Preparation Data Structures 11 graphs

Data StructuresGraphs

Andres Mendez-Vazquez

May 8, 2015

1 / 91

Images/cinvestav-1.jpg

Outline1 Graphs

Graphs EverywhereHistoryBasic Theory

2 Graph RepresentationMatrix RepresentationPossible Code for This RepresentationAdjacency List Representation

3 Traversing the GraphBreadth-first searchDepth-First Search

4 ApplicationsFinding a path between nodesConnected ComponentsSpanning TreesTopological Sorting

2 / 91

Outline1 Graphs

3 / 91

We are full of Graphs

4 / 91

State Machines for Branch Prediction in CPU’s

5 / 91

Social Networks

6 / 91

Outline1 Graphs

7 / 91

History

Something NotableGraph theory started with Euler who was asked to find a nice path acrossthe seven Königsberg bridges

The Actual City

8 / 91

HistorySomething NotableGraph theory started with Euler who was asked to find a nice path acrossthe seven Königsberg bridges

The Actual City

8 / 91

There is No-Solution

Using GraphsThe (Eulerian) path should cross over each of the seven bridges exactlyonce

Can you find Any possible solution?

9 / 91

There is No-Solution

Using GraphsThe (Eulerian) path should cross over each of the seven bridges exactlyonce

Can you find Any possible solution?

9 / 91

Necessary Condition

Euler discovered thatA necessary condition for the walk of the desired form is that the graph beconnected and have exactly zero or two nodes of odd degree.

Add an extra bridge

10 / 91

Necessary Condition

Euler discovered thatA necessary condition for the walk of the desired form is that the graph beconnected and have exactly zero or two nodes of odd degree.

Add an extra bridge

Add Extra Bridge

10 / 91

Studying Graphs

All the previous examples are telling usData Structures are required to design structures to hold the informationcoming from graphs!!!

Good RepresentationsThey will allow to handle the data structures with ease!!!

11 / 91

Studying Graphs

All the previous examples are telling usData Structures are required to design structures to hold the informationcoming from graphs!!!

Good RepresentationsThey will allow to handle the data structures with ease!!!

11 / 91

Outline1 Graphs

12 / 91

Basic TheoryDefinitionA Graph is composed of the following parts: Nodes and Edges

NodesThey can represent multiple things:

PeopleCitiesStates of Beingetc

EdgesThey can represent multiple things too:

Distance between citiesFriendshipsMatching StringsEtc

13 / 91

Basic Theory

DefinitionA graph G = (V ,E ) consists of a set of vertices (or nodes) V and a set ofedges E , where we assume that they are finite i.e. |V | = n and |E | = m.

Example

14 / 91

Basic Theory

DefinitionA graph G = (V ,E ) consists of a set of vertices (or nodes) V and a set ofedges E , where we assume that they are finite i.e. |V | = n and |E | = m.

Example

14 / 91

Properties

IncidentAn edge ek = (vi , vj) is incident with the vertices vi and vj .

A simple graph has no self-loops or multiple edges like below

15 / 91

Properties

IncidentAn edge ek = (vi , vj) is incident with the vertices vi and vj .

A simple graph has no self-loops or multiple edges like below

15 / 91

Some properties

DegreeThe degree d(v) of a vertex V is its number of incident edges

A self loopA self-loop counts for 2 in the degree function.

PropositionThe sum of the degrees of a graph G = (V ,E ) equals 2|E | = 2m (trivial).

16 / 91

Some properties

16 / 91

Some properties

16 / 91

CompleteA complete graph Kn is a simple graph with all n(n−1)/2 possible edges, likethe graph below for n = 2, 3, 4, 5.

Example

17 / 91

CompleteA complete graph Kn is a simple graph with all n(n−1)/2 possible edges, likethe graph below for n = 2, 3, 4, 5.

