NPTEL MOOC,JAN-FEB 2015
Week 3, Module 5
DESIGN AND ANALYSIS
OF ALGORITHMS
Applications of BFS and DFS
MADHAVAN MUKUND, CHENNAI MATHEMATICAL INSTITUTE
http://www.cmi.ac.in/~madhavan
Graphs, formally
G = (V,E)
Set of vertices V
Set of edges E
E is a subset of pairs (v,v’): E ⊆ V × V
Undirected graph: (v,v’) and (v’,v) are the same edge
Directed graph:
(v,v’) is an edge from v to v’
Does not guarantee that (v’,v) is also an edge
Exploring graph structure
Breadth first search
Level by level exploration
Depth first search
Explore each vertex as soon as it is visited
DFS numbering
What can we find out about a graph using BFS/
DFS?
Connectivity
1 2 1 2 3 4
3 5 6 7 8
4 5 9 10 11 12
Connected graph Disconnected graph
Connectivity
1 2 1 2 3 4
3 5 6 7 8
4 5 9 10 11 12
Connected graph Disconnected graph
Connected components
Identifying connected
components
Vertices {1,2,…,N}
Start BFS or DFS from 1
All nodes marked Visited form a connected component
Pick first unvisited node, say j, and run BFS or DFS
from j
Repeat till all nodes are visited
Update BFS/DFS to label each visited node with
component number
Connected components
1 2 3 4
5 6 7 8
9 10 11 12
Add a counter comp to number components
Increment counter each time a fresh BFS/DFS starts
Label each visited node j with component[j] = comp
Connected components
1 1 2 3 4
5 6 7 8
9 10 11 12
Add a counter comp to number components
Increment counter each time a fresh BFS/DFS starts
Label each visited node j with component[j] = comp
Connected components
1 1 2 3 4 2
5 6 7 8
9 10 11 12
Add a counter comp to number components
Increment counter each time a fresh BFS/DFS starts
Label each visited node j with component[j] = comp
Connected components
1 1 2 3 4 2
5 3 6 7 8
9 10 11 12
Add a counter comp to number components
Increment counter each time a fresh BFS/DFS starts
Label each visited node j with component[j] = comp
Cycles
1 2 1 2 3 4
3 5 6 7 8
4 5 9 10 11 12
Acyclic graph Graph with cycles
BFS tree
1 2 1 2 3 4
3 5 6 7 8
4 5 9 10 11 12
Edges explored by BFS form a tree
Acyclic graph = connected, with n-1 edges
BFS tree
1 2 1 2 3 4
3 5 6 7 8
4 5 9 10 11 12
Edges explored by BFS form a tree
Acyclic graph = connected, with n-1 edges
Any non-tree edge generates a cycle
DFS tree
1 2 3 4
5 6 7 8
9 10 11 12
DFS tree
1 2 3 4 1
5 6 7 8
9 10 11 12
DFS tree pre
0
1 2 3 4 1
5 6 7 8
9 10 11 12
DFS tree pre
0
1 2 3 4 1
5 6 7 8 2
1
9 10 11 12
DFS tree pre
0
1 2 3 4 1
5 6 7 8 2
1 2
post
9 10 11 12
DFS tree pre
0
1 2 3 4 1
5 6 7 8 2 5
1 2 3
post
9 10 11 12
DFS tree pre
0
1 2 3 4 1
5 6 7 8 2 5
1 2 3
post
9 10 11 12 9
4
DFS tree pre
0
1 2 3 4 1
5 6 7 8 2 5
1 2 3
post
9 10 11 12 9
4
10
5
DFS tree pre
0
1 2 3 4 1
5 6 7 8 2 5
1 2 3
post
9 10 11 12 9
4
10
5 6
DFS tree pre
0
1 2 3 4 1
5 6 7 8 2 5
1 2 3
post
9 10 11 12 9
4 7
10
5 6
DFS tree pre
0
1 2 3 4 1
5 6 7 8 2 5
1 2 3 8
post
