Graphs 2/21/2003 9:27 AM

Graphs
1843 ORD
SFO

802
43
17
337

1233
LAX DFW

Graphs version 1.3 1

Outline and Reading

Graphs (§6.1)
„ Definitions
„ Applications
„ Terminology
„ Properties
„ ADT
Data structures for graphs (§6.2)
„ Edge list structure
„ Adjacency list structure
„ Adjacency matrix structure

Graphs version 1.3 2

Graph
A graph is a pair (V, E), where
„ V is a set of nodes, called vertices
„ E is a collection of edges (pairs of vertices)
„ Vertices and edges are positions and store elements
Graphs useful for representing real-world relationships:
„ vertex = airport
„ edge = flight route, storing mileage
„ Abstract real-world problems into problems on graphs
849 PVD
1843 ORD 2
SFO 14
802

43 LGA
17
337

7
HNL 2555 138 10
99
1233
LAX DFW 1120
MIA
Graphs version 1.3 3

1
Graphs 2/21/2003 9:27 AM

Applications
cslab1a cslab1b

Electronic circuits math.brown.edu

„ Printed circuit board
„ Integrated circuit
cs.brown.edu
Transportation networks
„ Highway network brown.edu
„ Flight network qwest.net

Computer networks att.net

„ Local area network

„ Internet cox.net
„ Web John

Databases Paul
David

„ Entity-relationship diagram
Graphs version 1.3 4

Sample problems
What is cheapest way to fly from X to Y?
If airport X closes from bad weather, can I
still fly between every other pair of cities?
Many classes have prereqs; in what order can
I take the classes for my major?
How much traffic can flow between
intersection X and intersection Y
How can I minimize the amount of wiring
needed to connect some outlets together?

Graphs version 1.3 5

Edge Types
Directed edge
ordered pair of vertices (u,v)
„
flight
„ first vertex u is the origin ORD AA 1206 PVD
„ second vertex v is the destination
„ e.g., a flight
Undirected edge
unordered pair of vertices (u,v) 849
ORD PVD
„

„ e.g., a flight route miles

Directed graph
„ all the edges are directed
„ e.g., route network
Undirected graph
„ all the edges are undirected
„ e.g., flight network

Graphs version 1.3 6

2
Graphs 2/21/2003 9:27 AM

Terminology
End vertices (or endpoints) of an
edge
„ U and V are the endpoints of a V
Edges incident on a vertex a b
h j
„ a, d, and b are incident on V
Adjacent vertices U d X Z
U and V are adjacent
„
c e i
Degree of a vertex
„ X has degree 5 W g
Parallel edges (typically not used)
„ h and i are parallel edges f
Self-loop (typically not used) Y
„ j is a self-loop

Graphs version 1.3 7

Terminology (cont.)
Path
„ sequence of alternating vertices
and edges
V
„ begins and ends with some vertex a b
„ each edge is preceded and P1
followed by its endpoints d
Simple path
U X Z
P2 h
„ path such that all its vertices and c e
edges are distinct
Reachable W g
„ path exists
Examples
f
„ P1=(V,b,X,h,Z) is a simple path
Y
„ P2=(U,c,W,e,X,g,Y,f,W,d,V) is a
path that is not simple
„ Z is reachable from U
Graphs version 1.3 8

Terminology (cont.)
Cycle
„ circular sequence of alternating
vertices and edges
„ each edge is preceded and V
followed by its endpoints a b C3
„ edges traversed only in one
direction d
U X Z
Simple cycle C2 h
„ cycle such that all its vertices
c e
and edges are distinct
W g
Examples
„ C1=(V,b,X,g,Y,f,W,c,U,a,↵) is a C1
simple cycle f
„ C2=(U,c,W,e,X,g,Y,f,W,d,V,a,↵)
Y
is a cycle that is not simple
„ C3=(X,h,Z,h,X) is not a cycle

