Design and Analysis of Algorithms

Minimum Spanning trees

Presented By
Dr Muhammad Atif Tahir
From
Haidong Xue
 What are trees in undirected graphs?
• We are familiar with a rooted directed tree,
such as:
3
– Binary search tree
5
– Heap 6
4
• They look like: 2
– You know which one the root is
– Who the parent is
– Who the children are
 What are trees in undirected graphs?
• However, in undirected graphs, there is
another definition of trees
• Tree
– A undirected graph (V, E), where E is the set of
undirected edges
– All vertices are connected 3

– |E|=|V|-1
5
6
4

2
 What is a spanning tree?
•

b c d

a i e

h g f
 What is a spanning tree?
• Given the following graph, are they spanning
trees?
b c d

a i e

h g f

Is the red sub graph a tree? Yes

Is the red sub graph a spanning

tree?
 What is a spanning tree?
• Given the following graph, are they spanning
trees?
b c d

a i e

h g f

Is the red sub graph a tree?

Is the red sub graph a spanning

tree?
 What is a spanning tree?
• Given the following graph, are they spanning
trees?
b c d

a i e

h g f

Is the red sub graph a tree? No. Vertices are not all connected

Is the red sub graph a spanning

tree?
 What is a spanning tree?
• Given the following graph, are they spanning
trees?
b c d

a i e

h g f

Is the red sub graph a tree? Yes

Is the red sub graph a spanning Yes

tree?
 What is a spanning tree?
• Given the following graph, are they spanning
trees?
b c d

a i e

h g f

Is the red sub graph a tree? Yes

Is the red sub graph a spanning Yes

tree?

There could be more than one spanning trees

What is a minimum spanning tree?
• When edges are weighted
• A minimum spanning tree of G is:
– A spanning tree of G
– With a minimum sum of a edge weights sum among all
spanning trees

8 7
b c d
4 9
2
11 14
a i 4 e
7 6
8 10
h 1
g 2
f
What is a minimum spanning tree?
• Spanning trees:
8 7
b c d
4 9 The weight sum of this
2
11 14
spanning tree?
a i 4 e 48
7 6
8 10
h 1
g 2
f

b c d
The weight sum of this
4 9
2 spanning tree?
11 14
a i 4 e 47
7 6
8 10
h 1
g 2
f
What is a minimum spanning tree?
• Spanning trees:
8 7
b c d
4 9 The weight sum of this
2
11 14
spanning tree?
a i 4 e 37
7 6
8 10
h 1
g 2
f

8 7
b c d
The weight sum of this
4 9
2 spanning tree?
11 14
a i 4 e 37
7 6
8 10
These 2 are minimum spanning trees
h 1
g 2
f since there is no spanning tree with a
smaller total weight
 Minimum spanning tree problem
• Minimum spanning tree (MST) problem:
given a undirected graph G, find a spanning
tree with the minimum total edge weights.
• There are 2 classic greedy algorithms to solve
this problem
– Kruskal’s algorithm
– Prim’s algorithm
 Minimum spanning tree problem
• A application example of MST in computer
network
– What if PC3 wants to communicate with PC7?
– By a routing protocol:
• PC3 broadcasts this requirement to PC2, PC4
• PC2 broadcasts it to PC1, PC5 PC1 PC2 PC3
• PC1 broadcasts it to PC5, PC6
…
• PC5 broadcasts it to PC2 PC6 PC5 PC4

– At least, a package is in a deadlock:

PC2->PC1->PC5-PC2->PC1-> …………….
Find and use a MST can solve this problem, since there is no cycle
 Minimum spanning tree problem
• The MST algorithm can be considered as
choose |V|-1 edges from E to form a
minimum spanning tree
• Both Kruskal’s algorithm and Prim’s algorithm
proposed a greedy choice about how to
choose a edge
• They have greedy choice property because of
textbook Corollary 23.2
 Kruskal’s algorithm
• Input: G=(V, E)
• Output: a minimum spanning tree (V, A)
• Kruskal (G=(V, E))
Set each vertices in V as a vertex set;
Set A as an empty set;
while (there are more than 1 vertex set){
Add a smallest edge (u, v) to A where u, v are not in the same set (i.e.
avoid cycle);
Merge the set contains u with the set contains v;
}
• In most of the implementations, edges are sorted on
weights at the beginning, and then scan them once
• The running time = O(|E|lg|V|)
 Kruskal’s algorithm
Kruskal (G=(V, E)) Edge Weight
Set each vertices in V as a vertex set;
Set A as an empty set; (h, g) 1
while (there are more than 1 vertex set){ (i, c) 2
Add a smallest edge (u, v) to A where u, (g, f) 2
v are not in the same set;
Merge the set contains u with the set (c, f) 4
contains v; (a, b) 4
} (i, g) 6
8 7
b c d (h, i) 7

4 9
(c, d) 7
2
(a, h) 8
11 14
4
a i e (b, c) 8
7 6
8
(d, e) 9
10
(e, f) 10
h 1
g 2
f
(b, h) 11
Total weight: 37 (d, f) 14
 Prim’s algorithm
• Input: G=(V, E)
• Output: a minimum spanning tree (V, A)
• Prim(G=(V, E))
Set A as an empty set;
Choose a vertex as the current MST;
while (not all the vertices are added into the MST){
Choose the smallest edge crossing current MST vertices and
other vertices, and put it to MST ;
}
• The running time = O(|E|lg|V|)
 Prim’s algorithm
Prim(G=(V, E))
Set A as an empty set;
Choose a vertex as the current MST;
while (not all the vertices are added into the MST){
Choose the smallest safe edge, and put it to MST;
}

8 7
b c d
4 9
2
11 14
4
a i e
7 6
8 10

h 1
g 2
f

Total weight: 37
 Summary
• Prim’s algorithm and Kruskal’s algorithm have
the same asymptotically running time