9 10 11 12 9
4 7
10
5 6
DFS tree pre
0 9
1 2 3 4 1
5 6 7 8 2 5
1 2 3 8
post
9 10 11 12 9
4 7
10
5 6
DFS tree pre
0 9
1 2 3 4 1 3
5 6 7 8 2 5
1 2 3 8
post
9 10 11 12 9
4 7
10
5 6
DFS tree pre
0 9 10
1 2 3 4 1 3
5 6 7 8 2 5
1 2 3 8
post
9 10 11 12 9
4 7
10
5 6
DFS tree pre
0 9 10
1 2 3 4 1 3
5 6 7 8 2 5 4
1 2 3 8 11
post
9 10 11 12 9
4 7
10
5 6
DFS tree pre
0 9 10
1 2 3 4 1 3
5 6 7 8 2 5 4
1 2 3 8 11
post
9 10 11 12 9 8
4 7 12
10
5 6
DFS tree pre
0 9 10
1 2 3 4 1 3
5 6 7 8 2 5 4
1 2 3 8 11
post
9 10 11 12 9 8
4 7 12
10 7
5 6 13
DFS tree pre
0 9 10
1 2 3 4 1 3
5 6 7 8 2 5 4
1 2 3 8 11
post
9 10 11 12 9 8
4 7 12
10 7
5 6 13
11
14
DFS tree pre
0 9 10
1 2 3 4 1 3
5 6 7 8 2 5 4
1 2 3 8 11
post
9 10 11 12 9 8
4 7 12
10 7
5 6 13
11
14 15
DFS tree pre
0 9 10
1 2 3 4 1 3
5 6 7 8 2 5 4
1 2 3 8 11
post
9 10 11 12 9 8
4 7 12
10 7
5 6 13 16
11
14 15
DFS tree pre
0 9 10
1 2 3 4 1 3
5 6 7 8 2 5 4
1 2 3 8 11
post
9 10 11 12 9 8
4 7 12
10 7 12
5 6 13 16 17
11
14 15
DFS tree pre
0 9 10
1 2 3 4 1 3
5 6 7 8 2 5 4
1 2 3 8 11
post
9 10 11 12 9 8
4 7 12
10 7 12
5 6 13 16 17 18
11
14 15
DFS tree pre
0 9 10
1 2 3 4 1 3
5 6 7 8 2 5 4
1 2 3 8 11
post
9 10 11 12 9 8
4 7 12 19
10 7 12
5 6 13 16 17 18
11
14 15
DFS tree pre
0 9 10
1 2 3 4 1 3
5 6 7 8 2 5 4
1 2 3 8 11 20
post
9 10 11 12 9 8
4 7 12 19
10 7 12
5 6 13 16 17 18
11
14 15
DFS tree pre
0 9 10 21
1 2 3 4 1 3
5 6 7 8 2 5 4
1 2 3 8 11 20
post
9 10 11 12 9 8
4 7 12 19
10 7 12
5 6 13 16 17 18
11
14 15
DFS tree pre
0 9 10 21
1 2 3 4 1 3 6
5 6 7 8 2 5 4
1 2 3 8 11 20
post
9 10 11 12 9 8
4 7 12 19
10 7 12
5 6 13 16 17 18
11
14 15
DFS tree pre
0 9 10 21 22
1 2 3 4 1 3 6
5 6 7 8 2 5 4
1 2 3 8 11 20
post
9 10 11 12 9 8
4 7 12 19
10 7 12
5 6 13 16 17 18
11
14 15
DFS tree pre
0 9 10 21 22 23
1 2 3 4 1 3 6
5 6 7 8 2 5 4
1 2 3 8 11 20
post
9 10 11 12 9 8
4 7 12 19
10 7 12
5 6 13 16 17 18
11
14 15
DFS tree pre
0 9 10 21 22 23
1 2 3 4 1 3 6
5 6 7 8 2 5 4
1 2 3 8 11 20
post
9 10 11 12 9 8
4 7 12 19
10 7 12
5 6 13 16 17 18
Any non-tree edge generates a cycle
11
14 15
Directed cycles
2 1 3
5 6 4
7 8
Directed cycles
1
2 1 3
5 6 4
7 8
Directed cycles 0
1
2 1 3
5 6 4
7 8
Directed cycles 0
1
2 1 3
1
2
5 6 4
7 8
Directed cycles 0
1
2 1 3
1
2
5 6 4
2
5
7 8
Directed cycles 0
1
2 1 3
1
2
5 6 4
2
5
7 8
3
6
Directed cycles 0
1
2 1 3
1
2
5 6 4
2
5
7 8
3
6
4
7
Directed cycles 0
1
2 1 3
1
2
5 6 4
2
5
7 8
3
6
4 5
7
Directed cycles 0
1
2 1 3
1
2
5 6 4
2
5
7 8
3 6
6
4 5
7
Directed cycles 0
1
2 1 3
1
2
5 6 4
2
5
7 8
3 6 7
6 8
4 5
7
Directed cycles 0
1
2 1 3
1
2
5 6 4
2
5
7 8
3 6 7 8
6 8
4 5
7
Directed cycles 0
1
2 1 3
1
2
5 6 4
2 9
5
7 8
3 6 7 8
6 8
4 5
7
Directed cycles 0
1
2 1 3
1 10
2
5 6 4
2 9
5
7 8
3 6 7 8
6 8
4 5
7
Directed cycles 0
1
2 1 3
1 10 11
2 3
5 6 4
2 9
5
7 8
3 6 7 8
6 8
4 5
7
Directed cycles 0
1
2 1 3
1 10 11
2 3
5 6 4
2 9 12
5 4
7 8
3 6 7 8
6 8
4 5
7
Directed cycles 0
1
2 1 3
1 10 11
2 3
5 6 4