Graphs version 1.3 9

3
Graphs 2/21/2003 9:27 AM

Subgraphs
A subgraph S of a graph
G is a graph such that
„ The vertices of S are a
subset of the vertices of G
„ The edges of S are a Subgraph
subset of the edges of G
A spanning subgraph of G
is a subgraph that
contains all the vertices
of G

Spanning subgraph

Graphs version 1.3 10

Connectivity
A graph is
connected if there is
a path between
every pair of Connected graph
vertices
A connected
component of a
graph G is a
maximal connected
subgraph of G Non connected graph with two
connected components

Graphs version 1.3 11

Trees and Forests

A (free) tree is an
undirected graph T such
that
„ T is connected
„ T has no cycles Tree
Different definition than a
rooted tree
A forest is an undirected
graph without cycles
The connected
components of a forest
are trees
Forest

Graphs version 1.3 12

4
Graphs 2/21/2003 9:27 AM

Spanning Trees and Forests

A spanning tree of a
connected graph is a
spanning subgraph that is
a tree
A spanning tree is not
unique unless the graph is
a tree Graph
Spanning trees have
applications to the design
of communication
networks
A spanning forest of a
graph is a spanning
subgraph that is a forest
Spanning tree
Graphs version 1.3 13

Properties
Property 1 Notation
Σv deg(v) = 2m n number of vertices
Proof: each edge is m number of edges
counted twice deg(v) degree of vertex v
Property 2
In an undirected graph Example
(with no self-loops or
parallel edges) „ n = 4

m ≤ n (n − 1)/2 „ m = 6
Proof: at most one edge „ deg(v) = 3
for every unique
combination of 2 for all
vertices vertices
What is the bound on m
for a directed graph?Graphs version 1.3 14

Main Methods of the Graph ADT

Vertices and Edges Accessor methods
accessor method: „ Iterator vertices()
„ Object element() „ Iterator edges()
Update methods „ Iterator incidentEdges(v)
„ Vertex insertVertex(o) „ Vertex[2] endVertices(e)
„ Edge insertEdge(v, w, o) „ Vertex opposite(v, e)
„ void removeVertex(v) „ boolean areAdjacent(v, w)
„ void removeEdge(e) Methods for directed edges
„ Vertex origin(e)
Accessor methods „ Vertex destination(e)
„ int numVertices() „ boolean isDirected(e)
„ int numEdges() „ Edge insertDirectedEdge(v, w, o)

„ Vertex aVertex()

Graphs version 1.3 15

5
Graphs 2/21/2003 9:27 AM

Edge List Structure

Vertex object u
„ element a c
Edge object b d
v w z
„ element
„ origin vertex object
„ destination vertex object
„ directed boolean flag
Vertex sequence u v w z
„ sequence of vertex
objects
Edge sequence a b c d
„ sequence of edge objects

Graphs version 1.3 16

Graph ADT with Positions

Recall
„ Position = place where item is stored in a sequence
In Goodrich’s book:
„ A Vertex is a Position
„ An Edge is a Position
Features of Positions
„ Enables faster removal
„ Implementation slightly more complex
„ Unnecessary when removeVertex and removeEdge are not
used

Graphs version 1.3 17

Edge List Structure (w/ Positions)

Vertex object u
„ element a c
„ reference to position in b d
vertex sequence v w z
Edge object
„ element
„ origin vertex object
„ destination vertex object
„ reference to position in u v w z
edge sequence
Vertex sequence
„ sequence of vertex a b c d
objects
Edge sequence
„ sequence of edge objects

Graphs version 1.3 18

6
Graphs 2/21/2003 9:27 AM

Adjacency List Structure

a v b
Edge list
u w
structure
Each Vertex now
stores incidence
sequence u v w
„ sequence of
references to
edge objects of
incident edges

a b

Graphs version 1.3 19

Adjacency List Structure (w/

Positions)
Edge list structure a v b
(w/ Positions) u w
Incidence sequence
for each vertex
„ sequence of
references to edge
objects of incident u v w
edges
Augmented edge
objects
„ references to
associated
positions in
incidence a b
sequences of end
vertices

