Discrete Mathematics

PART 1
COMBINATORIAL THEORY
(Lý thuyết tổ hợp)

PART 2
GRAPH THEORY
(Lý thuyết đồ thị)
2
 Content of Part 2
Chapter 1. Fundamental concepts
Chapter 2. Graph representation
Chapter 3. Graph Traversal
Chapter 4. Tree and Spanning tree
Chapter 5. Shortest path problem
Chapter 6. Maximum flow problem
 Contents
1. Tree and its properties
2. Spanning tree
3. The minimal spanning tree

4
1. Tree and its properties
A tree is an undirected connected graph with no cycles.
Example 1. Which of the graphs are trees?

Solution: G1, G2 Note. G3: contains cycle {a,b,e,d,a}

G4: not connected

5
1. Tree and its properties
Forest: Graph containing no cycles that are not connected, but each
connected component is a tree.

T1 T2

T3

Forest F consists of 3 trees: T1, T2,, T3

6
1. Tree and its properties
Theorem. Given an undirected graph G = (V,E), the following conditions are
equivalent:
(1) G is a connected graph with no cycles. (Thus G is a tree by the above
definition).
(2) For every two vertices u, v ∈ V, there exists exactly one simple path from u
to v.
(3) G is connected, and removing any edge from G disconnects it (each edge of
G is a bridge).
(4) G has no cycles, and adding any edge to G gives rise to a cycle. (Thus G is a
maximal acyclic graph).
(5) G is connected and |E| = |V | − 1.

7
 Contents
1. Tree and its properties
2. Spanning tree
3. The minimal spanning tree

8
2.The spanning tree
Let G=(V, E) be an undirected connected graph with vertex set V.
Tree T=(V,F) where FÍ E is called spanning tree of G

Undirected Connected graph without cycle

b c b c b c

a d a d a d

e e e
G Spanning tree T1 Spanning tree T2

Graph G and its 2 spanning trees T1 and T2

2. The spanning tree
Theorem. Every undirected connected graph contains a
spanning tree.
Proof. Let G be an undirected connected graph.
• If G contains no cycle then G is its own spanning tree.
• If G contains a cycle: Removing any edge from the cycle gives a graph
which is still connected. If the new graph contains a cycle then again remove
one edge of the cycle. Continue this process until the resulting graph T
contains no cycles. We have not removed any vertices so T has the same
vertex set as G, and at each step of the above process we obtain a connected
graph. Therefore T is connected and it is a spanning tree for G.
2. The spanning tree
Theorem (Cayley). A complete graph Kn has nn-2 spanning trees.

(A complete graph is a simple undirected graph in which

every pair of distinct vertices is connected by a unique edge)
Arthur Cayley
b a b c (1821 – 1895)

b c a

a c c a b

K3 3 spanning trees of K3
 16 spanning trees of K4
1 2 1 2 1 2 1 2

3 4 3 4 3 4 3 4

1 2 1 2 1 2 1 2

3 4 3 4 3 4 3 4

1 2 1 2 1 2 1 2

3 4 3 4 3 4 3 4

1 2 1 2 1 2 1 2

3 4 3 4 3 4 3 4
 Contents
1. Tree and its properties
2. Spanning tree
3. The minimal spanning tree

13
Weighted Graphs and Minimum Spanning Trees
Let G=(V, E) be an undirected connected graph with vertex set V:
• For each edge (u,v) in E, we have a weight w(u,v) specifying the cost (length
of edge) to connect u and v.

For any subgraph H of G, we define the weight of H, denoted by w(H), to be the

sum of its edge weights:
w( H ) = å
eÎE ( H )
c(e)

A minimum spanning tree for G is a spanning tree T which has the smallest
weight 7 a
2 2 d 5
5 4
f b
1
1 g The weight of the
4 3 7 spanning tree = 14
c 4 e
Weighted Graphs and Minimum Spanning Trees
Every undirected connected weighted graph G has a minimum spanning tree. Since
G has only a finite number of spanning trees, one of them must have minimum
weight.
Note: a given undirected connected weighted may have more than one minimum
spanning tree.
Example: An undirected connected weighted graph with two minimal spanning
trees, both of weight 14
a 7 a 7
2 2 d 2 2 d
5 5
f 5 4 f 5 4
b 1 g b 1 g
1 3 2 1 3
2 7
c 4 e 7 c 4 e
As the number of spanning trees of G is very large (see Cayley's theorem),
we could not solve this problem by brute force.
 Applications of Minimum Spanning Trees: an example
Network design: telephone, electrical, hydraulic, TV cable, computer, road.
• Phone network design. You have a business with several offices; you want to lease
phone lines to connect them up with each other; and the phone company charges
different amounts of money to connect different pairs of cities. You want a set of
lines that connects all your offices with a minimum total cost. It should be a
spanning tree, since if a network isn’t a tree you can always remove some edges and
save money.
• In the design of electronic circuits, it is often necessary to connect pins by wiring
them together:
• To interconnect n pins, we can use n-1 wires,
each connecting 2 pins.
• We want to minimize the total length of the
wires.
• Minimum Spanning Trees can be used to
model this problem.
 Applications of Minimum Spanning Trees: an example
Suppose we want to build a railway system that connects n cities so that
passengers can travel between any two cities and the total cost of
construction must be minimal.

