82 views

Uploaded by Ajith Suresh

graphy theory full lecture notes

- Solution Manual Introduction to the Theory of Computation Sipser
- 9780324782011_PPT_ch12
- Graph theory
- Schaum's: Graph Theory
- Introduction to Graph Theory Second Edition Solution Manual (by Douglas B. West)
- Graph Theory & Algorithms - M. Ashraf Iqbal
- Intro to Graph Theory
- EE304Quiz01
- Graph Theory Douglas B West 2 Ed
- Graph Theory
- graph theory 1
- Introduction to Graph Theory -- D. B. West
- Matching
- Jana Comstock- A finite presentation of knotted trivalent graphs
- Numerical Analysis
- Static to Dynamic Electric Power Network
- Lec4
- Matching2016 B Halls Theorem
- 12_DS and Algorithm_session_17.pps
- Math assignment

You are on page 1of 99

GRAPH THEORY

Tero Harju

Department of Mathematics University of Turku FIN-20014 Turku, Finland

1994 – 2012

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 42 42 46 52 60 60 67 74

2

3

4

5

6

Directed Graphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 6.1 Digraphs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 6.2 Network Flows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

1 Introduction

Graph theory may be said to have its beginning in 1736 when EULER 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 KÖNIG 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. BIGGS, R.J. LLOYD 1986.

AND

R.J. WILSON, “Graph Theory 1736 – 1936”, Clarendon Press,

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. MURTY, “Graph Theory with Applications”, Macmillan, 1978. R. DIESTEL, “Graph Theory”, Springer-Verlag, 1997. F. HARARY, “Graph Theory”, Addison-Wesley, 1969. D.B. WEST, “Introduction to Graph Theory”, Prentice Hall, 1996. R.J. WILSON, “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. BIGGS, “Algebraic Graph Theory”, Cambridge University Press, (2nd ed.) 1993. C. GODSIL, G.F. ROYLE, “Algebraic Graph Theory”, Springer, 2001. and for computational aspects, see S. EVEN, “Graph Algorithms”, Computer Science Press, 1979.

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

1.1 Graphs and their plane ﬁgures

4

**1.1 Graphs and their plane ﬁgures
**

Let V be a ﬁnite set, and denote by E (V ) = {{u, v } | u, v ∈ V, u = v } . the 2-sets of V , i.e., subsets of two distinct elements. DEFINITION. A pair G = (V, E ) with E ⊆ E (V ) is called a graph (on V ). The elements of V are the vertices of G , and those of E the edges of G . The vertex set of a graph G is denoted by VG and its edge set by EG . Therefore G = (VG , EG ). In literature, graphs are also called simple graphs; vertices are called nodes or points; edges are called lines or links. The list of alternatives is long (but still ﬁnite). A pair {u, v } is usually written simply as uv . Notice that then uv = vu. In order to simplify notations, we also write v ∈ G and e ∈ G instead of v ∈ VG and e ∈ EG . DEFINITION. For a graph G , we denote νG = |VG | and ǫG = | EG | . The number νG of the vertices is called the order of G , and ǫG is the size of G . For an edge e = uv ∈ G , the vertices u and v are its ends. Vertices u and v are adjacent or neighbours, if uv ∈ G . Two edges e1 = uv and e2 = uw having a common end, are adjacent with each other. 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 . The ﬁgure on the right is a geometric representation of the graph G with VG = { v1 , v2 , v3 , v4 , v5 , v6 } and EG = { v1 v2 , v1 v3 , v2 v3 , v2 v4 , v5 v6 }.

v1 v3 v6

v2

v4

v5

Often we shall omit the identities (names v ) of the vertices in our ﬁgures, in which case the vertices are drawn as anonymous circles. Graphs can be generalized by allowing loops v v and parallel (or multiple) edges between vertices to obtain a multigraph G = (V, E , ψ), where E = {e1 , e2 , . . . , em } is a set (of symbols), and ψ : E → E (V ) ∪ { v v | v ∈ V } is a function that attaches an unordered pair of vertices to each e ∈ E : ψ(e) = uv . Note that we can have ψ(e1 ) = ψ(e2 ). This is drawn in the ﬁgure of b G by placing two (parallel) edges that connect the common ends. On the right there is (a drawing of) a multigraph G with vertices V = {a, b, c } and edges ψ(e1 ) = aa, ψ(e2 ) = a b, ψ(e3 ) = bc , and a c ψ(e4 ) = bc .

1.1 Graphs and their plane ﬁgures Later we concentrate on (simple) graphs. DEFINITION. We also study directed graphs or digraphs D = (V, E ), where the edges have a direction, that is, the edges are ordered: E ⊆ V × V . In this case, uv = vu.

5

The directed graphs have representations, where the edges are drawn as arrows. A digraph can contain edges uv and vu of opposite directions. Graphs and digraphs can also be coloured, labelled, and weighted: DEFINITION. A function α : VG → K is a vertex colouring of G by a set K of colours. A function α : EG → K is an edge colouring of G . Usually, K = [1, k] for some k ≥ 1. If K ⊆ (often K ⊆ ), then α is a weight function or a distance function.

Isomorphism of graphs

DEFINITION. Two graphs G and H are isomorphic, denoted by G ∼ = H , if there exists a bijection α : VG → VH such that uv ∈ EG ⇐⇒ α(u)α( v ) ∈ EH for all u, v ∈ G . Hence G and H are isomorphic if the vertices of H are renamings of those of G . Two isomorphic graphs enjoy the same graph theoretical properties, and they are often identiﬁed. In particular, all isomorphic graphs have the same plane ﬁgures (excepting the identities of the vertices). This shows in the ﬁgures, where we tend to replace the vertices by small circles, and talk of ‘the graph’ although there are, in fact, inﬁnitely many such graphs. Example 1.1. The following graphs are isomorphic. Indeed, the required isomorphism is given by v1 → 1, v2 → 3, v3 → 4, v4 → 2, v5 → 5.

v2 v5 v1 v4 v3 1 3 5 2 4

Isomorphism Problem. 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, and the number of all nonisomorphic graphs on n vertices. It tells that at least for computational purposes an efﬁcient algorithm for checking whether two graphs are isomorphic or not would be greatly appreciated. n 1 2 3 4 5 6 7 8 9

n

graphs 1 2 8 64 1024 32 768 2 097 152 268 435 456 236 > 6 · 1010 156 1044 12 346 274 668 nonisomorphic 1 2 4 11 34

1.1 Graphs and their plane ﬁgures

6

Other representations

Plane ﬁgures catch graphs for our eyes, but if a problem on graphs is to be programmed, then these ﬁgures are, to say the least, unsuitable. Integer matrices are ideal for computers, since every respectable programming language has array structures for these, and computers are good in crunching numbers. Let VG = { v1 , . . . , vn } be ordered. The adjacency matrix of 0 1 1 0 1 G is the n × n-matrix M with entries Mi j = 1 or Mi j = 0 1 0 0 1 1 according to whether vi v j ∈ G or vi v j ∈ / G . For instance, the 1 0 0 1 0 0 1 1 0 0 graph in Example 1.1 has an adjacency matrix on the right. Notice that the adjacency matrix is always symmetric (with 1 1 0 0 0 respect to its diagonal consisting of zeros). A graph has usually many different adjacency matrices, one for each ordering of its set VG of vertices. The following result is obvious from the deﬁnitions. Theorem 1.1. Two graphs G and H are isomorphic if and only if they have a common adjacency matrix. Moreover, two isomorphic graphs have exactly the same set of adjacency matrices. Graphs can also be represented by sets. For this, let = {X 1 , X 2 , . . . , X n } be a family of subsets of a set X , and deﬁne the intersection graph G as the graph with vertices X 1 , . . . , X n , and edges X i X j for all i and j (i = j ) with X i ∩ X j = . Theorem 1.2. Every graph is an intersection graph of some family of subsets. Proof. Let G be a graph, and deﬁne, for all v ∈ G , a set X v = {{ v, u} | vu ∈ G }. Then X u ∩ X v = if and only if uv ∈ G . ⊔ ⊓

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 , that is, s(G ) = min{|X | | G ∼ =G for some ⊆ 2X } .

How small can s(G ) be compared to the order νG (or the size ǫG ) of the graph? It was shown by KOU, STOCKMEYER AND WONG (1976) that it is algorithmically difﬁcult to determine the number s(G ) – the problem is NP-complete. Example 1.2. As yet another example, let A ⊆ be a ﬁnite set of natural numbers, and let GA = (A, E ) be the graph with rs ∈ E if and only if r and s (for r = s) have a common divisor > 1. As an exercise, we state: All graphs can be represented in the form GA for some set A of natural numbers.

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

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

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

dH ( v1 ) = d1 . Proof. .1). . . . dG ( vn ) = dn as in (1. E ) is trivial. dd1 +3 . Add a new vertex v1 and the edges v1 vi for all i ∈ [2. dn is graphical (when put into nonincreasing order). dn (with d1 ≥ 1 and n ≥ 2) is graphical if and only if d2 − 1. Consider the sequence s = 4. and they will be denoted by Kn . . 0 is graphical. 1. The last sequence corresponds to a graph with no edges. By Lemma 1. v1 v2 v6 v4 (1. and hence also our original sequence s is graphical. .2 Subgraphs 10 Theorem 1. 1. we can suppose that NG ( v1 ) = { v2 . All complete graphs of order n are isomorphic with each other.. νG = 1. A graph G is said to be regular. 3. all discrete graphs of order n are isomorphic with each other. dd1 +1 ]. . 2. .4 (HAVEL (1955). . . then G is r -regular or regular of degree r . But now the degree sequence of G − v1 is in (1. Clearly.2 and Theorem 1. dG ( vd1 +1 ) = dd1 +1 − 1. . 4. By Theorem 1. 0. . . The graph G = KV is the complete graph on V .1) v3 v5 Special graphs DEFINITION. . 1. 1 is graphical 2. . .3. d3 − 1.1. and dH ( vi ) = di for all i . (⇒) Assume dG ( vi ) = di . 0 is graphical 0. Then in the new graph H . Indeed. . otherwise G is nontrivial. if every two vertices are adjacent: E = E (V ).e. . 4.4. the graph G on the right has this degree sequence. if every vertex of G has the same degree. dG ( vd1+2 ) = dd1 +2 . HAKIMI (1962)). 3. d2 . .3. where EG = {e ∈ E (V ) | e ∈ / EG }. (⇐) Consider G of order n − 1 with vertices (and degrees) dG ( v2 ) = d2 − 1. A graph G = (V. . ⊔ ⊓ Example 1. if it has only one vertex. vd1 +1 }. . .1). . . The complement of G is the graph G on VG . dd1 +1 − 1. A sequence d1 . 2. 1. . The complements G = K V of the complete graphs are called discrete graphs. In a discrete graph EG = . s is graphical ⇐⇒ 3. dd1 +2 . i. If this degree is equal to r .

x y If n = 4. In particular. x y }. k]. with VQ k = k .5. the number of binary strings of length k. This kind of problems can be difﬁcult. k − 1]. For instance. uw }) ∪ {ux . 001.4. 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. 100. Let VH = VG ∪ { x . ǫKn = n(n − 1)/2. Indeed. and therefore ǫG ≤ n(n − 1)/2 for all graphs G that have order n. Example 1. to ﬁnd a maximal complete subgraph (a subgraph Km of maximum order) of a graph is unlikely to be even in NP .3 Paths and cycles 11 A discrete graph is 0-regular. It is a 3-regular graph of order 10. . Here ei and ei +1 are compatible in the sense that ei is adjacent to ei +1 for all i ∈ [1.1. Let k ≥ 1 be an integer.3 Paths and cycles The most fundamental notions in graph theory are practically oriented. where uv ∈ Q k if and only if the strings u and v differ in exactly one place. We show by induction that there exists a 3-regular graph G with νG = n. w u v 1. 010.6. and suppose that uv. y }. Q k is k-regular. because there are usually choices what branch to choose when leaving an intermediate point. . y w. then K4 is 3-regular. k 100 101 010 011 001 000 Example 1. For instance. 111}. called the k-cube. uw ∈ EG . 011. 101. 110 111 The order of Q k is νQ k = 2 . and a complete graph Kn is (n − 1)-regular. u y. Let n ≥ 4 be any even number. The sequence W = e1 e2 . Let G be a 3-regular graph of order 2m − 2. and EH = ( EG \ {uv. On the right we have the 3-cube. by the handshaking lemma. Also. ek is a walk of length k from u1 to uk+1 . Let ei = ui ui +1 ∈ G be edges of G for i ∈ [1. Walks DEFINITION. or simply the cube. x v. 110. Then H is 3-regular of order 2m. ǫQ k = k · 2k−1. Example 1. 3 = {000. and consider the set k of all binary strings of length k. Notice that all 3-regular graphs have even order by the handshaking lemma. Let Q k be the graph. and so. Many problems concerning (induced) subgraphs are algorithmically difﬁcult. . The graph on the right is the Petersen graph that we will meet several times (drawn differently). or at least nontrivial.

⋆ ⋆ ⋆ The join of two walks W1 : u − → v and W2 : v − → w is the walk W1 W2 : u − → w . we say that the path or cycle is even (odd). .. By repeating this argument. resp.. (Notice that in the above it may very well be that W1 or W3 is a trivial walk. we have a path as required. . all paths of length k are isomorphic. in W = W1 W2 W3 : u − → ui − → uj − → v the ⋆ ⋆ portion U1 = W1 W3 : u − → ui = u j − → v is a shorter walk.) ⊔ ⊓ . Each walk W : u − → v with u = v contains a path P : u − → v.3 Paths and cycles We write. W is a path. if it is closed. the inverse walk P −1 of a path P is a path (the inverse path of P ). − → uk − → u k +1 ⋆ 12 or → u k +1 . . and ui = u j for i = j except that u1 = uk+1 . W : u1 − k Write u − → v to say that there is a walk of some length from u to v . .3. Let i < j be indices such that ui = u j . if its length is 0. For a walk W : u = u1 − → . itself.) having k vertices is denoted by Pk (Ck . there is a ⋆ path P : u − → v that is obtained from W by removing edges and vertices. (Here the end v must be common to the walks. Proof. DEFINITION. W = e1 e2 . which is a path (cycle) of length k − 1 (k. Clearly.1. also W −1 : v = u k +1 − → . The length of a walk W is denoted by |W |. . if they have no vertices in common. if u1 = uk+1 .. if ui = u j for all i = j .). if they can share only their ends. − → uk+1 = v .. although we sometimes do not care to mention the edges ei on it. then W . − → uk+1 = v . − → u1 = u is a walk in G . W is a trivial path. ⋆ P5 ⋆ C6 Lemma 1. ek . A (sub)graph. that is. resp. A trivial path has no edges. U2 . When the procedure stops. we ⋆ obtain a sequence U1 . W is closed. is a path. W : u1 − → u2 − → . . A vertex u is an end of a path P . Clearly. Otherwise. W is a cycle. called the inverse walk of W . Here we understand that ⋆ W:u− → v is always a speciﬁc walk. and they are independent. Let W : u = u1 − → . If no such ⋆ ⋆ ⋆ i and j exist.. ek (ei = ui ui +1 ) be a walk. more informally. The same holds for cycles of ﬁxed length... . .. The join of two paths need not be a path. If k is even (odd). if P starts or ends in u. Let W = e1 e2 .) Paths P and Q are disjoint. Um of walks u − → v with |W | > |U1 | > · · · > |Um |.

where the weights present the rate of unreliability or frequency of action of the connections. and ∗ NG ( v ) = {u | dG ( v. if P = e1 e2 . connected. i. for every vertex v ∈ / H. then its weight is k α( P ) = i =1 α(ei ). If c (G ) = 1. . otherwise. G α is a graph G together with a weight function α : EG → on its edges. v ) < ∞ for all u. where the weights on edges are distances. In particular. that is. ek is a path. it is disconnected. then G is. v ) = ∞ by convention. In particular. Apparently. u) < ∞} is the connected component of G that contains v ∈ G .1. The minimum weighted distance between two vertices is α dG (u.. The maximal connected subgraphs of G are its connected components. or rates of ﬂow in the network. • a system of channels in (tele)communication or computer architecture. every connected component is an induced subgraph. let dG (u.3 Paths and cycles DEFINITION. • a model of chemical bonds. For H ⊆ G . . If there exists a walk (and hence a path) from u to v in G . If there are no walks u − → v . . v ) = min{α( P ) | P : u − → v} . ⋆ Shortest paths DEFINITION. A graph G is connected. for mail). the subgraph G [VH ∪ { v }] is disconnected. 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 . v ) = min{k | u − → v} k 13 be the distance between u and v . ⋆ In extremal problems we seek for optimal subgraphs H ⊆ G satisfying speciﬁc conditions. where the weights measure molecular attraction. travel expenses. v ∈ G . of course. In practice we encounter situations where G might represent • a distribution or transportation network (say. let α( H ) = e∈H α( e ) be the (total) weight of H .e. if dG (u. Denote c (G ) = the number of connected components of G . Let G α be an edge weighted graph. let dG (u. the connected components form a partition of G .

∞} = 3. . For this. • t ( v5 ) = min{4. These are the minimal weights from v0 to each vi . ui } be any vertex with the least value t (ui +1 ). and then one wants to ﬁnd a subgraph with the maximum weight. or all vertices (if we want to travel around). 4} = 3. and which connects two given vertices. others are ∞. t ( v2 ) = min{∞. t ( v4 ) = 4. ∞} = 4. 2} = 2. if the graph represents a network of pipelines. • t ( v2 ) = min{3. Thus choose u2 = v3 . others are ∞. Let us make the convention that α(uv ) = ∞. The algorithm stops. (ii) For i ∈ [0. . t (u0 ) = 0 and t ( v ) = ∞ for all v = u0 . 3 + 1} = 4. 4 + 1} = 4. t ( v4 ) = 2 + 3 = 5. t ( v5 ) = 2 + 2 = 4. 3 + 2} = 5. .3 Paths and cycles 14 In these examples we look for a subgraph with the smallest weight. α (iii) Conclusion: dG (u. where u is a ﬁxed starting point and v ∈ G . Thus set u3 = v2 . t ( v4 ) = min{5. α ﬁnd dG (u. 3} = 3. Dijkstra’s algorithm: (i) Set u0 = u. t ( v2 ) = 3. t (u0 ) = 0. • t ( v4 ) = min{5. The shortest path problem: Given a connected graph G with a weight function α : EG → . ui }. v ∈ G. t ( v3 ) = 3. Apply Dijkstra’s algorithm to the vertex v0 . v ) for given u. . We have obtained: t ( v1 ) = 2. call α(uv ) the length of uv . if uv ∈ / G. .7. • t ( v1 ) = min{∞. t (ui ) + α(ui v )} . t ( v3 ) = 2 + 1 = 3. Dijkstra’s algorithm solves the problem for every pair u. 3 + 1} = 4. νG − 1]: for each v ∈ / {u1 . v . Thus u1 = v1 . v ) = t ( v ). . v0 3 v2 1 v4 2 v1 2 2 3 2 2 1 v3 1 v5 • t ( v2 ) = min{3. Thus choose u4 = v4 . t ( v5 ) = min{4.1. the weights are volumes or capacities. replace t ( v ) by min{ t ( v ). Assume that G is a connected graph. Consider the following weighted graph G . t ( v5 ) = 4 . let G be a graph with an integer weight function α : EG → . Let ui +1 ∈ / {u1 . t (u1 ) + α(u1 v2 )} = min{3. Example 1. In this case. We consider the minimum problem. On the other hand. t ( v5 ) = min{4. • u0 = v0 . . . .

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

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

(⇒) Observe that if G is (X . . and let H = G [ F ]. and Y of those vertices with colour 2. y ) is odd} . Y1 ∪ Y2 ∪ · · · ∪ Yp ). . VG = X ∪ Y . otherwise. ⋆ ⋆ Let then u.2. Therefore X and Y are stable subsets. and colour it by 1. 17 Proof. then so are all its subgraphs. Example 2. since. i. Indeed. Thus | P2 | and |Q 2 | have the same parity and hence the sum −1 | P2 |+|Q 2 | is even. Start from any vertex v1 .) Now 1 1 1 1 d G ( v ) = ǫG . Assume that x is the last common vertex of P and Q: ⋆ ⋆ P = P1 P2 . X ∩ Y = . Denote F = EG ∩ {uv | u ∈ X . Yi )-bipartite. ⊔ ⊓ Checking whether a graph is bipartite is easy. First. By the maximum condition. Indeed. if v ∈ X . an odd cycle C2k+1 is not bipartite. v is on the wrong side. | P1 | = |Q 1 |. at some point one of the vertices gets both colours. the path P2 Q 2 is even. then each cycle of G is contained in one of the connected components G1 . if G is disconnected. Then colour the neighbours of v1 by 2. where P2 : x − → u and Q 2 : x − → w are independent. If Gi is (X i . Let v ∈ G be a chosen vertex. we note that it sufﬁces to show the claim for connected graphs. this can be done by using two ‘opposite’ colours. A graph G is bipartite if and only if G it has no odd cycles (as subgraph). Y )-bipartite. Consequently. and proceed by colouring all neighbours of an already coloured vertex by the opposite colour. then the pair X ′ = X \ { v }. Also. and G is bipartite as claimed. Since G is connected.2.1 Bipartite graphs and trees Theorem 2. Y ′ = Y ∪ { v } does better that the pair X . and it is bipartite. (That is. Gp of G .. Q = Q 1Q 2 . Assume thus that G is connected. w ∈ G be both in X or both in Y . . let VG = X ∪ Y be a partition such that the number of edges between X and Y is maximum. We show that each graph G has a bipartite subgraph H ⊆ G 1 such that ǫH ≥ 2 ǫG . P1 v x P2 Q2 1 2 2 2 1 u uw w Q1 1 2 1 1 2 If the whole graph can be coloured without contradiction.2 (ERDÖS (1965)). then G is (X . dH ( v ) ≥ dG ( v )/2. Since P and Q are ⋆ shortest paths. and in this case. where X consists of those vertices with colour 1. v ∈ Y } . Obviously H is a spanning subgraph.e. otherwise. P1 and Q 1 are shortest paths v − → x . Y . and so uw ∈ / EG by assumption. by the deﬁnition of distance. and let P : v − → u and Q : v − → w be (among the) shortest paths from v to u and w . However. then G has the bipartition (X 1 ∪ X 2 ∪ · · · ∪ X p . say 1 and 2. G is not bipartite. and deﬁne X = { x | dG ( v. dH ( v ) ≥ ǫH = 2 v∈H 2 v∈G 2 2 . Y )-bipartite. (⇐) Suppose that all cycles in G are even. x ) is even} and Y = { y | dG ( v. . Indeed.

⋆ ⋆ and there is a path P : v − → u in G − e. We note that. the ends u and v are not connected in G − e. (⇒) If there is a cycle in G containing e. An edge e ∈ G is a bridge of the graph G . eP : u − →v− → u is a cycle in G containing e. (⇐) If e = uv is not a bridge. that is. Example 2.3 and the deﬁnition of a tree.3.1 Bipartite graphs and trees 18 Bridges DEFINITION. An acyclic graph is also called a forest. for each edge e ∈ G . . An edge e ∈ G is a bridge if and only if e is not in any cycle of G. say C = PeQ. an edge e in a connected G is a bridge if and only if G −e is disconnected.3. the ﬁrst of which was given by CAYLEY in 1889: There are nn−2 trees on a vertex set V of n elements. e = uv is a bridge ⇐⇒ u. For (i). We omit the proof. ⋆ Let w ∈ G . A connected graph is a tree if and only if all its edges are bridges. and so e is not a bridge. By Theorem 2. A tree is a connected acyclic graph. For (ii). ⋆ ⋆ unless P : w − → u → v contains e = uv . and f is not a bridge of H . Proof. v in different connected components of G − e . then QP : v − → u is a path in G − e.1. A graph is called acyclic. Since e is a bridge. The following enumeration result for trees has many different proofs. if G −e has more connected components than G . On the right (only) the two horizontal lines are bridges. and therefore C is inside H . there exists a path P : w − → v in G . Let e be a bridge in a connected graph G. then f is a bridge of G. Proof. we have Corollary 2. Now. if c (G −e) > c (G ). This is a path of G − e. in which case the part w − → u is a path in G −e. Theorem 2. In particular. if f ∈ H belongs to a cycle C of G . then C does not contain e (since e is in no cycle). ⊓ ⊔ Lemma 2. ⊔ ⊓ ⋆ Trees DEFINITION. then u and v are in the same connected component of G − e.1. (i) Then c (G −e) = 2. and most importantly. If f ∈ H is a bridge of H. if it has no cycles. Since G is connected. (ii) Let H be a connected component of G − e. let e = uv .2.

