Skip to content

4.21 Graphs

Track the relations between object

Example: road map of cities and road that connect themimage

Simple Labelled Graph

A simple labelled graph \(\Gamma\) is a pair \(\Gamma=(V,E)\) where \(V\) is the set of vertices of \(\Gamma\) and \(E\) is a subset of unordered pairs of distinct element of \(V\), that is the edges of \(\Gamma\)

As the example above, edges can be (Chengdu, Guangzhou), (Guangzhou, Shantou),...​​

And this graph is same with above pic. We only care vertices and edges.

image

Vertex and Edge

A degree of a vertex \(v\in V\) of a graph \(\Gamma=(V,E)\) is \(\#\{e\in E:v\in e\}=\deg (v)\) (The number of edges the vertex \(v\) belongs to)

image

When a vertex \(v\) belongs to an edge \(e\), we say that \(v\) and \(e\) are incident

When edge \(e=(v_1,v_2)\), we say that \(v_1\) and \(v_2\) are adjacent.

The handshaking lemma

Let \(\Gamma=(E,V)\) be a finite (\(\#\text{vertices}<\infty\)) simple graph, then \(2\#E=\sum_{v\in V}\deg(v)\)

Example: Take above pic as an example

\(\sum\deg=1+2+3+4+1+1+0=12=2\#E\)

Proof

Let's count the number of pairs: \(P=\{(v,e)\in V\times E:v\text{ is incident to }e\}\)

If we fix \(v_0\in V\), \(\#\{(v_{0},e)\}=\deg v_{0}\). If we fix \(e_0\in E\), \(\#\{(v,e_0)\}=2\) (illustration:image)

Then \(\#P=\sum_{v_0\in V}\deg v_0=\sum_{e_0\in E}2=2\#E\)

Graph isomorphism

We say the labelled graphs \(\Gamma_{1}=(V_{1},E_{1})\) and \(\Gamma_2=(V_2,E_2)\) are labelled if there exist a bijection \(f:V_1\to V_2\) such that \((v_{1},v_{2})\in E\iff(f(v_1),f(v_2))\in E_2\)

(Graph isomorphism is an equivalence relation on the set of graphs)

Subgraphs

Let \(\Gamma=(V,E)\) and \(\Gamma'=(V',E')\) are two graphs, we say that \(\Gamma\) is a subgraph of \(\Gamma'\) ( \(\Gamma\leq \Gamma'\) ) if \(V\subseteq V'\) and \(E\subseteq E'\)

image

Notice: It is not required that if \(v_1\) and \(v_2\) are adjacent in \(\Gamma'\), they are adjacent in \(\Gamma\)

Induced Subgraph

Let \(\Gamma'=(V',E')\) be a graph such that \(V\subseteq V'\), then \(\Gamma'[x]\) is a subgraph of \(\Gamma'\) defined by the following conditions

  1. Set of vertices is \(V\)

  2. \(v_1\) and \(v_2\) are adjacent in \(\Gamma'[x]\) iff they are adjacent in \(\Gamma'\)

Some Important Classes of Graphs

  • Empty or discrete graph \(E_n\) (\(D_n\)), that is \(n\) vertices with \(0\) edges

image​ * Complete graph \(K_n\), that is \(n\) vertices and every vertices are adjacentimage​ * Cycle graph \(C_n\), that is \(n\) vertices and adjacent to form a cycleimage​ * Path graph \(P_n\), that is \(n\) vertices and adjacent to form a pathimage

UV-WALK

Let \(\Gamma =(V,E)\) be a graph and \(u\),\(v\) are two vertices of \(\Gamma\) (not necessary different)

A UV-WALK in \(\Gamma\) is a sequence walk of length \(n+1:\) \(u\overset{e_0}{-}w_{1}\overset{e_1}{-}w_{2}\overset{e_2}{-}w_{3}\overset{e_3}{-} ...\overset{e_{n-1}}{-}w_{n}\overset{e_{n}}{-}v\) where \(w_1,...,w_n\) are vertices of \(\Gamma\) and \(e_1,...,e_n\) are edges of \(\Gamma\)

Here, \(u\) is adjacent to \(w_1\) via \(e_0\), \(w_{i}\) is adjacent to \(w_{i+1}\) via \(e_{i}\), ..., \(w_n\) is adjacent to \(v\) via \(e_n\)

Example

image length: 7

We say that \(\Gamma=(V,E)\) is connected if \(\forall\) two vertices \(u,v\in V\), there is a uv-walk between them.

Acyclic

A graph \(\Gamma=(V,E)\) is acyclic if there are no subgraphs of \(\Gamma\) isomorphic to \(C_n\) for \(n\geq 3\)

image

A graph \(\Gamma=(V,E)\) is a tree if it is connected and acyclic:image

UV-path

A uv-path in \(\Gamma(V,E)\) is a subgraph of \(\Gamma\) isomorphic to \(P_n\) for some \(n\geq 2\) such that the degree \(1\) vertices of \(P_n\) are mapped to \(u\) and \(v\) respectively

image(See the black lines)

Remark: UV-walk can go through a repeated vertex but U

v-path cannot.

Lemma

If \(\Gamma\) is a graph and there is a uv-walk in \(\Gamma\), then there is a uv-path in \(\Gamma\)

Idea: If every two vertices in a walk are different, then we have a path

Suppose there is a vertex \(w\) in the walk, that is encountered more than once

Reduces the numbers of appearance of vertex \(w\) by \(1\), do until we have a path.image

Proposition

Let \(\Gamma=(V,E)\) be a graph and \(\min_{v\in V}\deg\left(v\right)=\delta\geq2\), then \(\Gamma\) has a path of length \(\delta\) and a cycle of length at least \(\delta+1\)

Proof

Let \(P=(v_0,...,v_k)\) be a longest path in \(\Gamma\), denote \(W=\{v_0,...,v_k\}\) and \(\#W=k+1\)

Notice that every vertex adjacent to \(v_0\) is an element of \(W\), otherwise we can make a longer path, which would produce a contradiction

Thus \(\#W\geq \deg(v_{0})\geq \delta\)


Let \(i\in \{1,...,k\}\) be the largest number such that \(v_i\) is adjacent to \(v_0\)

Clearly, \(i\geq \delta\), connect \(v_i\) to \(v_0\) forming a cycle​

Illustration: image

Corollary

Every tree has a leaf. (leaf is a degree \(1\) vertex)

Suppose \(Г\) has no leaves, then \(\delta \ge 2\)

Then by corollary, it has a cycle of at least \(\delta+1\)
Then it is not acyclic, by definition it is not a tree.

Proposition

Let \(\Gamma=(V,E)\) be a tree and \(\#V=n\), then \(\#E=n-1\)

Proof

Induction on \(n\)

Base case: \(n=1\), \(1\) vertex and \(0\) edges

Induction step: Suppose we know it for all trees up to \(n-1\) vertices.

Take a tree on \(n\) vertices, by corollary it has a leaf. Remove the leaf, it gives a tree.

Then applied hypothesis, \(n-1\) vertices, \(n-2\) degrees to \(n\) vertices, \(n-1\) degrees