Example

17 / 91

We need NICE representations

First OneMatrix Representation

Second OneAdjacency Representation

18 / 91

We need NICE representations

First OneMatrix Representation

Second OneAdjacency Representation

18 / 91

Outline1 Graphs

19 / 91

Adjacency Matrix Representation

This is the simplest oneIf we number the nodes of an undirected graph:

20 / 91

In a natural way the edges can be identified by the nodesFor example, the edge between 1 and 4 nodes gets named as (1,4)

ThenHow, we use this to represent the graph through a Matrix or and Array ofArrays??!!!

21 / 91

In a natural way the edges can be identified by the nodesFor example, the edge between 1 and 4 nodes gets named as (1,4)

ThenHow, we use this to represent the graph through a Matrix or and Array ofArrays??!!!

21 / 91

What about the following?

How do we indicate that an edge exist given the following matrix1 2 3 4 5 6

1 − − − − − −2 − − − − − −3 − − − − − −4 − − − − − −5 − − − − − −6 − − − − − −

You say it!!Use a 0 for no-edgeUse a 1 for edge

22 / 91

What about the following?

How do we indicate that an edge exist given the following matrix1 2 3 4 5 6

1 − − − − − −2 − − − − − −3 − − − − − −4 − − − − − −5 − − − − − −6 − − − − − −

You say it!!Use a 0 for no-edgeUse a 1 for edge

22 / 91

We have then...

Definition0/1 N × N matrix with N =Number of nodes or verticesA(i , j) = 1 iff (i , j) is an edge

23 / 91

We have then...For the previous example

1 2 3 4 5 61 0 0 0 1 0 02 0 0 0 1 0 03 0 0 0 1 0 04 1 1 1 0 1 05 0 0 0 1 0 16 0 0 0 0 1 0

24 / 91

Properties of the Matrix for Undirected Graphs

Property OneDiagonal entries are zero.

Property TwoAdjacency matrix of an undirected graph is symmetric:

A (i , j) = A (j , i) for all i and j

25 / 91

Properties of the Matrix for Undirected Graphs

Property OneDiagonal entries are zero.

Property TwoAdjacency matrix of an undirected graph is symmetric:

A (i , j) = A (j , i) for all i and j

25 / 91

What about direct Graphs!!!

Similar ideaUse a 0 for no-edgeUse a 1 for directed edge

26 / 91

Example

We have that

1 2 3 4 5 6 71 0 1 1 1 0 0 02 0 0 0 1 1 0 03 0 0 0 0 0 1 04 0 0 1 0 0 1 15 0 0 0 1 0 0 16 0 0 0 0 0 0 07 0 0 0 0 0 1 0

27 / 91

Outline1 Graphs

28 / 91

What about the code?Partial Codep u b l i c c l a s s Graph{

// P r i v a t e s t o r a g ep r i v a t e boo l ean a d j a c e n c y M a t r i x [ ] [ ] ;p r i v a t e i n t c e r t e xCount ;

// C o n s t r u c t o rp u b l i c Graph ( i n t ve r t exCount ) {

t h i s . v e r t exCount = ve r t exCount ;a d j a c e n c y M a t r i x = new

boo l ean [ ve r t exCount ] [ v e r t exCount ] ;}

//Some Methodsp u b l i c v o i d addEdge ( i n t i , i n t j ) {

i f ( i >= 0 && i < ve r t exCount && j > 0&& j < ve r t exCount ) {

a d j a c e n c y M a t r i x [ i ] [ j ] = t r u e ;a d j a c e n c y M a t r i x [ j ] [ i ] = t r u e ;}

}p u b l i c v o i d removeEdge ( i n t i , i n t j ) {

i f ( i >= 0 && i < ve r t exCount && j > 0&& j < ve r t exCount ) {

a d j a c e n c y M a t r i x [ i ] [ j ] = f a l s e ;a d j a c e n c y M a t r i x [ j ] [ i ] = f a l s e ;

29 / 91

What about Time Complexity?

Operations on a GraphMost of the basic operations in a graph are:

Adding an edge – O(1)

Deleting an edge – O(1)

Answering the question “is there an edge between i and j” – O(1)

Finding the successors of a given vertex – O(N)

Finding (if exists) a path between two vertices – O(N2)

30 / 91

Space Drawbacks of This Representation

We need the following amount of spaceIf you have N integers of 4 bytes each, we requiere

4× N × N = 4N2

Which is a killer!!!If your graph does not have an edge between any two pair of nodes

SoWhat to do?

