊔ ⊓ . t]. The following are equivalent. but this algorithm is somewhat involved.12 (P ETERSEN (1891)). Furthermore. which uses ‘blossom shrinkings’. Tutte’s theorem does not provide a good algorithm for constructing a perfect matching. by Theorem 3. This graph is the smallest regular graph with an odd degree that has no perfect matching. (ii) For every proper subset S ⊂ VG .11.11 we can give a short proof of P ETERSEN’s result for 3-regular graphs (1891). Hence the number of odd connected components of G −S satisﬁes t≤ 1 1 t ∑ m i ≤ 3 ∑ d G ( v) = |S | .

3 Matchings 41 Stable Marriages D EFINITION .14. each vertex x ∈ G supplies an order of preferences of the vertices of NG ( x). We omit the proof of the next theorem. In addition. the graph on the right does have a perfect matching (of 4 edges). 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. Then iterate the following process until all vertices are saturated: Choose a vertex x ∈ X that is unsaturated in Mi−1 . A stable matching need not saturate X and Y. There is a catch. Let y ∈ Y be the most preferred vertex for x such that y ∈ P( x).) Theorem 3. For instance. the procedure terminates.3. The algorithm by G ALE works as follows.n be a complete bipartite graph. . We write u < x v. a stable matching exists for all lists of preferences. For bipartite graphs G. That was the good news. Let G = Kn. 47}. is a perfect matching. 35. it is not the case that x and y prefer each other better than their matched companions: xv ∈ M and y < x v. v ∈ X. if x ∈ Y. A stable matching of G is the following: M = {28. (2) If y is not saturated. Let the bipartition be ( X. say M = Mt . The ﬁnal outcome. Then G has a perfect and stable matching for all lists of preferences. if for each unmatched pair xy ∈ M / (with x ∈ X and y ∈ Y). since a vertex x ∈ X takes part in the iteration at most n times (once for each y ∈ Y). if x prefers v to u. v ∈ Y. and P( x) = ∅ for all x ∈ X. then set Mi = ( Mi−1 \ {zy}) ∪ { xy}.8. which leaves 1 and 6 unmatched. / (1) Add y to P( x). First of all. Proof. Theorem 3. (3) If zy ∈ Mi−1 and z <y x. Y ). Example 3. (Here u.) A matching M of G is said to be stable. of course. or uy ∈ M and x <y u. then set Mi = Mi−1 ∪ { xy}. Set M0 = ∅. AND S HAPLEY (1962) Procedure. and u. Y ) of the vertex set. since the iteration continues until there are no unsaturated vertices x ∈ X.13. if x ∈ X. Consider a bipartite graph G with a bipartition ( X.

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

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

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

The following important theorem. Proof. α ( e ). ∆( G ) ≤ χ′ ( G ) ≤ ∆( G ) + 1. Let α be an optimal k-edge colouring of G. and if α is proper. for all vertices v. Theorem 4. Let α be an optimal ∆-edge colouring of a bipartite G. and in this case α is optimal. Notice that we always have cα (v) ≤ dG (v). . Let ∆ = ∆( G ). due to V IZING. Furthermore. By Lemma 4. H has a 2-edge colouring γ : E H → {i.4. By Lemma 4.3.4. Suppose on the contrary that χ′ ( G ) > ∆ + 1. Then the connected component H of G α [i. An edge colouring α of G is proper if and only if cα (v) = dG (v) for all vertices v ∈ G. Lemma 4. α is optimal.) We obtain a recolouring β of G as follows: β( e) = γ ( e ). is an odd cycle.2 (V IZING (1964)). Therefore ∑u∈ G c β (u) > ∑u∈ G cα (u). if e ∈ H . by the construction of β. c β (v) = cα (v) + 1. if e ∈ H . G would contain an odd cycle. Hence H is an odd cycle. and let α be an optimal (∆ + 1)-edge colouring of G. Therefore. j] that contains v. if it cannot be improved. shows that the edge chromatic number of a graph G misses ∆( G ) by at most one colour. Thus an improvement of a colouring is a change towards a proper colouring. For any graph G. and let v ∈ G. we have c β (u) ≥ cα (u) for all u = v.4. ⊔ ⊓ Vizing’s theorem In general we can have χ′ ( G ) > ∆( G ) as one of our examples did show. cα (v) = dG (v). / 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. Suppose the connected component H is not an odd cycle. j}. which contradicts the optimality of α.3. If G is bipartite.2. If there were a v ∈ G with cα (v) < dG (v). then χ′ ( G ) = ∆( G ). Proof. ⊔ ⊓ Theorem 4. But a bipartite graph does not contain such cycles. Note also that a graph G always has an optimal k-edge colouring. Suppose that the colour i is available for v. if 45 ∑ c β ( v) > ∑ cα ( v) . Lemma 4. Proof.1 (K ÖNIG (1916)). v∈ G v∈ G Also. α is a proper colouring. in which both i and j are incident with each vertex x with d H ( x) ≥ 2. where ∆ = ∆( G ). The next lemma is obvious. then cα (v) = dG (v).1 Edge colourings A k-edge colouring β is an improvement of α. and the colour j is incident with v at least twice. (We have renamed the colours 1 and 2 to i and j. then by Lemma 4. but it need not have any proper k-edge colourings. and ∆ = χ′ ( G ) as required. We need only to show that χ′ ( G ) ≤ ∆ + 1.

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

4. ir ] and H2 of G γ [i0 . that is. However. v5 − u5 . Since C ′ needs three colours.2 Ramsey Theory 47 By Claim 1. 2 3 1 1 1 1 2 3 2 3 4. the connected components H1 of G β [i0 . The answer is known (only) for some special classes of graphs. Observe that every vertex is adjacent to all colours 1. Indeed. 3. H1 is a cycle (vur −1 ) P1 (ur v) and H2 is a cycle (vur −1 ) P2 (ut v). Vizing’s theorem (nor its present proof) does not offer any characterization for the graphs. where ⋆ ⋆ both P1 : ur −1 − ur and P2 : ur −1 − ut are paths. Ramsey theory studies unavoidable patterns in combinatorics. χ′ ( G ) = 3 or 4. v4 − u4 . In fact. cannot be a colour of any edge in C Edge Colouring Problem. since P1 ends in ur while P2 ends in a different vertex ut . In graph theoretic terms this means that each colouring of the edges of K6 with 2 colours results in a monochromatic triangle.2. v3 − u3 . This is not possible. 2. By H OLYER (1981). then arbitrarily in two colours).4. → → → → → Hence v1 − u1 . A typical example of a Ramsey property is the following: given 6 persons each pair of whom are either friends or enemies. ir ] containing the vertex v are cycles. the claim follows. By Lemma 4. v2 − u2 . . by Vizing’ theorem. . Suppose 3 colours sufﬁce. . Let C : v1 − . − u5 − u1 the inner cycle of G such that vi ui ∈ EG → → → for all i. Now C uses one colour (say 1) once and the other two twice. 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. This contradiction proves the theorem. This can be done uniquely (up to permutations): v1 − v2 − v3 − v4 − v5 − v1 . when the word ‘optimal’ is forgotten: colour ﬁrst the edges as well as you can (if nothing better. there are then 3 persons who are mutual friends or mutual enemies. there is a colour i0 = b(v) that is available for v. the problem whether χ′ ( G ) is ∆( G ) or ∆( G ) + 1 is NP-complete. this means that 1 → → → → → ′ . We consider an instance of this theory mainly for edge colourings (that need not be proper). The proof of Vizing’s theorem can be used to obtain a proper colouring of G with at most ∆( G ) + 1 colours. ⊔ ⊓ Example 4. it is one of the famous open problems of graph theory to ﬁnd such a characterization. . We show that χ′ ( G ) = 4 for the Petersen graph. − v5 − v1 be → → → the outer cycle and C ′ : u1 − . However. .2 Ramsey Theory In general. the edges of P1 and P2 → → have the same colours with respect to β and γ (either i0 or ir ). for which χ′ ( G ) = ∆( G ) + 1.

K p H. and there is nothing to prove. and that every nondiscrete graph contains K2 . One can compute that the number ε H of edges of H is equal to r p−2 2 r n − 1− . G2 . H p−1 of order t (when t > 0). i ∈ [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. p) edges.3 is applied to triangles K3 . Let p ≥ 3. Note that a complete p-partite graph is completely determined by its discrete parts Gi . Hr of order t + 1 and p − 1 − r parts Hr +1 . . we can apply the induction hypothesis to obtain ε G− A ≤ T (n − p + 1. . 2 ⊔ ⊓ When Theorem 4. ( p − 1)( p − 2) = T (n. p) = 2( p − 1) 2 p−1 The next result shows that the above bound T (n. and νG− A = n − p + 1. . where uv ∈ G if and only if u and v belong to different parts. . K p G − A. Gi and Gj with i = j. If t = 0. p) .2) T (n. such that there are r parts H1 . p) = n(n − 1)/2. p) is optimal. . Let n = ( p − 1)t + r for 1 ≤ r ≤ p − 1 and t ≥ 0. and is thus determined by n and p. 1 2 4 νG edges.3 (T URÁN (1941)). A complete p-partite graph G consists of p discrete and disjoint induced subgraphs G1 .4.1 ( M ANTEL (1907)). p) + (n − p + 1)( p − 2) + which proves the claim. then G contains a complete subgraph K p . p]. . Theorem 4. then T (n. Because n − p + 1 = (t − 1)( p − 1) + r. (Here r is the positive residue of n modulo ( p − 1).) By its deﬁnition. . and p − 1 vertices of this K p induce a subgraph K p−1 ⊆ G. . Now G contains a complete subgraph G [ A] = K p−1. otherwise G [ A ∪ {v}] = K p . Corollary 4. Now ε G ≤ T (n − p + 1. Suppose then that t ≥ 1. . . and let G be a graph of order n such that ε G is maximum subject to the condition K p G. Each v ∈ A is adjacent to at most p − 2 vertices of A. We prove the claim by induction on t. (4. It is clear that every graph contains K1 . Proof. we have the following interesting case. p). G p ⊆ G. and let H = Hn. p − 1] and t ≥ 0.p be the complete ( p − 1)-partite graph of order n = t( p − 1) + r. . / Also. since adding any one edge to G results in a K p . where r ∈ [1. D EFINITION . If a graph G has ε G > triangle K3 . then G contains a . . If a graph G of order n has ε G > T (n.