Graphs version 1.3 20

Adjacency Matrix Structure

Edge list structure a v b
Augmented vertex u w
objects
„ Integer key (index)
associated with vertex
2D-array adjacency
array 0 u 1 v 2 w
„ Reference to edge
object for adjacent
0 1 2
vertices
„ Null for non 0 ∅ ∅
nonadjacent vertices 1 ∅
The “old fashioned” a ∅ ∅ b
2
version just has 0 for
no edge and 1 for edge

Graphs version 1.3 21

7
Graphs 2/21/2003 9:27 AM

Asymptotic Performance
n vertices, m edges Edge Adjacency Adjacency
List List Matrix
Space O(n + m) O(n + m) O(n2 )
Iterating through
O(m) O( deg(v) ) O(n)
incidentEdges(v)
O( min(deg(v),
areAdjacent (v, w) O(m) O(1)
deg(w)) )
insertVertex(o) O(1) O(1) O(n2 )
insertEdge(v, w, o) O(1) O(1) O(1)
removeVertex(v) O(m) O( deg(v) ) O(n2 )
removeEdge(e) O(1) O(1) O(1)
Notes: Assuming no parallel edges or self-loops
Using Positions (for removeVertex and removeEdge)
Graphs version 1.3 22

Depth-First Search
A

B D E

Graphs version 1.3 23

Outline and Reading

Depth-first search (§6.3.1)
„ Algorithm
„ Example
„ Properties
„ Analysis
Applications of DFS (§6.5)
„ Path finding
„ Cycle finding

Graphs version 1.3 24

8
Graphs 2/21/2003 9:27 AM

Depth-First Search
Depth-first search (DFS) is
„ general graph traversal technique
„ visits all the vertices and edges of G
„ with n vertices and m edges takes O(n + m ) time
„ a recursive traversal like Euler tour for binary trees
A DFS traversal of a graph G can be used to
„ Determines whether G is connected
„ Computes the connected components of G
„ Computes a spanning forest of G
„ Find and report a path between two given vertices
„ Find a cycle in the graph
Graphs version 1.3 25

Example
A
A unexplored vertex
A visited vertex
B D E
unexplored edge
discovery edge C
back edge

A A

B D E B D E

C C

Graphs version 1.3 26

Example (cont.)
A A

B D E B D E

C C

A A

B D E B D E

C C

Graphs version 1.3 27

9
Graphs 2/21/2003 9:27 AM

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

Graphs version 1.3 28

DFS and Maze Traversal

The DFS algorithm is
similar to a classic
strategy for exploring
a maze
„ We mark each
intersection, corner
and dead end (vertex)
visited
„ We mark each corridor
(edge ) traversed
„ We keep track of the
path back to the
entrance (start vertex)
by means of a rope
(recursion stack)
Graphs version 1.3 29

Another Example of Depth

First Search
2
1 6

3
5
7
4

Graphs version 1.3 30

10
Graphs 2/21/2003 9:27 AM

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

Graphs version 1.3 31

Analysis of DFS
Setting/getting a vertex/edge label takes O(1) time
Each vertex is labeled twice
„ once as UNEXPLORED
„ once as VISITED
Each edge is labeled twice
„ once as UNEXPLORED
„ once as DISCOVERY or BACK
DFS(G, v) called once for each vertex v
Inner loop in DFS(G, v) runs in O(deg(v)) time
„ Not counting time inside recursive calls
„ Assuming adjacency list implementation
DFS runs in O(n + m) time
„ Recall that Σv deg(v)Graphs
= 2m
version 1.3 32

Properties of DFS
Property 1
DFS(G, v) visits all the
vertices and edges in the
connected component of v A
Property 2
The discovery edges labeled
by DFS(G, v) form a B D E
spanning tree of the
connected component of v
called DFS tree, rooted at v
C
Property 3
DFS_Sweep(G) visits all
vertices and edges of G