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

and N by −. Then ǫG ≥ ǫ T = ν T − 1 = νG − 1. where vw is the unique edge having an end v . If k is odd. then these are divided into ⌊k⌋ pairs. where a play corresponds to an edge of the tree. T is connected. we have d T ( v ) = 1. In this case. Since T is acyclic. For each connected graph G. then one of the teams goes to the next round without having to play. Spanning trees Theorem 2. for which we have proved that | Ei | = |Vi |− 1. ǫ T − v = n − 2. Therefore. a connected graph G is a tree if and only if ǫG = νG − 1. Each edge can be designated at most ⋆ once. Consider a cup tournament of n teams. and from each pair only the winner continues. let the connected components of T be Ti = (Vi . P and N alternate turns so that P designates an edge by +. k k n − 1 = ǫT = i =1 (|Vi | − 1) = i =1 |Vi | − k = n − k . We need to show that T is connected. since the cup tournament is a tree. and hence they are trees. and it satisﬁes the condition (ii). Indeed. 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 . subgraph that is a tree).5. ν T = i =1 |Vi |. s.5. so are the connected graphs Ti . a spanning graph that is a tree. and so ǫ T = ǫ T −v + 1 = n − 1. By induction hypothesis. ⊔ ⊓ which gives that k = 1. Proof. If during a round there are k teams left in the tournament. Moreover. Let T be a spanning tree of G . Proof. and after the second round 4 teams continue. and the claim follows.3.1 Bipartite graphs and trees ⋆ 20 Let P : u − → v be a maximal path in T . ⊔ ⊓ Example 2. to designate at least one edge in each such path). k k Now. ⋆ P: u − → w − → v . that is.2. for i ∈ [1. It is P ’s purpose to designate a path s − → r (that is. If VT = VG . But then T is not maximal. and ǫ T = i =1 | Ei |.4. there exists an edge uv ∈ / EG such that u ∈ T and v ∈ / T . We say that a game (G . a contradiction. (iii)⇒(i) Assume (iii) holds for T . k]. Example 2. The answer to our problem is n − 1. and N tries to block all paths s − → r (that is. Each connected graph has a spanning tree. then 2. to designate all edges in one such ⋆ path). ⊔ ⊓ Corollary 2. r ) is . By Lemma 2.e. after the ﬁrst round 7 teams continue.2. that is.. So 13 plays are needed in this example. How many plays are needed to determine the winner? So if there are 14 teams. ǫG ≥ νG − 1. The second claim is also clear. The subgraph T − v is connected. Ei ). Let T ⊆ G be a maximum order subtree of G (i.

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

. we still have a spanning tree. e. No solutions are known for any n ≥ 7. Also. e2 = v4 v5 . A message that is sent to v is preceded by the address a( v ). then a cycle C containing ek is created. then necessarily n = k2 or n = k2 + 2 for some k ≥ 1. v ) = h(a(u). Consider trees T with weight functions α : E T → . If we add ek to T1 . The length of an address is the number of its bits. one by one. 2 1 5 8 4 TAYLOR (1977) proved: if T of order n exists. 2. When we replace e by ek in T1 . e4 = v2 v3 and e5 = v1 v2 . v1 3 v2 1 4 1 2 v4 1 v5 3 2 v6 2 2 2 v3 Problem. We deduce that α( T ) ≤ α( T1 ). The total weight of the spanning tree is thus 9. 01100) = 3 and h(10000.g. the algorithm produces the sequence: e1 = v2 v4 . Example 2. α must be injective. Let ek be the ﬁrst edge produced by the algorithm that is not in T1 . Example 2. Also. This would make it easier for a node computer to forward a message. 2 ? In such a weighted tree T α different paths have different n weights. where a( v ) and a(u) differ. there may exist several optimal spanning trees (with the same weight. the selection e1 = v2 v5 .1 Bipartite graphs and trees 22 Suppose T1 is any spanning tree of G . α(ek ) ≤ α(e). Also. a( v )). . where the vertices are the node computers. by the construction. a(u)) of two addresses of the same length is the number of places. 00000) = 1. Indeed. and the edges indicate the direct links. and each i ∈ [1. .6. Note that T2 has more edges in common with T than T1 . A computer network can be presented as a graph G . (Why is this so?) Does there exist a weighted tree T α of order n such that n the (total) weights of its paths are 1. 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 can transform T1 to T by replacing edges. 2 ] is a weight of one path. . e5 = v1 v2 gives another optimal solution (of weight 9). When applied to the weighted graph on the right. 010 000 100 110 111 . such that the total weight does not increase. e2 = v4 v5 . C must contain an edge e that is not in T . and therefore α( T2 ) ≤ α( T1 ). a bit string (of zeros and ones). e3 = v3 v6 . h(00010. if two vertices were adjacent. A graph G is said to be addressable. if it has an addressing a such that dG (u. e4 = v3 v6 .2. In particular. Repeating the above procedure. their addresses should differ by one symbol. e3 = v5 v6 . The outcome of Kruskal’s algorithm need not be unique. However.7. of course) for a graph. Each computer v has an address a( v ).. Each tree T of order n has n exactly 2 paths. The Hamming distance h(a( v ). say T2 .

v ). By the induction hypothesis. A star address will be a sequence of these three symbols. v )-separating set. . the addresses of the vertices of T can be chosen to be of length ν T − 1. “A Course in Combinatorics”. but where two vertices are always connected by at least two independent paths. . We may also wish to construct graphs that are as simple as possible. v ) is the number of places. Set the address of v1 to 1a2 . if G −S is disconnected. 2. we can address the tree T − v1 by addresses of length k − 1. we modify the addressing for general graphs by introducing a special symbol ∗ in addition to 0 and 1. see VAN LINT AND WILSON. We change this addressing: let ai be the address of vi in T − v1 . v ) = dG (u. A subset S ⊆ VG is a separating set. The proof goes by induction. the addresses of the vertices are simply 0 and 1.2. ∗00 ∗ ∗∗) = 0. h(10 ∗ ∗01. and change it to 0ai . Moreover. In order to gain more generality. where one has to make sure that a break-down of one connection does not affect the functionality of the network. and assume that d T ( v1 ) = 1 (a leaf) and v1 v2 ∈ T . Separating sets DEFINITION. We also say that S separates the vertices u and v and it is a (u. 0 ∗ ∗101) = 1 and h(1 ∗ ∗ ∗ ∗∗. a( v2 ) = 10 ∗ 0 . It is now easy to see that we have obtained an addressing for T as required. In the case ν T = 2. vk+1}. where cost is the criterion. A vertex v ∈ G is a cut vertex. Similarly. . We star address this graph as follows: a( v1 ) = 0000 . in a reliable network we require that a break-down of a node (computer) should not result in the inactivity of the whole network. We still want to have h(u. if u and v belong to different connected components of G −S . h(u. a( v4 ) = ∗ ∗ 11 . The special symbol ∗ does not affect h(u. v1 v2 v4 v3 These addresses have length 4. Can you design a star addressing with addresses of length 3? WINKLER proved in 1983 a rather unexpected result: The minimum star address length of a graph G is at most νG − 1. Let then VT = { v1 . . For the proof of this.2 Connectivity 23 We prove that every tree T is addressable. a( v3 ) = 1 ∗ 01 .2 Connectivity Spanning trees are often optimal solutions to problems. v ). The Hamming distance remains as it was. These problems occur especially in different aspects of fault tolerance and reliability of networks. that is. . where u and v have a different symbol 0 or 1. So. If ν T ≤ 2. The triangle K3 is not addressable. then the claim is obvious. if c (G − v ) > c (G ).

and • otherwise κ(G ) equals the minimum size of a separating set of G . Assume that G = K2 and thus that νG ≥ 3.2 Connectivity 24 If G is connected. Proof. • κ(G ) = 0. In other words. A graph G is k-connected. Let G be connected. • κ(G ) = νG − 1. { v } is a separating set. the subgraph H = G −( F \ {e}) is connected for given e ∈ F . F ⊆ EG } . Now VG \ {u. if G is disconnected.4. κ(G ) = 2 and κ′ (G ) = 2. since G is connected. either νGu ≥ 2 (and u is a cut vertex) or νGv ≥ 2 (and v is a cut vertex). If F be a bond of a connected graph G. if G is connected. if G is disconnected. The (vertex) connectivity number κ(G ) of G is deﬁned as κ(G ) = min{k | k = |S |. and thus c (G − F ) = 2. The following lemma is immediate. if κ(G ) ≥ k. that is. if G is a complete graph. Proof. v } is a separating set. then κ′ (G ) = 0. If S ⊆ VG separates u and v.1. By Lemma 2. c (H −e) = 2. Lemma 2. then the claim is clear. let κ′ (G ) = 0.2. On the right. The claim follows from this. if κ′ (G ) ≥ k. Let Gu = NG − e (u) ∗ and Gv = NG −e ( v ) be the connected components of G − e containing u and v . A graph G is k-edge connected. then it is a complete graph. Since G − F is disconnected.5. For trivial graphs. Now. Let κ′ (G ) = min{k | k = | F |. ⊔ ⊓ DEFINITION. Lemma 2. Notice that the minimum degree is δ(G ) = 3. then c (G − F ) = 2. An edge cut F of G consists of edges so that G − F is disconnected.3. Hence e is a bridge in H . Again. ∗ Proof. Lemma 2. and F is minimal. then it is 1-connected. S ⊆ VG } . If νG ≤ 2. G − F disconnected. then every path P : u − → v visits a vertex of S. and let uv ∈ / G . then v is a cut vertex if and only if G − v is disconnected.6. A minimal edge cut F ⊆ EG is a bond ( F \ {e} is not an edge cut for any e ∈ F ).8. If e = uv is a bridge. assume that G is not complete. If a connected graph G has no separating sets. DEFINITION. ⊔ ⊓ ⋆ Lemma 2. Example 2. since H −e = G − F . ⊔ ⊓ . For νG ≥ 3. then either G = K2 or one of u or v is a cut vertex. Clearly. G −S disconnected or trivial.

required for the induction. ⊔ ⊓ G . v )-separating set of the smallest size. (The choices for different edges need not be different.) Note that since f = e. By Lemma 2. there are k independent paths u − → v in G as required. .6 (WHITNEY (1932)). and let e = uv ∈ F . either end of f is different from u or v . Then the minimum number of vertices separating u and v is equal to the maximum number of independent paths from u to v. and the claim follows. The case for k = 1 is clear. Assume G is nontrivial. w k } is a (u. (1) Assume ﬁrst that u and v have a common neighbour w ∈ NG (u) ∩ NG ( v ). . κ′ (G ) ≤ δ(G ). Menger’s theorem Theorem 2. Then F is a bond. ⋆ and the induction hypothesis yields that there are k − 1 independent paths u − → v in G − w . Conversely.2 Connectivity Theorem 2. If κ′ (G ) = 1. then G has at least (and thus exactly) k independent ⋆ paths u − → v. then either G −S = K2 (= e). then every path u − → v of G visits S . if G −S is connected. In both of these cases. 25 Proof. Proof. Then necessarily w ∈ S . there is a vertex of G −S . whose removal results in a trivial or a disconnected graph. Thus |S | ≤ | F | − 1 = κ′ (G ) − 1. If a subset S ⊆ VG is (u. v )-separating. In conclusion. . H . In the smaller graph G −w the set S \ {w } is a minimum (u. we disconnect G . Hence |S | is at least the number of independent paths from u to v . Let S be the collection of these choices. If κ′ (G ) = 0. where e is a bridge. and in this case also κ(G ) = 0. Let F be an edge cut of G with | F | = κ′ (G ). either G = K2 or G has a cut vertex. For any graph G. . . F Now for each f ∈ F \ {e} choose an end different from u and v . . and therefore G −S ⊆ H is an induced subgraph of H ). and G −S does not contain edges from F \ {e}. v )-separating set. . since if we remove all edges with an end v . Let u.7 (MENGER (1927)). we use induction on m = νG + ǫG to show that if S = {w1 . also κ(G ) = 1. κ(G ) ≤ |S | + 1 ≤ κ′ (G ). κ(G ) ≤ κ′ (G ) ≤ δ(G ) . Consider the connected subgraph H = G − ( F \ { e }) = ( G − F ) + e . ⋆ . . . On the other hand. Assume then that κ′ (G ) ≥ 2.5. w2 . and G − F has two connected components. then S is a separating set and so κ(G ) ≤ |S | ≤ κ′ (G ) − 1 and we are done. ⋆ Together with the path u − →w− → v . .2. . . If G −S is disconnected. then G is disconnected. and this takes care of the small values of m. In both of these cases. e . . Clearly. then G is connected and contains a bridge. or either u or v (or both) is a cut vertex of G −S (since H −S = G −S . v ∈ G be nonadjacent vertices of a connected graph G.

⋆ If |S ′ | ≥ k. ⊔ ⊓ Theorem 2. and Q : y − → v.. let S ′ be a minimum set that separates u and v . and so y = v . ⋆ there are k independent paths u − → v in Gu .1) Suppose next that S NG (u) and S NG ( v ). since the path P is shortest and so u y ∈ /G (meaning that u is not adjacent to all of S y ). A graph G is k-connected if and only if every two vertices are connected by at least k independent paths..2) There remains the case. and they have size k. where for all (u. by the induction hypothesis. We noted that Gu is smaller than G . which contradicts the assumption (2). By our current assumption. y ∈ / S ′. by (2). then. f = x y . ⋆ ⋆ . all w i ∈ S have to be adjacent to some vertex in H v . moreover. and thus by the induction hypothesis.8 (MENGER (1927)). The graph Gu is connected and it is smaller than G . and deﬁne Gu to be the graph on Hu ∪ S ∪{ v } having the edges of G [Hu ∪ S ] together with vw i for all i ∈ [1. there are k independent paths u − → v in G − f . which is deﬁned similarly to Gu . Let Sx = S′ ∪ { x } and S y = S′ ∪ { y } . (Note that then. |S ′ | < k. wk . By the present assumption.1) rules out the case H v = { v }. that have only the end u in common. (2. This shows that ǫGu ≤ ǫG . we obtain k independent paths u − → wi − → v in G . in order for S to be a minimum separating set. Since k is the size of a minimum (u. and the claim is clear in this case. A symmetric argument applies to the graph Gv (with a new vertex u). These sets separate u and v in G (by the above fact). and so x . This is possible only if there exist k paths ⋆ u− → w i . k]. one for each i ∈ [1. v )-separating set of Gu . k]. on the other hand. This means that S ′ separates u and v in G . the vertices of S y are adjacent to v . The assumption (2) yields that u is adjacent to all of S x . In the smaller graph G − f . either S ⊆ NG (u) or S ⊆ NG ( v ). and. v )-separating set.2. (2. S ∩ NG ( v ) = or S ∩ NG (u) = . So |H v | ≥ 2 and νGu < νG . Let v be a new vertex. we have |S ′ | ≥ k. v )-separating sets S of k elements. where e = ux . by the assumption (2). Notice that. Every path u − →v in G −S ′ necessarily travels along the edge f = x y . But these are paths of G . since ux ∈ G . w2 u w1 v If S ′ is any (u. and denote by Hu = NG −S (u) and H v = NG −S ( v ) the connected components of G −S for u and v . then u and v are still connected in G −S ′ .2 Connectivity 26 ∗ ∗ (2) Assume then that NG (u) ∩ NG ( v ) = . the assumption (2. This ⋆ yields that there are k paths w i − → v that have only the end v in common. then S ′ will separate u from all w i ∈ S \ S ′ in G .) Let P = e f Q be a shortest path u − → v in G . | P | ≥ 3. also u is nonadjacent to some vertex of S . But now both u and v are adjacent to the vertices of S ′ . ⋆ If. Indeed. When we combine ⋆ ⋆ ⋆ these with the above paths u − → w i .

5. (The case for G1 −S is symmetric. κ(Q k ) ≤ δ(Q k ) = k. κ(G ) ≤ κ′ (G ) ≤ δ(G ). Say. vi ) of pairs of vertices for i = 1. Let S be a separating set of Q k with |S | ≤ k. ⋆ and it is asked whether there exist paths Pi : ui − → vi that have no vertices in common. then. also κ′ (Q k ) = k. . Indeed. for ﬁxed k. 1u) for all u ∈ k−1 . 1u) of Q k would guarantee that Q k −S is connected. .2 Connectivity 27 Proof. Example 2.9. So we can assume that G0 −S is disconnected. Let u. In the disjoint paths problem we are given a set (ui . This proves the claim.) By the induction hypothesis. Now u and v are connected by at most k − 2 independent paths in G − e. this vertex is u. . suppose that κ(G ) = k. u and v can be separated in G − e by a set S with |S | = k − 2. Recall the deﬁnition of the cube Q k from Example 1. A uv -disconnecting set is a set F ⊆ EG such that every path ⋆ u− → v contains an edge from F . the problem has a fast algorithm due to ROBERTSON and SEYMOUR (1986). ⊔ ⊓ We state without a proof the corresponding separation property for edge connectivity.10. which contradicts the assumption that κ(G ) = k. Then the maximum number of edge⋆ disjoint paths u − → v equals the minimum number k of edges in a uv-disconnecting set. and by Theorem 2.) . If any two vertices are connected by k independent paths. Corollary 2. and the edges (0u. We have κ′ (Q k ) = k for the k-cube. Theorem 2. Let G be a graph.7. v }. If both G0 −S and G1 −S were connected. κ(G0 ) = k − 1. The vertex w is separated ⋆ in G − e by S from u or from v . These are (isomorphic to) Q k−1. of course. If there were no vertices from G1 in S . also Q k −S would be connected. by Whitney’s theorem. The connectivity problems tend to be algorithmically difﬁcult. Consider the graph G − e. and it separates u from w in G .2. A graph G is k-edge connected if and only if every two vertices are connected by at least k edge disjoint paths.4. it must be that e = uv ∈ G . a contradiction. G1 −S is connected. This problem was shown to be NP-complete by KNUTH in 1975. otherwise there would be a path u − → v in (G − e)−S . The set S ∪ { v } has k − 1 elements. Extract from Q k the disjoint subgraphs: G0 induced by {0u | u ∈ k−1 } and G1 induced by {1u | u ∈ k−1 }.9. Hence |S | ≥ k . we show the claim by induction. Example 2. then it is clear that κ(G ) ≥ k. and hence S contains at least k − 1 vertices of G0 (and so |S | ≥ k − 1). 1u) necessarily remains in Q k −S . .7. Algorithmic Problem. We show that κ(Q k ) = k. Since κ(Q k ) = k = δ(Q k ). Since νG > k (because κ(G ) = k). v ∈ G with u = v in a graph G. DEFINITION. k. but that G has vertices u and v connected by at most k − 1 independent paths. By Theorem 2. In converse. 2. First of all. In converse. and Q k is obtained from the union of G0 and G1 by adding the 2k−1 edges (0u. since one pair (0u. (However. there exists a w ∈ G that is not in S ∪ {u.

Then the path W : u − → w (or w − → u) along C contains all vertices of S ∩ VC .2 Connectivity 28 Dirac’s fans DEFINITION.10 that there is a ( v. if they have only v in common. and thus no bridges. G has no cut vertices. It follows from Theorem 2. ⊔ ⊓ .2. Exercise. Let G be a k-connected graph for k ≥ 2. S )-fan. . and proves the claim. there exists a ( v. since κ(G ) ≥ 2. Let S ⊆ VG be such that |S | = k. Let the vertices of S ∩ VC be v1 . It follows that every edge. S )-fan of k paths..11 (DIRAC (1960)). Proof. v Theorem 2. Let v ∈ G and S ⊆ VG such that v ∈ / S in a graph G . and let C be a cycle of G that contains the maximum number of vertices of S . . vi +1 ) of C . vr listed in order around C so that each pair ( vi . otherwise. |VC |} paths. First of all. Then for any k vertices. 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. Proof. there exists a cycle of G containing them. r ]. S ⊔ ⊓ Theorem 2. Such a path is referred to as a segment of C .10 (DIRAC (1960)). . suppose v ∈ S is not on C . .. ∗ ∗ ∗ . This contradicts the choice of C . and thus every vertex of G belongs to a cycle. ⋆ ⋆ Therefore there are two paths P : v − → u and Q : v − → w in such a fan that end in the same ⋆ ⋆ segment ( vi . A set of paths from v to a vertex in S is called a ( v. VC )-fan of at least min{k. If C contains all vertices of S . then we are done. But now PW Q−1 is a cycle of G that contains v and all vi for i ∈ [1. vi +1 ) (with indices modulo r ) deﬁnes a path along C (except in the special case where r = 1).

. this problem concerns of travelling in a graph such that one tries to avoid using any edge twice. where in order to save time each street should be travelled only once. . en : v0 − → vn ⋆ ⋆ with ei = vi −1 vi be a longest trail in G . e n e1 . Such a trail is called an Euler tour. in optimizing distribution networks – such as delivering mail. . . A complete proof of the following Euler’s Theorem was ﬁrst given by HIERHOLZER in 1873. A connected graph G is eulerian. en is an Euler tour (and so EG = {e1 . dG (u) is even. another edge departs from v . Notice that if W = e1 e2 . . This contradiction to the choice of W proves the claim. However. A walk W = e1 e2 . (⇐) Assume G is a nontrivial connected graph such that dG ( v ) is even for all v ∈ G . en }).1 Eulerian graphs The ﬁrst proper problem in graph theory was the Königsberg bridge problem. It follows that all e = vn w ∈ G are among the edges of W . . . and it is longer than W . . there exists an edge f = vi u ∈ G for some i . W could be prolonged to We. Also. HIERHOLZER (1873)). . . en e1 . W is a closed trail. for instance. Every time an edge arrives at v . where one avoids lifting the pen off the paper while drawing the lines. . then. since W starts and ends at u. en is a trail. Let W = e1 e2 . In particular. e i f is a trail in G . A connected graph G is eulerian if and only if every vertex has an even degree. then dG ( vn ) = 2(k − 1) + 1 and that would be odd. v0 = vn . if it has a closed trail containing every edge of G . ei −1 is an Euler tour for all i ∈ [1. The same problem occurs in mechanical graph plotting. which is not in W . In general. now e i + 1 .) If W is not an Euler tour. An Euler trail of a graph G is a trail that visits every edge once.3 Tours and Matchings 3. (⇒) Suppose W : u − → u is an Euler tour. . if ei = e j for all i = j . Euler tours DEFINITION. . otherwise. Proof. Theorem 3. .1 (EULER (1736). also ei ei +1 . Let v (= u) be a vertex that occurs k times in W . e2 . for. In practice these eulerian problems occur. n]. . if it were vn = v0 and vn occurs k times in W . ⊔ ⊓ . since G is connected. . . and therefore dG ( v ) = 2k. (Indeed. . that is.

and there is a good algorithm for ﬁnding such a tour: Fleury’s algorithm: • Let v0 ∈ G be a chosen vertex. and therefore G has exactly two such vertices. . Let G be a graph with a weight function α : EG → + . ⊔ ⊓ ⋆ The Chinese postman The following problem is due to GUAN MEIGU (1962). then any Euler tour will do as a solution. The k-cube Q k is eulerian for even integers k. and ⋆ ⋆ hence it has an Euler tour. the post ofﬁce). and let W0 be the trivial path on v0 . where a postman wishes to plan his route to save the legs. ei }. ei has been constructed. ⊔ ⊓ . Theorem 3. Consider a village. each vertex w∈ / {u. Exercise. 2. unless there is no alternative.1. If G is eulerian. The Chinese postman problem is to ﬁnd a minimum weighted tour in G (starting from a given vertex. say u − →v− →w− → u. and the edges uw and vw . Assume then that G is connected and has at most two vertices of odd degree. i ]) so that (i) ei +1 has an end vi . where e j = v j −1 v j . 30 Theorem 3. as long as possible: suppose a trail Wi = e1 e2 . as is natural. .1. . v } has an even degree. then. Proof. . . Let H be a graph obtained from G by adding a vertex w . by the handshaking lemma.1. and (ii)ei +1 is not a bridge of Gi = G −{e1 .3. If G is eulerian.3.2. . • Repeat the following procedure for i = 1. Proof. Notice that. In H every vertex has an even degree. This problem is akin to Euler’s problem and to the shortest path problem. A connected graph has an Euler trail if and only if it has at most two vertices of odd degree. because such a tour traverses each edge exactly once and this is the best one can do. Otherwise. If G has no vertices of odd degree then. Here the beginning part u − → v is an Euler trail of G . G has an Euler trail. If G has an Euler trail u − → v . but still every street has to be walked through. because Q k is k-regular. then any trail of G constructed by Fleury’s algorithm is an Euler tour of G.1 Eulerian graphs Example 3. . Choose an edge ei +1 (= e j for j ∈ [1. say u and v . . the weights α(e) play no role in the eulerian case. In this case the weight of the optimal tour is the total weight of the graph G . by Theorem 3. every graph has an even number of vertices with odd degree. as in the proof of Theorem 3. .