G [ NG (v)] contains a complete subgraph B of order p − 1 or a stable subset S of order q. then it has a complete subgraph of order p or a stable subset of order q. and in the second case the same stable set of order q is good for G. any 2-edge colouring of Kn → [1. q). Proof. The Ramsey number R( p. q) = q. then G [ A] contains a complete subgraph of order p or a stable subset S of order q − 1. Then there exists a (smallest) integer R( p. ⊔ ⊓ . The number R( p. Let α be an edge colouring of G.4 and 4. It is clear that R( p. S ∪ {v} is a stable subset of G of q vertices. Let v ∈ G. q − 1) + R( p − 1. q). q − 1) (or both). Before proving this.6. and R( p. q) such that for all n ≥ R( p. q − 1) + R( p − 1. We use induction on p + q. It is clear that R( p. q ≥ 2. any graph G of order n contains a complete subgraph of order p or a stable set of order q. q). By the deﬁnition of Ramsey numbers. q) . Theorems 4. In the ﬁrst case.4 (R AMSEY (1930)).5 follow from the next result which shows (inductively) that an upper bound exists for the Ramsey numbers R( p. Theorem 4. 2] contains a 1-monochromatic K p or a 2-monochromatic Kq . q). q) is known as the Ramsey number for p and q. q ≥ 2 be any integers. q) exists for p = 2 or q = 2. This proves the claim. if G [ X ] is a discrete graph.2 Ramsey Theory 49 Ramsey’s theorem D EFINITION . and denote by A = VG \ ( NG (v) ∪ {v}) the set of vertices that are not adjacent to v. q) such that for all n ≥ R( p. q ≥ 2 be any integers. Let p. Since G has R( p. and it is thus exists for p + q ≤ 5. Recall that a subset X ⊆ VG is stable. q) exists for all p. 2) = p and R(2. q) − 1 vertices different from v. Theorem 4.5. we give an equivalent statement. A subgraph H ⊆ G is said to be (i-) monochromatic. Theorem 4. Be patient. Let p. If | A| ≥ R( p. q − 1) + R( p − 1. Assume ﬁrst that | NG (v)| ≥ R( p − 1. It is now sufﬁcient to show that if G is a graph of order R( p. Then there exists a (smallest) integer R( p. q) ≤ R( p. if all edges of H have the same colour i. q) or | A| ≥ R( p. and in the second case. The following theorem is one of the jewels of combinatorics.4. q − 1). either | NG (v)| ≥ R( p − 1. this will follow from Theorem 4. B ∪ {v} induces a complete subgraph K p in G. In the ﬁrst case. q).6 (E RDÖS and S ZEKERES (1935)). the same complete subgraph of order p is good for G.

4. q ≥ 2. . R( p. Proof. k − 1] be obtained from α by identifying the colours k − 1 and k: β( e) = 1 α(e) if α(e) < k − 1 . Theorem 4. . For all p. qk ). any k-edge colouring of Kn has an i-monochromatic Kqi for some i. We use induction on p + q for the general statement. 2000 on the Web . The case k = 2 is treated in Theorem 4.P. qk ) ≤ R(q1 .6 and the induction hypothesis.. . . q) p+q−3 + p−1 p+q−3 p−2 p+q−2 . . Let β : EKn → [1. By Theorem 4.7 ( E RDÖS and S ZEKERES (1935)). As can be read from the table1 .8. k] with k ≥ 2. where p = R(qk−1 . qk−2 . Then there exists an integer R = R(q1 . For p = 2 or q = 2. R( p. the claim is clear. . Assume that p. Generalizations Theorem 4. qk−2 . Electronic J. = ⊔ ⊓ In the table below we give some known values and estimates for the Ramsey numbers R( p. . we show that R(q1 . . q) ≤ p+q−2 . q − 1) + R( p − 1. q ≥ 3. . The proof is by induction on k. . k] be an edge colouring. q2 . p−1 50 Proof. Theorem 4. p). . Let n = R(q1 . of Combin. . For k > 2. and let α : EKn → [1. . Small Ramsey numbers. qk ) such that for all n ≥ R. Let qi ≥ 2 be integers for i ∈ [1. . It has been veriﬁed that 43 ≤ R(5.4.4 can be generalized as follows. . p−1 ≤ which is what we wanted. q). not so much is known about these numbers. q) ≤ R( p. . p) (where p = q) is for p = 5. S. but to determine the exact value is an open problem. 5) ≤ 49. k − 1 if α(e) = k − 1 or k . p). R ADZISZOWSKI.2 Ramsey Theory A concrete upper bound is given in the following result. 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.

However. R2 (K3. 3n/2 − 1 if n ≥ 6 and n even. . H α and thus monochromatic subgraph H p α Kn has a (k − 1)-monochromatic or a k-monochromatic subgraph. by Theorem 4. . . even in more general cases. we have Corollary 4. Kn has an i-monochromatic Kqi for some 1 ≤ i ≤ k − 2 α (and we are done. H2 . Below we list some of these results for cases.4.3. .4. 2 6 if n = 3 or n = 4. R2 (K1. ⊔ ⊓ β Since for each graph H. . and this proves the claim. . . . Then there exists an integer R( H1 .2 Ramsey Theory β 51 By the induction hypothesis.2. . Hk ) such that for all complete graphs Kn with n ≥ R( H1 . . since this subgraph is monochromatic in Kn ) or Kn has a (k − 1)β = K . . for some dear graphs (see R ADZISZOWSKI’s survey). . R2 (Cn ) = 2n − 1 if n ≥ 5 and n odd . The best known lower bound of R2 ( G ) for connected graphs was obtained by B URR AND E RDÖS (1976).3 ) = 10. Kn contains an i-monochromatic subgraph Hi for some i. Hk ) can be much smaller than their counter parts (for complete graphs) in Theorem 4. H ⊆ Km for m = νH . H2 . then R( T. In the latter case. any graph G of order at least R( T. .8. G. To this end. . . . that is. R2 (K2. We leave the following statement as an exercise: If T is a tree of order m. the generalized Ramsey numbers R( H1 . Hk ) α and for all k-edge colourings α of Kn . H2 . This generalization is trivial from Theorem 4.8. .3 ) = 18. if n is odd. G ) (k times G ). Kn ) = (m − 1)(n − 1) + 1 . Example 4.2. where the graphs are equal. Examples of Ramsey numbers∗ Some exact values are known in Corollary 4. . Let k ≥ 2 and H1 . . Kn ) contains a subgraph isomorphic to T. Here is a list of some special cases: n R2 ( Pn ) = n + − 1. . let Rk ( G ) = R( G.n ) = 2n − 1 2n if n is even. . H2 . or the complement of G contains a complete subgraph Kn . R2 ( G ) ≥ 4νG − 1 3 ( G connected). Hk be arbitrary graphs.

Sn of N S(n) . 10. Theorem 4. 4. but also. Indeed. . The following gives a short proof using Ramsey’s theorem. if |i − j| ∈ Sk . For each n ≥ 1. The following impressive theorem improves the results we have mentioned in this chapter and it has a difﬁcult proof. We observe that in no partition class there are three integers such that x + y = z.L. much less is known. there exists an integer S(n) such that any partition S1 .4. where x + y = z has a solution.9 (D EUBER. It is a cycle Cn−1 . consider the partition {1.. . that is. 7. The value R2 (K2. the only nontrivial result for complete graphs is: R3 (K3 ) = 17. It follows from Theorem 4.4. Needless to say that no exact values are known for Rk (Kn ) for k ≥ 4 and n ≥ 3. By Theorem 4. 128 ≤ R3 (K4 ) ≤ 235. we shortly describe Schur’s theorem. H AJNAL. . Sn of N S(n) has a class Si containing two integers x. where 3 occurs n times. and in general. t − i ∈ Sk for some k. E RDÖS. For three colours. . . . {5. 11. Then there exists a graph G such that any 2-edge colouring of G has an monochromatic induced subgraph H. then you are bound to ﬁnd a class. let S(n) = R(3. Let Wn denote the wheel on n vertices. S CHUR (1916) solved this problem in a general setting. there exists a graph G (well.g. e. R OTHSCHILD AND J. however. any sufﬁciently large complete graph) such that any 2-edge colouring of G has a monochromatic (induced) subgraph Kn . For the square C4 . R2 (K2. Example 4. G RAHAM . Note. For this. 3. B. .17 ) is either 65 or 66.) 1990. 12}. and let K be a complete on N S(n) . (2nd ed. where a vertex v with degree n − 1 is attached. there are three vertices i. and 385 ≤ R3 (K5 ). For a partition S1 . S PENCER. .4 that for any complete Kn . 9} of the set N13 = [1. 3). 8. Also. . Let H be any graph. deﬁne an edge colouring α of K by α(ij) = k. that in Corollary 4. y such that x + y ∈ Si .8. . P ÓSA. j − i.n ) ≤ 4n − 2. 13]. Then R2 (W5 ) = 15 and R2 (W6 ) = 17. . but no nontrivial upper bound is known for R3 (K5 ). K α has a monochromatic triangle. 13}.2 the monochromatic subgraph Hi is not required to be induced. 3. and R ÖDL (around 1973)).2 Ramsey Theory 52 The values R2 (K2.L. As an application of Ramsey’s theorem. Wiley. see R. In fact. But (t − j) + ( j − i) = t − i proves the claim. geometry and algebra. . Note that W4 = K4 . if you try to partition N14 into three classes. {2. we know that R3 (C4 ) = 11. .L. t such that 1 ≤ i < j < t ≤ S(n) with t − j. However. j. “Ramsey Theory”. There are quite many interesting corollaries to Ramsey’s theorem in various parts of mathematics including not only graph theory. 6.n ) are known for n ≤ 16.