Graphs version 1.3 33

11
Graphs 2/21/2003 9:27 AM

Connected Components &

DFS Spanning Forest
Algorithm ccDFS_Sweep(G)
Input graph G
Use DFS(G, v) to label all
Output labeling of the vertices
and edges of G based on
edges and vertices in one
component number. connected component
for all u ∈ G.vertices() (property 1)
setLabel(u, UNEXPLORED) DFS_Sweep can label all
for all e ∈ G.edges() connected components
setLabel(e, UNEXPLORED) (property 3)
comp_num ← 1 In a DFS_Sweep call, consider
for all v ∈ G.vertices() subgraph of
if getLabel(v) = UNEXPLORED
„ all vertices
perform DFS(G, v) search,
„ all DISCOVERY Edges
labeling vertices and edges
found in the search with By DFS property 2 and 3, this
comp_num. subgraph is a Spanning
comp_num ← comp_num +1 Forest of G.
Graphs version 1.3 34

More Properties of DFS

Classifying Edges by DFS
Edge (v,w) type:
tree = in the DFS tree
back = w is ancestor of v
in DFS tree A
forward = w is descendent
of v in DFS tree
cross = w is neither ancestor B D E
nor descendant of v
in DFS tree
(Assuming edge first explored from
v to w). C
Property 4: edges labeled BACK
are in fact back edges
Property 5: back edges form a
cycle
Graphs version 1.3 35

Path Finding
Specialize DFS to find a path
between two given vertices v Algorithm pathDFS(G, v, z)
and z setLabel(v, VISITED)
S.push(v)
Call DFS(G, v, z) where
if v = z
„ G is the graph return S.elements()
„ v is the start vertex for all e ∈ G.incidentEdges(v)
„ z is the destination vertex if getLabel(e) = UNEXPLORED
Use a stack S to keep track w ← opposite(v,e)
of the path between the if getLabel(w) = UNEXPLORED
start vertex and the current setLabel(e, DISCOVERY)
vertex S.push(e)
As soon as destination pathDFS(G, w, z)
vertex z is encountered, we S.pop(e)
return the path as the else
contents of the stack setLabel(e, BACK)
S.pop(v)
Graphs version 1.3 36

12
Graphs 2/21/2003 9:27 AM

Cycle Finding
Algorithm cycleDFS(G, v)
Specialize DFS to find a setLabel(v, VISITED)
simple cycle S.push(v)
We use a stack S to for all e ∈ G.incidentEdges(v)
if getLabel(e) = UNEXPLORED
keep track of the path w ← opposite(v,e)
between the start vertex S.push(e)
and the current vertex if getLabel(w) = UNEXPLORED
setLabel(e, DISCOVERY)
As soon as a back edge cycleDFS(G, w)
(v, w) is encountered, S.pop(e)
we return the cycle as else
T ← new empty stack
the portion of the stack repeat
from the top to vertex w o ← S.pop()
T.push(o)
until o = w
return T.elements()
S.pop(v)
Graphs version 1.3 37

abstract DFS Template

Algorithm DFS(G, v)
Use Template Method setLabel(v, VISITED)
Design Pattern to startVisit(v)
implement DFS for all e ∈ G.incidentEdges(v)
Extend template to if getLabel(e) = UNEXPLORED
implement any algorithm w ← opposite(v,e)
if getLabel(w) = UNEXPLORED
that uses DFS.
setLabel(e, DISCOVERY)
Extensions need to traverseDiscovery(e)
define the following: if (not isDone())
„ startVisit() DFS(G, w)
„ traverseDiscovery() else
setLabel(e, BACK)
„ traverseBack()
traverseBack(e)
„ isDone() finishVisit(v)
„ finishVisit()
„ a method that returns
results.

Graphs version 1.3 38

13