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

some interesting general conditions. This graph is not hamiltonian. Let ui be the last vertex of C that belongs to Gi . But then necessarily |X | = |Y |. Let G be a graph of order νG ≥ 3. but it does satisfy the condition c (G −S ) ≤ |S | for all S = . i ∈ [1. (⇒) This is trivial since if G has a Hamilton cycle C . however.4. There exists an i with 1 < i < n such that uvi ∈ G and vi −1 v ∈ G . Suppose thus that e is on C . Indeed. v ∈ G be such that dG (u) + dG ( v ) ≥ n. The case k = 1 is trivial. Theorem 3. Interestingly this graph is (X . Gi . then C is also a Hamilton cycle of G + uv . and v j = vt for all j = t .1) no good characterization is known for hamiltonian graphs. If uv ∈ G . then it is a Hamilton cycle of G . due to ORE. Let = S ⊆ VG . There are.3. then there is nothing to prove. otherwise. since X and Y are stable subsets. then for every nonempty subset S ⊆ VG . − → vn = v is a Hamilton path of G .1 is not sufﬁcient to ensure that a graph is hamiltonian. Proof. Proof. which appears in many places in graph theory as a counter example for various conditions. c (G −S ) = 3 > 2 = |S | for the set S of black vertices.. v ∈ G be such that dG (u) + dG ( v ) ≥ νG . k]. In G .2 Hamiltonian graphs 32 and thus the cycle visits the sets X . If C does not use the edge e. Example 3. Therefore the conclusion of Lemma 3. For.. If G is hamiltonian. c (G −S ) ≤ |S | . Denote n = νG . and hence suppose that k > 1. Assume G −S has k connected components. the problem to determine if G is hamiltonian is NP-complete. Now vi ∈ S for each i by the choice of ui . dG ( v ) < n − dG (u) would contradict the assumption. Then G is hamiltonian if and only if G + uv is hamiltonian. ⊔ ⊓ Example 3. and hence it is not hamiltonian. We may then assume ⋆ that C : u − →v− → u.3.4 (ORE (1962)). Unlike for eulerian graphs (Theorem 3. and let C : u − → u be a Hamilton cycle of G .1. Assume thus that uv ∈ / G. generalizes an earlier result by DIRAC (1952). because C is a cycle and ui vi ∈ G for all i . and let u. Thus |S | ≥ k as required. Therefore G does not satisfy the condition of Lemma 3. Y equally many times. Consider the graph on the right. It is also 3-regular. u ∈ S .1. Y )-bipartite of even order with |X | = |Y |. Let u. The following theorem. (⇐) Denote e = uv and suppose that G + e has a Hamilton cycle C . Consider the Petersen graph on the right. Lemma 3. and let vi be the vertex that follows ui in C . ⋆ . Now u = v1 − → v2 − → .

dG (u) + dG ( v ) ≥ νG . ⊔ ⊓ Theorem 3. If cl (G ) is hamiltonian. Proof. G1 . since ek ∈ H k . For the initial values. then so are Gk−1 . . . that is. . . if δ(G ) ≥ 1 2 G hamiltonian. . . This procedure stops when no new edges can be added to Gk for some k. Symmetrically. Therefore H ⊆ H ′ .2. Gk of graphs such that G0 = G and Gi +1 = Gi + uv . In particular. Gk = cl (G ) be a construction sequence of the closure of G . Let ek = uv be the ﬁrst edge such that ek = f i for all i . . and thus also dH ′ (u)+ dH ′ ( v ) ≥ n.5. In each step of the construction of cl (G ) there are usually alternatives which edge uv is to be added to the graph. ⊔ ⊓ Theorem 3. in Gk . and hence H ′ = H . say H = G + {e1 . .3. Let G be a graph of order νG ≥ 3. The Claim (ii) follows from (i). The result of this procedure is the closure of G . First. f i }. for all u.4. and it is denoted by cl (G ) (= Gk ) . .2 Hamiltonian graphs v1 ⋆ 33 v2 ◦ ◦ ⋆ vi−1 vi ◦ ◦ vn But now u = v1 − → vi −1 − → vn − → vn−1 − → vi +1 − → vi − → v1 = u is a Hamilton cycle in G . then G is hamiltonian. . Suppose there are two ways to close G . The closure cl (G ) is uniquely deﬁned for all graphs G of order νG ≥ 3. By the choice of ek . Proof. since each complete graph is hamiltonian. and therefore the above procedure is not deterministic. For a graph G . Let H i = G + {e1 . where the edges are added in the given orders. . . so is cl (G ). . G1 . . (ii) If cl (G ) is a complete graph. . . then G is and v. . Lemma 3. e r } and H ′ = G + { f1 . we deduce that H ′ ⊆ H . . . (i) G is hamiltonian if and only if its closure cl (G ) is hamiltonian. . the ﬁnal result cl (G ) is independent of the choices. G ⊆ cl (G ) and G spans cl (G ). which means that e = uv must be in H ′ . deﬁne inductively a sequence G0 . we have G = H0 = H0 . . fs } . . . . Suppose that for all nonadjacent vertices u ν . but ek ∈ / H k−1 . and thus if G is hamiltonian. a contradiction. we have H k−1 ⊆ H ′ . G1 and G0 by Theorem 3. . However. Then dHk−1 (u) + dHk−1 ( v ) ≥ n. Denote n = νG . where u and v are any vertices such that uv ∈ / Gi and dGi (u) + dGi ( v ) ≥ νG . Then G is hamiltonian.6. . v ∈ G either uv ∈ Gk or dGk (u) + dGk ( v ) < νG . . . Let G be a graph of order νG ≥ 3. ei } and H i′ = G + ′ { f1 . In the other direction. ⊔ ⊓ Closure DEFINITION. . let G = G0 .

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

then G is a path or a cycle. One can use the ﬁrst part of the proof to construct a maximum matching in an iterative manner starting from any matching M and from any M -augmented path. then M is a perfect matching. where we are given n applicants and m jobs. A matching M is a maximum matching. . On the right the horizontal edges form a perfect matching. A matching M of G is a maximum matching if and only if there are no M-augmented paths in G. and we should assign each applicant to a job he is qualiﬁed. A matching M saturates v ∈ G . For a graph G .8 (BERGE (1957)). a subset M ⊆ EG is a matching of G . If M saturates VG .3 Matchings In matching problems we are given an availability relation between the elements of a set. if it saturates every v ∈ A. Given a set X of boys and a set Y of girls. . . if v is an end of an edge in M . although you cannot add any edges to M to form a larger matching. The two ends of an edge e ∈ M are matched under M . An odd path P = e1 e2 . if for no matching M ′ . | M | < | M ′ |. Proof. This matching is not maximum because the graph has a matching of three edges. Exercise. ⊔ ⊓ We start with a result that gives a necessary and sufﬁcient condition for a matching to be maximum. and • the ends of P are not saturated. under what condition can each boy marry a girl who cares to marry him? This problem has many variations. The two vertical edges on the right constitute a matching M that is not a maximum matching. The problem is that an applicant may be qualiﬁed for several jobs. M saturates A ⊆ VG . Theorem 3. if M contains no adjacent edges. and a job may be suited for several applicants. DEFINITION.3. It is clear that every perfect matching is maximum. If G is connected with ∆(G ) ≤ 2. A special case of the matching problem is the marriage problem. e2k+1 is M -augmented. Also. One of them is the job assignment problem. which is stated as follows. Lemma 3. DEFINITION. Let M be a matching of G .3 Matchings 35 3. e2i +1 ∈ G − M and e2i ∈ M ). 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.3. Maximum matchings DEFINITION.

then the claim is clear. (⇐) Assume N is a maximum matching. v }. then choose an edge uv ∈ G with u ∈ X . Hence |N | > | M |. . .2) implies the existence of a matching that saturates every proper subset of X . If |S | > |NG (S )| for some S ⊆ X . This A cannot be a cycle. there is a connected component A of H . Hall’s theorem For a subset S ⊆ VG of a graph G . Let then |X | ≥ 2. e1 . which has more edges from N than from M . Indeed. N is a matching of G . ⊔ ⊓ Example 3. because v is an end of at most one edge in M and N . (3. and assume (3. and. . and S ⊆ X . but M is not. By Lemma 3. Hence A : u − → v is a path (of odd length). . Now.3 Matchings 36 Proof. and consider the induced subgraph H = G −{u. k]} . each connected component A of H is either a path or a cycle. Consider the subgraph H = G [ M △ N ] for the symmetric difference M △ N . Now. A is an M -augmented path.9 (HALL (1935)). the matching M = {(0u. and it is clearly perfect. e4 . . k]}) ∪ {e2i +1 | i ∈ [0. has 2k−1 edges. We have dH ( v ) ≤ 2 for each v ∈ H . and |N | = | M | + 1. Consider the k-cube Q k for k ≥ 1. Y )-bipartite graph. (⇒) Let M be a matching that saturates X .3.2). .2) Proof. We prove the claim by induction on |X |. 1u) | u ∈ k−1 }. . Therefore M is not a maximum matching. Since no v ∈ A can be an end of two edges from N or from M . the ends u and v are not saturated by M . . denote NG (S ) = { v | uv ∈ G for some u ∈ S } . which starts and ends with an edge from N . then not all x ∈ S can be matched with different y ∈ NG (S ). Y )-bipartite. Deﬁne N ⊆ EG by N = ( M \ {e2i | i ∈ [1. e3 . If |NG (S )| ≥ |S | + 1 for every nonempty S ⊆ X with S = X . and it thus contains equally many edges from ⋆ N and M . This proves the theorem. Each maximum matching of Q k has 2k−1 edges. (⇒) Let a matching M have an M -augmented path P = e1 e2 . consequently.3. If |X | = 1. . . Because A is a connected component of H . Then G contains a matching M saturating X if and only if |S | ≤ |NG (S )| for all S ⊆ X . because an alternating cycle has even length. If G is (X . known as the Theorem 3. (⇐) Let G satisfy Hall’s condition (3. Let G be a (X . since |N | > | M |. Here e2 . then NG (S ) ⊆ Y . e2k+1 in G . For all S ⊆ X \ {u}. e2 k ∈ M .5. The following result. each connected component (path or a cycle) A alternates between N and M . . e2 k + 1 ∈ / M .

vertices in X \ A are not adjacent to vertices of Y \ B . Now M ∪ {uv } is a matching saturating X in G . let S = [1. by Menger’s theorem. Since |X | = |Y |. k · |NG (S )| = | E2 | ≥ | E1 | = k ·|S |. By the induction hypothesis.3 Matchings 37 |NH (S )| ≥ |NG (S )| − 1 ≥ |S | . Proof.2). and thus that |X \ A| ≤ |NG (X \ A)| ≤ |B | using the condition (3. this matching is necessarily perfect. k · |X | = ǫG = k · |Y |. then G has a perfect matching. it sufﬁces to show that every set S that separates x and y in H has at least |X | vertices. As an example. and hence |X | = |Y |. 2. Let S = A ∪ B . if there were a subset S ⊆ X \ R such that |NH2 (S )| < |S |. 3} to . then we would have |NG (S ∪ R)| = |NH2 (S )| + |NH1 (R)| < |S | + |NG (R)| = |S | + |R| = |S ∪ R| (since S ∩ R = ). Y )-bipartite graph. Indeed. This proof of the direction (⇐) uses Menger’s theorem. S2 . and hence. satisﬁes (3. E1 ⊆ E2 . and hence we have NG (X \ A) ⊆ B . Denote by E1 the set of the edges with an end in S . Now.1 (FROBENIUS (1917)). the induced subgraph H2 = G [VG \ A]. H2 has a matching M2 that saturates X \ R (with the other ends in Y \ NG (R)). . H contains a matching M saturating X \ {u}. If G is a k-regular bipartite graph with k > 0. which contradicts (3. 6}. . By regularity.) A transversal (or a system of distinct representatives) of is a subset T ⊆ S of m distinct elements one from each Si . the set T = {1.2) (since G does). where A ⊆ X and B ⊆ Y . Also. by the induction hypothesis. and by E2 the set of the edges with an end in NG (S ). By Theorem 3. 2}. 3. Clearly. then it is impossible to ﬁnd a transversal for this new family. We conclude that |S | = |A| + |B | ≥ |X |. 4} is a transversal. Combining the matchings for H1 and H2 . . and let S1 = S2 = {1. Let H be the graph obtained from G by adding two new vertices x . 5. . G has a matching that saturates X . S3 . 6]. If we add the set S5 = {2. S2 . Sm } be a family of ﬁnite nonempty subsets of a set S . Suppose then that there exists a nonempty subset R ⊆ X with R = X such that |NG (R)| = |R|. S3 = {2. we get a matching M1 ∪ M2 saturating X in G . There exists a matching saturating X if (and only if) the number ⋆ of independent paths x − → y is equal to |X |. For = {S1 . Let = {S1 . y such that x is adjacent to each v ∈ X and y is adjacent to each v ∈ Y .9. H1 contains a matching M1 that saturates R (with the other ends in NG (R)). and hence.2) for G . and so |NG (S )| ≥ |S |. The induced subgraph H1 = G [R ∪ NG (R)] satisﬁes (3. x X \A A Y \B B y ⊔ ⊓ Corollary 3. Let S ⊆ X . by the induction hypothesis. as was required.3. For this. ⊔ ⊓ Applications of Hall’s theorem DEFINITION. Therefore. . S4 }. Let G be k-regular (X . for A = R ∪ NG (R). ⊔ ⊓ Second proof. 4.2). 3} and S4 = {1. (Si need not be distinct.

3.3 Matchings

38

The connection of transversals to the Marriage Theorem is as follows. Let S = Y and X = [1, m]. Form an (X , Y )-bipartite graph G such that there is an edge (i , s) if and only if s ∈ Si . The possible transversals T of are then obtained from the matchings M saturating X in G by taking the ends in Y of the edges of M . Corollary 3.2. Let be a family of ﬁnite nonempty sets. Then has a transversal if and only if the union of any k of the subsets Si of contains at least k elements. Example 3.6. An m × n latin rectangle is an m × n integer matrix M with entries Mi j ∈ [1, n] such that the entries in the same row and in the same column are different. Moreover, if m = n, then M is a latin square. Note that in a m × n latin rectangle M , we always have that m ≤ n. We show the following: Let M be an m × n latin rectangle (with m < n). Then M can be extended to a latin square by the addition of n − m new rows. The claim follows when we show that M can be extended to an (m + 1) × n latin rectangle. Let Ai ⊆ [1, n] be the set of those elements that do not occur in the i -th column of M . Clearly, |Ai | = n − m for each i , and hence i ∈ I |Ai | = | I |(n − m) for all subsets I ⊆ [1, n]. Now ∪i ∈ I Ai ≥ | I |, since otherwise at least one element from the union would be in more than n − m of the sets Ai with i ∈ I . However, each row has all the n elements, and therefore each i is missing from exactly n − m columns. By Marriage Theorem, the family {A1 , A2 , . . . , An } has a transversal, and this transversal can be added as a new row to M . This proves the claim.

Tutte’s theorem

The next theorem is a classic characterization of perfect matchings. DEFINITION. A connected component of a graph G is said to be odd (even), if it has an odd (even) number of vertices. Denote by codd (G ) the number of odd connected components in G. Denote by m(G ) be the number of edges in a maximum matching of a graph G . Theorem 3.10 (Tutte-Berge Formula). Each maximum matching of a graph G has m(G ) = min

S ⊆ VG

νG + |S | − codd (G −S ) 2

(3.3)

elements. Note that the condition in (ii) includes the case, where S = . Proof. We prove the result for connected graphs. The result then follows for disconnected graphs by adding the formulas for the connected components.

3.3 Matchings We observe ﬁrst that ≤ holds in (3.3), since, for all S ⊆ VG , m(G ) ≤ |S | + m(G −S ) ≤ |S | + |VG \ S | − codd (G −S ) 2 = νG + |S | − codd (G −S ) 2 .

39

Indeed, each odd component of G −S must have at least one unsaturated vertex. The proof proceeds by induction on νG . If νG = 1, then the claim is trivial. Suppose that νG ≥ 2. Assume ﬁrst that there exists a vertex v ∈ G such that v is saturated by all maximum matchings. Then m(G − v ) = m(G ) − 1. For a subset S ′ ⊆ G − v , denote S = S ′ ∪ { v }. By the induction hypothesis, for all S ′ ⊆ G − v , m(G ) − 1 ≥ = 1 2 1 2 (νG − 1) + |S ′ | − codd (G −(S ′ ∪ { v })) (νG + |S | − codd (G −S )) − 1.

The claim follows from this. Suppose then that for each vertex v , there is a maximum matching that does not saturate v . We claim that m(G ) = (νG − 1)/2. Suppose to the contrary, and let M be a maximum matching having two different unsaturated vertices u and v , and choose M so that the distance dG (u, v ) is as small as possible. Now dG (u, v ) ≥ 2, since otherwise uv ∈ G could be added to M , contradicting the maximality of M . Let w be an intermediate vertex on a ⋆ shortest path u − → v . By assumption, there exists a maximum matching N that does not saturate w . We can choose N such that the intersection M ∩ N is maximal. Since dG (u, w ) < dG (u, v ) and dG (w, v ) < dG (u, v ), N saturates both u and v . The (maximum) matchings N and M leave equally many vertices unsaturated, and hence there exists another vertex x = w saturated by M but which is unsaturated by N . Let e = x y ∈ M . If y is also unsaturated by N , then N ∪ {e} is a matching, contradicting maximality of N . It also follows that y = w . Therefore there exists an edge e′ = yz in N , where z = x . But now N ′ = N ∪ {e} \ {e′ } is a maximum matching that does not saturate w . However, N ∩ M ⊂ N ′ ∩ M contradicts the choice of N . Therefore, every maximum matching leaves exactly one vertex unsaturated, i.e., m(G ) = (νG − 1)/2. In this case, for S = , the right hand side of (3.3) gets value (νG − 1)/2, and hence, by the beginning of the proof, this must be the minimum of the right hand side. ⊔ ⊓ For perfect matchings we have the following corollary, since for a perfect matching we have m(G ) = (1/2)νG . Theorem 3.11 (TUTTE (1947)). Let G be a nontrivial graph. The following are equivalent. (i) G has a perfect matching. (ii) For every proper subset S ⊂ VG , codd (G −S ) ≤ |S |. Tutte’s theorem does not provide a good algorithm for constructing a perfect matching, because the theorem requires ‘too many cases’. Its applications are mainly in the proofs

3.3 Matchings

40

of other results that are related to matchings. There is a good algorithm due to EDMONDS (1965), which uses ‘blossom shrinkings’, but this algorithm is somewhat involved. Example 3.7. The simplest connected graph that has no perfect matching is the path P3 . Here removing the middle vertex creates two odd components. The next 3-regular graph (known as the Sylvester graph) does not have a perfect matching, because removing the black vertex results in a graph with three odd connected components. This graph is the smallest regular graph with an odd degree that has no perfect matching. Using Theorem 3.11 we can give a short proof of PETERSEN’s result for 3-regular graphs (1891). Theorem 3.12 (PETERSEN (1891)). If G is a bridgeless 3-regular graph, then it has a perfect matching. Proof. Let S be a proper subset of VG , and let Gi , i ∈ [1, t ], be the odd connected components of G −S . Denote by mi the number of edges with one end in Gi and the other in S . Since G is 3-regular, d G ( v ) = 3 · ν Gi

v ∈ Gi

and

v ∈S

d G ( v ) = 3 · |S | .

**The ﬁrst of these implies that mi =
**

v ∈ Gi

d G ( v ) − 2 · ǫ Gi

is odd. Furthermore, mi = 1, because G has no bridges, and therefore mi ≥ 3. Hence the number of odd connected components of G −S satisﬁes t≤ 1 3

t

mi ≤

i =1

1 3

v ∈S

d G ( v ) = |S | , ⊔ ⊓

and so, by Theorem 3.11, G has a perfect matching.

Stable Marriages

DEFINITION. Consider a bipartite graph G with a bipartition (X , Y ) of the vertex set. In addition, each vertex x ∈ G supplies an order of preferences of the vertices of NG ( x ). We write u < x v , if x prefers v to u. (Here u, v ∈ Y , if x ∈ X , and u, v ∈ X , if x ∈ Y .) A matching M of G is said to be stable, if for each unmatched pair x y ∈ / M (with x ∈ X and y ∈ Y ), it is not the case that x and y prefer each other better than their matched companions: x v ∈ M and y < x v, or u y ∈ M and x < y u.

uz ∈ Mk−1 and x <z u. and P ( x ) = for all x ∈ X . but y < x z for some z ∈ Y . but xz was not added). for some u ∈ X .3. Note ﬁrst that.14. (You should check that there is no stable matching containing the edges 15 and 26. the procedure terminates. ⊔ ⊓ . if x y ∈ Mi and z y ∈ M j for some x = z and i < j . 35. by (3). Assume the that x y ∈ M . First of all. The ﬁnal outcome.) Theorem 3. Set M0 = . Hence x took part in the iteration at an earlier step Mk . Also. Then iterate the following process until all vertices are saturated: Choose a vertex x ∈ X that is unsaturated in Mi −1 . if x < y v for some v ∈ X . then set Mi = Mi −1 ∪ { x y }. 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 41 4: 7 < 5 8: 3 < 2 Then there is no stable matchings of four edges. is a perfect matching.13. Then x y is added to the matching at some step. the matching M = M t is stable. There is a catch. then set Mi = ( Mi −1 \ {z y }) ∪ { x y }. For bipartite graphs G. Let y ∈ Y be the most preferred vertex for x such that y ∈ / P ( x ). Similarly. since a vertex x ∈ X takes part in the iteration at most n times (once for each y ∈ Y ). That was the good news. then y < v z for the vertex z ∈ Y such that vz ∈ M . a stable matching exists for all lists of preferences. Let the bipartition be (X .n be a complete bipartite graph. which leaves 1 and 6 unmatched. A stable matching of G is the following: M = {28. which means that z ∈ P ( x ) at this step (otherwise x would have ‘proposed’ z ).3 Matchings We omit the proof of the next theorem. (3)If z y ∈ Mi −1 and z < y x . Then G has a perfect and stable matching for all lists of preferences.8. Y ). Let G = Kn. (2)If y is not saturated. then x < y z . of course. 47}. k < i (where z was put to the list P ( x ). Theorem 3. AND SHAPLEY (1962) works as Procedure. Proof. A stable matching need not saturate X and Y . Example 3. For instance. say M = M t . The algorithm by GALE follows. (1)Add y to P ( x ). x y ∈ Mi . since the iteration continues until there are no unsaturated vertices x ∈ X . the graph on the right does have a perfect matching (of 4 edges). Thus. and so in M the vertex z is matched to some w with x <z w .