31 / 91

4× N × N = 4N2

SoWhat to do?

31 / 91

4× N × N = 4N2

SoWhat to do?

31 / 91

Possible Solutions

If you have an undirected graphFor an undirected graph, may store only lower or upper triangle(exclude diagonal).Space used (Assume Integers): 4× N(N−1)

2 = 2N (N − 1)

BetterUse Sparse Matrix Representations

32 / 91

Possible Solutions

2 = 2N (N − 1)

32 / 91

Possible Solutions

2 = 2N (N − 1)

32 / 91

We use the sparse Representation of Matrices!!!

Example: Array of Row Chains

33 / 91

We use the sparse Representation of Matrices!!!

Example: Orthogonal Lists

34 / 91

Code For Sparse Representation

Node and Link Representationsp u b l i c c l a s s Node {

p r i v a t e Node next ;p r i v a t e i n t c o l ;

p r i v a t e i n t Value

// Node c o n s t r u c t o rp u b l i c Node ( ){nex t . . .c o l . . .Va lue . . . }// Here the e x t r a// P u b l i c methods. . .

}p u b l i c c l a s s L i n k e d L i s t {

p r i v a t e Node head ;p r i v a t e i n t l i s t C o u n t ;

// L i n k e d L i s t c o n s t r u c t o rp u b l i c L i n k e d L i s t ( ) {head = new Node ( n u l l ) ;l i s t C o u n t = 0 ;}

35 / 91

Code For Sparse Representation

Graph Representationp u b l i c c l a s s G r a p h A d j L i s t S p a r s e {p u b l i c i n t NumNodes ;p u b l i c L i n k L i s t [ ] graph ;

// c o n s t r u c tp u b l i c G r a p h A d j L i s t S p a r s e ( i n t n ) {

NumNodes = n ;graph = new L i n k L i s t [ NumNodes ] ;f o r ( i n t i = 0 ; i < nodes ; i ++)

graph [ i ] = new L i n k L i s t ( ) ;}

// More Methods Here. . .

36 / 91

Outline1 Graphs

37 / 91

Adjacency List Representation

DefinitionAdjacency list for vertex i is a linear list of vertices adjacent from vertex i .

BasicallyAn array of N adjacency lists.

ThusEach adjacency list is a chain.

38 / 91

For the previous example

39 / 91

Properties

Space for storageFor undirected or directed graphs O (V + E )

Search: Successful or UnsuccessfulO(1 + degree(v))

In additionAdjacency lists can readily be adapted to represent weighted graphs

Weight function w : E → RThe weight w(u, v) of the edge (u, v) ∈ E is simply stored withvertex v in u’s adjacency list

40 / 91

Properties

40 / 91

Properties

40 / 91

Properties

40 / 91

Possible Disadvantage

When looking to see if an edge existThere is no quicker way to determine if a given edge (u,v)

41 / 91

Traversing the Graph

Why?Do you have any examples?

YesSearch for paths satisfying various constraints

I Shortest Path

Visit some sets of verticesI Tours

Search for subgraphsI Isomorphisms

42 / 91

I Shortest Path

42 / 91

I Shortest Path

42 / 91

I Shortest Path

42 / 91

I Shortest Path

42 / 91

I Shortest Path

42 / 91

I Shortest Path

42 / 91

Outline1 Graphs

43 / 91

Breadth-first search

DefinitionGiven a graph G = (V ,E ) and a source vertex s, breadth-first searchsystematically explores the edges of Gto “discover” every vertex that isreachable from the vertex s

Something NotableA vertex is discovered the first time it is encountered during the search

44 / 91

Breadth-first search

DefinitionGiven a graph G = (V ,E ) and a source vertex s, breadth-first searchsystematically explores the edges of Gto “discover” every vertex that isreachable from the vertex s

Something NotableA vertex is discovered the first time it is encountered during the search

