Data Structures and Algorithms

(ESO207)

Lecture 26
• Depth First Search (DFS) Traversal
• DFS Tree
• Novel application: computing biconnected components of a graph

1
 DFS traversal of G
DFS(v)
{ Visited(v)  true; DFN[v]  dfn ++;
For each neighbor w of v
{ if (Visited(w) = false)
{ DFS(w) ;
……..;
}
……..;
}
}

DFS-traversal(G)
{
dfn  0;
For each vertex vϵ V { Visited(v) false }
For each vertex v ϵ V { If (Visited(v ) = false) DFS(v) }
}
2
 Depth First Numbering (DFN)

DFN[x] :
y v The number at which x gets visited during DFS
1 0
traversal.
10 w
f 2
3
b
4h 12 11 d
g
5 r
u
8

6 9z
s 7
c

3
DFS tree

4
 DFS(v) computes a tree rooted at v

y v
Can
Is any tree
a DFS rooted tree
unique
be obtained through
for a graph ?
w
f DFS ?
b
h d No
No
g

u r

z
s
c

A DFS tree rooted at v

5
How will an edge appear in DFS traversal ?

• as a tree-edge.

If the edge is a non-tree edge :

• Edge between ancestor and descendant in

DFS tree.

any other possibility ?

No

6
How will an edge appear in DFS traversal ?

It can never happen

x
y

7
How will an edge appear in DFS traversal ?

x
y

8
How will an edge appear in DFS traversal ?

y v

w
f
b
h d
g

x
y
z
s
c

9
 How will an edge appear in DFS traversal ?

A short proof:
Let (x,y) be a non-tree edge.
Let x get visited before y.
Question:
If we remove all vertices visited prior to x, does y still lie in the connected
component of x ?
Answer: yes.

DFS pursued from x will have a path to y in DFS tree.
Hence x must be ancestor of y in the DFS tree.

10
 Always remember
the following picture for DFS traversal

non-tree edge  back edge

This is called DFS representation of the graph.

It plays a key role in the design of every
efficient algorithm.

11
A novel application of DFS traversal

Determining if a graph G is biconnected

12
Definition: A connected graph is said to be biconnected
if there does not exist any vertex whose removal disconnects the graph.

Motivation: To design robust networks

(immune to any single node failure).

13
 Is this graph biconnected ?
y v

w No.
f
b
h d
g

u r

z
s
c

14
 A trivial algorithms for checking
bi-connectedness of a graph

• For each vertex v, determine if G\{v} is connected

(One may use either BFS or DFS traversal here)

Time complexity of the trivial algorithm : O()

15
An O() time algorithm

A single DFS traversal

16
 An O() time algorithm

• A formal characterization of the problem.

(articulation points)

• Exploring relationship between articulation point & DFS tree.

• Using the relation cleverly to design an efficient algorithm.

17
 This graph is NOT biconnected
y v

w
f
b
h d
g
The removal of any of {v,f,u} can destroy
u r connectivity.

z v,f,u are called the articulation points of G.

s
c

18
 A formal definition of articulaton point

Definition: A vertex x is said to be articulation point if

u,v different from x
such that every path between u and v passes through x.

u x v

Observation: A graph is biconnected if none of its vertices is an articulation point.

AIM:
Design an algorithm to compute all articulation points in a given graph.

19
Articulation points and DFS traversal

Don’t Focus on the graph. Instead,

focus on the DFS tree representation of the graph.

20
 Some observations

v
Question: When can a leaf node be an a.p. ?
Answer: Never

Question: When can root be an a.p. ?

21
 Some observations

v
Question: When can a leaf node be an a.p. ?
Answer: Never

Question: When can root be an a.p. ?

22
 Some observations

v
Question: When can a leaf node be an a.p. ?
Answer: Never

Question: When can root be an a.p. ?

Answer: Iff it has two or more children.

23
 Some observations

u x v

AIM:
To find necessary and sufficient conditions for an internal node to be articulation point.

How will x look like

in DFS tree ?

24
 conditions for an internal node to be articulation
point.
root

Where will u and v

be relative to x ?
At least one of u and v
must be descendant of x. x

25
Case 1: Exactly one of u and v is a descendant of x in DFS tree
root

v
y u—w—y—v :
Can there be a a u-v path not passing through x
back edge from
T1 to ancestor of
x?
x

No

u
w

T1 26
 Case 2: both u and v are descendants of x in DFS tree
root

Can u and v belong to

T1 simultaneously? v u

No T1 27
 Case 2: both u and v are descendants of x in DFS tree
root

z u—w—z—y—q—v :
a u-v path not passing through x
y
At least one of T1 and
T2 have no back edge x
to ancestor of x

u v
w q

T1 T2 28
 Necessary condition for x to be articulation point
root

Necessary condition:
x has at least one child y s.t.
there is no back edge
v from subtree(y) to ancestor of x.

x
Is this condition
u y sufficient also ?

yes.

29
 Articulation points and DFS
Let G=(V,E) be a connected graph.
Perform DFS traversal from any graph and get a DFS tree T.
• No leaf of T is an articulation point.
• root of T is an articulation point if and only if it has more than one child.
• For any internal node … ??

Theorem1 : An internal node x is articulation point if and only if

it has a child y such that
there is no back edge
from subtree(y) to any ancestor of x.

30
Efficient algorithm for Articulation points

Use Theorem 1
Exploit recursive nature of DFS

Ponder over it before

31
next lecture 