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

. We deﬁne a new graph G0 by adding a vertex v0 to G and connecting v0 to each v ∈ G of odd degree. Proof. and a graph with χ ′ (G ) = 4.1. Hence the ends of the edges ei for i ∈ [2. and hence G0 is eulerian. 2 v0 1 1 1 2 2 ⊔ ⊓ . Then there exists a 2-edge colouring (that need not be proper). Otherwise. The three numbers in Lemma 4. A star. Assume that G is nontrivial. where ei = vi vi +1 (and vt +1 = v1 ).1 Edge colourings 43 Example 4. if i is even . 2. in which both colours are incident with each vertex v with d G ( v ) ≥ 2. let cα ( v ) = |{i | v is incident with i ∈ [1. e t be an Euler tour of G .1 can be equal. then a 2-edge colouring exists as required. α restricted to EG is a colouring of G as required by the claim. Let e0 e1 . But often the inequalities are strict. Now. (2) Suppose then that G is not eulerian. All vertices are among these ends. where ei = vi vi +1 . In G0 every vertex has even degree including v0 (by the handshaking lemma). when G = K1. otherwise. the claim is trivial. k]}| . If G is an even cycle. The condition dG ( v1 ) ≥ 4 guarantees this for v1 . . G has a vertex v1 with dG ( v1 ) ≥ 4. for instance. Let G be a connected graph that is not an odd cycle. if i is odd . DEFINITION. By the previous case. Lemma 4. (1) Suppose ﬁrst that G is eulerian. t − 1] are incident with both colours. Deﬁne α( e i ) = 1. This happens. Hence the claim holds in the eulerian case.n is a star.4. Optimal colourings We show that for bipartite graphs the lower bound is always optimal: χ ′ (G ) = ∆(G ). there is a required colouring α of G0 as above. Let e1 e2 . e t be an eulerian tour of G0 . .2. since now dG ( v ) is even for all v . 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 . For a k-edge colouring α of G . .

α is a proper colouring. Suppose that the colour i is available for v. Suppose on the contrary that χ ′ (G ) > ∆ + 1. then by Lemma 4. ⊔ ⊓ Theorem 4. Let α be an optimal ∆-edge colouring of a bipartite G . if e ∈ H . in which both i and j are incident with each vertex x with dH ( x ) ≥ 2. we have cβ (u) ≥ cα (u) for all u = v . β (e) = α(e). then cα ( v ) = dG ( v ). α is optimal.1 (KÖNIG (1916)). Notice that we always have cα ( v ) ≤ dG ( v ). Also. Furthermore.4. where ∆ = ∆(G ). Theorem 4. if cβ ( v ) > v∈G v∈G 44 cα ( v ) . The following important theorem. Suppose the connected component H is not an odd cycle. (We have renamed the colours 1 and 2 to i and j . Lemma 4. is an odd cycle. and let α be an optimal (∆ + 1)-edge colouring of G . shows that the edge chromatic number of a graph G misses ∆(G ) by at most one colour. Therefore u∈G cβ (u) > u∈G cα (u). An edge colouring α of G is proper if and only if cα ( v ) = dG ( v ) for all vertices v ∈ G. cα ( v ) = dG ( v ).) We obtain a recolouring β of G as follows: γ(e).1 Edge colourings A k-edge colouring β is an improvement of α. if e ∈ / H.4. but it need not have any proper k-edge colourings. Lemma 4. for all vertices v . We need only to show that χ ′ (G ) ≤ ∆ + 1. and the colour j is incident with v at least twice. Proof.2 (VIZING (1964)). But a bipartite graph does not contain such cycles. and if α is proper. j ] that contains v. ⊔ ⊓ Vizing’s theorem In general we can have χ ′ (G ) > ∆(G ) as one of our examples did show. Since dH ( v ) ≥ 2 (by the assumption on the colour j ) and in β both colours i and j are now incident with v . Proof. By Lemma 4. By Lemma 4. if it cannot be improved.3. Then the connected component H of G α [i . For any graph G. If G is bipartite. Proof. Note also that a graph G always has an optimal k-edge colouring. and let v ∈ G. which contradicts the optimality of α. and ∆ = χ ′ (G ) as required. Hence H is an odd cycle.3. If there were a v ∈ G with cα ( v ) < dG ( v ). ∆(G ) ≤ χ ′ (G ) ≤ ∆(G ) + 1.4. cβ ( v ) = cα ( v ) + 1. Thus an improvement of a colouring is a change towards a proper colouring.2. then χ ′ (G ) = ∆(G ). H has a 2-edge colouring γ : EH → {i . due to VIZING. Therefore. G would contain an odd cycle. . Let ∆ = ∆(G ). and in this case α is optimal. j }. The next lemma is obvious. Let α be an optimal k-edge colouring of G. by the construction of β .

cγ (u) ≥ cβ (u) follows as for β . Claim 3. Claim 2. let u1 be as in (4. . Assume we have already found the vertices u1 . Therefore. and thus that cγ (u t ) ≥ cβ (u t ).. u j . contrary to the claim.1) ir −1 u r −1 . . each uℓ gains a new colour iℓ+1 (and may loose the colour iℓ ). cβ ( v ) = cα ( v ) and cβ (u) ≥ cα (u) for all u. . it ut i1 i1 it ut (4. and proves Claim 2. Now. because dG ( v ) is ﬁnite. and obtain in this way an improvement of α. For all other vertices. Suppose.. u2 . β (e) = α(e). say α( vu1 ) = i1 = α( v x ) . . For each u ∈ G. u2 ur . Indeed. ir = i t +1 v u2 i3 u1 i2 i1 x ur . Let β be a recolouring of G such that for 1 ≤ j ≤ r − 1.. and so Claim 1. with j ≥ 1. . Indeed. there exists an available colour b(u) for u. . the fact i r = i t +1 ensures that i r is a new colour incident with u t . . for each uℓ either its number of colours remains the same or it increases by one. i t +1 = i r . 45 Moreover. since each u j (1 ≤ j ≤ r − 1) gains a new colour ji +1 although it may loose one of its old colours. Let t be the smallest index such that for some r < t . ir = i t +1 v i2 u1 x ur . Let then the colouring γ be obtained from β by recolouring the edges vu j by i j +1 for r ≤ j ≤ t . . and hence a colour i1 that is incident with v at least twice. Here v gains a new colour i j +1 .1). vu t is recoloured by i r = i t +1 . j ]. such that the claim holds for these. that v is not incident with b(u j ) = i j +1 . and for all other edges e. . . γ is an optimal (∆ + 1)-edge colouring of G. We can recolour the edges vuℓ by iℓ+1 for ℓ ∈ [1. Such an index t exists. by the counter hypothesis. . Indeed. such that α( vu j ) = i j and i j +1 = b(u j ) . Claim 4.1 Edge colourings We have (trivially) dG (u) < ∆ + 1 < χ ′ (G ) for all u ∈ G . There is a sequence of vertices u1 . Also. α is not a proper colouring. . . . . ir u r −1 ut . β is an optimal (∆ + 1)-edge colouring of G. u2 i3 u1 x u r −1 . This contradicts the optimality of α. β ( vu j ) = i j +1 . . and hence there exists a v ∈ G with cα ( v ) < dG ( v ). ir v i2 i1 ir +1 ir .4.

. 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. However. In graph theoretic terms this means that each colouring of the edges of K6 with 2 colours results in a monochromatic triangle. 3.2 Ramsey Theory In general. the claim follows. 2. for which χ ′ (G ) = ∆(G )+ 1.4. Indeed.2. by Vizing’ theorem. Observe that every vertex is adjacent to all colours 1. − → v5 − → v1 be the outer cycle and C ′ : u1 − → . when the word ‘optimal’ is forgotten: colour ﬁrst the edges as well as you can (if nothing better. In fact. this means that 1 cannot be ′ ′ a colour of any edge in C . there are then 3 persons who are mutual friends or mutual enemies. since P1 ends in u r while P2 ends in a different vertex u t . Now C uses one colour (say 1) once and the other two twice. v4 − → u4 . We show that χ ′ (G ) = 4 for the Petersen graph. A typical example of a Ramsey property is the following: given 6 persons each pair of whom are either friends or enemies. It is clear that every graph contains K1 . Vizing’s theorem (nor its present proof) does not offer any characterization for the graphs. i r ] containing the vertex v are cycles. Edge Colouring Problem. that is. the problem whether χ ′ (G ) is ∆(G ) or ∆(G ) + 1 is NP-complete. This is not possible. the connected components H1 of G β [i0 . then arbitrarily in two colours). This contradiction proves the theorem. However. it is one of the famous open problems of graph theory to ﬁnd such a characterization. χ ′ (G ) = 3 or 4. − → u5 − → u1 the inner cycle of G such that vi ui ∈ EG for all i .. v2 − → u2 . We consider an instance of this theory mainly for edge colourings (that need not be proper)... This can be done uniquely (up to permutations): → v1 .4. the edges of P1 and P2 have the same colours with respect to β and γ (either i0 or i r ). By Lemma 4. ⊔ ⊓ Example 4. there is a colour i0 = b( v ) that is available for v . v3 − → u3 . . v5 − → u5 . Let C : v1 − → . 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. By HOLYER (1981).2 Ramsey Theory 46 By Claim 1. where both P1 : u r −1 − → ur ⋆ and P2 : ur −1 − → u t are paths. The proof of Vizing’s theorem can be used to obtain a proper colouring of G with at most ∆(G ) + 1 colours. i r ] and H2 of G γ [i0 . Suppose 3 colours sufﬁce. Ramsey theory studies unavoidable patterns in combinatorics. ⋆ H1 is a cycle ( vu r −1 ) P1 (u r v ) and H2 is a cycle ( vu r −1 ) P2 (u t v ). Since C needs three colours. 2 3 1 1 1 1 2 3 2 3 4. and that every nondiscrete graph contains K2 . The answer is known (only) for some special classes of graphs. → v5 − → v4 − → v3 − → v2 − v1 − Hence v1 − → u1 .

If a graph G has ǫG > 4 νG edges.2) The next result shows that the above bound T (n. Now ǫG ≤ T (n − p + 1. (Here r is the positive residue of n modulo ( p − 1). we can apply the induction hypothesis to obtain ǫG −A ≤ T (n − p + 1. . 1 2 Corollary 4. Theorem 4. Let n = ( p − 1) t + r for 1 ≤ r ≤ p − 1 and t ≥ 0. . . G2 . K p G −A. p ) . H p−1 of order t (when t > 0). p) edges. Also. When Theorem 4. otherwise G [A ∪ { v }] = K p . then G contains a complete subgraph K p . A complete p-partite graph G consists of p discrete and disjoint induced subgraphs G1 . If a graph G of order n has ǫG > T (n. p) is optimal. 47 Let p ≥ 3. then T (n.1 ( MANTEL (1907)). p − 1] and t ≥ 0. .3 (TURÁN (1941)). and let G be a graph of order n such that ǫG is maximum subject to the condition K p G . Now G contains a complete subgraph G [A] = K p−1 . . If t = 0. We prove the claim by induction on t . One can compute that the number ǫH of edges of H is equal to T ( n. since adding any one edge to G results in a K p . . p) = n(n − 1)/2. . p ) = p−2 2( p − 1) n2 − r 2 1− r p−1 . and νG −A = n − p + 1. Note that a complete p-partite graph is completely determined by its discrete parts Gi . Proof. Gi and G j with i = j .) By its deﬁnition. Each v ∈ / A is adjacent to at most p − 2 vertices of A. where r ∈ [1.p be the complete ( p − 1)-partite graph of order n = t ( p − 1)+ r . . such that there are r parts H1 . Suppose then that t ≥ 1. p) + (n − p + 1)( p − 2) + which proves the claim. and let H = H n. then G contains a triangle K3 . . and there is nothing to prove. we have the following interesting case. ( p − 1)( p − 2) 2 = T ( n. . H r of order t + 1 and p − 1 − r parts H r +1 . p]. and is thus determined by n and p. ⊔ ⊓ . p).3 is applied to triangles K3 .4. . K p H .2 Ramsey Theory DEFINITION. (4. where uv ∈ G if and only if u and v belong to different parts. Because n − p + 1 = ( t − 1)( p − 1) + r . Gp ⊆ G . . and p − 1 vertices of this K p induce a subgraph K p−1 ⊆ G . i ∈ [1.

Theorem 4. q − 1). G [NG ( v )] contains a complete subgraph B of order p − 1 or a stable subset S of order q. q). In the ﬁrst case. q) − 1 vertices different from v . q) such that for all n ≥ R( p. q) = q. q − 1) + R( p − 1. If |A| ≥ R( p.4.6. either |NG ( v )| ≥ R( p − 1. Theorems 4. q ≥ 2 be any integers. if all edges of H have the same colour i . and it is thus exists for p + q ≤ 5. Before proving this. and denote by A = VG \ (NG ( v ) ∪{ v }) the set of vertices that are not adjacent to v . then G [A] contains a complete subgraph of order p or a stable subset S of order q − 1. q − 1) (or both). we give an equivalent statement. Theorem 4. q) or |A| ≥ R( p. In the ﬁrst case. q) exists for all p. q − 1)+ R( p − 1. ⊔ ⊓ .4 (RAMSEY (1930)).6 (ERDÖS and SZEKERES (1935)). Since G has R( p. A subgraph H ⊆ G is said to be (i -) monochromatic. q) is known as the Ramsey number for p and q. q ≥ 2.5. 2) = p and R(2. Proof. By the deﬁnition of Ramsey numbers. It is clear that R( p. It is clear that R( p. Assume ﬁrst that |NG ( v )| ≥ R( p − 1.4 and 4. 2] contains a 1-monochromatic K p or a 2-monochromatic Kq . q) such that for all n ≥ R( p.2 Ramsey Theory 48 Ramsey’s theorem DEFINITION. any 2-edge colouring of Kn → [1. q). The Ramsey number R( p. q − 1) + R( p − 1. It is now sufﬁcient to show that if G is a graph of order R( p. The number R( p. Then there exists a (smallest) integer R( p. q ≥ 2 be any integers. Be patient. q) ≤ R( p. q) exists for p = 2 or q = 2. The following theorem is one of the jewels of combinatorics. We use induction on p + q. and in the second case.5 follow from the next result which shows (inductively) that an upper bound exists for the Ramsey numbers R( p. Then there exists a (smallest) integer R( p. and in the second case the same stable set of order q is good for G . This proves the claim. the same complete subgraph of order p is good for G . then it has a complete subgraph of order p or a stable subset of order q. Let p. Theorem 4. q). this will follow from Theorem 4. q). B ∪ { v } induces a complete subgraph K p in G . Let α be an edge colouring of G . Let p. S ∪ { v } is a stable subset of G of q vertices. and R( p. Let v ∈ G . q). Recall that a subset X ⊆ VG is stable. q) . any graph G of order n contains a complete subgraph of order p or a stable set of order q. if G [X ] is a discrete graph.

Proof. q ≥ 3. R( p.4. any k-edge colouring of Kn has an i-monochromatic Kqi for some i. 5) ≤ 49. . It has been veriﬁed that 43 ≤ R(5. . q2 . Let β : EKn → [1. and let α : EKn → [1. k − 1] be obtained from α by identifying the colours k − 1 and k: β (e) = 1 α( e ) k−1 if α(e) < k − 1 .7 ( ERDÖS and SZEKERES (1935)). R( p . p). Let qi ≥ 2 be integers for i ∈ [1. We use induction on p + q for the general statement. By Theorem 4. q ≥ 2. The case k = 2 is treated in Theorem 4. q − 1) + R( p − 1. .P . . of Combin. For k > 2. .8. . For all p. Then there exists an integer R = R(q1 . The proof is by induction on k. Generalizations Theorem 4. . not so much is known about these numbers. 49 Proof. qk ) such that for all n ≥ R. Small Ramsey numbers. qk−2.2 Ramsey Theory A concrete upper bound is given in the following result. p \q 3 4 5 3 4 5 6 7 8 9 10 6 9 14 18 23 28 36 40-43 9 18 25 35-41 49-61 55-84 69-115 80-149 14 25 43-49 58-87 80-143 95-216 121-316 141-442 The ﬁrst unknown R( p. As can be read from the table1 . Let n = R(q1 .4 can be generalized as follows. . q). Theorem 4. p+q−3 p−1 + p+q−3 p−2 = p+q−2 p−1 . p) (where p = q) is for p = 5. k] be an edge colouring. . q) ≤ which is what we wanted. For p = 2 or q = 2. the claim is clear. . k] with k ≥ 2. we show that R(q1 . Theorem 4. S. if α(e) = k − 1 or k . . Electronic J. qk ). . p). q) ≤ R( p. .6 and the induction hypothesis. ⊔ ⊓ In the table below we give some known values and estimates for the Ramsey numbers R( p. but to determine the exact value is an open problem. RADZISZOWSKI. . . Assume that p. . where p = R(qk−1 . 2000 on the Web .4. qk ) ≤ R(q1 .. q ) ≤ p+q−2 p−1 . qk−2 .

let R k ( G ) = R( G . . . Then there exists an integer R(H1 . Kn) = (m − 1)(n − 1) + 1 . 6 R2 (Cn ) = 2n − 1 if n ≥ 5 and n odd . H k ) and for all α k-edge colourings α of Kn . We leave the following statement as an exercise: If T is a tree of order m.2. H k be arbitrary graphs. H α and thus Kn has a (k − 1)monochromatic or a k-monochromatic subgraph. H2 . even in more general cases. . . for some dear graphs (see RADZISZOWSKI’s survey). Kn contains an i-monochromatic subgraph H i for some i. and this proves the claim. 4ν G − 1 (G connected). R2 (G ) ≥ 3 Here is a list of some special cases: n R2 ( Pn ) = n + − 1. . . . R2 (K1. .8. or the complement of G contains a complete subgraph Kn . 2 if n = 3 or n = 4. . the generalized Ramsey numbers R(H1 .n ) = 2n − 1 2n if n is even. To this end. H k ) can be much smaller than their counter parts (for complete graphs) in Theorem 4. AND The best known lower bound of R2 (G ) for connected graphs was obtained by BURR ERDÖS (1976). In the latter case. Kn has an i -monochromatic Kqi for some 1 ≤ i ≤ k − 2 (and we α are done. where the graphs are equal. . G ) (k times G ). that is. G . Let k ≥ 2 and H1 . . H2 . 3n/2 − 1 if n ≥ 6 and n even. . since this subgraph is monochromatic in Kn ) or Kn has a (k − 1)-monochromatic β α subgraph H = K p .3 ) = 18. .4. Examples of Ramsey numbers∗ Some exact values are known in Corollary 4. Example 4. Below we list some of these results for cases. .3 ) = 10. R2 (K2. However. if n is odd. . H2 . H2 . any graph G of order at least R( T. we have Corollary 4. H ⊆ Km for m = νH . . . then R( T. .4.2. Kn ) contains a subgraph isomorphic to T .8. by Theorem 4. ⊔ ⊓ β Since for each graph H .2 Ramsey Theory β 50 By the induction hypothesis. H k ) such that for all complete graphs Kn with n ≥ R(H1 . R2 (K3. . .3. This generalization is trivial from Theorem 4. .

We observe that in no partition class there are three integers such that x + y = z . The following impressive theorem improves the results we have mentioned in this chapter and it has a difﬁcult proof. For three colours.2 the monochromatic subgraph H i is not required to be induced. . Needless to say that no exact values are known for R k (Kn ) for k ≥ 4 and n ≥ 3. we know that R3 (C4 ) = 11.17 ) is either 65 or 66. For the square C4 . Then R2 (W5 ) = 15 and R2 (W6 ) = 17. we shortly describe Schur’s theorem. the only nontrivial result for complete graphs is: R3 (K3 ) = 17. 4. t such that 1 ≤ i < j < t ≤ S (n) with t − j . 3). Sn of class Si containing two integers x . 6. HAJNAL. Example 4. that in Corollary 4. S ( n) has a Indeed. j .4 that for any complete Kn . As an application of Ramsey’s theorem. any sufﬁciently large complete graph) such that any 2-edge colouring of G has a monochromatic (induced) subgraph Kn . For each n ≥ 1. e. 8. 13]. B. but also. Let Wn denote the wheel on n vertices. SCHUR (1916) solved this problem in a general setting. 10. Let H be any graph.2 Ramsey Theory 51 The values R2 (K2.. However. . and RÖDL (around 1973)). and 385 ≤ R3 (K5 ). if you try to partition 14 into three classes. where 3 occurs n times. however. .) 1990. . The following gives a short proof using Ramsey’s theorem. and let K be a complete on S (n) . and in general. .L. . t − i ∈ Sk for some k. ERDÖS. see R. y such that x + y ∈ Si . R2 (K2. It follows from Theorem 4. . {5. Wiley.4. Theorem 4. PÓSA. Also. . Note. S n of S (n) .n ) ≤ 4n − 2. . Note that W4 = K4 . (2nd ed. 7. 3. SPENCER. ROTHSCHILD AND J. then you are bound to ﬁnd a class. For a partition S1 . there exists an integer S (n) such that any partition S1 .L. It is a cycle Cn−1 . 3. much less is known.n ) are known for n ≤ 16. consider the partition {1. that is. 9} of the set 13 = [1. In fact.9 (DEUBER. K α has a monochromatic triangle. 11. deﬁne an edge colouring α of K by α(i j ) = k. There are quite many interesting corollaries to Ramsey’s theorem in various parts of mathematics including not only graph theory. but no nontrivial upper bound is known for R3 (K5 ). 12}. there are three vertices i . let S (n) = R(3. {2. The value R2 (K2. . “Ramsey Theory”.L. .8. where a vertex v with degree n − 1 is attached. By Theorem 4. . there exists a graph G (well. Then there exists a graph G such that any 2-edge colouring of G has an monochromatic induced subgraph H. . 13}. j − i . 128 ≤ R3 (K4 ) ≤ 235. GRAHAM.4.g. But ( t − j ) + ( j − i ) = t − i proves the claim. For this. where x + y = z has a solution. if |i − j | ∈ Sk . geometry and algebra.

n]. Indeed. k] is proper. we shall concentrate on proper vertex colourings. . is 3-chromatic. then uv ∈ G implies that α(u) = α( v ). It follows that πα is a proper colouring. A graph G is k-colourable. The graph on the right. vn }. if there is a proper k-colouring for G . k]. Let α be a proper k-colouring of G. The chromatic number DEFINITION. the ‘names’ of the colours are immaterial: Lemma 4.3 Vertex colourings The vertices of a graph G can also be classiﬁed using colourings. if α : VG → [1. . . v2 . A colour i ∈ [1. . A k-colouring (or a k-vertex colouring) of a graph G is a mapping α : VG → [1. and let π be any permutation of the colours. These colourings tell that certain vertices have a common property (or that they are similar in some respect). u2 . ⊔ ⊓ Example 4. which is often called a wheel (of order 7). If χ (G ) = k. k] provides a partition {V1 . . V2 . where adjacent vertices get different colours. .5. We show that there are triangle-free graphs with arbitrarily large chromatic numbers. if adjacent vertices obtain a different colour: for all uv ∈ G . and if π : [1. if they share the same colour. Let G t be a new graph obtained by adding n + 1 new vertices v and u1 . . Proof. 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. Then the colouring β = πα is a proper k-colouring of G. k] is said to be available for a vertex v . The (vertex) chromatic number χ (G ) of G is deﬁnes as χ (G ) = min{k | there exists a proper k-colouring of G } . we have α(u) = α( v ). . if it has no subgraphs isomorphic to K3 . Each proper vertex colouring α : VG → [1. where Vi = { v | α( v ) = i }. k] → [1. . . The following construction is due to GRÖTZEL: Let G be any triangle-free graph with VG = { v1 .4. . Example 4. . The colouring α is proper. Again. if no neighbour of v is coloured by i . . and hence also that πα(u) = πα( v ). A graph is triangle-free. a graph G is 2-colourable if and only if it is bipartite. In this chapter. Vk } of the vertex set VG .3 Vertex colourings 52 4.6. k] is a bijection.5. then G is k-chromatic. By the deﬁnitions.