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

Then α(ui ) = k. Therefore χ( G t ) = k + 1. Let G be any graph with k = χ( G ). Since G is critical. we obtain larger triangle -free graphs with high chromatic numbers. Connectivity claim follows from this observation. and hence there is an available colour i for v. but δ( G ) = dG (v) ≤ k − 2 for v ∈ G. so does {vi . v j . Secondly. If {ui . there is a proper (k − 1)-colouring of G −v. This gives a proper (k − 1)-colouring to G. But there are k colours. since in Kn −(uv) the vertices u and v can gain the same colour. ⊔ ⊓ The case (iii) of the next theorem is due to S ZEKERES Theorem 4.4. vk } by the deﬁnition of G t . A k-chromatic graph G is said to be k-critical. then a proper (k − 1)-colouring is obtained for G. vk } induces a triangle. but the latter triangle is already in G. Example 4. If α is a proper k-colouring of G. Assume that α is a proper k-colouring of G t . AND W ILF (1968). and δ( G ) ≥ k − 1. then it is connected. Note that for any graph G with the connected components G1 . χ( G t ) > k. Proof. extend it by setting α(ui ) = α(vi ) and α(v) = k + 1. un }. Now v is adjacent to only δ( G ) < k − 1 vertices. . Recolour each vi by α(ui ). . For the chromatic number we notice ﬁrst that χ( G t ) ≤ (k + 1). Now using inductively the above construction starting from the triangle-free graph K2 . (i) G has a k-critical subgraph H. a contradiction. The 3-critical graphs are exactly the odd cycles C2n+1 for n ≥ 1. and thus must have a cycle of odd length. a contradiction. The graph K2 = P2 is the only 2-critical graph. If G is k-critical for k ≥ 2. . . since a 3-chromatic G is not bipartite. say with α(v) = k. .11. . Gm . G t is triangle-free and it is k + 1-chromatic Indeed. v j .7. a contradiction. . let U = {u1 . Let then G be k-critical. G2 . and so a triangle contains at most (and thus exactly) one vertex ui ∈ U. .10. 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. . We show ﬁrst that G t is triangle-free. Theorem 4. Now. (ii) G has at least k vertices of degree ≥ k − 1. (iii) k ≤ 1 + max H ⊆ G δ( H ). χ( G ) = max{χ( Gi ) | i ∈ [1. U is stable.3 Vertex colourings 54 Claim. Each Kn is n-critical. Critical graphs D EFINITION . m]} . if χ( H ) < k for all H ⊆ G with H = G. If we recolour v by i.

Denote Gi = G [ Ai ∪ {v}]. Lemma 4. every k-critical graph H must have at least k vertices. v} is a separating set.6. Proof. Lemma 4. and let Ai . ⊔ ⊓ Lemma 4. χ( G ) − 1 ≤ δ( H ). χ( G ) ≤ ∆( G ) + 1. there exists a proper colouring α : VG → [1. Moreover. for i ∈ [1. . . ⊔ ⊓ Brooks’ theorem For edge colourings we have Vizing’s theorem. then {u. The claim follows. if uv ∈ G. but no such strong results are known for vertex colouring. By Theorem 4. which proves this claim. Then α is proper. By Lemma 4. if dG (u. we always have χ( G ) ≤ ∆( G ) + 1. the chromatic number χ( G ) usually takes much lower values – as seen in the bipartite case. let H ⊆ G be k-critical.10. The next proof of Brook’s theorem is by L OVÁSZ (1975) as modiﬁed by B RYANT (1996). let H ⊆ G be k-critical. we may take αi (v) = 1 for all i. ⊔ ⊓ Although. m]. . then {u. clearly. (ii) For all u. v ∈ G. Let G be a 2-connected graph. and α(vi ) = min{ j | α(vt ) = j for all t < i with vi vt ∈ G } . For (ii).10. vn } be ordered in some way.8. Then the following are equivalent: (i) G is a complete graph or a cycle. Let v be a cut vertex of a connected graph G. v) = 2. Suppose each Gi has a proper k-colouring αi . / (iii) For all u. We use greedy colouring to prove the claim. and α(vi ) ≤ dG (vi ) + 1 for all i. be the connected components of G −v. . In particular. a critical graph does not have cut vertices.7. For (iii). . and deﬁne α : VG → N inductively as follows: α(v1 ) = 1.5. By Theorem 4. These k-colourings give a k-colouring of G. Of course.4.3 Vertex colourings 55 Proof. v ∈ G. For all graphs G. m]}. ∆( G ) + 1] such that α(v) ≤ dG (v) + 1 for all vertices v ∈ G. 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. In fact. Let VG = {v1 . Then χ( G ) = max{χ( Gi ) | i ∈ [1. v} is a separating set. because. also dG (v) ≥ k − 1 for every v ∈ H. the maximum value ∆( G ) + 1 is obtained only in two special cases as was shown by B ROOKS in 1941. d H (v) ≥ k − 1 for every v ∈ H. The claim follows from this. Proof. For (i).

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

where vi v j ∈ E just in case vi and v j are incompatible. Unfortunately. indeed. (However. no good algorithms are known for determining χ( G ). we ﬁnd at least one j > i such that vi v j ∈ G (possibly v j = w). Already the problem if χ( G ) = 3 is NP-complete. there exist v1 . α(vi ) ≤ ∆( G ) for all i ∈ [1. for all 1 ≤ i < n. . Then colour v1 . vn }.3). v2 } is connected. . the chromatic number problem is NP-complete. v2 ) = 2. . d H ( vi . w = vn has two neighbours. 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. Suppose we have n objects V = {v1 . k] a proper colouring}| . . This shows that G has a proper ∆( G )colouring. k] for sufﬁciently large natural numbers k. where χ G (k) = |{α | α : VG → [1. w ) .8.3 Vertex colourings 57 By Lemma 4. . n − 1]. and so α(vn ) ≤ ∆( G ). E). or worse. The chromatic polynomial of G is the function χ G : N → N. some of which are not compatible (like chemicals that react with each other. v2 ∈ G with dG (v1 . . . In particular. Therefore for each i ∈ [1. (4. . of the same colour 1. . .4. . and. and we would like to colour the vertices of G properly using as few colours as possible. . Indeed. vn } such / that vn = w. v4 .3) The colouring α is proper. 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 } . wv2 ∈ G with v1 v2 ∈ G. as we have seen. | NG (vi ) ∩ {v1 . v1 and v2 . n − 1].8. graph theorists who will ﬁght during a conference). . consequently. such that H = G −{v1 . This problem requires that we ﬁnd χ( G ). say v1 w. and. . Order VH = {v3 . and for all i ≥ 3. ⊔ ⊓ Example 4. the problem whether χ( G ) = 2 has a fast algorithm. In graph theoretical terminology we consider the graph G = (V. . . and since vn has at most ∆( G ) neighbours.5 to be certain on this point. there is an available colour for vn . see Lemma 4. By (4. v2 . w ) ≥ d H ( vi+1 . . χ( G ) ≤ ∆( G ). Also. D EFINITION . vi−1 }| < dG (vi ) ≤ ∆( G ) .) The chromatic polynomial A given graph G has many different proper vertex colourings α : VG → [1.

Example 4. and. indeed. The considered method for checking the number of possibilities to colour a ‘next vertex’ is exceptional. . v2 .3 Vertex colourings 58 This notion was introduced by B IRKHOFF (1912). Let G be a graph. after which k − 1 colours are available for v2 . u e v x . On the other hand. v4 }. then we easily compute the chromatic number χ( G ) just by evaluating χ G (k) for k = 1. and by replacing all edges wu and wv by wx. e = uv ∈ G. Then v3 has k − 2 and ﬁnally v4 has k − 3 available colours. and so χK4 (k) = k(k − 1)(k − 2)(k − 3) . (Of course. and let e ∈ G. B IRKHOFF AND L EWIS (1946). if we can ﬁnd the chromatic polynomial of G. no loops or parallel edges are allowed in the new graph G ∗ e. D EFINITION . The vertex v1 can be ﬁrst given any of the k colours. v}) ∪ { x} is obtained from G by contracting the edge e. 2. Remark. Therefore. and the vertices u and v are deleted. Let G be a graph. Consider the complete graph K4 on {v1 . and thus any k-colouring is a proper colouring. Therefore there are k(k − 1)(k − 2)(k − 3) different ways to properly colour K4 with k colours. f has no end u or v} ∪ {wx | wu ∈ G or wv ∈ G } . Let k ≥ χ(K4 ) = 4.) Theorem 4.13. when EG∗e = { f | f ∈ EG . . Therefore χ K 4 ( k ) = k4 .9. The graph G ∗ e on VG∗e = (VG \ {u. in the discrete graph K 4 has no edges.4. v3 . and for more nonregular graphs it should be avoided. . then clearly χ G (k) = 0. If k < χ( G ).13 will give the tools for constructing χ G . until we hit a nonzero value. Then χ G ( k ) = χ G − e ( k ) − χ G ∗ e ( k ). but its applications have turned out to be elsewhere. Theorem 4. χ( G ) = min{k | χ G (k) = 0} . and let x = x(uv) be a new contracted vertex. to attack the famous 4-Colour Theorem. Hence G ∗ e is obtained by introducing a new vertex x.

f } − ( G − e) ∗ f G−e . then α corresponds to a unique proper k-colouring of G ∗ e. Hence the number of such colourings is χ G∗e (k). k] of G −e can be divided into two disjoint cases. χ T ∗e (k) = k(k − 1)n−2 . χ T (k) = χ T −e (k) − χ T ∗e (k).3 Vertex colourings 59 Proof.10. and thus. e G = G−e − G∗e f = G − {e. when we set α( x) = α(u) for the contracted vertex x = x(uv). χ Gm (k) . Then χ T (k) = k(k − 1)n−1 .13. by the induction hypothesis. Proof. The graph T −e consists of the isolated v and a tree of order n − 1. Let T be a tree of order n. Indeed. where v is a leaf. .9. Here T ∗ e is a tree of order n − 1. χ T −e (k) = k · k(k − 1)n−2 .4. . Let the graph G have the connected components G1 . Proof. also P1 − P2 is a polynomial. By Theorem 4. Therefore χ T ( k ) = k ( k − 1) n − 1 . Let e = uv. By Lemma 4. Suppose that n ≥ 3. Then we have the following reductions. which together show that χ G−e (k) = χ G (k) + χ G∗e (k): (1) If α(u) = α(v). ⊔ ⊓ Theorem 4. χKn (k) = kn for the discrete graph. The proof is by induction on ε G . The claim follows from Theorem 4. The chromatic polynomial is a polynomial. Consider the graph G of order 4 from the above. The proper k-colourings α : VG → [1. and so Lemma 4. and for two polynomials P1 and P2 . the claim is obvious. and the induction hypothesis.15. For n ≤ 2. Theorem 4. Hence the number of such colourings is χ G (k). then α corresponds to a unique proper k-colouring of G. Gm . . ⊔ ⊓ Example 4. . ⊔ ⊓ The connected components of a graph can be coloured independently. . .13. namely α.14.9. namely α. and let e = vu ∈ T. G2 . Then χ G (k) = χ G1 (k)χ G2 (k) . (2) If α(u) = α(v). We use induction on n. since there G −e and G ∗ e have less edges than G.

