CS2040 – Data Structures

and Algorithms
Lecture 12 – The Foundations ~ Graphs
[email protected]
 Journey so far
• Sorting

• Lists With Thanks to

• HashTable Prof Ket Fah Chong

• Binary Heap Prof Roger Zimmermann

• UFDS

• Ordered Map
2
What do you remember thus far? ☺
• Go to: https://menti.com (code: 29 66 92 5)

3
 Road Ahead

• Graphs, graphs and more graphs ☺

• Lots of very cool algorithms ☺

4
Outline of this Lecture
A. Motivation on why you should learn graph
• Graph terminologies

B. Three Graph Data Structures

• Adjacency Matrix
• Adjacency List
• Edge List
• https://visualgo.net/en/graphds

C. Some Graph Data Structure Applications

D. This lecture is setup for the rest of the module on graph DSAs
5
Graph Terminologies (1)
• Extension from what you already know: (Binary) Tree
• Vertex, Edge, Direction (of Edge), Weight (of Edge)

• But in a general graph, there is no notion of

• Root
• Parent/Child
• Ancestor/Descendant

6
Graph (?)
15

6 50

7 23 77
4

7
Graph (?)
7 23

6 50

15
77
4

5
8
 Note: definitions here might be
Graph is… a bit different from CS1231/S

• Graph is a set of N vertices where some [0 .. NC2] pairs of the vertices

are connected by edges (3 types – undirected, directed, bi-directed)
• We will ignore “multi graph” where there can be more than
one edge
5 (of any edge type) between a pair of vertices
Weighted Undirected Edge
Weighted Bi-directed Edge
7
0 1 2 (made of 2 directed edges
going in opposite directions)
Undirected Edge
Thus vertex 0 and vertex 3 2 A graph with a combination
are called “adjacent” or 3 4 5 of undirected+directed edge
they are neighbours of each 3 is called a mixed graph
other
Directed Edge
A vertex, usually labeled from 0 to N-1 9
Example

10
Graph Terminologies (2)
• Sparse/Dense
Sparse Graph
• Sparse = not so many edges
• Dense = many edges
• No guideline for “how many” In/out degree
of vertex 5 = 3

• Complete Graph
• Simple graph with N vertices Dense Graph
and NC2 edges
Complete Graph
7 vertices,
• In/Out Degree of a vertex 7C2 = 21 edges

• Number of in/out edges

from a vertex
11
Graph Terminologies (3)
• (Simple) Path
• Sequence of vertices connected by a sequence of undirected edges
• Simple = no repeated vertex
• A path with only 1 vertex and no edge is an empty path

• (Simple) Directed Path

• Same as (Simple) Path with the added restriction that the edges in the
path are directed and in the same direction

• Path Length/Cost
• In unweighted graph, usually number of edges in the path
• In weighted graph, usually sum of edge weight in the path
12
Graph Terminologies (4)
• (Simple) Cycle
• Path that starts and ends with the same vertex and with no repeated
vertices except start/end vertex and no repeated edges

• Involves 3 or more unique vertices

• (Simple) Directed Cycle

• Same as (Simple) Cycle with the added restriction that the edges in the
cycle are directed and in the same direction

• Involves 2 or more unique vertices

13
Transportation Network

Simple cycle
Sengkang LRT, west loop)

Simple path (from Clementi to

Outram Park MRT) with length
7 (in terms of number of hops)

14
Internet/Computer/Communication Networks

15
Graph Terminologies (5)
• Component 3 components in this graph
• A maximal group of vertices in an • Disconnected graph (since it has > 1
undirected graph that can visit component)
each other via some path • Vertices 1-2-3-4 are reachable from
vertex 0
• Connected graph
• Vertices 5, 6-7-8 are unreachable from
• Undirected graph with 1 component
vertex 0
• Reachable/Unreachable Vertex • {7-6-8 5} is a sub graph of this graph
• See example

• Sub Graph
• Subset of vertices (and their
connecting edges) of the
original graph
16
Graph Terminologies (6)
• Directed Acyclic Graph (DAG) Out degree of vertex 0 = 2
• Directed graph that has no cycle

• Tree (bottom left)

• Connected graph – one
In degree of vertex 2 = 2
unique path between any pair
of vertices

• Bipartite Graph (bottom right)

• Undirected graph where we can so that
partition the vertices into two sets so
that there are no edges between
members of the same set
17
Graph Data Structures

https://visualgo.net/en/graphds

18
Adjacency Matrix
• A 2D array (AdjMatrix)

• AdjMatrix[i][j] = 1, if there exist an edge i ➔ j in G, otherwise 0

• For weighted graph, AdjMatrix[i][j] contains the weight of edge i ➔

j, not just binary values {1, 0}

• Space Complexity (V = number of vertices in G)

• O(V2)

19
Example of Adjacency Matrix

3 4 5 Adjacency Matrix
0 1 2 3 4 5 6
0 0 1 1 0 0 0 0
1 1 0 1 1 0 0 0
2 1 1 0 0 1 0 0
1 2 6 3 0 1 0 0 1 0 0
4 0 0 1 1 0 1 0
5 0 0 0 0 1 0 1
0 6 0 0 0 0 0 1 0

20
Adjacency List
• An array of V lists (AdjList)
• One element for each vertex

• For each vertex i, AdjList[i] = list of i’s neighbours

• For weighted graph, stores pair (neighbour, weight)
• Can use same strategy for unweighted graph
• Connected to neighbour? Set weight to 1 (unit weight), 0 otherwise

• Space Complexity (E = number of edges in G)

• O(V + E) (E = O(V2))