11. Therefore χ (G t ) = k + 1. Connectivity claim follows from this observation. ⊔ ⊓ The case (iii) of the next theorem is due to SZEKERES Theorem 4. a contradiction. so does { vi . then a proper (k − 1)-colouring is obtained for G . let U = {u1 . 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 . since in Kn −(uv ) the vertices u and v can gain the same colour. and hence there is an available colour i for v . vk } induces a triangle. Theorem 4. Let then G be k-critical. and so a triangle contains at most (and thus exactly) one vertex ui ∈ U . vk } by the deﬁnition of G t . Let G be any graph with k = χ (G ). If we recolour v by i . (iii) k ≤ 1 + maxH ⊆G δ(H ). m]} . If G is k-critical for k ≥ 2. This gives a proper (k − 1)-colouring to G . Note that for any graph G with the connected components G1 . there is a proper (k − 1)-colouring of G − v . un }. For the chromatic number we notice ﬁrst that χ (G t ) ≤ (k + 1).4. . . A k-chromatic graph G is said to be k-critical. The graph K2 = P2 is the only 2-critical graph.10. Critical graphs DEFINITION. G2 . . . . Secondly. but the latter triangle is already in G . Each Kn is n-critical. U is stable. extend it by setting α(ui ) = α( vi ) and α( v ) = k + 1. G t is triangle-free and it is k + 1-chromatic Indeed. if χ (H ) < k for all H ⊆ G with H = G. then it is connected. We show ﬁrst that G t is triangle-free. .3 Vertex colourings 53 Claim. but δ(G ) = dG ( v ) ≤ k − 2 for v ∈ G . v j . a contradiction. Recolour each vi by α(ui ). Example 4. Gm . χ (G t ) > k. If α is a proper k-colouring of G . Now using inductively the above construction starting from the triangle-free graph K2 . . Then α(ui ) = k. Now v is adjacent to only δ(G ) < k − 1 vertices. AND WILF (1968). χ (G ) = max{χ (Gi ) | i ∈ [1. and δ(G ) ≥ k − 1. But there are k colours. If {ui . . we obtain larger triangle -free graphs with high chromatic numbers. Assume that α is a proper k-colouring of G t .7. Now. Since G is critical. (ii) G has at least k vertices of degree ≥ k − 1. (i) G has a k-critical subgraph H. since a 3-chromatic G is not bipartite. Proof. 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. say with α( v ) = k. v j .

The claim follows. also dG ( v ) ≥ k − 1 for every v ∈ H . a critical graph does not have cut vertices. Let v be a cut vertex of a connected graph G. let H ⊆ G be k-critical. ⊔ ⊓ Brooks’ theorem For edge colourings we have Vizing’s theorem. . The claim follows from this. . Proof. v ∈ G. clearly. For (iii). Then α is proper. (ii) For all u. because. and let Ai . Then χ (G ) = max{χ (Gi ) | i ∈ [1. χ (G ) − 1 ≤ δ(H ). v } is a separating set. v } is a separating set. but no such strong results are known for vertex colouring. and α( vi ) = min{ j | α( vt ) = j for all t < i with vi vt ∈ G } .7.4. and deﬁne α : VG → inductively as follows: α( v1 ) = 1. Assume then that (iii) holds. By Theorem 4. every kcritical graph H must have at least k vertices. (iii) For all u. vn } be ordered in some way. for i ∈ [1. Suppose each Gi has a proper k-colouring αi . For (ii). Of course. and α( vi ) ≤ dG ( vi ) + 1 for all i . v ∈ G. Let VG = { v1 . Moreover. ⊔ ⊓ Although.10. By Theorem 4.6. the chromatic number χ (G ) usually takes much lower values – as seen in the bipartite case.5.3 Vertex colourings 54 Proof. Then the following are equivalent: (i) G is a complete graph or a cycle. . For all graphs G. Lemma 4. be the connected components of G − v. Lemma 4. It is clear that (i) implies (ii). then {u. then {u.10. . 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. Proof. We need only to show that (iii) implies (i). The next proof of Brook’s theorem is by LOVÁSZ (1975) as modiﬁed by BRYANT (1996). v ) = 2. Denote Gi = G [Ai ∪ { v }]. By Lemma 4. there exists a proper colouring α : VG → [1. . Proof. if uv ∈ / G. the maximum value ∆(G ) + 1 is obtained only in two special cases as was shown by BROOKS in 1941. We use greedy colouring to prove the claim. if dG (u. m]}. In fact. let H ⊆ G be k-critical. We shall show that either G is a complete graph or dG ( v ) = 2 for all v ∈ G . which proves this claim. from which the theorem follows. For (i). χ (G ) ≤ ∆(G ) + 1. we always have χ (G ) ≤ ∆(G ) + 1. ∆(G ) + 1] such that α( v ) ≤ dG ( v ) + 1 for all vertices v ∈ G. m]. These k-colourings give a k-colouring of G . and that (ii) implies (iii). Let G be a 2-connected graph. we may take αi ( v ) = 1 for all i . In particular. ⊔ ⊓ Lemma 4.8. dH ( v ) ≥ k − 1 for every v ∈ H .

(⇐=) Indeed. v2 ∈ G with dG ( v1 . there exists an x ∈ W with x = w such that xu ∈ G or x v ∈ G . and for all i ≥ 3. we ﬁnd at least one j > i such that vi v j ∈ G (possibly v j = w ). and x (or y . then G is complete. and ∆(G ) > ∆(H ). Suppose on the contrary that |W | ≥ 2. and hence that also u. we must have χ (H ) = ∆(H ) + 1. y ) = 2. Order VH = { v3 . such that H = G −{ v1 . v ∈ W1 . and let H ⊂ G be a k-critical proper subgraph. v } ∪ U . Consequently. dH ( vi .3 Vertex colourings 55 First of all. since G is connected. Suppose now that G is neither complete nor a cycle (odd or even). By Lemma 4. and all paths from a vertex of W to a vertex of U go through either u or v . proves the claim. assume the claim holds for k-critical graphs. Indeed. . v ∈ NG (w ) such that uv ∈ / G . . . ∆(C2k+1) = 2. and therefore there exists an edge uv ∈ G with u ∈ H and v ∈ / H . Then χ (G ) = ∆(G ) + 1 if and only if either G is an odd cycle or a complete graph. and thus H is a complete graph or an odd cycle. y } is a separating set. and thus that ∆(G ) = 2 as required. But then dG (u) > dH (u). v2 } is connected. Since w is not a cut vertex (since G has no cut vertices). w ) . (=⇒) Assume that k = χ (G ). χ (C2k+1 ) = 3. . Assume then that there are different vertices u. This means that dG (u. w v2 ∈ G with v1 v2 ∈ / G. v ) = 2 (the shortest path is u − →w− → v ). vn } such that vn = w . But then dG ( v ) > dG (w ). Hence dG ( x .4. (Since uw. Since χ (H ) = k = ∆(G ) + 1 > ∆(H ). Let G be connected. and χ (Kn ) = n.6. w ) ≥ dH ( vi +1 . v4 . for all 1 ≤ i < n. dG (w ) = ∆(G ).8. ∆(Kn ) = n − 1. respectively) would be a cut vertex of G . If G is an even cycle. say xu ∈ G . Now G is connected. This contradiction. { x . Therefore for each i ∈ [1. We may suppose that G is k-critical. dG ( v ) ≥ 2 for all v . then all paths from z to u must pass through x (or y . respectively). v } is a separating set of G . and by (iii). there is a partition VG = W ∪ {u. then k = 2 = ∆(G ). there exists a y ∈ U such that u y ∈ G . By Lemma 4. Thus VG = W1 ∪ { x . since G is 2-connected. Let w be a vertex of maximum degree. it is 2-connected. If the neighbourhood NG (w ) induces a complete subgraph. n − 1]. Note that U1 ⊆ W ∪ U . otherwise. where all paths from W1 to U1 pass through x or y . there exists a vertex u ∈ / NG (w ) ∪{w } such that u is adjacent to a vertex v ∈ NG (w ). Proof.12 ( BROOKS (1941)). there exist v1 . and this contradicts the choice of w . In particular. {u. u w x v y There exists a vertex z ∈ U1 . y } ∪ U1 . Let k = ∆(G ) + 1. Since v is not a cut vertex. respectively). where w ∈ W . y }). and by (iii). since H = Kn or H = Cn . We claim that W = {w }. If z ∈ W (or z ∈ U . Indeed. We show that χ (G ) ≤ ∆(G ). vw ∈ VG −{ x . say v1 w. ⊔ ⊓ Theorem 4. Let then G be any k-critical graph for k ≥ 2. Assume that w ∈ W1 . . v2 ) = 2.

k] for sufﬁciently large natural numbers k. where χG (k) = |{α | α : VG → [1. . χ (G ) = min{k | χG (k) = 0} . Then colour v1 .3 Vertex colourings |NG ( vi ) ∩ { v1 . as we have seen. 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 } . but its applications have turned out to be elsewhere. and so α( vn) ≤ ∆(G ).5 to be certain on this point. if we can ﬁnd the chromatic polynomial of G .) The chromatic polynomial A given graph G has many different proper vertex colourings α : VG → [1. indeed. . of the same colour 1. indeed. vn }. χ (G ) ≤ ∆(G ). E ). In graph theoretical terminology we consider the graph G = (V. w = vn has two neighbours. Indeed. Unfortunately. Also. . This problem requires that we ﬁnd χ (G ). . where vi v j ∈ E just in case vi and v j are incompatible. This notion was introduced by BIRKHOFF (1912).3). .13 will give the tools for constructing χG . some of which are not compatible (like chemicals that react with each other.4. . By (4. the problem whether χ (G ) = 2 has a fast algorithm. k] a proper colouring}| . then we easily compute the chromatic number χ (G ) just by evaluating χG (k) for k = 1. and. no good algorithms are known for determining χ (G ). If k < χ (G ). . Theorem 4. the chromatic number problem is NP-complete. or worse. BIRKHOFF AND LEWIS (1946). . Already the problem if χ (G ) = 3 is NP-complete. vi −1 }| < dG ( vi ) ≤ ∆(G ) . and since vn has at most ∆(G ) neighbours. (However. there is an available colour for vn . 56 (4. consequently. Therefore. . . graph theorists who will ﬁght during a conference).8. 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. and. to attack the famous 4-Colour Theorem. and we would like to colour the vertices of G properly using as few colours as possible. . and. Suppose we have n objects V = { v1 . then clearly χG (k) = 0. n − 1]. ⊔ ⊓ Example 4. . see Lemma 4. . 2. DEFINITION. v1 and v2 . α( vi ) ≤ ∆(G ) for all i ∈ [1.3) The colouring α is proper. This shows that G has a proper ∆(G )-colouring. . v2 . . The chromatic polynomial of G is the function χG : → . until we hit a nonzero value. .

then α corresponds to a unique proper k-colouring of G . Hence G ∗ e is obtained by introducing a new vertex x . Proof. since there G − e and G ∗ e have less edges than G . and so χK4 (k) = k(k − 1)(k − 2)(k − 3) .3 Vertex colourings 57 Example 4. Then χ G ( k ) = χ G − e ( k ) − χ G ∗ e ( k ). ⊔ ⊓ Theorem 4. when EG ∗e = { f | f ∈ EG .9.13. e = uv ∈ G . Indeed. On the other hand. and the vertices u and v are deleted.14. The proof is by induction on ǫG . also P1 − P2 is a polynomial. v2 .4. and let x = x (uv ) be a new contracted vertex. χK n (k) = k n for the discrete graph. Then v3 has k − 2 and ﬁnally v4 has k − 3 available colours. and so u e v x . namely α. (Of course. The considered method for checking the number of possibilities to colour a ‘next vertex’ is exceptional. (2) If α(u) = α( v ). Proof. Therefore χK 4 (k) = k4 . and for two polynomials P1 and P2 . DEFINITION. after which k − 1 colours are available for v2 . namely α. ⊔ ⊓ The connected components of a graph can be coloured independently. v3 . and by replacing all edges wu and w v by w x . and thus any k-colouring is a proper colouring.) Theorem 4. and let e ∈ G. Remark. then α corresponds to a unique proper k-colouring of G ∗ e. and for more nonregular graphs it should be avoided.13. Let G be a graph. in the discrete graph K 4 has no edges. The proper k-colourings α : VG → [1. f has no end u or v } ∪ {w x | wu ∈ G or w v ∈ G } . which together show that χG −e (k) = χG (k) + χG ∗e (k): (1) If α(u) = α( v ). v }) ∪ { x } is obtained from G by contracting the edge e. The claim follows from Theorem 4. Hence the number of such colourings is χG ∗e (k). no loops or parallel edges are allowed in the new graph G ∗ e. Let k ≥ χ (K4 ) = 4. The graph G ∗ e on VG ∗e = (VG \ {u. Therefore there are k(k − 1)(k − 2)(k − 3) different ways to properly colour K4 with k colours. Consider the complete graph K4 on { v1 . when we set α( x ) = α(u) for the contracted vertex x = x (uv ). Let e = uv . Let G be a graph. k] of G −e can be divided into two disjoint cases. The vertex v1 can be ﬁrst given any of the k colours. The chromatic polynomial is a polynomial. Hence the number of such colourings is χG (k). v4 }.

by the induction hypothesis.4. 58 Proof. In our example. f } (k) − χ(G −e)∗ f (k) = k ( k − 1)3 − k ( k − 1)2 = k ( k − 1)2 ( k − 2) . Consider the graph G of order 4 from the above. . the claim is obvious. Let T be a tree of order n. for 3 colours. We use induction on n. which would tell from any polynomial P (k) whether it is a chromatic .13. and thus.9. and let e = vu ∈ T .9. as an exercise). For n ≤ 2. and so χG (k) = χG −e (k) − χG ∗e (k) = k(k − 1)2 (k − 2) − k(k − 1)(k − 2) = k(k − 1)(k − 2)2 = k4 − 5k3 + 8k2 − 4k . we know the chromatic polynomials for trees (and complete graphs. since the reduction method provided by Theorem 4. f } − (G − e) ∗ f Theorem 4. there is known no characterization. Suppose that n ≥ 3. Also.10. Therefore χ T (k) = k(k − 1)n−1. χ T ∗e (k) = k(k − 1)n−2. It is difﬁcult to determine χG of a given graph. Then we have the following reductions.15. Theorem 4. However. . By Theorem 4. Gm .3 Vertex colourings Lemma 4. . Chromatic Polynomial Problems. ⊔ ⊓ Example 4.13 is time consuming.13 reduces the computation of χG to the discrete graphs. Let the graph G have the connected components G1 . we have obtained χG −e (k) = χG −{e. χ T −e (k) = k · k(k − 1)n−2. G2 . and so there is no need to prolong the reductions beyond these. By Lemma 4. Then χ T (k) = k(k − 1)n−1 . The graph T − e consists of the isolated v and a tree of order n − 1. Then χG (k) = χG1 (k)χG2 (k) . . there are 6 proper colourings of the given graph. where v is a leaf. χGm (k) . Here T ∗ e is a tree of order n − 1. χ T (k) = χ T −e (k) − χ T ∗e (k). . For instance. and the induction hypothesis. . e G = G−e − G∗e f = G−e G − {e.

The coefﬁcients alternate in sign. χG ( x ) = 0 for all real numbers 0 < x < 1. but it seems to satisfy the general properties (that are known or conjectured) of these polynomials. REED (1968) conjectured that the coefﬁcients of a chromatic polynomial should ﬁrst increase and then decrease in absolute value. REED (1968) and TUTTE (1974) proved that for each G of order νG = n: • • • • • • • The degree of χG (k) equals n. . The coefﬁcient of k n equals 1. the polynomial k4 − 3k3 + 3k2 is not a chromatic polynomial of any graph. χG (m) ≤ m(m − 1)n − 1 for all positive integers m. The constant term is 0. when G is connected.4. The coefﬁcient of k n−1 equals −ǫG .3 Vertex colourings 59 polynomial of some graph. For instance.

the planar graphs. . Also. 1990. Planar graphs come into use also in some parts of mathematics. A subdivision H of a graph G is obtained from G by a sequence of subdivisions. Most of them are based on an algorithm designed by AUSLANDER AND PARTER (1961) see Section 6.5 Graphs on Surfaces 5. a graph G can be drawn in arbitrarily many different ways. “Implementing Discrete Mathematics: Combinatorics and Graph Theory with Mathematica”. Indeed. Deﬁnition DEFINITION. especially in group theory and topology. the algorithms are all rather difﬁcult to implement.4 is a planar graph. There are. if it has a plane ﬁgure P (G ).5 of S. We restrict ourselves in this chapter to the most natural of these. important families of graphs. SKIENA. that rely on the (topological or geometrical) properties of the drawings of 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. e. The geometry of the plane will be treated intuitively. DEFINITION. in the design of electrical (or similar) circuits. There are fast algorithms (linear time algorithms) for testing whether a graph is planar or not.g. where one tries to (or has to) avoid crossing the wires or laser beams. the surface graphs. However. Addison-Wesley. A planar graph will be a graph that can be drawn in the plane so that no two edges intersect with each other. Such graphs are used. when it is replaced by a path u − →x− →v of length two by introducing a new vertex x .. called the plane embedding of G .1 Planar graphs The plane representations of graphs are by no means unique. but may be apparent from another. the properties of a graph are not necessarily immediate from one representation. however. An edge e = uv ∈ G is subdivided. A graph G is a planar graph.

(vi) An edge that is not a bridge belongs to the boundary of exactly two faces. every point x ∈ F has a disk centred at x and contained in F . This means. Lemma 5.) that are not on the exterior boundary are interior vertices interior edges. in particular. F2 F1 F0 F3 Embeddings P (G ) satisfy some properties that we accepts at face value. Vertices (edges. Then F is a region. if it is bounded. that every planar graph has a plane embedding inside any geometric circle of arbitrarily small radius.1 Planar graphs The following result is clear.). The vertices (edges.2. then so is any drawing P ′ (G ) which is obtained from P (G ) by an injective mapping of the plane that preserves continuous curves. and P (G ) one of its plane embeddings. A face of P (G ) is an interior face. that is.) on the exterior boundary are called exterior vertices exterior edges. Let P (G ) be a plane embedding of a planar graph G. Lemma 5. resp. resp. its interior contains) at least one internal face of P (G ). (i) Two different faces F1 and F2 are disjoint. called the faces of P (G ). Let F be an open set of the plane × . The edges that surround a face F constitute the boundary ∂ ( F ) of F . if any two points x . (iii) Each edge e belongs to the boundary of at most two faces. (iv) Each cycle of G surrounds (that is. (v) A bridge of G belongs to the boundary of only one face. The boundary ∂ ( F ) of a region F consists of those points for which every neighbourhood contains points from F and its complement. The exterior boundary is the boundary of the exterior face. The (unique) face that is unbounded is called the exterior face of P (G ). and it is divided into a ﬁnite number of disjoint regions.5. For instance. the way we draw a graph G in the plane does not change its maximum degree or its chromatic number. (ii) P (G ) has a unique exterior face. y ∈ F can be joined by a continuous curve the points of which are all in F . DEFINITION. A graph is planar if and only if its subdivisions are planar. 61 Geometric properties It is clear that the graph theoretical properties of G are inherited by all of its plane embeddings. More importantly. resp. We recall ﬁrst some elements of the plane geometry. If P (G ) is a plane embedding of a graph G . Let G be a planar graph. resp. Regard now each edge e = uv ∈ G as a line from u to v . .).1. The set ( × ) \ EG is open. and their boundaries can intersect only on edges. or inside any geometric triangle. there are – as we shall see – some nontrivial topological (or geometric) properties that are shared by the plane embeddings.

1 (Euler’s formula). ⊔ ⊓ In particular. since each P (G ) has an exterior face. . By the induction hypothesis.3. then ǫG ≤ 2νG − 4. then. This is clear from Lemma 5. and hence νG − (ǫG − 1)+(ϕ − 1) = 2.1. since the two faces of P (G ) that are separated by e are merged into one face of P (G − e). since νG νG ǫG = i =1 ǫ Gi ≤ 3 i =1 ν G i − 6 k = 3ν G − 6 k ≤ 3ν G − 6 . The subgraph G −e is planar with a plane embedding P (G − e) = P (G )− e obtained by simply erasing the edge e. we have ǫG = νG − 1. then it holds for G . If G is disconnected with connected components Gi . there are no cycles in G . if and only if the connected components of G are trees. Corollary 5. Proof. then ǫG ≤ 3νG − 6. notice that ϕ ≥ 1. The next general form of Euler’s formula was proved by LEGENDRE (1794). if G has no triangles C3 .4. we have the following invariant property of planar graphs. Then every plane embedding of G has the same number of faces: ϕ G = ǫG − ν G + 2 ⊔ ⊓ Maximal planar graphs Lemma 5. It is thus sufﬁcient to prove the claim for connected planar graphs. where ϕ is the number of faces of P (G ). νG −e − ǫG −e +(ϕ − 1) = 2. If G is a planar graph of order νG ≥ 3. and the claim follows.1 Planar graphs 62 Euler’s formula Lemma 5. First. Suppose then that the claim is true for all plane embeddings with less than ϕ faces for ϕ ≥ 2. that is. In this case.5. Let G be a connected planar graph.3. and the claim holds. If ϕ = 1. A plane embedding P (G ) of a planar graph G has no interior faces if and only if G is acyclic. Let P (G ) be a plane embedding of a connected planar graph such that P (G ) has ϕ faces. Now P (G − e) has ϕ − 1 faces. Proof. and let P (G ) be any of its plane embeddings. Let G be a planar graph. Theorem 5. Moreover. for i ∈ [1. by Theorem 2.2. and if the claim holds for these smaller (necessarily planar) graphs Gi . Proof. We shall prove the claim by induction on the number of faces ϕ of a plane embedding P (G ). Then ν G − ǫG + ϕ = 2 . by Lemma 5. and since G is connected. k]. it is a tree. Let e ∈ G be an edge that is not a bridge.4.

then there is nothing to prove. Proof. The result is a plane embedding of a graph H with VH = VG ∪ { x } (that has G as its induced subgraph).2 (HEAWOOD (1890)). the result is a maximal planar graph.4. If νG ≤ 2. each face F of P (G ) contains at least four edges on its boundary (when G is connected and ǫG ≥ 4). if an edge is removed from K3. then δ(G ) ≤ 5.) Add a new vertex x inside F . ⊔ ⊓ By the previous lemma. Theorem 5. a planar graph of order 5 has at most 9 edges. ⊔ ⊓ DEFINITION.1 Planar graphs 63 Also. and the graph remains to be planar. The second claim is proved similarly except that. Hence 3ϕ ≤ 2ǫG . the result is not maximal! Lemma 5. Proof.2. we have |K | < 4 as required. the case where ǫG ≤ 2 is clear.4. Proof.3 are not planar graphs.4. Suppose νG ≥ 3. A planar graph G is maximal.5. . Theorem 5. in this case. Clearly.3 . then G is triangulated.3 has 6 vertices and 9 edges. and since H is planar. However.5. If G is a maximal planar graph with νG ≥ 3. The induced subgraph H [K ∪ { x }] is complete. Theorem 5. but K3. that is. By the handshaking lemma and the previous lemma.1. Let F be a face of a plane embedding P (G ) that has at least four edges on its boundary. ⊔ ⊓ It follows that δ(G ) ≤ 5. every face of a plane embedding P (G ) has a boundary of exactly three edges. if we remove one edge from K5 . if a face has a boundary of at least four edges. For a maximal planar graph G of order νG ≥ 3. since each edge lies on at most two faces. Assume that the set of the boundary vertices of F induces a complete subgraph K . By the second claim of Lemma 5. If G is a planar graph. Hence we have proved Corollary 5. ⊔ ⊓ An upper bound for δ(G ) for planar graphs was achieved by HEAWOOD. δ(G ) · νG ≤ v∈G dG ( v ) = 2ǫG ≤ 6νG − 12 . a triangle-free planar graph of order 6 has at most 8 edges. The edges of K are either on the boundary or they are not inside F (since F is a face. if G + e is nonplanar for every e ∈ / G. Example 5. By Lemma 5. and connect the vertices of K to x . ǫG = 3νG − 6 . Then there are two nonadjacent vertices on the boundary of F . Suppose thus that ǫG ≥ 3. The ﬁrst claim follows from Euler’s formula. K5 and K3. then an edge can be added to the graph (inside the face). Each face F of an embedding P (G ) contains at least three edges on its boundary ∂ ( F ). but K5 has 5 vertices and 10 edges.3.

then G is not planar. if it is a subdivision of K5 or K3. This theorem (due to KURATOWSKI in 1930) is one of the jewels of graph theory.2. We prove this result along the lines of THOMASSEN (1981) using 3-connectivity. AND M. On the other hand. ⊔ ⊓ . QUINTAS. Let E ⊆ EG be the set of the boundary edges of a face F in a plane embedding of G.5 (KURATOWSKI (1930)). 3. Then there exists a plane embedding P (G ). Then use geometric inversion with respect to this circle.M. Lemma 5. if G contains a subdivision of K5 or K3.1. L. In fact. Proof. Hence 3ϕ = 2ǫG . Therefore Theorem 5. 0000 1010 1001 0100 1110 1101 0001 0011 1000 1100 0010 DEFINITION. and therefore they are nonplanar. KENNEDY. The subgraph of Q 4 that is a subdivision of K3.W.3 . 356 – 368.3 as a subgraph. and thus by Theorem 5. The claim follows from Euler’s formula. and also independently by FRINK AND SMITH (1930). see J.3 are the smallest nonplanar graphs. since there are now no bridges. A graph is planar if and only if it contains no subdivision of K5 or K3. 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.5 will give a simple criterion for planarity of graphs. The cube Q k is planar only for k = 1. The graphs K5 and K3.3 . Historia Math.3 as a subgraph. and. the theorem was proven earlier by PONTRYAGIN (1927-1928). SYSLO. Indeed. We prove the converse of this result in what follows. by Lemma 5. Each face F of an embedding P (G ) is a triangle having three edges on its boundary. ⊔ ⊓ Kuratowski’s theorem Theorem 5. each Q k with k ≥ 4 has Q 4 as a subgraph. where the edges of E are exterior edges. 2.1 Planar graphs 64 Proof.3 is given below.6. This will map the given face as the exterior face of the image plane embedding. This is a geometric proof. For history of the result. Example 5.V . the graph Q 4 contains a subdivision of K3. 12 (1985).5. The theorem on planar graphs. A graph G is called a Kuratowski graph.5 it is not planar.