44 / 91

Defining Some Basic Terms

ColorGiven a node u, we have that

I color is a field indicatingF WHITE Never visited nodeF GRAY Node pointer is at the Queue QF BLACK Node has been visited and processed

45 / 91

Distance dGiven a node u, we have that

I d is a field indicating the distance from the node source s so far

46 / 91

Predecesor π

Given a node u, we have thatI π is a field indicating who is the immediate predecessor of u in the

path from s to u

47 / 91

Pseudo CodeBFS(G, s)

1 for each vertex u ∈ G.V − {s}2 u.color = WHITE3 u.d =∞4 u.π = NIL5 s.color =GRAY6 s.d = 07 s.π = NIL8 Q = ∅9 Enqueue(Q, s)10 while Q = ∅11 u =Dequeue(Q)

12 for each v ∈ G.Adj [u]13 if v .color ==WHITE14 v .color =GRAY15 v .d = u.d + 116 Enqueue(Q, v)17 u.color = BLACK

48 / 91

Change the Order of Recursion

It is like a wave going from a node

49 / 91

Example

What do you see?

50 / 91

Example

What do you see?

51 / 91

Example

What do you see?

52 / 91

Example

What do you see?

53 / 91

Example

What do you see?

54 / 91

Example

What do you see?

55 / 91

Example

What do you see?

56 / 91

Example

What do you see?

57 / 91

Complexity

What about the outer loop?O(V ) Enqueue / Dequeue operations – Each adjacency list is processedonly once.

What about the inner loop?The sum of the lengths of f all the adjacency lists is Θ(E ) so the scanningtakes O(E )

58 / 91

Complexity

What about the outer loop?O(V ) Enqueue / Dequeue operations – Each adjacency list is processedonly once.

What about the inner loop?The sum of the lengths of f all the adjacency lists is Θ(E ) so the scanningtakes O(E )

58 / 91

Complexity

Overhead of CreationO (V )

ThenTotal complexity O (V + E )

59 / 91

Complexity

Overhead of CreationO (V )

ThenTotal complexity O (V + E )

59 / 91

Properties

Something NotableBreadth-first search constructs a breadth-first tree, initially containing onlyits root, which is the source vertex s

ThusWe say that u is the predecessor or parent of v in the breadth-first tree.

60 / 91

Properties

Something NotableBreadth-first search constructs a breadth-first tree, initially containing onlyits root, which is the source vertex s

ThusWe say that u is the predecessor or parent of v in the breadth-first tree.

60 / 91

Breadth-First Search Property

Something NotableAll vertices reachable from the start vertex (including the start vertex) arevisited.

61 / 91

Outline1 Graphs

62 / 91

Depth-first search

Given GPick an unvisited vertex v, remember the rest.

I Recurse on vertices adjacent to v

63 / 91

The Pseudo-code

Code for DFSDFS(G)

1 for each vertex u ∈ G .V2 u.color = WHITE3 u.π = NIL4 time = 05 for each vertex u ∈ G .V6 if u.color = WHITE7 DFS-VISIT(G , u)

64 / 91

The Pseudo-code

Code for DFSDFS(G)

64 / 91

The Pseudo-code

Code for DFSDFS(G)

64 / 91

The Pseudo-code

Code for DFSDFS(G)

64 / 91

The Pseudo-code

Code for DFSDFS(G)

64 / 91

The Pseudo-code

Code for DFSDFS(G)

64 / 91

The Pseudo-code

Code for DFSDFS(G)

64 / 91

Now, we have the following

Code for DFS-VISITDFS-VISIT(G , u)

1 time = time + 12 u.d = time3 u.color = GRAY4 for each vertex v ∈ G .Adj [u]

5 if v .color == WHITE6 v .π = u7 DFS-VISIT(G , v)

8 u.color = BLACK9 time = time + 110 u.f = time

65 / 91

Example

What do we do?

66 / 91

Example

What do we do?

67 / 91

Example

What do we do?

68 / 91

Example

What do we do?

69 / 91

Example

What do we do?

70 / 91

Example

What do we do?

71 / 91

Example

What do we do?

72 / 91

Example

What do we do?