Also. f } (k) − χ( G−e)∗ f (k) = k ( k − 1) 3 − k ( k − 1) 2 = k ( k − 1) 2 ( k − 2) . and so there is no need to prolong the reductions beyond these. The constant term is 0. there are 6 proper colourings of the given graph. but it seems to satisfy the general properties (that are known or conjectured) of these polynomials. . we have obtained χ G−e (k) = χ G−{e. since the reduction method provided by Theorem 4.13 reduces the computation of χ G to the discrete graphs. for 3 colours. which would tell from any polynomial P(k) whether it is a chromatic polynomial of some graph. when G is connected. as an exercise). there is known no characterization.3 Vertex colourings 60 Theorem 4. It is difﬁcult to determine χ G of a given graph. However. χ G ( x) = 0 for all real numbers 0 < x < 1. we know the chromatic polynomials for trees (and complete graphs. Chromatic Polynomial Problems. For instance. The coefﬁcient of kn−1 equals −ε G . In our example. χ G (m) ≤ m(m − 1)n − 1 for all positive integers m. The coefﬁcients alternate in sign. the polynomial k4 − 3k3 + 3k2 is not a chromatic polynomial of any graph.13 is time consuming. 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 . R EED (1968) conjectured that the coefﬁcients of a chromatic polynomial should ﬁrst increase and then decrease in absolute value. For instance. The coefﬁcient of kn equals 1. R EED (1968) and T UTTE (1974) proved that for each G of order νG = n: • • • • • • • The degree of χ G (k) equals n.4.

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

(vi) An edge that is not a bridge belongs to the boundary of exactly two faces. resp. there are – as we shall see – some nontrivial topological (or geometric) properties that are shared by the plane embeddings. if it is bounded. resp. resp. Lemma 5. The set (R × R ) \ EG is open. (i) Two different faces F1 and F2 are disjoint. Let G be a planar graph. The boundary ∂( F ) of a region F consists of those points for which every neighbourhood contains points from F and its complement. Lemma 5. Regard now each edge e = uv ∈ G as a line from u to v.1 Planar graphs 62 The following result is clear. The exterior boundary is the boundary of the exterior face. More importantly. We recall ﬁrst some elements of the plane geometry. the way we draw a graph G in the plane does not change its maximum degree or its chromatic number. Vertices (edges. (ii) P( G ) has a unique exterior face.) on the exterior boundary are called exterior vertices exterior edges. A face of P( G ) is an interior face. every point x ∈ F has a disk centred at x and contained in F.). and it is divided into a ﬁnite number of disjoint regions.5.2. resp. its interior contains) at least one internal face of P( G ). .1.). The (unique) face that is unbounded is called the exterior face of P( G ). that is. if any two points x. and P( G ) one of its plane embeddings. Then F is a region. For instance. F2 F1 F0 F3 Embeddings P( G ) satisfy some properties that we accepts at face value. A graph is planar if and only if its subdivisions are planar.) that are not on the exterior boundary are interior vertices interior edges. D EFINITION . The edges that surround a face F constitute the boundary ∂( F ) of F. called the faces of P( G ). Let P( G ) be a plane embedding of a planar graph G. Geometric properties It is clear that the graph theoretical properties of G are inherited by all of its plane embeddings. and their boundaries can intersect only on edges. (iv) Each cycle of G surrounds (that is. (v) A bridge of G belongs to the boundary of only one face. y ∈ F can be joined by a continuous curve the points of which are all in F. (iii) Each edge e belongs to the boundary of at most two faces. Let F be an open set of the plane R × R. The vertices (edges.

if and only if the connected components of G are trees.2.4. ⊔ ⊓ In particular. and hence νG − (ε G − 1) + ( ϕ − 1) = 2. If ϕ = 1. Moreover. then. and since G is connected. Proof. If G is a planar graph of order νG ≥ 3. This is clear from Lemma 5. Let e ∈ G be an edge that is not a bridge. in particular.3. νG− e − ε G−e + ( ϕ − 1) = 2. and the claim follows.5. then ε G ≤ 2νG − 4. or inside any geometric triangle.1. there are no cycles in G. Then νG − ε G + ϕ = 2 . Let G be a planar graph. First. We shall prove the claim by induction on the number of faces ϕ of a plane embedding P( G ). A plane embedding P( G ) of a planar graph G has no interior faces if and only if G is acyclic.1 (Euler’s formula). we have ε G = νG − 1. since each P( G ) has an exterior face. Suppose then that the claim is true for all plane embeddings with less than ϕ faces for ϕ ≥ 2. and let P( G ) be any of its plane embeddings.3. then ε G ≤ 3νG − 6. Then every plane embedding of G has the same number of faces: ϕ G = ε G − νG + 2 ⊔ ⊓ Maximal planar graphs Lemma 5. Euler’s formula Lemma 5. then so is any drawing P′ ( G ) which is obtained from P( G ) by an injective mapping of the plane that preserves continuous curves. The subgraph G −e is planar with a plane embedding P( G −e) = P( G )−e obtained by simply erasing the edge e. By the induction hypothesis. by Lemma 5. if G has no triangles C3 . it is a tree.4. where ϕ is the number of faces of P( G ). that every planar graph has a plane embedding inside any geometric circle of arbitrarily small radius.1 Planar graphs 63 If P( G ) is a plane embedding of a graph G. Corollary 5. Proof. This means. The next general form of Euler’s formula was proved by L EGENDRE (1794). by Theorem 2. since the two faces of P( G ) that are separated by e are merged into one face of P( G −e). Now P( G −e) has ϕ − 1 faces. . Theorem 5. that is. notice that ϕ ≥ 1. In this case. Let G be a connected planar graph. Let P( G ) be a plane embedding of a connected planar graph such that P( G ) has ϕ faces. we have the following invariant property of planar graphs. and the claim holds.

we have |K | < 4 as required. A planar graph G is maximal. If G is disconnected with connected components Gi .1. The ﬁrst claim follows from Euler’s formula. By the second claim of Lemma 5. If νG ≤ 2. Suppose thus that ε G ≥ 3. the case where ε G ≤ 2 is clear. K5 and K3. if we remove one edge from K5 . Let F be a face of a plane embedding P( G ) that has at least four edges on its boundary. i=1 νG νG It is thus sufﬁcient to prove the claim for connected planar graphs. since εG = i=1 ∑ ε Gi ≤ 3 ∑ νGi − 6k = 3νG − 6k ≤ 3νG − 6 . Suppose νG ≥ 3. Proof.) Add a new vertex x inside F. Theorem 5. δ( G ) · νG ≤ It follows that δ( G ) ≤ 5. Proof. if an edge is removed from K3. a triangle-free planar graph of order 6 has at most 8 edges. However. v∈ G ⊔ ⊓ . and since H is planar.5.2 (H EAWOOD (1890)). The result is a plane embedding of a graph H with VH = VG ∪ { x} (that has G as its induced subgraph). then δ( G ) ≤ 5. for i ∈ [1. If G is a planar graph. The edges of K are either on the boundary or they are not inside F (since F is a face. then there is nothing to prove. Each face F of an embedding P( G ) contains at least three edges on its boundary ∂( F ). the result is a maximal planar graph.4. since each edge lies on at most two faces. Assume that the set of the boundary vertices of F induces a complete subgraph K. the result is not maximal! Lemma 5.4. ⊔ ⊓ D EFINITION . k]. each face F of P( G ) contains at least four edges on its boundary (when G is connected and ε G ≥ 4). but K5 has 5 vertices and 10 edges. By Lemma 5. Then there are two nonadjacent vertices on the boundary of F. Also. then it holds for G. in this case. ⊔ ⊓ An upper bound for δ( G ) for planar graphs was achieved by H EAWOOD. By the handshaking lemma and the previous lemma. ⊔ ⊓ ∑ dG (v) = 2ε G ≤ 6νG − 12 .3 are not planar graphs. Theorem 5. / Example 5. Clearly. Hence 3ϕ ≤ 2ε G . if G + e is nonplanar for every e ∈ G. a planar graph of order 5 has at most 9 edges.3. and connect the vertices of K to x.3 has 6 vertices and 9 edges. but K3.5. Proof. The second claim is proved similarly except that.1 Planar graphs 64 Proof.3 . and if the claim holds for these smaller (necessarily planar) graphs Gi . The induced subgraph H [K ∪ { x}] is complete.