Therefore y is connected to y also in G − T . Assume then that G has a separating set S = {u. (Otherwise {u. By the minimality assumption. and this will contradict the choice of G .8. On the contrary suppose that for any e ∈ G . G −R has a connected component B such that u. and so B ⊆ A. Let H i = Gi + uv . It can be that w ∈ {u. Hence |B | < |A|. where uv is an exterior edge. ⊔ ⊓ Lemma 5. It is now H1 H2 easy to glue H1 and H2 together on the edge uv to obtain a plane embedding of G + uv . since G is 3-connected. Let G be a 3-connected graph of order νG ≥ 5. y. ⊔ ⊓ . Since G is minimal nonplanar with respect to ǫG . v } would separate G .7. by the minimality ⋆ of G . There exists a vertex y ∈ A with z y ∈ G . We can glue these plane embeddings together at v to obtain a plane embedding of G . assume that v is a cut vertex of G . since G2 ⊆ H2 . v }. by Lemma 5. w } separates G .5. and thus G contains a Kuratowski graph. y ′ ∈ A. Let e = uv . Let G1 and G2 be any subgraphs of G such that EG = EG1 ∪ EG2 . there are paths u − → v in G1 and G2 . The inclusion is proper. S = VG1 ∩ VG2 . and so. v ∈ / B .1 Planar graphs 65 Lemma 5. otherwise. that is. For each y ′ ∈ B . This contradiction shows that G is 3-connected. the subgraphs Gi = G [Ai ∪{ v }] have plane embeddings P (Gi ). S would separate G already). Necessarily x ∈ S . but by symmetry we can suppose that w = u. Then there exists an edge e ∈ G such that the contraction G ∗ e is 3-connected. Assume that e and S are chosen such that G − T has a connected component A with the least possible number of vertices. Then G is 3-connected. w }. B A y u T v z Since uv ∈ G . and thus of G . z } separates G . where A1 A2 v is an exterior vertex. Ak be the connected components of G − v . . there exists a vertex w such that R = {z . Indeed. H1 contains a Kuratowski graph H . Now ǫH1 < ǫG . and this contradicts the choice of A. Therefore T = {u. then. G is connected. both u and v are adjacent to a vertex of each connected component of G −S . Since G is 2-connected (by the above). We show then that G is 2-connected. ⋆ there exists a path P : u − → y ′ in G −{z . there is a path u − → v in H2 . We conclude that H1 or H2 is nonplanar. and both G1 and G2 contain a connected ⋆ component of G −S . Let G be a nonplanar graph without Kuratowski graphs of minimal total size ǫG + νG . and let x = x (uv ) be the contracted vertex. However. as in the above. v. . z } (for. and hence. and hence this P goes ′ through y . . they have plane embeddings. This path can be regarded as a subdivision of uv . say H1 . since y ∈ / B . Proof.) If both H1 and H2 are planar. and let A1 . . (Maybe uv ∈ G . the graph G ∗ e has a separating set S with |S | = 2.) The graph G ∗ (z y ) is not 3-connected by assumption.6. say S = { x . v }. On the contrary. Proof.

Now since G −{u. and both u and v have 3 neighbours in the subgraph of G corresponding to H . The proof consists of several cases depending on the Kuratowski graph. clearly. by symmetry.5. Let ⋆ NG ( v ) \ {u} = { v1 . G contains a subdivision of K3. Proof. . Let Pi . where H is a subdivision of K5 . If dH ( x ) = 2. Proof. If for some e ∈ G the contraction G ∗ e has a Kuratowski subgraph. We obtain a plane embedding of G −u by drawing (straight) edges v vi for 1 ≤ i ≤ k. vi +1 ) and by drawing the edges with an end u inside this triangle).3 . Therefore we can assume that νG ≥ 5. z . Let H be a Kuratowski graph of G ∗ e. where x = x (uv ) is the contracted vertex for e = uv .i +1 (i + 1 is taken modulo k) for some i . The proof is by induction on νG . Assume. then one easily sees that G contains a Kuratowski graph. By Lemma 5. Suppose then that dH ( x ) = 3 or 4. Here C is a cycle of G (since G is 3-connected). vi . Every 3-connected graph G without Kuratowski subgraphs is planar. The only 3-connected graph of order 4 is the planar graph K4 . . v }. . then. vk } in order along the cycle C . v2 . (1) If NG (u) \ { v } ∈ Pi . If there exists at most one edge x y ∈ H such that u y ∈ G (or v y ∈ G ). v } = (G ∗ e)− x . there exists an edge e = uv ∈ G such that G ∗ e (with a contracted vertex x ) is 3-connected.8. Now. {u. that dG ( v ) ≤ dG (u). In this case. Let G be a graph. There remains only one case. v j }. P (G ∗ e)− x is a plane embedding of G −{u. y (2) Assume there are y. where y. By Lemma 5. j : vi − → v j be the path along C from vi to v j . G has a plane embedding (obtained from P (G )−u by putting u inside the triangle ( v. then so does G. z ∈ / { vi . then the claim is obviously true. and let C be the boundary of the face of P (G ∗ e)− x containing x (in P (G ∗ e)). vi +1 } ∪ { v. and hence G ∗ e has a plane embedding P (G ∗ e) by the induction hypothesis. G ∗ e has no Kuratowski subgraphs. Consider the part P (G ∗ e)− x .10. y } form a subdivision of K3.9.9. z u v . and NG (u) ⊆ VC ∪ { v } and NG ( v ) ⊆ VC ∪ {u}. . a Kuratowski graph cannot be created by contractions. We do not consider the details of these cases.3 . z ∈ NG (u) \ { v } such that y ∈ Pi j and z ∈ / Pi j for some i and j . and how the subdivision is made.1 Planar graphs By the next lemma. vi . 66 Lemma 5. ⊔ ⊓ v2 x v1 v3 v1 v4 v2 u v v3 v4 Lemma 5.

v2 . Therefore G is planar if and only if ×(G ) = 0. If G is a planar graph. we can assume that NG (u) \ { v } ⊆ NG ( v ). ×(K5 ) = 1.3.5. ⊔ ⊓ 67 u v Proof of Theorem 5. ⊔ ⊓ The proof of the following theorem is partly geometric in nature. For the equality. Now c ≥ 3. We show that ×(K6 ) = 3. This results in a planar graph G on c + 6 vertices and 2c + 15 edges. Therefore. The 4-Colour Conjecture has had a deep inﬂuence on the theory of graphs during the last 150 years.2 Colouring planar graphs The most famous problem in the history of graph theory is that of the chromatic number of planar graphs. v3 give a subdivision of K5 . then χ (G ) ≤ 4. This contradiction proves the claim. Proof. χ (G − v ) ≤ 6. dG ( v ) = dG (u) > 3. Add a new vertex at each crossing. then χ (G ) ≤ 5.5. The problem was known as the 4-Colour Conjecture for more than 120 years.6 (HEAWOOD (1890)). G is 3-connected. The 5-colour theorem We prove HEAWOOD’s result (1890) that each planar graph is properly 5-colourable. Theorem 5. Let X (K6 ) be a drawing of K6 using c crossings so that two edges cross at most once. 5.7. suppose that G is a nonplanar graph that has a minimal size ǫG such that G does not contain a Kuratowski subgraph. The proof is by induction on νG . By Theorem 5. For this we need to show that ×(K6 ) ≥ 3. If G is a planar graph. Any graph G can be drawn in the plane so that three of its edges never intersect at the same point. Since dG ( v ) ≤ 5. then χ (G ) ≤ 6. By Theorem 5. By the induction hypothesis. v.3 and Lemma 5. one is invited to design a drawing with exactly 3 crossings. Lemma 5. and by Lemma 5. since ǫG = 2c + 15 ≤ 3(c + 6) − 6 = 3νG − 6. On the contrary. by Lemma 5. and it requires the assistance of a computer. until it was solved by APPEL AND HAKEN in 1976: if G is a planar graph. NG (u) \ { v } = NG ( v ) \ {u} by the assumption dG ( v ) ≤ dG (u). and so χ (G ) ≤ 6. it is planar. the claim holds for νG ≤ 6.2. we need to show that each nonplanar graph G contains a Kuratowski subgraph. But now u.2 Colouring planar graphs By (1) and (2). v1 .1. and. ⊔ ⊓ Example 5.10. by (1). Clearly. Also. a planar graph G has a vertex v with dG ( v ) ≤ 5.11. The crossing number ×(G ) is the minimum number of intersections of its edges in such plane drawings of G . Then. . The solution of the 4-Colour Theorem is difﬁcult. there is a colour i available for v in the 6-colouring of G − v . for instance.

and therefore for each i ∈ [1. V3 and V4 . ⋆ Let Pi j : vi − → v j be a path in G [i . Hence a country is a . V2 and V4 . E2 ) are bipartite. This is a contradiction. Then the edges of G can be partitioned into two subsets E1 and E2 such that (V. and we wish to colour each country so that no neighbouring countries obtain the same colour. ⊔ ⊓ The ﬁnal word on the chromatic number of planar graphs was proved by APPEL HAKEN in 1976. then χ (G ) ≤ 4. Theorem 5. each planar graph can be decomposed into two bipartite graphs. j ]. V1 and V4 . where vi and v j have the same colour i . By assumption. and then recolour v with the remaining colour j ). j ] (for.8. The edges of this graph represent the boundaries between countries. and let C = ( v v1 ) P13 ( v3 v ). Such a colouring exists. Let E2 be the rest of the edges. since the vertices of P24 are coloured by 2 and 4. By Theorem 5. By the geometric assumption. countries that have only one point in common are not allowed in the map colouring. δ(H ) ≥ k − 1. Suppose the claim does not hold. exactly one of v2 . χ (G ) > 5. Suppose these neighbours vi of v occur in the plane in the geometric order of the ﬁgure. they are between the sets V1 and V3 .2. v4 lies inside the region enclosed by the cycle C . Recall that for k-critical graphs H . otherwise we interchange the colours i and j in the connected component containing v j to obtain a recolouring of G .5. χ (G ) ≤ 4. ⊔ ⊓ AND Map colouring∗ The 4-Colour Conjecture was originally stated for maps. because G is 6-critical. dG ( v ) = 5. since G is planar. E1 ) and (V. How many colours are needed? A border between two countries is assumed to have a positive length – in particular. Theorem 5.7 (4-Colour Theorem). V2 and V3 . The vertices vi and v j are in the same connected component of G [i . The deﬁne E1 as the subset of the edges of G that are between the sets V1 and V2 . E ) be a 4-chromatic graph. Let G = (V. v5 v v4 v1 v2 v3 P13 Consider the subgraph G [i . E2 ) are both bipartite. there exists a neighbour vi ∈ NG ( v ) such that α( vi ) = i . Formally. since the sets Vi are stable. Let α be a proper 5-colouring of G − v . If G is a planar graph. E1 ) and (V. but C contains no such colours. and let G be a 6-critical planar graph. By the following theorem. Let Vi = α−1 (i ) be the set of vertices coloured by i in a proper 4-colouring α of G . Proof. and thus there exists a vertex v with dG ( v ) = δ(G ) ≥ 5. that is. It is clear that (V.2 Colouring planar graphs 68 Proof. we deﬁne a map as a connected planar (embedding of a) graph with no bridges. Now. In the map-colouring problem we are given several countries with common borders. 5]. the path P24 must meet C at some vertex of C . j ] ⊆ G made of colours i and j .

Map-colouring can be used in rather generic topological setting. where he claimed a proof of the 4CC. . . During the following 120 years many outstanding mathematicians tried to solve the problem. Also. Using this notion of a dual graph. The same year ALFRED KEMPE published a paper. consider ﬁnitely many simple closed curves in the plane. 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. . is bipartite. and the edges correspond to the neighbourhood relation. a student of DE MORGAN at University College of London. and it cannot be coloured using only 3 colours. The regions are 2-colourable. TAIT published two papers on the 4CC in the 1880’s that contained clever ideas. otherwise. because a bridge in such a map would be a boundary inside a country. and some of them even thought that they had been successful. As we shall see in the next chapter. . vn } such that vi v j ∈ G if and only if the countries Fi and F j are neighbours. That is. The basic idea in KEMPE’s argument (known later as Kempe chains) was the same as later used by HEAWOOD to prove the 5-Colour Theorem. 2 1 2 2 2 1 2 1 1 2 1 1 History of the 4-Colour Theorem That four colours sufﬁce planar maps was conjectured around 1850 by FRANCIS GUTHRIE. if the region is inside an odd number of curves. For more than 10 years KEMPE’s proof was considered to be valid. In 1890 HEAWOOD showed that KEMPE’s proof had serious gaps. . In 1879 CAYLEY pointed out some difﬁculties that lie in the conjecture. BIRKHOFF introduced one of the basic notions (reducibility) needed . colour it by 2. . 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. The illustrated map can be 4-coloured. but also some further errors. colour a region by 1. F2 . . and.6).2 Colouring planar graphs 69 face of the map. v2 . For instance. he proved that if the number of edges around each region is divisible by 3. then the map is 4-colourable. Let F1 .5. HEAWOOD discovered the number of colours needed for all maps on other surfaces than the plane. To see this. where the maps are deﬁned by curves in the plane. by adding edges to divide the faces into triangles. As an example. Fn be the countries of a map M . because every two faces have a common border. These curves divide the plane into regions. and two neighbouring countries share a common edge (not just a single vertex). and deﬁne a graph G with VG = { v1 . One can triangulate any planar graph G (drawn in the plane). . We deny bridges. (Theorem 5. It is easy to see that G is a planar graph. the graph where the vertices correspond to the regions.

3} x3 {2. The proof used 1200 hours of computer time. The lists for the vertices shown in the ﬁgure show that χℓ (K3. and in 1922 FRANKLIN showed that the 4CC holds for maps with at most 25 regions. x 2 . According to HEESCH one might be satisﬁed with a set of 8900 conﬁgurations to prove the 4CC. m] of G is a (Λ-)list colouring.3 ) > 2.3 is not 2choosable. List colouring DEFINITION.3 be (X .C. From Euler’s formula we see that for the sum of the charges. They based the proof on reducibility using Kempe chains. one can ‘redistribute’ the charges so that no v comes up with a positive charge. Let G be a graph so that each of its vertices v is given a list (set) Λ( v ) of colours. α( v ) ∈ Λ( v ). The following book contains the ideas of the proof of the 4-Colour Theorem. if for a triangulation. and ended up with an unavoidable set with over 1900 conﬁgurations and some 300 discharging rules. Y ). There were difﬁculties with his approach that were solved in 1976 by APPEL AND HAKEN. to 35 by WINN (1940). Indeed. 2} x2 {1. The list chromatic number χℓ (G ) is the smallest integer k such that G has a Λ-list colouring for all lists of size k. {1. The ﬁnal notion for the solution was due to HEESCH. that does not contain a conﬁguration from S . where X = { x 1 .5. A proper colouring α : VG → [1. T. 3} . 2} {1. x 3 } and Y = { y1 . v Now. let the bipartition of K3. The search for avoidable sets began in 1904 with work of WEINICKE. if χℓ (G ) ≤ k. if each vertex v gets a colour from its list. Also. “The Four-Color Problem”. KAINEN. The bipartite graph K3. SANDERS. Dover. who in 1969 introduced discharging. This consists of assigning to a vertex v the charge 6 − dG ( v ). SEYMOUR AND THOMAS (1997) uses 633 conﬁgurations and 32 discharging rules. G is k-choosable. (KOCH assisted with the computer calculations. to 39 by ORE AND STEMPLE (1970).2 Colouring planar graphs 70 in the proof of the 4CC. a conﬁguration is a part that is contained inside a cycle. |Λ( v )| = k}.4. 3} y3 Example 5. Because of these simpliﬁcations also the computer time is much less than in the original proof. This number was increased to 27 by REYNOLDS (1926). we have (6 − dG ( v )) = 12.L. if it is not contained in a triangulation of a minimal counter example to the 4CC. y3 }. 1986. y1 x1 {1. An unavoidable set is a set of conﬁgurations such that any triangulation must contain one of the conﬁgurations in the set.) A simpliﬁed proof by ROBERTSON. SAATY AND P . y2 . a given set S of conﬁgurations can be proved to be unavoidable. to 95 by MAYER (1976). In a triangulation. 3} y2 {2. A conﬁguration is said to be reducible.

This gives a Λ′ -list colouring for G . and that it is given by a neartriangulation. Let NG ( v ) = { x . (If the boundary of a face has more than 3 edges. Proof. there are two colours r. Note that the exterior boundary is unchanged by a triangulation of the interior faces. RUBIN AND TAYLOR (1979) that χℓ (G ) ≤ ∆(G ) + 1 . and otherwise Λ′ (z ) = Λ(z ). . Since Λ′ (z ) ⊆ Λ(z ) for all z . H has a Λ′ -list colouring α. where the interior faces are triangles. it was proved by VIZING (1976) and ERDÖS. We deﬁne new lists for H as follows: Λ′ ( vi ) ⊆ Λ( vi ) \ { r. ⊔ ⊓ v1 .) For νG ≤ 3. we choose α( v ) = r or s such that α( v ) = α( y ). y }. the smallest such a graph was produced by MIRZAKHANI (1996). then we can add an edge inside the face. and it is of order 63. and |Λ( v )| = 5 for all v∈ / C. v1 . Then χℓ (G ) ≤ 5. C : v → x − → v.2 Colouring planar graphs 71 Obviously χ (G ) ≤ χℓ (G ). y }. vk v2 Consider the subgraph H = G − v .) This is because adding edges to a graph can only make the list colouring more difﬁcult. Now νH = νG − 1. Since |Λ( v )| = 3. an embedding. where y ∈ C . However. and by the induction hypothesis (with c still ﬁxed for x ). Let x ∈ C be a vertex. s ∈ Λ( v ) that differ from the colour c of x . It can y x be that NG ( v ) = { x . In fact.. THOMASSEN proved a stronger statement: Theorem 5. . in which case x y ∈ G . vk .10. Let Λ consist of lists such that |Λ( v )| = 3 for all v ∈ C.5. and vi are orv dered such that the faces are triangles as in the ﬁgure.. ⋆ H is the cycle x → v1 → · · · → vk → y − → x . that is. the claim is obvious. k]. but equality does not hold in general. At the moment. (Thus we prove a stronger statement than claimed.9 (THOMASSEN (1994)). s} such that |Λ′ ( vi )| = 3 for each i ∈ [1. Then G has a Λ-list colouring α. Suppose then that νG ≥ 4. since proper colourings are special cases of list colourings. Let G be a planar graph and let C be the cycle that is the boundary of the exterior face. For planar graphs we do not have a ‘4-list colour theorem’. Indeed. it was shown by VOIGT (1993) that there exists a planar graph with χℓ (G ) = 5. We can assume that the planar graph G is connected. The proof is by induction on νG under the additional constraint that one of the vertices of C has a ﬁxed colour. . For the vertex v . Theorem 5. Let v ∈ C be a vertex adjacent ⋆ to x . The exterior boundary of . . we have that α is a Λ-list colouring of G . for which we ﬁx a colour c ∈ Λ( x ). Let G be a planar graph.

The string graph of S is the graph G = (S . we draw a graph by putting an edge between two midpoints of kissing circles. It is known that all planar graphs are string graphs (this is a trivial result). −1. Line Segment Problem.12 (KOEBE (1936)). Theorem 5. 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. all complete graphs are such graphs. For this. the proof of which can be deduced from the above proof of Kuratowski’s theorem. where uv ∈ E if and only if the curves u and v intersect. A positive answer to this 4-slope problem for planar graphs would prove the 4-Colour Theorem. consider a set S of curves of the plane (that are continuous between their end points). The following improvement of the above theorem is due to KOEBE (1936). if they intersect in one point and their interiors do not intersect. 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.2 Colouring planar graphs 72 Straight lines and kissing circles∗ We state an interesting result of WAGNER. · · · · · · · · Theorem 5. A graph is planar if and only if it is a kissing graph of circles. The above question remains open even in the case when the slopes of the lines are +1. A graph is a line segment graph if it is a string graph for a set L of straight line segments in the plane. Indeed. The result is known as Fáry’s Theorem.5. A planar graph G has a plane embedding. 0 and ∞. For a set of circles. This raises a difﬁcult problem: Integer Length Problem. Graphs can be represented as plane ﬁgures in many different ways. and it was rediscovered independently by ANDREEV (1970) and THURSTON (1985). but this is not the case.11 (WAGNER (1936)). +1 −1 0 ∞ . At ﬁrst it might seem that every graph is a string graph. E ). where the edges are straight lines.

In particular. A family of graphs is said to be minor closed. G and G L and L H imply G ∼ = H. A property of graphs is said to be inherited by minors. (2) planar graphs (and their generalizations to other surfaces). WAGNER proved a minor version of Kuratowski’s theorem: . if all minors of a graph G satisfy whenever G does. denoted by H G . . if H is isomorphic to a graph obtained from a subgraph of G by successively contracting edges. . the subgraph order of trees ( T1 ⊆ T2 ) is not a well-quasi-order. it is reﬂexive and transitive. . G2 .2 Colouring planar graphs 73 The Minor Theorem∗ DEFINITION. Indeed. one of the graphs is a (proper) minor of another. in any inﬁnite family of graphs. The conditions (i) and (iii) ensure that the relation is a quasi-order. However. every inﬁnite sequence G1 . A graph H is a minor of G . It turns out to be a well-quasi-order. A recent result of ROBERTSON AND SEYMOUR (1983-2000) on graph minors is (one of) the deepest results of graph theory. Theorem 5.5. G imply H G . and thus it is not minor closed. namely. that is. the family of those graphs that satisfy this property.13 (Minor Theorem). DEFINITION. if every minor H of a graph G ∈ is also in . The following families of graphs are minor closed: the family of (1) all graphs. The acyclic graphs include all trees. of graphs has two graphs Gi and G j with i < j such that Gi G j . More importantly. the proof of Theorem 5. H G . The following properties of the minor relation are easily established: (i) G (ii) H (iii) H G. Each property of graphs deﬁnes a family of graphs. (3) acyclic graphs. that is. e G a subgraph a contraction Note that every subgraph H ⊆ G is a minor. the family of trees is not closed under taking subgraphs. The proof goes beyond these lectures. The minor order is a well-quasi-order on graphs.13 is around 500 pages long.

