Lecture

Graph Algorithms
Minimum Spanning Tree
A minimum spanning tree is a tree formed
from a subset of the edges in a given
undirected graph, with two properties:
1. It spans the graph, i.e., it includes every

vertex in the graph, and

2. It is a minimum, i.e., the total weight of
all the edges is as low as possible.
Example of Spanning Tree

Source: http://www.cs.nthu.edu.tw/~wkhon/algo09/lectures/lecture26.pdf
Example of Minimum Spanning Tree
Problem Statement

Given a weighted graph G(V,E) with

unique positive edge weights, find the
Minimum Spanning Tree of it.

Itis perhaps one of the best studied

problem in combinatorial optimization.
Algorithms you have done?

Kruskal’s algorithm :
Prim’s algorithm:
History
Boruvka 1926–O(m log n)
Kruskal 1956- O ( m log n)
Prim 1957- O ( m log n)
Yao 1975 – O ( m log log n)
Cheriton and Tarjan 1976 – O ( m log log n)
Fredman and Tarjan 1987 –O (m (m, n))
Gabow 1986 – O(m log  (m, n))
Chazelle 1995 – O ( m ( m, n))
Minimum Spanning Forest
A forest F is an acyclic subgraph of G that
consists of a collection of disjoint trees in G;
treat isolated vertices in F as tree of size 1.
A spanning forest is a forest whose trees are
spanning trees for the connected component
of the graph G.
A spanning forest is a spanning tree if and
only if the graph is connected.
A Minimum Spanning Forest is a forest
with minimum weight.
Boruvka
For each vertex of G(V,E), select the minimum-
weight edge incident to the vertex. Contract all
the selected edges, replacing by a single vertex
each connected component defined by the
selected edges and deleting all resulting isolated
vertices, loops (edges both of whose endpoints
are the same), and all but the lowest-weight edge
among each set of multiple edges. Call this new
graph after contraction as G’.
This is called a Boruvka phase.
Apply recursively Boruvka phase on G’.
Example
A G
3 2.1 3.2 1.5
2.2 5.1
B C H I

1 3.1 5 2.5 2.6

2.3 4
2.4 3.3
D E L J
4.1
1.3 1.2 1.4 2.7
F K
First Phase
A G
3 2.1 3.2 1.5
2.2 5.1
B C H I

1 3.1 5 2.5 2.6

2.3 4
2.4 3.3
D E L J
4.1
1.3 1.2 2.7
F 1.4 K

Edges selected = {AC, BD, EF, GH,

GI, GJ, LK}
After Contraction
A,C G
2.2 5.1 3.2 1.5

B H I

1 3.1 5 2.5 2.6

2.3 4
2.4 3.3
D E L J
4.1
1.3 1.2 2.7
F 1.4 K
After Contraction
A,C G
5.1 3.2 1.5

H I
2.2
2.3 4 5 2.5 2.6
2.4 3.3
B,D E L J
4.1
1.3 1.2 2.7
F 1.4 K
After Contraction
A,C G
5.1 3.2 1.5

H I
2.2
2.3 4 5 2.5 2.6

1.3 3.3
B,D E,F L J
4.1
2.7
1.4 K
After Contraction
A,C G,H
5.1 1.5

I
4
2.2
2.3 2.5 2.6

1.3 3.3
B,D E,F L J
4.1
2.7
1.4 K
After Contraction
A,C G,H,I
5.1

4
2.2
2.3 2.5

1.3 3.3
B,D E,F L J
4.1
2.7
1.4 K
After Contraction
A,C G,H,I,J
5.1

2.2 2.7
2.3 3.3

B,D 1.3 E,F L

4.1
1.4 K
End of Phase I
A,C G,H,I,J
5.1

2.2 2.7
2.3

B,D 1.3 E,F L,K

4.1

Edges selected = {AC, BD, EF, GH,

GI, GJ, LK}
Apply Phase II
A,C G,H,I,J
5.1

2.2 2.7
2.3

B,D 1.3 E,F L,K

4.1

Edges selected = {AC, BD, EF, GH,

GI, GJ, LK}
Apply Phase II
A,C G,H,I,J
5.1

2.2 2.7
2.3
1.3
B,D E,F L,K
4.1

Edges selected = {AC, BD, EF, GH,

GI, GJ, LK}
Apply Phase II
A,C G,H,I,J
5.1

2.2 2.7
2.3
1.3
B,D E,F L,K
4.1

Edges selected = {AC, BD, EF, GH,

GI, GJ, LK, BC, DF, JK}
After Contraction
A,C,B,D G,H,I,J
5.1

2.7
1.3

E,F L,K
4.1

Edges selected = {AC, BD, EF, GH,

GI, GJ, LK, BC, DF, JK}
Last Phase
A,C,B,D,E,F G,H,I,J,L,K
4.1

Edges selected = {AC, BD, EF, GH,

GI, GJ, LK, BC, DF, JK}
After Contraction
A,C,B,D,E,F G,H,I,J,L,K
4.1

Edges selected = {AC, BD, EF, GH,

GI, GJ, LK, BC, DF, JK,EL}
MST
Edges selected = {AC, BD, EF, GH, GI,
GJ, LK, BC, DF, JK,EL}

A G
3 2.1 3.2 1.5
2.2 5.1
B C H I

1 3.1 5 2.5
2.3 4
2.4 3.3
D E L J
4.1
1.3 1.2 1.4 2.7
F K
MST
Edges selected = {AC, BD, EF, GH, GI,
GJ, LK, BC, DF, JK,EL}

A G
2.1 3.2 1.5
2.2 C
B H I

1 2.5

D E L J
4.1
1.3 1.2 1.4 2.7
F K
Boruvka
 Claim: The graph G’ obtained after one Bouruvka
Phase has at most vertices.

 Proof: Note for each vertex we mark an incident

edge. But some of these edges can be marked twice.
So the number of marked edges is larger than . When
a marked edge is collapsed, it eliminates a distinct
vertex. Thus or more than vertices are eliminated.
Hence G’ has at most vertices.

 The number of edges in G’ is no more than since no

new edges are created during the process.
BORUVKA_STEP
Require: Input: Gi = (Vi;Ei)
Ensure: Output: a forest F of MST edges and a contracted graph G’
F  ;
for all {u,v}  Ei do
if w({u,v}) < u.minEdgeWeight then
u.minEdgeWeight  w({u,v})
u.minEdge  {u, v}
end if
if w({u,v}) < v.minEdgeWeight then
v.minEdgeWeight  w({u,v})
v.minEdge  {u, v}
end if
end for
for all v  V do
put v.minEdge in F
end for
Gi+1 = contract (Gi; F)
return Gi+1, F
CONTRACT
Require: Input: G = (V;E), a subgraph H of edges to contract
Ensure: Output: G’, the contracted graph
If not every vertex in V is represented in H, put the missing vertices in
V’  , E’  ;
G’ = (V’;E’)
connectedComponents  find-connected-components(H)
i 0
for all C  connectedComponents do
put i in V’
for all v  C do
v.component  i
end for
i i+1
end for
for all {u, v}  E do
If (u.component != v.component)
put {u.component, v.component} in E’
end for
return G’ = (V’;E’)
Running Time of Boruvka

One can show that a Boruvka phase on a

graph G with n vertices and m edges can
be implemented in time O(n+m).

At most log (n) Boruvka phases will be

required because after each phase the
resultant graph has less than vertices.
So worst case running time O(m log n).