It is clear that it corresponds to a graph where vertices are the cities and
the edges are the railroads connecting the cities, the length on each edge
is the cost of building a railway connecting the two cities.

Therefore, it leads to the problem of finding the smallest spanning tree

on a complete graph Kn
Build a railway system

Length of spanning tree: 17284

Length of spanning tree: 15932

 Applications of Minimum Spanning Trees: an example
Once upon a time there was a city that had no roads. The mayor of the city decided that
some of the streets must be paved, but didn’t want to spend more money than necessary
because the city also wanted to build a swimming pool. The mayor therefore specified two
conditions:
• Enough streets must be paved so that it is possible for everyone to travel from their
house to anyone else house only along paved roads,
• The paving should cost as little as possible. Here is the layout of the city. The number of
paving stones between each house represents the cost of paving that route.
Find the best route that connects all the houses, but uses as few counters (paving stones) as
possible
 Applications of Minimum Spanning Trees: examples
Other practical applications based on minimal spanning trees include:
• Taxonomy
• Cluster analysis: clustering points in the plane, single-linkage clustering,
graph-theoretic clustering, and clustering gene expression data.
• Constructing trees for broadcasting in computer networks. On Ethernet
networks this is accomplished by means of the Spanning tree protocol.
• Image registration and segmentation.
• Curvilinear feature extraction in computer vision.
• Handwriting recognition of mathematical expressions.
• Circuit design: implementing efficient multiple constant multiplications, as
used in finite impulse response filters.
• Regionalization of socio-geographic areas, the grouping of areas into
homogeneous, contiguous regions.
• Comparing ecotoxicology data.
• Topological observability in power systems.
• Measuring homogeneity of two-dimensional materials.
 General scheme of the algorithm to find MST
Initialize: The minimum spanning tree T = Æ
Each step of the algorithm: one edge e which is the “safe” edge is chosen, subject only
to the restriction that if adding edge e into T then T is still a tree (no cycle is created).

Generic-MST(G, c)
T = Æ
//T is the subset edges of some minimum spanning tree
while T is not the spanning tree do
Finding edge (u, v) is “safe” edge for T
T = T È{(u, v)}
return T

Edge with smallest weight and insert

it into T does not create cycle
Set T is always a subset of edges of some minimum spanning tree. This property is
called the invariant Property.
An edge (u,v) is a safe edge for T if adding the edge to T does not destroy the invariant.
 General scheme of the algorithm to find MST
Initialize: The minimum spanning tree T = Æ
Each step of the algorithm: one edge e which is the “safe” edge is chosen, subject only
to the restriction that if adding edge e into T then T is still a tree (no cycle is created).

Generic-MST(G, c)
T = Æ
//T is the subset edges of some minimum spanning tree
while T is not the spanning tree do
Finding edge (u, v) is “safe” edge for T
T = T È{(u, v)}
return T
How to find the “safe” edge ???:
The criteria to choose an edge at each step decides the process of two following
minimum spanning tree algorithms (both use greedy strategies):
1. Kruskal
2. PRIM
 How to find the “safe” edge?
T : set of edges of some spanning tree
Initialize: T = Æ

Kruskal algorithm
v T is forest.
v The “safe” edge added to T at each iteration is the edge with smallest
weight among edges connecting its connected components.

Prim algorithm
v T is tree.
v The “safe” edge added to T at each iteration is the edge with smallest weight
among edges connecting the tree T to other vertex not in the tree.
Kruskal algorithm
Generic-MST(G, c)
T = Æ
//T is the subset edges of a minimum spanning tree
while T is not the spanning tree do
Finding edge (u, v) is “safe” edge for T
T = T È{(u, v)} Joseph Kruskal
return T (1928 - ~)

Kruskal algorithm:
v T is forest (empty).
v The “safe” edge included in T at each iteration is the edge with smallest weight
among edges connecting its connected components.

void Kruskal ( ) {
Sort m edges of the graph e1,e2, . . . , em in ascending order of weight;
T = Æ; // T: set of edges of the minimum spanning tree
for (i = 1; i <=m; i++)
if (T È{ei} does not contain cycle)
T = T È {ei};
}
Kruskal algorithm: an example

7
a
2 2 Cycle (e,b,d,e)
d 5
Cycle (a,b,f,a)
5 4
f b 1 g
4 1 3
Cycle (a,b,c,f,a)
7
c e
4
Cycle (b,c,e,b)

Weight of the minimum spanning tree:

2+2+1+3+1+5=14
Computation time
void Kruskal ( ){
Sort m edges of the graph e1,e2, . . . , em in ascending order of weight;
T = Æ; // T: set of edges of the minimum spanning tree
for (i = 1; i <=m; i++)
if (T È{ei} does not contain cycle)
T = T È {ei};
}