There are quite many interesting surfaces many of which are rather difﬁcult to draw. However. see the next chapters. This problem is solved by the above results. Consider the space 3 . and it is known to hold for all r ≤ 6. These are surfaces that have both an inside and an outside. which has its (usual) distance function d ( x .3 G. ⊔ ⊓ SEYMOUR (1998) proved the Wagner’s conjecture: Theorem 5. A graph G is nonplanar if and only if K5 Proof.14 (WAGNER (1937)). ROBERTSON AND 74 G or K3. 5.5. if it can be drawn in the plane without crossing edges. The cases for r = 5 and 6 follow from the 4-Colour Theorem. y ) ∈ of its points.15 (Minor Theorem 2). Then there exists a ﬁnite set of graphs such that G satisﬁes if and only if G does not have a minor from . Let be a property of graphs inherited by minors. then G has a complete graph K r as its minor.15 concerns embeddings of graphs on surfaces. since the property ‘knotless’ is inherited by minors: there exists a fast algorithm to do the job. Exercise. A plane is an important special case of a surface. Background on surfaces We shall ﬁrst have a quick look at the general surfaces and their classiﬁcation without going into the details. . Kχ ( G ) G. By Theorem 5. if χ (G ) ≥ r. this algorithm is not known! Hadwiger’s Problem. one can test (with a fast algorithm) whether a graph can be embedded onto a surface. This number is the genus of the surface. We shall study the ‘easy surfaces’ – those that are compact and orientable. 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.15. There are also non-orientable compact surfaces such as the Klein bottle and the projective plane. The conjecture is trivial for r = 2. and can be entirely characterized by the number of holes in them. One of the impressive application of Theorem 5. that is.3 Genus of a graph A graph is planar.3 Genus of a graph Theorem 5. HADWIGER conjectured in 1943 that for every graph G . Every graph can be drawn in the 3-dimensional space without crossing edges. In this section we study shortly drawing graphs in other surfaces.

. Let X be a compact surface. All these deformations should be such that they can be undone.5. Such a permutation induces a direction for the edges of the triangle. and shrinking without tearing it apart or gluing any of its parts together. . and in this case a surface X is compact. (Such a deformation is not allowed by topological equivalence. Equivalently. Then (i) if X is orientable. Theorem 5. orientability can be described as follows.) The next result is known as the classiﬁcation theorem of compact surfaces.18 (DEHN AND HEEGAARD (1907)). (ii) if X is nonorientable. It is often difﬁcult to imagine how a ﬁgure (say. There is a helpful. A compact surface X is orientable if and only if it has no subsets that are topologically equivalent to the Möbius band. 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 | d ist (a. is a triangulation of X if X = ∪m i =1 Ti and any nonempty intersection Ti ∩ T j with i = j is either a vertex or an edge. a graph) can be drawn in a surface. two ﬁgures are topologically equivalent if one can be deformed to the other by bending. . then X is topologically equivalent to a connected sum of projective planes: Pn = P # P # . stretching. Theorem 5.16.e.3 Genus of a graph 75 Two ﬁgures (i. i = 1. where S1 is a torus. A subset of a compact surface X is a triangle if it is topologically equivalent to a triangle in the plane. m. which consists of a polygon in the plane such that . 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. The following is due to RADÓ (1925). . . We deal with surfaces of the real space. stating that every compact surface (orientable or not) has a description by a plane model. In particular.17. # P for some n ≥ 1. Note that the plane is not compact. 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. x ) < 1}. A surface is orientable if it admits an oriented triangulation. 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. Theorem 5.. since it it not bounded. and difﬁcult to prove. squeezing. #S1 for some n ≥ 1. if X is closed and bounded. 2. . Every compact surface has a triangulation. 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. Each triangle of a surface can be oriented by choosing an order for its vertices up to cyclic permutations. A set of points X is a surface. . result due to RADÓ (1920). A ﬁnite set of triangles Ti . where P is a projective plane. then it is topologically equivalent to a sphere S = S0 or a connected sum of tori: Sn = S1 #S1 # . .

) A plane model.5. By studying the relations of the representing words. represented by words W1 and W2 .3 Genus of a graph • each edge of the polygon is labelled by a letter. the Klein bottle by a ba−1 b and the projective plane by a bb−1 a. It is nonoriented if and only if. a compact surface is obtained by gluing together the edges having the same label in the direction that they have. and • each edge is given an orientation (clockwise or counter clockwise). Theorem 5. These surfaces. the torus by a ba−1 b−1 . and thus a compact surface. (This corresponds to the Möbius band. for some label a. 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). and concatenating the labels with the convention that a−1 is chosen if the direction of the edge is counter clockwise. have many other plane models and representing words as well. is obtained by concatenating these words to W1 W2 . can also be represented by a (circular) word by reading the model clockwise. the edges labelled by a have the same direction when read clockwise. a a b Sphere b b a Torus a b b a Klein bottle a b a a b b Projective plane From a plane model one can easily determine if the surface is oriented or not. A word representing a connected sum of two surfaces.18 can be proved. 76 Given a plane model M . . • each letter is a label of exactly two edges of the polygon. the sphere is represented by the word a bb−1 a−1 . Klein bottle Drawing a graph (or any ﬁgure) in a surface can be elaborated compared to drawing in a plane model. as do the other surfaces. Hence.

On the other hand.19. ¡ ¢£ Another way to see this is to use a projection of the sphere to a plane: Theorem 5. and this bounded area can be ironed on the surface of a large enough sphere. if a graph can be embedded in one sphere. then it can be drawn in a bounded area of the plane (without crossing edges). Recall that in the plane model for the Klein bottle the vertical edges of the square have the same direction. If a sphere is deformed by pressing or stretching. According to the next theorem a sphere has exactly the same embeddings as do the plane.5. and stretch it open until it looks like a region of the plane. if G is embeddable in a sphere S0 . and hence G is planar. its embeddability properties will remain the same. We then puncture the sphere at this area. Most importantly. The black dots indicate. Let S0 be (the surface of) a sphere. A graph G has an S0 -embedding if and only if it is planar. where the lines enter and leave the edges of the plane model. On the right we have drawn K6 in the Klein bottle. In topological terms the surface has been distorted by a continuous transformation. where there are no points of the edges. In general. if S is a surface. . This is sometimes convenient. then there is a small area of the sphere. if G can be drawn in S without crossing edges. In the one direction the claim is obvious: if G is a planar graph. e3 b e2 e1 e1 e2 b e3 e4 a e5 Sphere DEFINITION. Clearly.3 Genus of a graph e4 a e5 77 Example 5. a planar graph drawn in a sphere has no exterior face – all faces are bounded (by edges). then a graph G has an S -embedding.5. In this process no crossings of edges can be created. Therefore instead of planar embeddings we can equally well study embeddings of graphs in a sphere. since the sphere is closed and it has no boundaries. then it can be embedded in any sphere – the size of the sphere is of no importance.

6.5. If a graph G is S1 -embeddable. However. Genus Let Sn (n ≥ 0) be a sphere with n holes in it.3 Genus of a graph 78 Torus Consider next a surface which is obtained from the sphere S0 by pressing a hole in it. once again an Sn can be transformed (topologically) into a sphere with n handles. ¡¢ ¡¢ £¤¥ £¤¥ . because we do not put any restrictions on the places of the holes (except that we must not tear the surface into disjoint parts). The drawing of an S4 can already be quite complicated.3 have genus 1. This is a torus S1 (or an orientable surface of genus 1). K7 has genus 1 as can be seen from the plane model (of the torus) on the right. ¡ ¢£ ¡ ¡ ¡ ¢£ ¢£ ¢£ ¡ ¡ ¢£ ¢£ 1 1 5 2 6 3 7 4 1 1 Example 5. Also. The smallest nonplanar graphs K5 and K3. 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. then it can be drawn in any one of the above surfaces without crossing edges. The S1 -embeddable graphs are said to have genus equal to 1.

In general. If G is a planar graph. g (K4 ) = 0. For planar graphs. The same handle can be utilized by several edges. If G is a connected graph. A face of an embedding P (G ) in a surface is a 2-cell.20.3 Genus of a graph 79 DEFINITION.3 ) = 1. We deﬁne the genus g (G ) of a graph G as the smallest integer n. clearly g (G ) ≤ n. . 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). DEFINITION.5. and the above formula is the Euler’s formula for planar graphs. if every simple closed curve (that does not intersect with itself) can be continuously deformed to a single point. g (K3. we have g (K5 ) = 1. only that G can be embedded in some Sn . ¡¢ £¤¥ We should note that the above argument does not determine g (G ). then ν G − ǫG + ϕ G = 2 − 2 g ( G ) . For K5 . Also. in particular. for which G is Sn -embeddable. we have g (G ) = 0. However. The external face of G becomes an ‘ordinary face’ after G has been drawn in S0 . and. a face of an embedding of G in Sn (with g (G ) = n) is a region of Sn surrounded by edges of G . Theorem 5. Theorem 5. The next theorem states that any graph G can be embedded in some surface Sn with n ≥ 0. Every graph has a genus. since K5 is nonplanar. and thus the genus g (G ) of G exists. Let again ϕG denote the number of faces of an embedding of G in Sn . Indeed. We omit the proof of the next generalization of Euler’s formula. Euler’s formula with genus∗ ¡¢ £¤¥ The drawing of a planar graph G in a sphere has the advantage that the faces of the embedding are not divided into internal and external. then g (G ) = 0.21. 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. but is embeddable in a torus. This result has an easy intuitive veriﬁcation.

If n ≥ 3.24 (HEAWOOD (1890)). By the previous lemma. every nonbridge is on the boundary of exactly two faces. Combining these we obtain that 3ǫG − 3νG + 6 − 6 g (G ) ≤ 2ǫG . let G = K8 . then g (Kn ) ≥ (n − 3)(n − 4) 12 . n(n − 1). We deduce that K8 cannot be embedded in the surface S1 of the torus. ϕG = ǫG − νG + 2 − 2 g (G ). In this case νG = 8. and the claim follows. ⊔ ⊓ Theorem 5. The number of edges in Kn is equal to ǫG = 1 2 g (Kn ) ≥ (1/6)ǫG − (1/2)(n − 2) = (1/12)(n − 3)(n − 4) . For a connected G with νG ≥ 3 we have 3ϕG ≤ 2ǫG . embeddings. For the complete graphs Kn a good lower bound was found early. and. ǫG = 28. g (G ) ≤ g (Kn ). as before.22 (YOUNGS (1963)).5. g (G ) ≥ 2. then clearly g (H ) ≤ g (G ). and the genus of the torus is 1. In this case we need the knowledge that ϕG is counted in a surface that determines the genus of G (and in no surface with a larger genus).23. Now every face has a border of at least three edges. since H is obtained from G by omitting vertices and edges. Assume thus that νG ≥ 4. Proof. 3ϕG ≤ 2ǫG . If H ⊆ G .23. For a graph G of order n. For a connected G with νG ≥ 3.13. If n ≥ 3. Lemma 5. we can compute lower bounds for the genus g (G ) without drawing any .12. then the claim is trivial. ⊔ ⊓ This result was dramatically improved to obtain Theorem 5. We omit the proof of the general condition discovered by YOUNGS: Theorem 5. If νG = 3. and so g (G ) ≥ 5 3 Since the genus is always an integer. 6 2 1 Proof. ⊔ ⊓ By this theorem. As an example. The faces of an embedding of a connected graph G in a surface of genus g (G ) are 2-cells. g (G ) ≥ 1 ǫ G − (ν G − 2) . we obtain Proof. and by the generalized Euler’s formula. But this is because g (K4 ) = 0. In particular. then g (Kn ) = (n − 3)(n − 4) 12 . Theorem 5. Lemma 5. By Theorem 5.3 Genus of a graph 80 The complete graph K4 can be embedded in a torus such that it has a face that is not a 2-cell. .25 (RINGEL AND YOUNGS (1968)).

25.28 (HEAWOOD).26. Theorem 5. HEAWOOD proved it ‘only’ for g ≥ 1. the proof of the 4-Colour Theorem. If g (G ) = g ≥ 1. Notice that for g = 0 this theorem would be the 4-colour theorem. we know the exact genus for the complete bipartite graphs: Theorem 5. For the complete bipartite graphs. g (Km. then g (G ) = 1. Theorem 5. If a nonplanar graph G can be embedded in a torus. Such a drawing is called spatial embedding of the graph. and χ (G ) ≤ ⌊(7 + 1 + 48 g )/2⌋ = 7. By Theorem 5. Using the result of RINGEL AND YOUNGS and some elementary computations we can prove that the above theorem is the best possible. 81 Also. g (G ) ≤ (n − 3)(n − 4) 12 . Chromatic numbers∗ For the planar graphs G . it is surprising that the generalization of the 4-Colour Theorem for genus ≥ 1 is much easier.n ) = (m − 2)(n − 2) 4 . χ (G ) ≤ 4.5. Indeed. such an embedding can be achieved by putting all vertices of G on a line. g (K7 ) = 1. is extremely long and difﬁcult. Moreover. but g (K8 ) = 2. HEAWOOD proved a hundred years ago: Theorem 5.3 Genus of a graph Therefore g (K6 ) = ⌈3 · 2/12⌉ = ⌈1/2⌉ = 1.29.27 ( RINGEL (1965)). Also. and then drawing the edges in different planes that . For all graphs G of order n ≥ 3. for G = K7 we have that χ (K7 ) = 7 and g (K7 ) = 1. This in mind. there exists a graph G with genus g (G ) = g so that χ (G ) = 7+ 1 + 48 g 2 . For each g ≥ 1. Three dimensions∗ Every graph can be drawn without crossing edges in the 3-dimensional space. then χ (G ) ≤ 7+ 1 + 48 g 2 .

A spatial embedding of a graph G is said to have linked cycles. which is curious. and drawing the edges as straight lines crossing the sphere inside. This family of forbidden graphs was originally found by SACHS (without proof).3 Genus of a graph 82 contain the line. It was shown by ROBERTSON.5. if two cycles of G form a link (they cannot be separated in the space). Every graph in the set has 15 edges. the vertices of G can be put in a sphere. THOMASSEN. For further results and proofs concerning graphs in surfaces. . SEYMOUR AND THOMAS (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. Johns Hopkins. and it contains K6 and the Petersen graph. see B. 2001. By CONWAY and GORDON in 1983 every spatial embedding of K6 contains linked cycles. Alternatively. “Graphs on Surfaces”. MOHAR AND C.

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

where the longest directed paths have lengths χ (G ) − 1. A digraph D has a directed path of length χ ( D)− 1. Then I dD ( v ) = | D| = v∈D v∈D O dD ( v ).1 Digraphs DEFINITION. since D−A does not contain directed cycles. Indeed. This shows that α is a proper colouring of U ( D). (You offer and accept a handshake. Directed paths The relationship between paths and directed paths is in general rather complicated. This digraph has a path of length ﬁve. k ≥ χ ( D) − 1. if a longest directed path from v has length i − 1 in D−A. Consider then an edge e = vu ∈ A. then α(u) = α( v ). then ⋆ α(u) = α( v ).1 (ROY (1967). Proof.GALLAI (1968)). if e = uv ∈ D−A. if α( v ) = i . ⋆ First we observe that if P = e1 e2 . ( D−A) + e contains a directed ⋆ ⋆ cycle C : u − →v− → u. assign a colour α( v ) = i . . Let D be a digraph. ⊔ ⊓ . . ⋆ and PQ is a directed path. Let k = χ (G ) and let α be a proper k-colouring of G . Theorem 6. Proof. α(u) = i = α( v ). the so obtained orientation D has no directed paths of length ≥ k − 1. The indegree and the outdegree of a vertex are deﬁned as follows I dD ( v ) = |{e ∈ D | e = x v }|. Here 1 ≤ i ≤ k + 1. Let A ⊆ E D be a minimal set of edges such that the subdigraph D−A contains no directed cycles. k].6.1. ⊔ ⊓ The bound χ ( D) − 1 is the best possible in the following sense: Theorem 6. that is. 84 We have the following handshaking lemma. but its directed paths are of length one. then there exists a directed path Q : v − → w of length i − 1.2. In particular. We orient each edge uv ∈ G by setting uv ∈ D. As usual the set of colours is [1. Let k be the length of the longest directed path in D−A. There is a nice connection between the lengths of directed paths and the chromatic number χ ( D) = χ (U ( D)). and therefore χ ( D) ≤ k + 1. Clearly. For each vertex v ∈ D. O dD ( v ) = |{e ∈ D | e = v x }| . e r ( r ≥ 1) is any directed path u − → v in D−A. Since PQ : u − → w. Every graph G has an orientation D. By the minimality of A. and hence α(u) = α( v ).) Lemma 6. if α(u) < α( v ). where the part u − → v is a directed path in D−A.

On the other hand.2). ⋆ ⋆ (6. if it has no directed cycles. it cannot have both ways u − → v and v − → u. and the proof is completed. The acyclic orientations D that extend in both ways are exactly those that contain neither u − → v nor v − → u as a directed path. We conclude that a(G ) = a(G −e) + b. Now χG (k) is a polynomial that satisﬁes the recurrence χG (k) = χG −e (k) − χG ∗e (k). we choose e1 . a(G ) = (−1)n χG −e (−1) + (−1)n−1 χG ∗e (−1) = (−1)n χG (−1) . where b is the number of acyclic orientations D of G − e that extend in both ways e1 and e2 .2) Each acyclic orientation of G ∗ e corresponds in a natural way to an acyclic orientation D of G − e that satisﬁes (6. but the result may not be di-connected. Proof. by the induction hypothesis. For (6.3 (STANLEY (1973)). we observe that every acyclic orientation of G gives an acyclic orientation of G − e. for otherwise someone is not able to drive home. Note that since D is ⋆ ⋆ acyclic. if D is an acyclic orientation of G − e for e = uv . ⊔ ⊓ One-way trafﬁc Every graph can be oriented. In the one-way trafﬁc problem the resulting orientation should be di-connected. The proof is by induction on ǫG . if a(G ) = a(G −e ) + a(G ∗ e ) then. it extends to an acyclic ⋆ orientation of G by putting e1 : u → v or e2 : v → u. Let a(G ) be the number of acyclic orientations of G .6. ROBBINS’ theorem solves this problem. and a(G ) = 1 = (−1)n (−1)n = (−1)n χG (−1) as required. .1 Digraphs 85 DEFINITION. if there is a di-connected oriented graph D such that G = U ( D). DEFINITION. if G is discrete. A graph G is di-orientable. The next result is charming. An orientation D of an undirected graph G is acyclic. Therefore b = a(G ∗ e). and if D has no directed path v − → u. ⋆ we choose e2 . First. we show that a(G ) satisﬁes the same recurrence. To prove the claim. then χG (k) = k n . where χG is the chromatic polynomial of G. Let G be a graph of order n. since χG (−1) measures the number of proper colourings of G using −1 colours! Theorem 6.1) (6. Indeed. Indeed. Then the number of the acyclic orientations of G is a(G ) = (−1)n χG (−1) . if D has no directed path u − → v.1).

1 Digraphs 86 Theorem 6. and ai is followed by a j exactly ri j times. Consider a multidigraph D with VD = A for which there are ri j edges ai a j . then the word a1 a2 a1 a1 a2 is a solution to the problem. for directed graphs. Suppose then that G has no bridges. j ≤ n. and a cycle is always diorientable. It is rather obvious that a directed Euler trail of D gives a solution to the sequencing problem. Let D be a digraph. The following two results are left as exercises. . Then D has a directed Euler trail if and I O only if d D ( v ) = dD ( v ) for all vertices v with possibly excepting two vertices x . A directed Euler trail of D is a directed walk that uses each edge exactly once. a2 . s1 = 3. C . This proves the claim. then we are done. A connected graph G is di-orientable if and only if G has no bridges. G . where w is the last vertex inside H . the letters are A. Example 6. The edge e is not a bridge and thus there exists a cycle P P′ Q C ′ = ePQ : v − →u− →w− →v in G . Proof. Hence G has a cycle C . A directed Euler tour of D is a directed closed walk that uses each edge exactly once. (1) Let D be a digraph such that U ( D) is connected.2.1. T. Otherwise. there exists an edge e = vu ∈ G such that u ∈ H u v e but v ∈ / H (because G is connected). that is. then any orientation of G has at least two di-components (both sides of the bridge). In a problem instance. ⊔ ⊓ Example 6. In the di-orientation DH of H there is a directed path P ′ : u − → w . . and r11 = 1. r12 = 2. r21 = 1. In conclusion. The above results hold equally well for multidigraphs. y for which I O | d D ( v ) − d D ( v )| = 1. Now. Then D has a directed Euler tour if and I O only if d D ( v ) = dD ( v ) for all vertices v. r22 = 0. we orient e : v − →u ⋆ ⋆ ′ and the edges of Q in the direction Q : w − → v to obtain a directed cycle eP Q : v − →u− → ⋆ w− → v . and the question is: does there exist a word w in which each letter ai occurs exactly si times. . Let then H ⊆ G be maximal such that it has a di-orientation DH . In the case of DNA. where we allow parallel directed edges between the vertices. Consider words over an alphabet A = {a1 .6. ⋆ ⋆ ⋆ w . that is. The following problem was ﬁrst studied by HUTCHINSON AND WILF (1975) with a motivation from DNA sequencing. we are given nonnegative integers si and ri j for 1 ≤ i . For instance. which contradicts the maximality assumption on H . each word w is a sequence of letters. . if n = 2.4 (ROBBINS (1939)). If G has a bridge e. (2) Let D be a digraph such that U ( D) is connected. G [VH ∪ VC ] has a di-orientation. If H = G . an } of n letters.

Every tournament has a directed Hamilton path. Consider a tournament of n teams that play once against each other.3. Here the new digraph has the edge vu instead of uv .1 Digraphs 87 Tournaments DEFINITION. A tournament T is an orientation of a complete graph. Example 6. if v comes out with the most victories in the tournament. a tournament T of order n has a directed path of length n − 1. which contradicts the maximality assumption made for v .5 (RÉDEI (1934)). Example 6. there is a directed path v − → u of length at most two.4. DEFINITION. and thus d + 1 vertices in { y | x y ∈ T } = A ∪ { v }.6. and let d T ( v ) = d be the maximum outdegree in T . and hence by Theorem 6. where the vertices correspond to the teams vi . It follows that x v ∈ T and xu ∈ T for all u with vu ∈ T . Suppose that there exists an x . ⊔ ⊓ Problem. Proof.) Theorem 6. and there is an edge vi v j . Theorem 6. Let v be a vertex of a tournament T of maximum outdegree. It follows that the outdegree of x is d + 1. (Isomorphism of directed graphs is deﬁned in the obvious way. 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. There are four tournaments of four vertices that are not isomorphic with each other.6 states that a winner v either defeated a team u or v defeated a team that defeated u. Theorem 6. A team v is a winner (there may be more than one winner). if vi won v j in their mutual game. Let T be an orientation of Kn . ⋆ Then for all u. O Proof. This is then a directed Hamilton path visiting each vertex once.6 (LAUDAU (1953)). . The chromatic number of Kn is χ (Kn ) = n. But there are d vertices in A = { y | v y ∈ T }. ⊔ ⊓ The vertices of a tournament can be easily reached from one vertex (sometimes called the king). and suppose that each game has a winner. for which the directed distance from v to x is at least three. The situation can be presented as a tournament.1.

. and let T be the scoring digraph as in the ﬁgure. s1 (3) = 3. s2 (2) = 5. It can be proved (using matrix methods) that for a di-connected tournament with at least four teams. − → vin . 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 (m + k)th-level scorings for all k ≥ 1. then the level scoring should be carried out with respect to the di-components. O Let s1 ( j ) = d T ( j ) be the winning number of the team j (the number of teams beaten by j ).but 4 and 6 have the same score. . Consider a tournament of six teams 1 2 1. s1 (5) = 2. we have s2 (1) = 8. . . a tournament may have several directed Hamilton paths. s2 (5) = 4. s2 (6) = 3 . s1 (4) = 2. s2 (3) = 9. where it is a winner. s1 (1) = 4. Here 1 − → 2 − → 4 − → 5 − → 6 − → 3 is a directed 3 6 Hamilton path.1 Digraphs 88 A ranking of a tournament is a linear ordering of the teams vi1 > vi2 > · · · > vin that should reﬂect the scoring of the teams. it is a looser. s1 (6) = 1 . is team 1 the winner? If so.5. it seems that 3 is the winner. We continue by deﬁning inductively the mth-level scoring by sm ( j ) = ji ∈ T sm−1 (i ) . s1 (2) = 3. One way of ranking a tournament could be by a Hamilton path: the ordering can be obtained from a directed Hamilton path P : vi1 − → vi2 − → .6. 2. So. but this extends to a directed Hamilton cycle (by adding 3 − → 1)! So for every team there is a 4 5 Hamilton path. ji ∈ T This tells us how good teams j beat. In our example the level scoring gives 1 − →3− →2− →5− →4− → 6 as the ranking of the tournament. However. 6... Example 6. Now. If T is not di-connected. In our example. and some of these may do unjust for the ‘real’ winner. is 2 or 3 next? Deﬁne the second-level scoring for each team by s2 ( j ) = s1 ( i ) . . s2 (4) = 3. and in another. In the above tournament.