If G is a maximal planar graph with νG ≥ 3.4.3 .1 Planar graphs 65 By the previous lemma. the theorem was proven earlier by P ONTRYAGIN (1927-1928). Therefore Theorem 5. AND M. The graphs K5 and K3. since there are now no bridges. Hence 3ϕ = 2ε G . if G contains a subdivision of K5 or K3. The claim follows from Euler’s formula. Proof. K ENNEDY. and thus by Theorem 5. ε G = 3νG − 6 .1. For history of the result. On the other hand.V. We prove this result along the lines of T HOMASSEN (1981) using 3-connectivity. Example 5.5 will give a simple criterion for planarity of graphs.3 as a subgraph. The subgraph of Q4 that is a subdivision of K3. L.5 (K URATOWSKI (1930)).3 is given below. Theorem 5. A graph is planar if and only if it contains no subdivision of K5 or K3. then an edge can be added to the graph (inside the face).3 as a subgraph. For a maximal planar graph G of order νG ≥ 3. the graph Q4 contains a subdivision of K3. Hence we have proved Corollary 5. Q UINTAS . ⊔ ⊓ Kuratowski’s theorem Theorem 5. The cube Qk is planar only for k = 1. The theorem on planar graphs.5. 356 – 368.2. see J. then G is not planar. if a face has a boundary of at least four edges. In fact. each Qk with k ≥ 4 has Q4 as a subgraph. and therefore they are nonplanar. 2. then G is triangulated. We prove the converse of this result in what follows. every face of a plane embedding P( G ) has a boundary of exactly three edges. and also independently by F RINK AND S MITH (1930). by Lemma 5. S YSLO. Each face F of an embedding P( G ) is a triangle having three edges on its boundary.M. This theorem (due to K URATOWSKI in 1930) is one of the jewels of graph theory. Indeed. and the graph remains to be planar. that is.5 it is not planar. 0000 0100 1010 1110 1101 1001 0001 0011 1000 1100 0010 . and.3 are the smallest nonplanar graphs.2. 3.W. Historia Math. 12 (1985).

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

The inclusion is proper. If there exists at most one edge xy ∈ H such that uy ∈ G (or vy ∈ G). v} would separate G. Consider . there exists a path P : u − y′ in G −{z.8. and how the subdivision is made.3 . where x = x(uv) is the contracted vertex for e = uv. as in the above. w}. G − R has a connected component B such that u. There exists a vertex y ∈ A with zy ∈ G. By Lemma 5. there exists a vertex w such that R = {z. since G is 3-connected. and this contradicts / the choice of A. G contains a subdivision of K3. and so B ⊆ A. Assume that e and S are chosen such that G − T has a connected component A with the least possible number of vertices. Proof. The proof is by induction on νG . Therefore y′ is connected to y also in G − T. If d H ( x) = 2. that is. there exists an edge e = uv ∈ G such that G ∗ e (with a contracted vertex x) is 3-connected. a Kuratowski graph cannot be created by contractions. For each / ⋆ y′ ∈ B. Let H be a Kuratowski graph of G ∗ e. and hence → this P goes through y. (Otherwise {u. then the claim is obviously true.) The graph G ∗ (zy) is not 3connected by assumption. The only 3-connected graph of order 4 is the planar graph K4 . ⊔ ⊓ By the next lemma. It can be that w ∈ {u. w} separates G. Hence | B| < | A|. B A y u T v z Since uv ∈ G. We do not consider the details of these cases. y′ ∈ A.1 Planar graphs 67 {u. Every 3-connected graph G without Kuratowski subgraphs is planar. since y ∈ B. where H is a subdivision of K5 . Therefore we can assume that νG ≥ 5. ⊔ ⊓ v2 x v1 v3 v1 v4 v2 u v v3 v4 Lemma 5. In this case. Lemma 5. G ∗ e has no Kuratowski subgraphs. The proof consists of several cases depending on the Kuratowski graph. Proof. If for some e ∈ G the contraction G ∗ e has a Kuratowski subgraph.5. z} separates G. then so does G. By Lemma 5. v}.10. v ∈ B. Suppose then that d H ( x) = 3 or 4. and hence. and hence G ∗ e has a plane embedding P( G ∗ e) by the induction hypothesis. then one easily sees that G contains a Kuratowski graph. but by symmetry we can suppose that w = u. and both u and v have 3 neighbours in the subgraph of G corresponding to H.9.9. There remains only one case. v. Let G be a graph. y.

/ / {u.2 Colouring planar graphs 68 the part P( G ∗ e)− x. it is planar. ⊔ ⊓ Example 5. v} = ( G ∗ e)− x. where y. and. Now. Now since G −{u. . G is 3-connected.7. by symmetry.i+1 (i + 1 is taken modulo k) for some i. P( G ∗ e)− x is a plane embedding of G −{u.5. z. . Here C is a cycle of G (since G is 3-connected). Then. . But now u. y} form a subdivision of K3.5. Let X (K6 ) be a drawing of K6 using c crossings so that two edges cross at most once. For the equality. Assume. v3 give a subdivision of K5 . vk } in order along the cycle C. v1 . we can assume that NG (u) \ {v} ⊆ NG (v). suppose that G is a nonplanar graph that has a minimal size ε G such that G does not contain a Kuratowski subgraph. z u v By (1) and (2). Therefore G is planar if and only if ×( G ) = 0. y (2) Assume there are y. clearly. vi . by (1). v}. The crossing number ×( G ) is the minimum number of intersections of its edges in such plane drawings of G. We show that ×(K6 ) = 3. dG (v) = dG (u) > 3. v. . On the contrary. This results in a planar graph G on c + 6 vertices and 2c + 15 edges. This contradiction proves the claim.1. Now c ≥ 3. (1) If NG (u) \ {v} ∈ Pi. v j }. until it was solved by A PPEL AND H AKEN in 1976: if G is a . and NG (u) ⊆ VC ∪ {v} and NG (v) ⊆ VC ∪ {u}.3 and Lemma 5.j : vi − v j → be the path along C from vi to v j . and by Lemma 5. ⊔ ⊓ u v Proof of Theorem 5. since ε G = 2c + 15 ≤ 3(c + 6) − 6 = 3νG − 6. v2 . v2 . G has a plane embedding (obtained from P( G )−u by putting u inside the triangle (v. The problem was known as the 4-Colour Conjecture for more than 120 years. and let C be the boundary of the face of P( G ∗ e)− x containing x (in P( G ∗ e)). that dG (v) ≤ ⋆ dG (u). then.3. vi . 5. Add a new vertex at each crossing. vi+1 } ∪ {v. By Theorem 5. NG (u) \ {v} = NG (v) \ {u} by the assumption dG (v) ≤ dG (u). ×(K5 ) = 1. for instance. vi+1 ) and by drawing the edges with an end u inside this triangle). Also.2 Colouring planar graphs The most famous problem in the history of graph theory is that of the chromatic number of planar graphs. We obtain a plane embedding of G −u by drawing (straight) edges vvi for 1 ≤ i ≤ k. For this we need to show that ×(K6 ) ≥ 3. Therefore. z ∈ {vi . Let Pi. by Lemma 5. we need to show that each nonplanar graph G contains a Kuratowski subgraph. z ∈ NG (u) \ {v} such that y ∈ Pij and z ∈ Pij for some i and j. Any graph G can be drawn in the plane so that three of its edges never intersect at the same point.10.3. one is invited to design a drawing with exactly 3 crossings. Let NG (v) \ {u} = {v1 .

By Theorem 5.2. Theorem 5. since the vertices of P24 are coloured by 2 and 4. v4 lies inside the region enclosed by the cycle C. but C contains no such colours. This is a contradiction. The vertices vi and v j are in the same connected component of G [i. v5 v v4 v1 v2 v3 P13 Consider the subgraph G [i. Let α be a proper 5-colouring of G −v. ⋆ Let Pij : vi − v j be a path in G [i. Proof. and so χ( G ) ≤ 6. a planar graph G has a vertex v with dG (v) ≤ 5. Now. ⊔ ⊓ The ﬁnal word on the chromatic number of planar graphs was proved by A PPEL AND H AKEN in 1976. The proof is by induction on νG . exactly one of v2 . Suppose the claim does not hold. and let G be a 6-critical planar graph. Recall that for k-critical graphs H. Lemma 5. Suppose these neighbours vi of v occur in the plane in the geometric order of the ﬁgure. By the geometric → assumption. If G is a planar graph.6 (H EAWOOD (1890)).7 (4-Colour Theorem). Clearly. and thus there exists a vertex v with dG (v) = δ( G ) ≥ 5. then χ( G ) ≤ 4. χ( G ) > 5. the claim holds for νG ≤ 6. then χ( G ) ≤ 5. where vi and v j have the same colour i. dG (v) = 5. ⊔ ⊓ The proof of the following theorem is partly geometric in nature. By the following theorem. χ( G −v) ≤ 6. If G is a planar graph. then χ( G ) ≤ 4. . j].11. and then recolour v with the remaining colour j).5. Since dG (v) ≤ 5. By Theorem 5. since G is planar. and it requires the assistance of a computer. Such a colouring exists. Proof. and let C = (vv1 ) P13 (v3 v). there is a colour i available for v in the 6-colouring of G −v.2. The 5-colour theorem We prove H EAWOOD’s result (1890) that each planar graph is properly 5-colourable. By the induction hypothesis. By assumption. If G is a planar graph. j] ⊆ G made of colours i and j. then χ( G ) ≤ 6. The solution of the 4-Colour Theorem is difﬁcult. δ( H ) ≥ k − 1. the path P24 must meet C at some vertex of C. and therefore for each i ∈ [1. Theorem 5. there exists a neighbour vi ∈ NG (v) such that α(vi ) = i. The 4-Colour Conjecture has had a deep inﬂuence on the theory of graphs during the last 150 years. each planar graph can be decomposed into two bipartite graphs. 5]. otherwise we interchange the colours i and j in the connected component containing v j to obtain a recolouring of G. because G is 6-critical.2 Colouring planar graphs 69 planar graph. j] (for.