21
Example of Adjacency List

3 4 5
Adjacency List
0: 1 2
1: 0 2 3
2: 0 1 4
1 2 6 3: 1 4
4: 2 3 5
5: 4 6
6: 5
0

22
Edge List
• An array of E edges (EdgeList)
• One element for each edge

• For each edge i, EdgeList[i] = integer triple {u, v, w(u, v)}

• For unweighted graph, the weight can be stored as 0 (or 1), or

simply store an (integer) pair
• Space Complexity: O(E)

23
Example of Edge List
7 9 Edge List
3 4 5
0: 0 1 1
1: 0 2 4
5 5 3
2: 1 2 4
3: 1 3 5
4
1 2 6 4: 2 4 5
5: 3 4 7
1 4 6: 4 5 9

0 7: 5 6 3

24
Java Implementation (1)
Adjacency Matrix: Simple built-in 2D array
int V = NUM_V; // NUM_V has been set before
int[][] AdjMatrix = new int[V][V];

Adjacency List: With Java Collections framework

ArrayList < ArrayList < IntegerPair > >
AdjList = new ArrayList < ArrayList < IntegerPair > >();
// IntegerPair is a simple integer pair class
// to store pair info, see the next slide

Edge List: Also, with Java Collections framework

ArrayList < IntegerTriple > EdgeList =
new ArrayList < IntegerTriple >();
// IntegerTriple is similar to IntegerPair

PS: This is one implementation, there are other ways ☺

25
Graph Data Structures

What can we do with them?

Some basic calculations for now, but more later ☺

26
Counting Vertices

Adjacency Matrix Edge List

Adjacency List
0 1 2 3 4 5 6 0: 0 1 1
0: 1 2
0 0 1 1 0 0 0 0 1: 0 2 4
1: 0 2 3
1 1 0 1 1 0 0 0 2: 1 2 4
2: 0 1 4
2 1 1 0 0 1 0 0 3: 1 3 5
3: 1 4
3 0 1 0 0 1 0 0 4: 2 4 5
4: 2 3 5
4 0 0 1 1 0 1 0 5: 3 4 7
5: 4 6
5 0 0 0 0 1 0 1 6: 4 5 9
6: 5
6 0 0 0 0 0 1 0 7: 5 6 3

27
Counting Vertices
• Trivial for both AdjMatrix and AdjList: V ➔ number of rows!

• Sometimes this number is stored in separate variable so that

we do not have to re-compute this every time, that is, O(1),
especially if the graph never changes after it is created

• To think about: How about EdgeList?

28
Enumerating Neighbours

Adjacency Matrix Edge List

Adjacency List
0 1 2 3 4 5 6 0: 0 1 1
0: 1 2
0 0 1 1 0 0 0 0 1: 0 2 4
1: 0 2 3
1 1 0 1 1 0 0 0 2: 1 2 4
2: 0 1 4
2 1 1 0 0 1 0 0 3: 1 3 5
3: 1 4
3 0 1 0 0 1 0 0 4: 2 4 5
4: 2 3 5
4 0 0 1 1 0 1 0 5: 3 4 7
5: 4 6
5 0 0 0 0 1 0 1 6: 4 5 9
6: 5
6 0 0 0 0 0 1 0 7: 5 6 3

O(V) O(k)

29
Enumerating Neighbours
• O(V) for AdjMatrix: scan AdjMatrix[v][j], j  [0..V-1]

• O(k) for AdjList: scan AdjList[v]

• k is the number of neighbours of vertex v (output-sensitive algorithm)

• This is an important difference between AdjMatrix versus AdjList

• It affects the performance of many graph algorithms. Remember this!

• Usually, the neighbours are listed in increasing vertex number

• Again, what about EdgeList?

30
Counting Edges

Adjacency Matrix Edge List

Adjacency List
0 1 2 3 4 5 6 0: 0 1 1
0: 1 2
0 0 1 1 0 0 0 0 1: 0 2 4
1: 0 2 3
1 1 0 1 1 0 0 0 2: 1 2 4
2: 0 1 4
2 1 1 0 0 1 0 0 3: 1 3 5
3: 1 4
3 0 1 0 0 1 0 0 4: 2 4 5
4: 2 3 5
4 0 0 1 1 0 1 0 5: 3 4 7
5: 4 6
5 0 0 0 0 1 0 1 6: 4 5 9
6: 5
6 0 0 0 0 0 1 0 7: 5 6 3

O(V2) O(V) O(1)

31
Counting Edges
• O(1) for EdgeList
• Undirected/Bidirected edges may be listed once (or twice) in EdgeList,
depending on the need

• O(V2) for AdjMatrix: count non-zero entries in AdjMatrix

• O(V) for AdjList: sum the length of all V lists

• Sometimes this number is stored in separate variable so that

we do not have to re-compute this every time, i.e. O(1),
especially if the graph never changes after it is created
32
Existence of Edges
• Given vertices u and v, does edge(u,v) exist?

• O(1) for AdjMatrix: see if AdjMatrix[u][v] is non zero

• O(k) for AdjList: see if AdjList[u] contains v

• How about EdgeList?

33
Trade-Off
Adjacency Matrix Adjacency List

• Pros • Pros
• Existence of edge i – j can be • O(k) to enumerate k neighbours
found in O(1) of a vertex
• Good for dense graph/ Floyd • Good for sparse graph/Dijkstra’s/
Warshall’s DFS/BFS, O(V+E) space
• Cons • Cons
• O(V) to enumerate neighbours of • O(k) to check the existence of
a vertex edge i – j
• O(V2) space • A small overhead in maintaining
the list (for sparse graph)

34
Live Quiz ☺
• On Zoom

• And a question asking your opinion

35
 Feedback
https://forms.office.com/r/jcLS2bxjth

36