• Step 1. Sort the edges in ascending order of weight

– Could use heap sort/merge sort : O(m log m)
• Iterations: Each iteration we need to check whether T È { ei } contains the cycle?
– Could use DFS to check with time O(m+n).
– Total time: O(m(m+n))
Computation time: O(m log m + m(m+n)) where n, m is the number of vertices and
edges of the graph, respectively.
PRIM algorithm
Generic-MST(G, c)
T = Æ
//T is the subset edges of a minimum spanning tree
while T is not the spanning tree do
Finding edge (u, v) is “safe” edge for T Robert Clay Prim
T = T È{(u, v)} (1921 - ~)
return T

PRIM algorithm:
v T is tree (initialize: T has one vertex).
v The “safe” edge included in T at each iteration is the edge with smallest weight
among edges connecting a vertex of T to other vertex not in T
PRIM algorithm: an example
7
a
2 2 d 5 select
5 4
f b 1 g
4 1 3
7
c e
4
Edges could be selected
PRIM algorithm:
v T is tree (initialize: T has one vertex).
v The “safe” edge included in T at each iteration is the edge with smallest weight
among edges connecting a vertex of T to other vertex not in T
PRIM algorithm: an example
7
a
2 2 d 5
5 4
f b 1 g
4 1 3
7
c e
4

PRIM algorithm:
v T is tree (initialize: T has one vertex).
v The “safe” edge included in T at each iteration is the edge with smallest weight
among edges connecting a vertex of T to other vertex not in T
PRIM algorithm: an example
7
a
2 2 d 5
5 4
f b 1 g
4 1 3
7
c e
4

PRIM algorithm:
v T is tree (initialize: T has one vertex).
v The “safe” edge included in T at each iteration is the edge with smallest weight
among edges connecting a vertex of T to other vertex not in T
PRIM algorithm: an example
7
a
2 2 d 5
5 4
f b 1 g
4 1 3
7
c e
4

PRIM algorithm:
v T is tree (initialize: T has one vertex).
v The “safe” edge included in T at each iteration is the edge with smallest weight
among edges connecting a vertex of T to other vertex not in T
PRIM algorithm: an example
7
a
2 2 d 5
5 4
f b 1 g
4 1 3
7
c e
4

PRIM algorithm:
v T is tree (initialize: T has one vertex).
v The “safe” edge included in T at each iteration is the edge with smallest weight
among edges connecting a vertex of T to other vertex not in T
PRIM algorithm: an example
7
a
2 2 d 5
5 4
f b 1 g
4 1 3
7
c e
4

Minimum spanning tree with edges: (g,d), (d,e), (e,b), (b,c), (b,a), (a,f)
The weight: 14 5+1+3+1+2+2 = 14
PRIM algorithm
void Prim(G, C) //G: graph; C: weight matrix
{
Select an arbitrary vertex r ÎV; Tree T with only one vertex r
Initialize: tree T=(V(T), E(T)) where V(T)= {r}and E(T)=Æ;
while (T has < n vertices)Among edges connecting a vertex of T to other vertex
{ not in T: find the edge with minimum weight
(u, v) is the minimum weight where u Î V(T) and vÎV(G) – V(T)
E(T) ¬ E(T) È { (u, v) };
V(T) ¬ V(T) È { v } add vertex v and edge (u,v) to tree T
}
}
PRIM: Implementation
• Graph with weight matrix C = {c[i,j], i, j = 1, 2,..., n}.
• At each iteration: to select quickly a vertex and an edge to add to spanning tree,
vertices of graph are labeled:
Label of a vertex v Î V \ V(T) in the form [d[v], near[v]] :
– d[v] :use to record the distance from vertex v to vertex set V(T):
d[v] := min{ c[v, w] : w Î V(T) } ( = c[v, z])
Edge with minimum edge among edges connecting vertex v to other vertex of V(T)
– near[v] := z record the vertex of T that is nearest to vertex v
12 w1 w2
v wi
5
d(v) = 2 w3 w4
near[v] = w5 2 wj

w5 V(T)
void Prim ( ) {
// Initialize:
V(T) = { r }; E(T) = Æ ;
d[r] = 0; near[r] = r;
for v Î V \ V (T)
{
d[v] = c[r,v]; near[v] = r;
} Prepare data for finding “safe” edge process
// Iteration: d[v]: the edge with minimum weight connecting vertex v (not yet in
for (k=2; k<=n;k++) the spanning tree T to other vertex of T
{
Find v Î V \ V(T) satisfying: d[v] = min { d[i] : i Î V \ V(T) };
V(T) = V(T) È { v }; E(T) = E(T) È { ( v, near[v] ) } ;
for v’Î V \ V(T)
if (d[v’] > c[v,v’]) Because we have just changed the spanning
{ tree T : vertex v has just been added to T
d[v’] = c[v,v’] ; near[v’] = v; à Need to update label of vertices not yet in
} T if necessary
}
T is the minimum spanning tree;
}
Computation time: O(|V| 2)