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

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

since proper colourings are special cases of list colourings. and it is of order 63. “The Four-Color Problem”. K AINEN. 3} y3 x1 x2 x3 {1. However. and |Λ(v)| = 5 for all v ∈ C. (If the .2 Colouring planar graphs 72 According to H EESCH one might be satisﬁed with a set of 8900 conﬁgurations to prove the 4CC.4. x3 } and Y = {y1 . The following book contains the ideas of the proof of the 4-Colour Theorem. where X = { x1 . 3} {2. Theorem 5. S EYMOUR AND T HOMAS (1997) uses 633 conﬁgurations and 32 discharging rules. Example 5.L. 2} {1. A proper colouring α : VG → [1. where the interior faces are triangles.3 ) > 2.5. Also. List colouring D EFINITION . Let G be a planar graph and let C be the cycle that is the boundary of the exterior face. We can assume that the planar graph G is connected.9 (T HOMASSEN (1994)).C.) A simpliﬁed proof by R OBERTSON . m] of G is a (Λ-)list colouring. y3 }. 3} y2 {2. and that it is given by a near-triangulation. Let Λ consist of lists such that |Λ(v)| = 3 for all v ∈ C. Then χℓ ( G ) ≤ 5. The proof used 1200 hours of computer time. The bipartite graph K3. Indeed. 3} Obviously χ( G ) ≤ χℓ ( G ). if each vertex v gets a colour from its list. Because of these simpliﬁcations also the computer time is much less than in the original proof. 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. it was shown by V OIGT (1993) that there exists a planar graph with χℓ ( G ) = 5. Y ). Let G be a graph so that each of its vertices v is given a list (set) Λ(v) of colours. T. Dover.3 is not 2choosable. Indeed. 2} y1 {1.3 be ( X. The lists for the vertices shown in the ﬁgure show that χℓ (K3. They based the proof on reducibility using Kempe chains. y2 . S AATY AND P. but equality does not hold in general. let the bipartition of K3. For planar graphs we do not have a ‘4-list colour theorem’. |Λ(v)| = k}. it was proved by V IZING (1976) and E RDÖS . x2 . In fact.10. an embedding. if χℓ ( G ) ≤ k. Then G has a Λ-list colouring α. α(v) ∈ Λ(v). S ANDERS . There were difﬁculties with his approach that were solved in 1976 by A PPEL AND H AKEN. Let G be a planar graph. and ended up with an unavoidable set with over 1900 conﬁgurations and some 300 discharging rules. At the moment. R UBIN AND TAYLOR (1979) that χ ℓ ( G ) ≤ ∆( G ) + 1 . {1. G is k-choosable. (K OCH assisted with the computer calculations. T HOMASSEN proved a stronger statement: Theorem 5. / Proof. 1986. the smallest such a graph was produced by M IRZAKHANI (1996).

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

of graphs has two graphs Gi and Gj with i < j such that Gi Gj . if H is isomorphic to a graph obtained from a subgraph of G by successively contracting edges. . A graph is a line segment graph if it is a string graph for a set L of straight line segments in the plane. every inﬁnite sequence G1 . 0 and ∞. all complete graphs are such graphs. Indeed.2 Colouring planar graphs 74 Graphs can be represented as plane ﬁgures in many different ways. −1 0 ∞ The Minor Theorem∗ D EFINITION . The following properties of the minor relation are easily established: (i) G (ii) H (iii) H G. A graph H is a minor of G. H G. . The above question remains open even in the case when the slopes of the lines are +1.13 is around 500 pages long. For this.5. At ﬁrst it might seem that every graph is a string graph. = G imply H G. +1 A positive answer to this 4-slope problem for planar graphs would prove the 4-Colour Theorem. that is. G and G L and L H imply G ∼ H. The conditions (i) and (iii) ensure that the relation is a quasi-order. consider a set S of curves of the plane (that are continuous between their end points). It turns out to be a well-quasi-order. The proof goes beyond these lectures. it is reﬂexive and transitive. −1. 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. e G a subgraph a contraction Note that every subgraph H ⊆ G is a minor. The string graph of S is the graph G = (S. Line Segment Problem. . that is. Indeed. denoted by H G. A recent result of R OBERTSON AND S EYMOUR (1983-2000) on graph minors is (one of) the deepest results of graph theory. G2 . but this is not the case. . E). the proof of Theorem 5. where uv ∈ E if and only if the curves u and v intersect. It is known that all planar graphs are string graphs (this is a trivial result).

The cases for r = 5 and 6 follow from the 4-Colour Theorem. namely. Each property P of graphs deﬁnes a family of graphs. since the property ‘knotless’ is inherited by minors: there exists a fast algorithm to do the job. By Theorem 5. the subgraph order of trees (T1 ⊆ T2 ) is not a well-quasi-order. (3) acyclic graphs. The conjecture is trivial for r = 2. The acyclic graphs include all trees.3 ⊔ ⊓ S EYMOUR (1998) proved the Wagner’s conjecture: Theorem 5.2 Colouring planar graphs 75 Theorem 5. The minor order is a well-quasi-order on graphs. The following families of graphs are minor closed: the family of (1) all graphs. A property P of graphs is said to be inherited by minors. the family of those graphs that satisfy this property.5. if χ( G ) ≥ r. More importantly. Exercise. A graph G is nonplanar if and only if K5 G. one can test (with a fast algorithm) whether a graph can be embedded onto a surface. if every minor H of a graph G ∈ F is also in F . 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 .13 (Minor Theorem). this algorithm is not known! Hadwiger’s Problem. the family of trees is not closed under taking subgraphs. WAGNER proved a minor version of Kuratowski’s theorem: Theorem 5. if all minors of a graph G satisfy P whenever G does. Kχ( G) G. Every graph can be drawn in the 3-dimensional space without crossing edges.15 (Minor Theorem 2). D EFINITION . R OBERTSON AND G or K3. that is. A family F of graphs is said to be minor closed.15. (2) planar graphs (and their generalizations to other surfaces).15 concerns embeddings of graphs on surfaces. In particular. Proof. However. . see the next chapters. Let P be a property of graphs inherited by minors. This problem is solved by the above results. then G has a complete graph Kr as its minor. 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. in any inﬁnite family F of graphs.14 (WAGNER (1937)). One of the impressive application of Theorem 5. and it is known to hold for all r ≤ 6. and thus it is not minor closed. one of the graphs is a (proper) minor of another. H ADWIGER conjectured in 1943 that for every graph G.

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

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

In the one direction the claim is obvious: if G is a planar graph. Another way to see this is to use a projection of the sphere to a plane: ¡ ¢£ . We then puncture the sphere at this area. By studying the relations of the representing words. On the other hand. According to the next theorem a sphere has exactly the same embeddings as do the plane. On the right we have drawn K6 in the Klein bottle. and hence G is planar. A word representing a connected sum of two surfaces. as do the other surfaces. then it can be drawn in a bounded area of the plane (without crossing edges). and stretch it open until it looks like a region of the plane. then there is a small area of the sphere. have many other plane models and representing words as well. The black dots indicate. Clearly. where the lines enter and leave the edges of the plane model. if G is embeddable in a sphere S0 . In general. if S is a surface. if a graph can be embedded in one sphere. 78 Klein bottle Drawing a graph (or any ﬁgure) in a surface can be elaborated compared to drawing in a plane model. In this process no crossings of edges can be created. Recall that in the plane model for the Klein bottle the vertical edges of the square have the same direction. if G can be drawn in S without crossing edges. where there are no points of the edges. is obtained by concatenating these words to W1 W2 .5.5. 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 a graph G has an S-embedding. e3 b e2 e1 e1 e2 b e3 e4 a e5 Sphere D EFINITION .18 can be proved. Theorem 5. then it can be embedded in any sphere – the size of the sphere is of no importance.3 Genus of a graph These surfaces. represented by words W1 and W2 . e4 a e5 Example 5. Let S0 be (the surface of) a sphere. and this bounded area can be ironed on the surface of a large enough sphere.

This is sometimes convenient. . This is a torus S1 (or an orientable surface of genus 1). Torus Consider next a surface which is obtained from the sphere S0 by pressing a hole in it. The S1 -embeddable graphs are said to have genus equal to 1. 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. a planar graph drawn in a sphere has no exterior face – all faces are bounded (by edges). 79 Therefore instead of planar embeddings we can equally well study embeddings of graphs in a sphere.6.3 Genus of a graph Theorem 5. its embeddability properties will remain the same. K7 has genus 1 as can be seen from the plane model (of the torus) on the right.3 have genus 1. A graph G has an S0 -embedding if and only if it is planar. If a graph G is S1 -embeddable.19. ¡ ¢£ ¡ ¡ ¡ ¢£ ¢£ ¢£ ¡ ¡ ¢£ ¢£ 1 1 5 2 6 3 7 4 1 1 Example 5.5. Also. then it can be drawn in any one of the above surfaces without crossing edges. If a sphere is deformed by pressing or stretching. In topological terms the surface has been distorted by a continuous transformation. The smallest nonplanar graphs K5 and K3. since the sphere is closed and it has no boundaries. Most importantly.

For planar graphs. Theorem 5. We deﬁne the genus g( G ) of a graph G as the smallest integer n. The drawing of an S4 can already be quite complicated. because we do not put any restrictions on the places of the holes (except that we must not tear the surface into disjoint parts). since K5 is nonplanar. However. The same handle can be utilized by several edges. 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. for which G is Sn -embeddable. we have g( G ) = 0. but is embeddable in a torus. ¡¢ £¤¥ ¡¢ £¤¥ . in particular. and. once again an Sn can be transformed (topologically) into a sphere with n handles. g(K4 ) = 0. Indeed. 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 next theorem states that any graph G can be embedded in some surface Sn with n ≥ 0. we have g(K5 ) = 1.3 Genus of a graph 80 Genus Let Sn (n ≥ 0) be a sphere with n holes in it. Every graph has a genus. We should note that the above argument does not determine g( G ). clearly g( G ) ≤ n. ¡¢ ¡¢ £¤¥ £¤¥ D EFINITION .20. This result has an easy intuitive veriﬁcation. Also. only that G can be embedded in some Sn . However. g(K3.3 ) = 1. For K5 . and thus the genus g( G ) of G exists.5.

