Subgraphs

 A subgraph S of a graph
G is a graph such that
Depth-First Search  The vertices of S are a
subset of the vertices of G
 The edges of S are a Subgraph
A
subset of the edges of G
 A spanning subgraph of G
B D E
is a subgraph that
contains all the vertices
C
of G

Spanning subgraph

© 2010 Goodrich, Tamassia Depth-First Search 1 © 2010 Goodrich, Tamassia Depth-First Search 2

Connectivity Trees and Forests

 A (free) tree is an
 A graph is undirected graph T such
connected if there is that
a path between  T is connected
every pair of Connected graph  T has no cycles
Tree
This definition of tree is
vertices different from the one of
 A connected a rooted tree
component of a  A forest is an undirected
graph G is a graph without cycles
maximal connected  The connected
subgraph of G Non connected graph with two components of a forest
connected components are trees Forest

© 2010 Goodrich, Tamassia Depth-First Search 3 © 2010 Goodrich, Tamassia Depth-First Search 4
 Spanning Trees and Forests Depth-First Search
 A spanning tree of a  Depth-first search (DFS)  DFS on a graph with n
connected graph is a is a general technique vertices and m edges
spanning subgraph that is for traversing a graph takes O(n + m ) time
a tree
 A DFS traversal of a  DFS can be further
A spanning tree is not
graph G extended to solve other

unique unless the graph is
a tree Graph  Visits all the vertices and graph problems
 Spanning trees have edges of G  Find and report a path
applications to the design  Determines whether G is between two given
of communication connected vertices
networks  Computes the connected  Find a cycle in the graph
 A spanning forest of a components of G  Depth-first search is to
graph is a spanning Computes a spanning
subgraph that is a forest

graphs what Euler tour
forest of G
is to binary trees
Spanning tree
© 2010 Goodrich, Tamassia Depth-First Search 5 © 2010 Goodrich, Tamassia Depth-First Search 6

DFS Algorithm Example

The algorithm uses a mechanism A

A unexplored vertex
for setting and getting “labels” of Algorithm DFS(G, v)
vertices and edges Input graph G and a start vertex v of G
A visited vertex
B D E
Algorithm DFS(G) Output labeling of the edges of G unexplored edge
in the connected component of v
Input graph G
as discovery edges and back edges discovery edge C
Output labeling of the edges of G
as discovery edges and v.setLabel(VISITED) back edge
back edges for all e ∈ G.incidentEdges(v)
for all u ∈ G.vertices() if e.getLabel() = UNEXPLORED
u.setLabel(UNEXPLORED) w ← e.opposite(v) A A
for all e ∈ G.edges() if w.getLabel() = UNEXPLORED
e.setLabel(UNEXPLORED) e.setLabel(DISCOVERY)
B D E B D E
for all v ∈ G.vertices() DFS(G, w)
if v.getLabel() = UNEXPLORED else
DFS(G, v) e.setLabel(BACK) C C

© 2010 Goodrich, Tamassia Depth-First Search 7 © 2010 Goodrich, Tamassia Depth-First Search 8
 Example (cont.) DFS and Maze Traversal
A A  The DFS algorithm is
similar to a classic
B D E B D E strategy for exploring
a maze
C C  We mark each
intersection, corner
and dead end (vertex)
visited
 We mark each corridor
A A
(edge ) traversed
 We keep track of the
B D E B D E path back to the
entrance (start vertex)
by means of a rope
C C (recursion stack)
© 2010 Goodrich, Tamassia Depth-First Search 9 © 2010 Goodrich, Tamassia Depth-First Search 10

Properties of DFS Analysis of DFS

Property 1  Setting/getting a vertex/edge label takes O(1) time
DFS(G, v) visits all the  Each vertex is labeled twice
vertices and edges in  once as UNEXPLORED
the connected A  once as VISITED
component of v  Each edge is labeled twice
once as UNEXPLORED
Property 2 B D E


 once as DISCOVERY or BACK

The discovery edges
 Method incidentEdges is called once for each vertex
labeled by DFS(G, v)
form a spanning tree of C  DFS runs in O(n + m) time provided the graph is
the connected represented by the adjacency list structure
component of v  Recall that Σv deg(v) = 2m

© 2010 Goodrich, Tamassia Depth-First Search 11 © 2010 Goodrich, Tamassia Depth-First Search 12
 Path Finding Cycle Finding
 We can specialize the DFS  We can specialize the
Algorithm cycleDFS(G, v, z)
algorithm to find a path Algorithm pathDFS(G, v, z) v.setLabel(VISITED)
DFS algorithm to find a S.push(v)
between two given v.setLabel(VISITED)
for all e ∈ v.incidentEdges()
vertices u and z using the S.push(v) simple cycle using the
if e.getLabel() = UNEXPLORED
template method pattern if v = z template method pattern w ← e.opposite(v)
return S.elements() We use a stack S to
 We call DFS(G, u) with u for all e ∈ v.incidentEdges()
 S.push(e)
as the start vertex keep track of the path if w.getLabel() = UNEXPLORED
if e.getLabel() = UNEXPLORED
 We use a stack S to keep w ← e.opposite(v) between the start vertex e.setLabel(DISCOVERY)
pathDFS(G, w, z)
track of the path between if w.getLabel() = UNEXPLORED and the current vertex S.pop(e)
the start vertex and the e.setLabel(DISCOVERY)  As soon as a back edge else
current vertex T ← new empty stack
S.push(e) (v, w) is encountered,
 As soon as destination pathDFS(G, w, z) repeat
we return the cycle as o ← S.pop()
vertex z is encountered, S.pop(e)
we return the path as the else the portion of the stack T.push(o)
until o = w
contents of the stack e.setLabel(BACK) from the top to vertex w return T.elements()
S.pop(v) S.pop(v)

© 2010 Goodrich, Tamassia Depth-First Search 13 © 2010 Goodrich, Tamassia Depth-First Search 14