(Structural) Graph Theory
Adam Hesterberg
Based on Paul Seymour’s notes and work
Blue MOP 2011, June 28th
1 Definitions
1.1 Definitions you must know
Graph (= multigraph), simple graph, vertex, edge, adjacent, loop, parallel edge, clique/complete graph,
degree, subgraph, connected, component (= connected component), tree, forest, spanning tree, path, cy-
cle, Eulerian cycle, Hamiltonian cycle, bipartite graph, stable set (= independent set), matching, perfect
matching, planar graph, 𝑘-regular graph, digraph.
1.2 Other useful definitions
Definition 1.1. A cut-vertex of a connected graph is a vertex whose deletion disconnects the graph. A
cut-edge or cut-set (of vertices or edges, usually the former) is similar.
Definition 1.2. A graph 𝐺 is 𝑘-connected iff ∣𝑉 (𝐺)∣ ≥ 𝑘 + 1 and for every 𝑋 ⊂ 𝑉 (𝐺) with ∣𝑋∣ < 𝑘, 𝐺 ∖ 𝑋
is connected.
Definition 1.3. The line graph 𝐿(𝐺) of a graph 𝐺 is the graph with 𝑉 (𝐿(𝐺)) = 𝐸(𝐺), with an edge for
for every pair of incidences of two edges of 𝐺 on the same vertex of 𝐺.
Definition 1.4. A graph 𝐺 is 𝑘-edge-connected iff its line graph is 𝑘-connected. Alternately, 𝐺 is 𝑘-edge-
connected iff for every 𝑋 ⊂ 𝐸(𝐺) with ∣𝑋∣ < 𝑘, 𝐺 ∖ 𝑋 is connected.
Definition 1.5. A separation of 𝐺 is a pair (𝐴, 𝐵) of subsets of 𝑉 (𝐺) with 𝐴 ∪ 𝐵 = 𝑉 (𝐺), such that there
is no edge between 𝐴 ∖ 𝐵 and 𝐵 ∖ 𝐴. Its order is ∣𝐴 ∩ 𝐵∣.
2 Useful Theorems
Theorem 2.1. (Erdős) If 𝐺 is a graph with no stable set of size 𝑡, then there’s a graph 𝐻 with 𝑉 (𝐺) = 𝑉 (𝐻)
and at most 𝑡 − 1 components, each of which is a complete graph, such that ∀𝑣, deg𝐻 (𝑣) ≤ deg𝐺 (𝑣).
Theorem 2.2. (Menger’s Theorem) Let 𝑄, 𝑅 ⊂ 𝑉 (𝐺), and let 𝑘 ≥ 0. Then there are 𝑘 pairwise vertex-
disjoint paths from 𝑄 to 𝑅 unlesss there’s a separation (𝐴, 𝐵) of 𝐺 of order < 𝐾 with 𝑄 ⊂ 𝐴 and 𝑅 ⊂ 𝐵.
Theorem 2.3. (Tutte’s Theorem) Let 𝑜𝑑𝑑(𝑋) be the number of components of 𝑋 with an odd number of
vertices. Then 𝐺 has a perfect matching unlesss there exists 𝑋 ⊂ 𝑉 (𝐺) with 𝑜𝑑𝑑(𝐺 ∖ 𝑋) > ∣𝑋∣.
1
3 Matchings
1. (König’s Theorem) Let 𝐺 be bipartite, and 𝑘 ≥ 0 an integer. Then 𝐺 has a matching of size at least
𝑘 unlesss there exists 𝑋 ⊂ 𝑉 (𝐺) with ∣𝑋∣ < 𝑘 such that 𝑋 meets every edge of 𝐺.
2. Let 𝐺 be a loopless graph in which every vertex has positive degree. Let 𝑋 be the largest matching in
𝐺, and let 𝑌 be the smallest set of edges of 𝐺 whose union contains 𝑉 (𝐺). Show that ∣𝑋∣+∣𝑌 ∣ = ∣𝑉 (𝐺).
3. Show that every 2-edge-connected cubic graph has a perfect matching.
4. Let 𝐺 be a 𝑑-regular bipartite graph. Show that 𝐸(𝐺) can be partitioned into perfect matchings.
4 Minors
Definition 4.1. If 𝑒 ∈ 𝐸(𝐺), then 𝐺/𝑒 (“𝐺 contract 𝑒”) is the graph formed by deleting 𝑒 and identifying
its endpoints.
Definition 4.2. A graph 𝐻 is a minor of a graph 𝐺 iff it’s obtainable from a subgraph of 𝐺 by contracting
edges.
That is, to get a minor, one first deletes vertices and edges, then contracts edges. Note that contraction
and deletion commute, so one can do so any order.
Theorem 4.3 (Wagner’s Theorem). A graph 𝐺 is planar unlesss it has a 𝐾5 or 𝐾3,3 minor.
Theorem 4.4 (Kuratowski’s Theorem). A graph 𝐺 is planar unlesss it has a subdivision of 𝐾5 or 𝐾3,3
(that is, with edges turned into paths) as a subgraph.
Theorem 4.5 (Seymour). If 𝐺 is an infinite set of graphs, then one is a minor of another.
1. Prove that every 3-connected graph has a 𝐾4 minor.
2. Prove that a graph 𝐺 can be drawn in the plane with all vertices in the same region unlesss 𝐺 has a
𝐾4 or 𝐾3,2 minor.
3. Prove that if a graph 𝐺 has no 𝐾5 minor, then it’s 4-colorable. (You may assume the Four-color
Theorem.)
4. Prove that every simple graph with average degree at least 2𝑝 has a 𝐾𝑝+2 minor.
5. Prove that if 𝐺 is nonnull and loopless and ∣𝐸(𝐺)∣ ≥ 2∣𝑉 (𝐺)∣ − 1, then 𝐺 has a graph with three
parallel edges as a minor.
6. Find all 2-connected graphs with no 𝐶5 minor.
7. Find all 2-connected graphs with no 𝐾4 ∖ 𝑒 minor.
8. (Kotzig’s Theorem, also 2009 MOP K6.2/B6.4) Let G be a connected graph that has a perfect matching.
Prove that if for any edge 𝑒 of the perfect matching, 𝐺 ∖ 𝑒 is connected, then 𝐺 has another perfect
matching.