23. every nonbridge is on the boundary of exactly two faces. For a connected G with νG ≥ 3. ε G = 28. Combining these we obtain that 3ε G − 3νG + 6 − 6g( G ) ≤ 2ε 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. then g( G ) = 0. g( G ) ≥ 2. As an example.12. a face of an embedding of G in Sn (with g( G ) = n) is a region of Sn surrounded by edges of G. and the claim follows. and so 5 g( G ) ≥ 3 . Theorem 5. 6 2 Proof. we can compute lower bounds for the genus g( G ) without drawing any embeddings. If H ⊆ G. as before. ϕ G = ε G − νG + 2 − 2g( G ). then the claim is trivial. In general. But this is because g(K4 ) = 0. If G is a planar graph. 1 1 ε G − (νG − 2) . In this case νG = 8. 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 the genus of the torus is 1. A face of an embedding P( G ) in a surface is a 2-cell. We deduce that K8 cannot be embedded in the surface S1 of the torus.22 (Y OUNGS (1963)). In particular. and the above formula is the Euler’s formula for planar graphs. Let again ϕ G denote the number of faces of an embedding of G in Sn . By the previous lemma. ⊔ ⊓ g( G ) ≥ By this theorem. We omit the proof of the next generalization of Euler’s formula. Proof. and. D EFINITION .5. Assume thus that νG ≥ 4. If G is a connected graph. then clearly g( H ) ≤ g( G ). The external face of G becomes an ‘ordinary face’ after G has been drawn in S0 . For a connected G with νG ≥ 3 we have 3ϕ G ≤ 2ε G .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.21. Since the genus is always an integer. . We omit the proof of the general condition discovered by Y OUNGS: Theorem 5. if every simple closed curve (that does not intersect with itself) can be continuously deformed to a single point. since H is obtained from G by omitting vertices and edges. ⊔ ⊓ Theorem 5. Now every face has a border of at least three edges. and by the generalized Euler’s formula. 3ϕ G ≤ 2ε G . Lemma 5. then νG − ε G + ϕ G = 2 − 2g( G ) . If νG = 3. let G = K8 .

For all graphs G of order n ≥ 3. 12 Proof.13. is extremely long and difﬁcult. g(Km. but g(K8 ) = 2.3 Genus of a graph Lemma 5. then g( K n ) = (n − 3)(n − 4) .28 (H EAWOOD ). . χ( G ) ≤ 4. For the complete bipartite graphs.25. H EAWOOD proved it ‘only’ for g ≥ 1.24 (H EAWOOD (1890)). it is surprising that the generalization of the 4-Colour Theorem for genus ≥ 1 is much easier. 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. If g( G ) = g ≥ 1. The number of edges in Kn is equal to ε G = 1 n(n − 1). the proof of the 4-Colour Theorem. g(K7 ) = 1. For a graph G of order n. For the complete graphs Kn a good lower bound was found early. 2 Notice that for g = 0 this theorem would be the 4-colour theorem. then χ( G ) ≤ 7+ 1 + 48g . g( G ) ≤ (n − 3)(n − 4) . 12 Also. we know the exact genus for the complete bipartite graphs: Theorem 5. 12 Therefore g(K6 ) = ⌈3 · 2/12⌉ = ⌈1/2⌉ = 1. By Theorem 5.n ) = (m − 2)(n − 2) .27 ( R INGEL (1965)). This in mind. Also. Theorem 5.26. H EAWOOD proved a hundred years ago: Theorem 5. Theorem 5. then g( K n ) ≥ 82 (n − 3)(n − 4) . ⊔ ⊓ This result was dramatically improved to obtain Theorem 5. we 2 obtain g(Kn ) ≥ (1/6)ε G − (1/2)(n − 2) = (1/12)(n − 3)(n − 4) . If n ≥ 3. g( G ) ≤ g(Kn ). 4 Chromatic numbers∗ For the planar graphs G.23. If n ≥ 3.5.25 (R INGEL AND Y OUNGS (1968)).

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

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

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

where b is the number of acyclic orientations D of G −e that extend in both ways e1 and e2 . if D is an acyclic orientation of G −e for e = uv. Indeed. Let k = χ( G ) and let α be a proper k-colouring of G. To prove the claim. Note that since D is acyclic. ⊔ ⊓ D EFINITION . we choose e2 . Clearly. 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) . it extends to an acyclic orientation of G by putting e1 : u → v or e2 : v → u. Now χ G (k) is a polynomial that satisﬁes the recurrence χ G (k) = χ G−e (k) − χ G∗e (k). and a( G ) = 1 = (−1)n (−1)n = (−1)n χ G (−1) as required.2) Each acyclic orientation of G ∗ e corresponds in a natural way to an acyclic orientation D of G −e that satisﬁes (6. We orient each edge uv ∈ G by setting uv ∈ D. we choose → → ⋆ ⋆ e1 . if a( G ) = a( G − e) + a( G ∗ e) then. we show that a( G ) satisﬁes the same recurrence. the so obtained orientation D has no directed paths of length ≥ k − 1.1 Digraphs 86 Proof. if α(u) < α(v). if it has no directed cycles. if D has no ⋆ ⋆ directed path u − v.2). then χ G (k) = kn . The next result is charming. Let G be a graph of order n. it cannot have both ways u − v and v − u. → → (6. → → We conclude that a( G ) = a( G − e) + b.3 (S TANLEY (1973)). since χ G (−1) measures the number of proper colourings of G using −1 colours! Theorem 6. and if D has no directed path v − u. As usual the set of colours is [1. where χ G is the chromatic polynomial of G. and the proof is completed. Then the number of the acyclic orientations of G is a( G ) = (−1)n χ G (−1) . On the other hand. Indeed.1). Proof. Therefore b = a( G ∗ e). if G is discrete. An orientation D of an undirected graph G is acyclic. k]. ⊔ ⊓ . Let a( G ) be the number of acyclic orientations of G. by the induction hypothesis. For (6.6. First. we observe that every acyclic orientation of G gives an acyclic orientation of G −e.1) (6. The acyclic orientations D that extend in both ways are exactly those that contain ⋆ ⋆ neither u − v nor v − u as a directed path.

A connected graph G is di-orientable if and only if G has no bridges. 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. which eP → → → H C contradicts the maximality assumption on H. ⊔ ⊓ Example 6. R OBBINS’ theorem solves this problem. Theorem 6. that is. Proof. Then D has a directed Euler tour if and I only if dD (v) = dO (v) for all vertices v. 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). If G has a bridge e. where we allow parallel directed edges between the vertices. where w is the last vertex inside H. if there is a di-connected oriented graph D such that G = U ( D ). Example 6. Hence G has a cycle C. 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. there exists an edge e = vu ∈ G such that u ∈ H but v ∈ H (because G is connected). but the result may not be di-connected. . A directed Euler tour of D is a directed closed walk that uses each edge exactly once. y for which D I ( v) − dO ( v)| = 1.4 (R OBBINS (1939)). This proves the claim. Suppose then that G has no bridges. D (2) Let D be a digraph such that U ( D ) is connected. each word w is a sequence of letters.6. 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. . ⋆ In the di-orientation D H of H there is a directed path P′ : u − w. In conclusion. G [V ∪ V ] has a di-orientation. . |d D D The above results hold equally well for multidigraphs. A directed Euler trail of D is a directed walk that uses each edge exactly once. then we are done. In the case . that is. .1 Digraphs 87 One-way trafﬁc Every graph can be oriented. A graph G is di-orientable. The following two results are left as exercises. and a cycle is always di-orientable. Consider words over an alphabet A = {a1 . The following problem was ﬁrst studied by H UTCHINSON AND W ILF (1975) with a motivation from DNA sequencing. (1) Let D be a digraph such that U ( D ) is connected. Let D be a digraph. D EFINITION .2. Otherwise.1. In the oneway trafﬁc problem the resulting orientation should be di-connected. a2 . for directed graphs. If H = G. an } of n letters. Now. Let then H ⊆ G be maximal such that it has a di-orientation D H .

1 Digraphs 88 of DNA. A tournament T is an orientation of a complete graph. Here the new digraph has the edge vu instead of uv. . j ≤ n.1.5 (R ÉDEI (1934)). Proof.3. r12 = 2. and r11 = 1. Every tournament has a directed Hamilton path. for which the directed distance from v to x is at least three. → Proof. But there are d vertices in A = {y | vy ∈ T }. and let dO (v) = d be the maximum outdegree in T T.) Theorem 6. For instance.6. There are four tournaments of four vertices that are not isomorphic with each other. Tournaments D EFINITION . a tournament T of order n has a directed path of length n − 1. r22 = 0. Consider a multidigraph D with VD = A for which there are rij edges ai a j . 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. there is a directed path v − u of length at most two. T. s1 = 3. ⋆ Then for all u. The chromatic number of Kn is χ(Kn ) = n. and the question is: does there exist a word w in which each letter ai occurs exactly si times. and hence by Theorem 6. r21 = 1. C. if n = 2. Let v be a vertex of a tournament T of maximum outdegree. Suppose that there exists an x. It follows that the outdegree of x is d + 1. ⊔ ⊓ Problem. we are given nonnegative integers si and rij for 1 ≤ i. 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 ai is followed by a j exactly rij times. then the word a1 a2 a1 a1 a2 is a solution to the problem. Example 6.6 (L AUDAU (1953)). the letters are A. In a problem instance. Theorem 6. Let T be an orientation of Kn . G. which contradicts the maximality assumption made for v. and thus d + 1 vertices in {y | xy ∈ T } = A ∪ {v}. (Isomorphism of directed graphs is deﬁned in the obvious way. This is then a directed Hamilton path visiting each vertex once. ⊔ ⊓ The vertices of a tournament can be easily reached from one vertex (sometimes called the king).

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

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

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

A ⋆ path P : s − r is ( f -)improvable. and / the capacities of the edges are not exceeded. then it has no improvable paths. If f is a maximum ﬂow of N. r. if there is no cut [ R] with α[ R] < α[S]. if e is not in P . A cut [S] is a minimum cut. s. Proof. For a subset S ⊂ VN with s ∈ S and r ∈ S. let the cut by S be / [S] = [S. Let N = ( D. since P has exactly one edge sv ∈ N for the source s. we have ( f ′ )− (v) = ( f ′ )+ (v). S] (= {uv ∈ N | u ∈ S. the bold path has value ι( P) = 1. Let N be a network. → On the right. if ι( P) > 0. Now val( f ′ ) = val( f ) + ι( P). Let f be a ﬂow in a network N.2 Network Flows 4 92 5 s 3 4 4 3 4 0 4 1 r D EFINITION . r}.3. val( f ) = val( f S ) = f + (S) − f − (S) . α) be a network. 5 s 6 4 4 5 2 4 4 2 5 r . if ι( P) > 0. 4 4 5 s 5 4 3 2 0 4 2 r Then f ′ is a ﬂow. For a ﬂow f and a cut [S] of N. since at each intermediate vertex v ∈ {s. if e is against P . Hence. In our original network the capacity of the cut for the indicated vertices is equal to 10.4. Deﬁne f ( e) + ι( P ) ′ f ( e) = f ( e) − ι( P ) f ( e) if e is along P . / The capacity of the cut [S] is the sum α[S ] = α+ (S ) = ∑ e ∈[S ] α( e) . Example 6. Lemma 6.7. and therefore this path is improvable. Lemma 6.6. v ∈ S}) . ⊔ ⊓ Max-Flow Min-Cut Theorem D EFINITION . then we can improve the ﬂow.