6. Harris in the 1950s. and • a capacity function α : V × V → + (nonnegative real numbers). f + (u) = v∈N f (uv ) and f − (u) = v∈N f ( vu) . for which α(e) = 0. 4 If N = ( D.A] f (e) . 4 5 s 4 4 3 3 0 4 1 r .A] f (e) and f − (A) = e∈[A. A network N consists of • an underlying digraph D = (V. The previous network has a ﬂow that is indicated on the right. E ). r. Denote VN = V and EN = E .E. The second condition states that there should be no loss. 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). • two distinct vertices s and r . Flows DEFINITION. v ∈ / A} . In modelling a transportation network by a digraph. The connection to Kirchhoff’s Current Law (1847) is immediate. then the value α(e) gives the capacity ( x m3 /min) of the pipe e. According to this law. and + such that f − ( v ) = f + ( v ) for all v ∈ / {s . we must make sure that the number of goods remains the same at each crossing of the roads. called the source and the sink of N . if 5 s 6 4 4 5 4 2 5 4 2 r [A. We adopt the following notations: any function such that f (e) = 0. The problem setting for such networks was proposed by T. Example 6. A ﬂow in a network N is a function f : VN × VN → 0 ≤ f (e) ≤ α(e) for all e.6.2 Network Flows Various transportation networks or water pipelines are conveniently represented by weighted directed graphs. Goods are transported from speciﬁc places (warehouses) to ﬁnal locations (marketing places) through a network of roads. in every electrical network the amount of current ﬂowing in a vertex equals the amount ﬂowing out that vertex. A] = {e ∈ D | e = uv. if e ∈ / E. r } . α) is a network of water pipes. In particular. u ∈ A. s.2 Network Flows 89 6. DEFINITION. These networks usually possess also some additional requirements. Let A ⊆ VN be a set of vertices. and f : VN × VN → e∈ / N . f + (A) = e∈[A.

2.. 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. en be an undirected path in N where an edge ei is along P . if ι ( P ) > 0. Let N be a network. Let A ⊆ N \ {s. Let N = ( D. 4 DEFINITION. r. if ei = vi +1 vi ∈ N . Every network N has a zero ﬂow deﬁned by f (e) = 0 for all e. then val( fA) = 0. and therefore this path is improvable. . and against P . v ∈ A cancel each out. then it has no improvable paths. α) be a network with a ﬂow f . Let f be a ﬂow in a network N .6. (i) If A ⊆ N \ {s.3. r }. . Then 0= v ∈A ( f + ( v ) − f − ( v )) = v ∈A f +(v) − v ∈A f − ( v ) = f + (A) − f − (A) = val( fA). because the junctions cannot handle this much water.g. f (e) if e is against P . ⊔ ⊓ Improvable ﬂows Let f be a ﬂow in a network N . s. if ei = vi vi +1 ∈ N . If f is a maximum ﬂow of N . in the above ﬁgure the source should not try to feed the network the full capacity (11 m3 /min) of its pipes. Lemma 6. Lemma 6. and let P = e1 e2 . A path ⋆ P: s − → r is ( f -)improvable.2 Network Flows 90 A ﬂow f in N is something that the network can handle. A ﬂow f of a network N is a maximum ﬂow. DEFINITION. Proof. The second claim is also clear. E. For a ﬂow f and each subset A ⊆ VN . if there does not exist any ﬂow f ′ such that val( f ) < val( f ′ ). (ii) val( f ) = −val( f r ). In the above example. We deﬁne a nonnegative number ι ( P ) for P as follows: ι ( P ) = min ι (e). deﬁne the resultant ﬂow from A and the value of f as the numbers val( fA) = f + (A) − f − (A) and val( f ) = val( fs ) (= f + (s) − f − (s)) . s 5 3 4 4 3 4 0 1 r 4 On the right. the bold path has value ι ( P ) = 1. . r }. val( f ) = 9. ei where ι (e) = α(e) − f (e) if e is along P . where the third equality holds since the values of the edges uv with u.

⊔ ⊓ Theorem 6.2 Network Flows Proof. r }. equality holds if and only if for each u ∈ S and v ∈ / S. and the capacities of the edges are not exceeded. For a ﬂow f and a cut [S ] of N .6. since at each intermediate vertex v ∈ / ′ − ′ + {s. since P has exactly one edge sv ∈ N for the source s. S ] (= {uv ∈ N | u ∈ S . s. Hence val( fS ) =val( fs ) − v ∈S I f (sv ) + v ∈S I f ( vs) f ( vs) v ∈S I + val( fSI ) + v ∈S I f (sv ) − =val( fs ) = val( f ) . 5 4 5 2 4 6 4 2 5 4 r Example 6. For a ﬂow f and any cut [S ] of N . Furthermore. Now val( f ′ ) = val( f )+ι ( P ). . ⊔ ⊓ Max-Flow Min-Cut Theorem DEFINITION. Lemma 6. Let N = ( D. if ι ( P ) > 0. Hence. and val( f ) = val( fs ). In our original network the capacity of the cut for the indicated vertices is equal to 10. ′ f (e) = f (e) − ι ( P ) if e is against P .7. Proof. s val( f ) = val( fS ) = f + (S ) − f − (S ) . f (e) if e is not in P . we have ( f ) ( v ) = ( f ) ( v ). Now val(S I ) = 0 (since S I ⊆ N \ {s. 5 s 5 4 3 91 4 4 2 2 0 4 r Then f ′ is a ﬂow. r. [S ] = [S . Deﬁne f (e) + ι ( P ) if e is along P . A cut [S ] is a minimum cut. then we can improve the ﬂow. v ∈ The capacity of the cut [S ] is the sum α [S ] = α + (S ) = e ∈[S ] α( e ) . Let S I = S \ {s}. For a subset S ⊂ VN with s ∈ S and r ∈ / S . let the cut by S be / S }) .7. val( f ) ≤ α[S ].4. α) be a network. if there is no cut [R] with α[R] < α[S ]. r }).

and let S I = {u ∈ N | for some path P : s − → u. By Lemma 6.3. (ii) if e = vu ∈ N . there exists a path ⋆ ⋆ P: s − → u with ι ( P ) > 0. STEIN. FEINSHANNON. v ∈ / S. A ﬂow f of a network N is maximum if and only if there are no f -improvable paths in N . Moreover. a maximum ﬂow cannot have improvable paths. Proof. ι ( Pe) = 0 for the path Pe : s − → v . since v ∈ / S . assume that N contains no f -improvable paths. ι ( P ) > 0} . for an edge e = vu ∈ N with v ∈ / S and u ∈ S . ⊔ ⊓ Theorem 6. then f is a maximum ﬂow and [S ] a minimum cut. If f is a ﬂow and [S ] a cut such that val( f ) = α[S ]. The following main result of network ﬂows was proved independently by ELIAS. then val( f ) ≤ α[S ] .1. This proves the claim. Since u ∈ S . where u ∈ S and v ∈ / S . Conversely. ⊔ ⊓ Theorem 6. by FORD AND FULKERSON. By the deﬁnition of a ﬂow. By the same argument. then f (e) = 0.4. and so f (e) = α(e). Corollary 6. Proof.9. Therefore ι (e) = 0. By Theorem 6. the equality val( f ) = α[S ] holds if and only if (1) f + (S ) = α[S ] and (2) f − (S ) = 0. Also. Let N be a network.6. The present approach is due to Ford and Fulkerson. f (e) = 0. if f is a maximum ﬂow and [S ] a minimum cut. Corollary 6. we have val( f ) = α[S ]. and f − (S ) ≥ 0. Set S = S I ∪ {s}. In particular. This holds if and only if f (e) = α(e) for all e ∈ [S ] (since f (e) ≤ α(e)). val( f ) = val( fS ) = f + (S ) − f − (S ). as required.2 Network Flows (i) if e = uv ∈ N .8.1 implies now that f is a maximum ﬂow (and [S ] is a minimum cut). By Lemma 6.7. then f (e) = α(e). f + (S ) = e ∈[S ] 92 f (e) ≤ e ∈[S ] α ( e ) = α [S ] . and (2) f (e) = 0 for all e = vu with u ∈ S . and by ROBACKER in 1955 – 56. has integer ⋆ . where the capacity function α : V × V → values. and hence val( f ) ≤ α[S ]. Consider an edge e = uv ∈ N . Then N has a maximum ﬂow with integer values.

For this. Y ). and consider a network N with vertices {s. ⊔ ⊓ Corollary 6. if x y ∈ G for x ∈ X .9. y r ∈ N (α( y r ) = 1) for all x ∈ X . ⊔ ⊓ Corollary 6.6.8 showed also Theorem 6.10 gives Theorem 3.5. Theorem 3. where the capacities of the edges are equal to one (α(e) = 1 for all e ∈ N ). Now Theorem 6. Next we apply the theorem to unit networks. ⊔ ⊓ The next corollary is Menger’s Theorem for edge connectivity. The claim follows then easily from Corollary 6.2 Network Flows 93 Proof. the value val( f ) of a maximum ﬂow equals the capacity α[S ] of a minimum cut. (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 proof of Theorem 6. whose removal destroys ⋆ all the directed connections u − → v from D. Let the edges (with their capacities) be s x ∈ N (α(s x ) = 1).10. Let f0 be the zero ﬂow. We mention a connection to the Marriage Theorem. Applications to graphs⋆ The Max-Flow Min-Cut Theorem is a strong result. Exercise. f0 (e) = 0 for all e ∈ V × V . y ∈ Y .9.5 and Corollary 6.3. The maximum number of edge⋆ disjoint directed paths u − → v equals the minimum number of edges.3 by increasing and decreasing the values of the edges by integers only. Proof. whose removal destroys all the connections ⋆ u− → v from G. Lemma 6. y ∈ Y together with the edges x y ∈ N (α( x y ) = |X | + 1). Then G has a matching that saturates X if and only if N has a maximum ﬂow of value |X |. 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. and many of our previous results follow from it. We obtain results for (directed) graphs. In a network N . A network N with source s and sink r is obtained by setting the capacities equal to 1. let G be a bipartite graph with a bipartition (X .2. Let N be a unit network with source s and sink r.2. Proof.10 (Max-Flow Min-Cut). A maximum ﬂow is constructed using Lemma 6. The maximum number of edge-disjoint ⋆ paths u − → v equals the minimum number of edges. Let u and v be two vertices of a digraph D. Let u and v be two vertices of a graph G. (i) The value val( f ) of a maximum ﬂow equals the maximum number of edge-disjoint directed ⋆ paths s − → r. Proof. The claim follows from Lemma 6. r } ∪ X ∪ Y .

Proof. A k-ﬂow (H . k − 1] such that for all vertices v ∈ V . ⊔ ⊓ Seymour’s 6-ﬂows∗ DEFINITION. For convenience. The Petersen graph has a nowhere zero 5-ﬂow but does not have any nowhere 4-ﬂows. e∈[A. A k-ﬂow is nowhere zero. let α(e−1 ) = −α(e) for all e ∈ H in the orientation H of G so that the condition (6. (6.11. and indeed. Tutte’s conjecture resembles the 4-Colour Theorem. In particular. 1 3 2 2 The condition (6.6.8. Lemma 6. If G has a nowhere zero k-ﬂow for some k.4) Example 6. α( e ) = 0 . the sum of the incoming values equals the sum of the outgoing values. In the k-ﬂows we do not have any source or sink.3. if α(e) = 0 for all e ∈ H . the conjecture is known to hold for the planar graphs. In order to fully appreciate Seymour’s result. and so 5 is the best one can think of. e= vu∈ H 1 2 1 (6. It was conjectured by TUTTE (1954) that every bridgeless graph has a nowhere zero 5-ﬂow. A graph G is k-edge connected if and only if any two distinct vertices of G are connected by at least k independent paths.6.3) becomes α( e ) = 0 . α( e ) = e= vu∈ H f =uv ∈ H α( f ) . SEYMOUR’s remarkable result reads as follows: .3) that is.A] (6. Theorem 6. we mention that it was proved as late as 1976 (by JAEGER) that every bridgeless G has a nowhere zero k-ﬂow for some integer k. A graph with a nowhere zero 4-ﬂow. The claim follows immediately from Corollary 6. Tutte’s Problem. then G has no bridges. α) of an undirected graph G is an orientation H of G together with an edge colouring α : EH → [0.2 Network Flows 94 Corollary 6.5) since the values of the edges inside A cancel out each other.4. The proof of this uses the 4-Colour Theorem.4) generalizes to the subsets A ⊆ VG in a natural way.

we have that f (G ) ≤ 3. Let D be the orientation of G corresponding to the direction of W . A connected graph G has a ﬂow number f (G ) = 2 if and only if it is eulerian. Since W arrives and leaves each vertex equally many times. and set α(e) = 2. y3 ∈ Y such that x y2 . We have f (K4 ) = 4.6. If an edge uv ∈ D. the function α is a nowhere zero 2-ﬂow. Proof. Then necessarily the degrees of the vertices are even. By Corollary 3. Suppose G is eulerian. we have f (G ) ≤ 3. a 3-regular graph has a perfect matching M .12. ⊔ ⊓ . if n is even . Proof.13. and if n > 4. x y3 ∈ / M . Orient the edges e ∈ M from X to Y .9. ⊔ ⊓ Example 6. let α be a nowhere zero 2-ﬂow of an orientation D 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. For each 3-regular bipartite graph G . and so G is eulerian. Since each x ∈ X has exactly one neighbour y1 ∈ Y such that x y1 ∈ M .1. Theorem 6. and set α(e) = 1. Theorem 6. 95 ⊔ ⊓ DEFINITION. let G be (X . Conversely. Orient the edges e ∈ / M from Y to X . let α(e) = 1.11 (SEYMOUR’s (1981)).2 Network Flows Theorem 6. then f (Kn ) = Proof. Y )-bipartite. Indeed. Exercise. Omitted. and two neighbours y2 . Every bridgeless graph has a nowhere zero 6-ﬂow. 2 3 if n is odd . and consider an Euler tour W of G .

22 adjacency matrix. 67 cube.edge. 61 bridge. 13 disconnecting set. boundary. 18 capacity. 52 complement G . 53 crossing number. 57 critical. 56 closed walk. 61 fan. 52 chromatic polynomial χG . 13 distance function. 5 edge. 89 ﬂow number. trail. 9 Hamilton path . 12 distance. 85 addressable. 10 disjoint walks. 3 2-sets. 24 boundary. 29 even component. 62 eulerian. 61 face. tour. 5. 70 chromatic number χ (G ). 42 edge colouring. 16 complete graph Kn . 89 choosable. 23 cycle. 79 graph. 29 Euler’s formula. 12 exterior: face. 31 . 75 connectivity number κ(G ). 18 genus. 16 bond. 24 contracted vertex. 4 Euler trail . 83 di-orientable. 12 end (of an edge). 77 end (of a path). 10 connected. 24 connected (component). 28 ﬂow. 12 colouring.Index N P -complete problems. 83 directed Euler tour. cycle. 6 adjacent. 42. 4 graphical sequence. 38 even cycle. 91 cut vertex. 5 disconnected. 85 digraph. 18 digraph.vertex. 10 complete bipartite Km. 13 connected sum. 4 edge chromatic number χ ′ (G ). 95 forest. 52 bipartite. path. 42 edge cut. 35 available colour. 4 acyclic. cycle. 24 embedding. 83 directed graphs (digraph).k . 4 augmented path. 27 discrete graph. 7 di-connected di-component. 86 walk. 11 cut (in a network). 91 capacity function. 12 degree dG ( v ).

70 list colouring. 78 tournament. 87 trail. 44 incident colours. 12 induced subdigraph. 75 oriented graph. 70 loop. 7 network. 35 maximal planar graph. 32 planar graph. 42 indegree. 10 97 . 52 Ramsey number. 72 Kuratowski graph. 4 neighbourhood NG ( v ). 83 inverse walk (path). 82 list chromatic number. 71 neighbour. 4 Möbius band. 90 matching. 4 near-triangulation. 35 Petersen graph. 11. 37 tree. 73 monochromatic. 16 subdigraph. 75 improvable (path). 22 homeomorphic. 7 weighted distance. 38 leaf. 90 improvement (colouring). 52 triangulation. 83 induced subgraph. 89 size ǫG . 38 odd cycle. 72 linked cycles. 3 path. 75 torus. 68 matching. 64 latin rectangle. 4 spanning subgraph. 90 same parity. 75 orientation. 8 interior: face. 4 partition. 13 minor. 61 intersection graph. 75 symmetric difference. 84 independent paths. 60 subgraph. edge. 77 sphere with a handle. 83 subdivision. 7 line segment graph. 12 optimal colouring. 6 inverse pair. 7 isomorphic. 48 ranking. 10 resultant ﬂow. 75 triangle-free. 84 parallel edges. 18 triangle. 20 spatial embedding. 42. 10 odd component. 12 perfect matching. 7 ﬂow. 75 map. 20 sink and source. 35 separates. 83 outdegree. 8 spanning tree. 60 plane model. 88 regular graph. 3 topologically equivalent. 48 multigraph. 81 sphere. 23 separating set. 63 maximum degree ∆(G ). 35 minimum cut. 38 latin square. 75 proper colouring. 31 Hamming distance. 4 orientable surface. 3 saturate (matching).Index hamiltonian. 44 order νG . 91 degree δ(G ). vertex. 89 nontrivial graph. 83 oriented. 75 trivial graph. 87 kiss (circles). 78 stable matching. 29 transversal. 12 isolated vertex. 60 plane embedding. 5 join of walks. 40 stable set. 12 king. 23 Shannon’s switching game. 8 surface.

11 weight. 89 underlying graph. 4 vertex colouring. 12 2-cell. 13 weight function. 90 98 . 79 2-switch. 9 underlying digraph. 5 walk. 83 unit networks. 5 wheel. 93 vertex. 51 winning number. 88 zero ﬂow.Index trivial path.

- Solution Manual Introduction to the Theory of Computation SipserUploaded byVishwaraj Anand
- 9780324782011_PPT_ch12Uploaded byBobby Stanley
- Graph theoryUploaded bygirish
- Schaum's: Graph TheoryUploaded byLyndsae Vine
- Introduction to Graph Theory Second Edition Solution Manual (by Douglas B. West)Uploaded bySuchan Khankluay
- Graph Theory & Algorithms - M. Ashraf IqbalUploaded byAbdul Rehman Naeem
- Intro to Graph TheoryUploaded byapi-19523062
- EE304Quiz01Uploaded byJoyprakash Lairenlakpam
- Graph Theory Douglas B West 2 EdUploaded bysumi_11
- Graph TheoryUploaded byIlyes Altair
- graph theory 1Uploaded bysmusala7970
- Introduction to Graph Theory -- D. B. WestUploaded byazfarhussain8776
- MatchingUploaded byBotnaru Nicoleta
- Jana Comstock- A finite presentation of knotted trivalent graphsUploaded byGmonicon
- Numerical AnalysisUploaded byAmirAli Mojtahed
- Static to Dynamic Electric Power NetworkUploaded byA.K.A. Haji
- Lec4Uploaded bycu4kiss
- Matching2016 B Halls TheoremUploaded byAniket More
- 12_DS and Algorithm_session_17.ppsUploaded byVikas Singh
- Math assignmentUploaded byitsjessicatime
- lessons 1-3 translations reflections orientationsUploaded byapi-264837349
- Art and Geometry.pdfUploaded byIon C. Andronache
- 42DirectedGraphs-2x2Uploaded byTiến Phương Nguyễn
- Analysing the configuration of integrated multi-modal urban networksUploaded byjalopesgil
- Writing Thesis ReportUploaded bykhairus sadiq
- ispUploaded byKait Reinhardt
- BruteForceExhaustiveSearchUploaded byPuneet Mehta
- GRAPH-HOC 040401Uploaded byMegan Bell
- Data InterpretationUploaded bynskumar_19986252
- graphOPEN MP.pdfUploaded byDaengTatto

- Computer_Networks_OSL Lab Manual.pdfUploaded byAjith Suresh
- Main SolutionUploaded byAjith Suresh
- Advanced algorithms at a glanceUploaded byAjith Suresh
- Advanced algorithms at a glanceUploaded byAjith Suresh
- DESIGN AND ANALYSIS OF ALGORITHMS LABORATORY 10CSL47Uploaded byPradyot SN
- Half(L)ExampleUploaded byAjith Suresh
- Fellowship Hike FAQUploaded byAjith Suresh
- Cryptography Note S8 NoteUploaded byAjith Suresh
- sol2Uploaded byAjith Suresh
- CA LP New_01Uploaded byAjith Suresh
- Graph Theory SD8 Class NoteUploaded byAjith Suresh
- ADUni - Theory of Computation - Problem Set 01 SolutionsUploaded byAjith Suresh
- ShutterburgUploaded byAjith Suresh
- ps02solUploaded byAjith Suresh
- CloudMoV Cloud-Based Mobile Social TVUploaded byAjith Suresh
- DbtUploaded byAjith Suresh
- s7_btech_1Uploaded byAkhil Rathu
- Neural_Computing___An_Introduction.pdfUploaded byAjith Suresh
- bib_gitUploaded byAjith Suresh
- bib_gitUploaded byAjith Suresh
- Html5 for BeginnersUploaded byRamesh Dobbala
- (Now Update!!!) Nokia e6 p.a.s (Rm-609 111.140.0058) by Zahidaina_link in PostUploaded byAjith Suresh
- RAMAYANAUploaded bySudhir Srinivasan
- Single pass assembler in c language.pdfUploaded byAjith Suresh
- ENGINEERING LAB MANUALS_ CS2308 -SYSTEM SOFTWARE LM.pdfUploaded byAjith Suresh
- Top-Down Parsing.pdfUploaded byAjith Suresh
- Html5 for Beginners Learn by VideoUploaded byAjith Suresh
- Safe in Flight Mode Technology.pdfUploaded byAjith Suresh

- Swarm Course OutlineUploaded byzaig
- The Prime Glossary_ Sieve of Eratosthenes.pdfUploaded bymiles1280
- Spring 2010 FinalTerm OPKST CS502 Solved 11082010Uploaded byatali
- Doc-186158Uploaded bypoonam_goyal26
- Networks in Transportation TheoryUploaded bysandi nurmansyah
- Upcasting and Down CastingUploaded byatwalsunny
- Harmanis - DevelopmentOfCSUploaded byjosegalavizc
- m7l17Uploaded bypravin
- Polymorhpism AdditionalUploaded byErik Stodola
- Test prepation for algorithm.docxUploaded byM. Yasir zia aulakh
- (Foundations of Computing) Joseph A. Goguen, Grant Malcolm-Algebraic Semantics of Imperative Programs-The MIT Press (1996).pdfUploaded byFabio Arevalo
- Predmicate TuteUploaded byNavodya Nuwanthi
- 14 Other Schduling ModelsUploaded bydugguz
- 3.Finite AutomataUploaded bytejshahi1
- randomized algorithmsUploaded byHarman Patel
- Advanced PascalUploaded byMichalis Irakleous
- 08 r05310505 Principles of Programming LangaugesUploaded byandhracolleges
- Data Structures and Algorithms in C++-Uploaded byDhruvin Bhatt
- Peakdet_ Peak Detection Using MATLAB (Non-Derivative Local Extremum, Maximum, Minimum)Uploaded byrnjlmv83
- LR GrammarsUploaded bypseudoworkoholic
- Course Handout DSUploaded byvaibhav0206
- 4.Design Analysis of Algorithm QBUploaded byVigneshwar Suresh
- Automatic Type Promotion in Expressions in javaUploaded byarunverma47
- Network Routing Problem in MultigraphUploaded byIRJCS-INTERNATIONAL RESEARCH JOURNAL OF COMPUTER SCIENCE
- Soal ujian MITUploaded byMuhammad Al Kahfi
- AlgorithmsUploaded bydeveshag
- Summary Primitive Types and ExpressionsUploaded bysaracatunga
- Optimization HandoutUploaded byKaseh Dhaah
- WikiBooks - FSharpUploaded byperson
- QKD RSAUploaded bydemonones