2 9 12 13
5 4
7 8
3 6 7 8
6 8
4 5
7
Directed cycles 0
1
2 1 3
1 10 11 14
2 3
5 6 4
2 9 12 13
5 4
7 8
3 6 7 8
6 8
4 5
7
Directed cycles 0 15
1
2 1 3
1 10 11 14
2 3
5 6 4
2 9 12 13
5 4
7 8
3 6 7 8
6 8
4 5
7
Directed cycles 0 15
1
2 1 3
1 10 11 14
2 3
5 6 4
2 9 12 13
5 4
7 8
3 6 7 8
6 8
Tree edge
4 5
7
Directed cycles 0 15
1
2 1 3
1 10 11 14
2 3
5 6 4
2 9 12 13
5 4
7 8
3 6 7 8
6 8
Tree edge
4 5
7
Directed cycles 0 15
1
2 1 3
1 10 11 14
2 3
5 6 4
2 9 12 13
5 4
7 8
3 6 7 8
6 8
Tree edge
Forward edge 4 5
7
Directed cycles 0 15
1
2 1 3
1 10 11 14
2 3
5 6 4
2 9 12 13
5 4
7 8
3 6 7 8
6 8
Tree edge
Forward edge 4 5
7
Directed cycles 0 15
1
2 1 3
1 10 11 14
2 3
5 6 4
2 9 12 13
5 4
7 8
3 6 7 8
6 8
Tree edge
Forward edge 4 5
Back edge 7
Directed cycles 0 15
1
2 1 3
1 10 11 14
2 3
5 6 4
2 9 12 13
5 4
7 8
3 6 7 8
6 8
Tree edge
Forward edge 4 5
Back edge 7
Directed cycles 0 15
1
2 1 3
1 10 11 14
2 3
5 6 4
2 9 12 13
5 4
7 8
3 6 7 8
6 8
Tree edge
Forward edge 4 5
Back edge 7
Cross edge
Directed cycles 0 15
1
2 1 3
1 10 11 14
2 3
5 6 4
2 9 12 13
5 4
7 8
3 6 7 8
6 8
Tree edge
Forward edge 4 5
Back edge 7
Cross edge
Directed cycles
A directed graph has a cycle if and only if DFS reveals a
back edge
Can classify edges using pre and post numbers
Tree/Forward edge (u,v) :
Interval [pre(u),post(u)] contains [(pre(v),post(v)]
Backward edge (u,v):
Interval [pre(v),post(v)] contains [(pre(u),post(u)]
Cross edge (u,v):
Intervals [(pre(u),post(u)] and [(pre(v),post(v)] disjoint
Directed acyclic graphs
Directed graphs without cycles are useful for
modelling dependencies
Courses with prerequisites
Edge (Algebra,Calculus) indicates that Algebra
is a prerequisite for Calculus
Will look at Directed Acyclic Graphs (DAGs) soon
Connectivity in directed
graphs
Need to take directions into account
Nodes i and j are strongly connected if there is a
path from i to j and a path from j to i
Directed graph can be decomposed into strongly
connected components (SCCs)
All pairs of nodes in an SCC are strongly
connected
Computing SCCs
2 1 3 DFS numbering (pre and
post) can be used to
compute SCCs
5 6 4
[Dasgupta,
7 8 Papadimitriou,Vazirani]
Computing SCCs
2 1 3 DFS numbering (pre and
post) can be used to
compute SCCs
5 6 4
[Dasgupta,
7 8 Papadimitriou,Vazirani]
Computing SCCs
2 1 3 DFS numbering (pre and
post) can be used to
compute SCCs
5 6 4
[Dasgupta,
7 8 Papadimitriou,Vazirani]
Computing SCCs
2 1 3 DFS numbering (pre and
post) can be used to
compute SCCs
5 6 4
[Dasgupta,
7 8 Papadimitriou,Vazirani]
Computing SCCs
2 1 3 DFS numbering (pre and
post) can be used to
compute SCCs
5 6 4
[Dasgupta,
7 8 Papadimitriou,Vazirani]
Other properties
A number of other structural properties can be
inferred from DFS numbering
Articulation points (vertices)
Removing such a vertex disconnects the graph
Bridges (edges)
Removing such an edge disconnects the graph