/ (i) if e = uv ∈ N. Proof. ⊔ ⊓ Theorem 6.6. then f (e) = 0. val( f ) = val( f S ) = f + (S) − f − (S). Conversely.1.7. then val( f ) ≤ α[S] . a maximum ﬂow cannot have improvable paths. v ∈ S. by F ORD AND F ULKERSON. f + (S) = ∑ e ∈[S ] f ( e) ≤ ∑ e ∈[S ] α( e) = α[S ] . In particular. and let ⋆ S I = {u ∈ N | for some path P : s − u. Theorem 6. / This proves the claim. and hence val( f ) ≤ α[S]. By the deﬁnition of a ﬂow. val( f ) ≤ α[S]. S HANNON.4. as required. If f is a ﬂow and [S] a cut such that val( f ) = α[S]. then f (e) = α(e). and f − (S) ≥ 0. By Lemma 6. r}).3. 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 ) . The present approach is due to Ford and Fulkerson. F EINSTEIN . then f is a maximum ﬂow and [S] a minimum cut. if f is a maximum ﬂow and [S] a minimum cut. and val( f ) = val( f s ).2 Network Flows 93 Proof. Proof. and (2) f (e) = 0 for all e = vu with u ∈ S. Let S I = S \ {s}. assume that N contains no f -improvable paths. By Lemma 6.8. equality holds if and only if for each u ∈ S and v ∈ S. and by R OBACKER in 1955 – 56. the equality val( f ) = α[S] holds if and only if (1) f + (S) = α[S] and (2) f − (S) = 0. (ii) if e = vu ∈ N. Now val(S I ) = 0 (since S I ⊆ N \ {s. For a ﬂow f and any cut [S] of N. ι( P) > 0} . A ﬂow f of a network N is maximum if and only if there are no f -improvable paths in N. → ⊔ ⊓ . This holds if and only if f (e) = α(e) for all e ∈ [S] (since f (e) ≤ α(e)). Furthermore. Also. Corollary 6. The following main result of network ﬂows was proved independently by E LIAS .

since v ∈ S. and consider a network N with vertices {s. for an edge e = vu ∈ N with v ∈ S and u ∈ S. ⊔ ⊓ Corollary 6. In a network N. → Proof. Next we apply the theorem to unit networks.6. y ∈ Y together with the edges xy ∈ N (α( xy) = | X | + 1). Y ).10 (Max-Flow Min-Cut). and so f (e) = α(e). Corollary 6. Lemma 6. f0 (e) = 0 for all e ∈ V × V. where the capacity function α : V × V → N has integer values. → (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. The maximum number of edge⋆ disjoint directed paths u − v equals the minimum number of edges. if xy ∈ G for x ∈ X. ⊔ ⊓ The proof of Theorem 6. Moreover. We mention a connection to the Marriage Theorem. Now Theorem 6. ι( Pe) = 0 for the path Pe : s − v.9. r} ∪ X ∪ Y. → . Since u ∈ S. / By Theorem 6.3 by increasing and decreasing the values of the edges by integers only. Let N be a network. Applications to graphs⋆ The Max-Flow Min-Cut Theorem is a strong result.7. whose removal destroys → ⋆ all the directed connections u − v from D. we have val( f ) = α[S]. (i) The value val( f ) of a maximum ﬂow equals the maximum number of edge-disjoint di⋆ rected paths s − r. Proof. Let the edges (with their capacities) be sx ∈ N (α(sx) = 1). where u ∈ S and v ∈ S.5. f (e) = 0. Consider an edge e = uv ∈ N.2. and many of our previous results follow from it.1 implies now that f is a maximum ﬂow (and [S] is a minimum cut). the value val( f ) of a maximum ﬂow equals the capacity α[S] of a minimum cut.2 Network Flows 94 Set S = S I ∪ {s}. let G be a bipartite graph with a bipartition ( X. Let N be a unit network with source s and sink r. there exists a / ⋆ ⋆ path P : s − u with ι( P) > 0. yr ∈ N (α(yr) = 1) for all x ∈ X. Theorem 3. Let f 0 be the zero ﬂow. → / → Therefore ι(e) = 0. Let u and v be two vertices of a digraph D. ⊔ ⊓ Theorem 6. For this. y ∈ Y. By the same argument. We obtain results for (directed) graphs. Then N has a maximum ﬂow with integer values. A maximum ﬂow is constructed using Lemma 6.9. Exercise.10 gives Theorem 3. where the capacities of the edges are equal to one (α(e) = 1 for all e ∈ N).9.8 showed also Theorem 6. Then G has a matching that saturates X if and only if N has a maximum ﬂow of value | X |.

A] α( e) = 0 . A network N with source s and sink r is obtained by setting the capacities equal to 1. α) of an undirected graph G is an orientation H of G together with an edge colouring α : E H → [0. The claim follows immediately from Corollary 6.3) becomes ∑ e = vu ∈ H α( e) = 0 . → Proof. The claim follows from Lemma 6.6. 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.4) Example 6. if α(e) = 0 for all e ∈ H. 1 3 2 2 The condition (6.5) since the values of the edges inside A cancel out each other. In the k-ﬂows we do not have any source or sink. ⊔ ⊓ Corollary 6. k − 1] such that for all vertices v ∈ V. whose removal destroys all the → ⋆ connections u − v from G. A k-ﬂow is nowhere zero. Let u and v be two vertices of a graph G.5 and Corollary 6. ⊔ ⊓ Seymour’s 6-ﬂows∗ D EFINITION . Proof. (6. A k-ﬂow ( H.3) that is. A graph with a nowhere zero 4-ﬂow. Corollary 6. For convenience.2 Network Flows 95 Proof. . In particular.3.2.3. The claim follows then easily from Corollary 6. ∑ e = vu ∈ H α( e) = ∑ f = uv∈ H α( f ) . (6. 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. 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. let α(e−1 ) = −α(e) for all e ∈ H in the orientation H of G so that the condition (6.8. 1 2 1 (6.4) generalizes to the subsets A ⊆ VG in a natural way. ∑ e ∈[ A.4. ⊔ ⊓ The next corollary is Menger’s Theorem for edge connectivity.

⊔ ⊓ D EFINITION .13.1. if n is even . S EYMOUR’s remarkable result reads as follows: Theorem 6. and set α(e) = 1. Orient the edges e ∈ M from X to Y. and so 5 is the best one can think of. ⊔ ⊓ . and if n > 4.12. Proof. The Petersen graph has a nowhere zero 5-ﬂow but does not have any nowhere 4-ﬂows.11. 96 Tutte’s Problem. let G be ( X. and two neighbours y2 . a 3-regular graph has a perfect matching M. If an edge uv ∈ D. 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. Omitted. If G has a nowhere zero k-ﬂow for some k. Theorem 6. we have f ( G ) ≤ 3. Every bridgeless graph has a nowhere zero 6-ﬂow. In order to fully appreciate Seymour’s result.9. By Corollary 3. Since W arrives and leaves each vertex equally many times. then f (Kn ) = Proof. The proof of this uses the 4-Colour Theorem. Since each x ∈ X has exactly one neighbour y1 ∈ Y such that xy1 ∈ M. Conversely. A connected graph G has a ﬂow number f ( G ) = 2 if and only if it is eulerian. Y )-bipartite.6. the function α is a nowhere zero 2-ﬂow.2 Network Flows Lemma 6. Exercise. For each 3-regular bipartite graph G. ⊔ ⊓ Example 6. and so G is eulerian. and indeed. Then necessarily the degrees of the vertices are even.6.11 (S EYMOUR’s (1981)). Theorem 6. let α(e) = 1. Orient the edges e ∈ M from / Y to X. we have that / f ( G ) ≤ 3. Indeed. and set α(e) = 2. We have f (K4 ) = 4. Theorem 6. and consider an Euler tour W of G. The ﬂow number f ( G ) of a bridgeless graph G is the least integer k for which G has a nowhere zero k-ﬂow. Proof. 2 3 if n is odd . let α be a nowhere zero 2-ﬂow of an orientation D of G. Suppose G is eulerian. Let D be the orientation of G corresponding to the direction of W. It was conjectured by T UTTE (1954) that every bridgeless graph has a nowhere zero 5-ﬂow. Tutte’s conjecture resembles the 4-Colour Theorem. y3 ∈ Y such that xy2 . then G has no bridges. the conjecture is known to hold for the planar graphs. xy3 ∈ M.

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

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

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

Sign up to vote on this title

UsefulNot useful- Graph Theory
- Graph Theory With Applications
- Graph Theory With Applications - J. Bondy, U. Murty
- Intro to Graph Theory
- Lecture Notes on Graph Theory
- Introduction to Graph Theory Second Edition Solution Manual (by Douglas B. West)
- Graph Theory
- Graph Theory
- GT English
- Introduction to Graph Theory -- D. B. West
- Graph Theory & Algorithms - M. Ashraf Iqbal
- graph theory by narsing deo
- Graph Theory
- C. Vasudev - Combinatorics and Graph Theory - New Age - 577p
- Graph Theory Douglas B West 2 Ed
- graph theory 1
- Schaum's
- A Course in Combinatorics Second Edition
- GRAPH THEORY
- Graph Theory Bondy Murty
- Graph Theory Lecture
- Ring Theory
- 3.2Graph%20Theory.pdf
- 2_graph_theory
- Graph theory
- graph theory
- Graph Theory Book
- Handbook of Graph Theory
- graph theory
- 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.

scribd