73 / 91

Complexity

Analysis1 The loops on lines 1–3 and lines 5–7 of DFS take Θ (V ).2 The procedure DFS-VISIT is called exactly once for each vertex

v ∈ V .3 During an execution of DFS-VISIT(G , v) the loop on lines 4–7

executes |Adj (v)| times.4 But

∑v∈V |Adj (v)| = O (E ) we have that the cost of executing g

lines 4–7 of DFS-VISIT is Θ (E ) .

ThenDFS complexity is Θ (V + E )

74 / 91

Complexity

74 / 91

Complexity

74 / 91

Complexity

74 / 91

Complexity

74 / 91

Applications

We have severalFinding a path between nodesStrongly Connected ComponentsSpanning TreesTopological Sort - The Program (or Project) Evaluation and Review(PERT)Computer Vision AlgorithmsArtificial Intelligence AlgorithmsImportance in Social NetworkRank Algorithms for GoogleEtc.

75 / 91

Applications

75 / 91

Applications

75 / 91

Applications

75 / 91

Applications

75 / 91

Applications

75 / 91

Applications

75 / 91

Applications

75 / 91

Applications

75 / 91

Outline1 Graphs

76 / 91

Finding a path between nodes

We do the followingStart a breadth-first search at vertex v .Terminate when vertex u is visited or when Q becomes empty(whichever occurs first).

Time ComplexityO(V 2)when adjacency matrix used.

O(V + E ) when adjacency lists used.

77 / 91

This allow to use the Algorithm for finding The ShortestPath

ClearlyThis is the unweighted version or all weights are equal!!!

We have the following functionδ (s, v)= shortest path from s to v

We claim thatUpon termination of BFS, every vertex v ∈ V reachable from s hasdistance(v) = δ(s, v)

78 / 91

Outline1 Graphs

79 / 91

Connected Components

DefinitionA connected component (or just component) of an undirected graph is asubgraph in which any two vertices are connected to each other by paths.

Example

80 / 91

Connected ComponentsDefinitionA connected component (or just component) of an undirected graph is asubgraph in which any two vertices are connected to each other by paths.

Example

80 / 91

Procedure

FirstStart a breadth-first search at any as yet unvisited vertex of the graph.

ThusNewly visited vertices (plus edges between them) define a component.

RepeatRepeat until all vertices are visited.

81 / 91

Procedure

81 / 91

Procedure

81 / 91

O (V 2)

When adjacency matrix used

O(V + E )

When adjacency lists used (E is number of edges)

82 / 91

O (V 2)

O(V + E )

82 / 91

Outline1 Graphs

83 / 91

Spanning TreeDefinitionA spanning tree of a graph G = (V ,E ) is a acyclic graph where foru, v ∈ V , there is a path between them

Example

84 / 91

Spanning TreeDefinitionA spanning tree of a graph G = (V ,E ) is a acyclic graph where foru, v ∈ V , there is a path between them

Example

84 / 91

Procedure

FirstStart a breadth-first search at any vertex of the graph.

ThusIf graph is connected, the n − 1 edges used to get to unvisited verticesdefine a spanning tree (breadth-first spanning tree).

85 / 91

Procedure

FirstStart a breadth-first search at any vertex of the graph.

ThusIf graph is connected, the n − 1 edges used to get to unvisited verticesdefine a spanning tree (breadth-first spanning tree).

85 / 91

O (V 2)

O(V + E )

86 / 91

O (V 2)

O(V + E )

86 / 91

Outline1 Graphs

87 / 91

Topological Sorting

DefinitionsA topological sort (sometimes abbreviated topsort or toposort) ortopological ordering of a directed graph is a linear ordering of its verticessuch that for every directed edge (u, v) from vertex u to vertex y , u comesbefore v in the ordering.

From Industrial EngineeringThe canonical application of topological sorting (topological order) isin scheduling a sequence of jobs or tasks based on their dependencies.Topological sorting algorithms were first studied in the early 1960s inthe context of the PERT technique for scheduling in projectmanagement (Jarnagin 1960).

88 / 91

Topological Sorting

DefinitionsA topological sort (sometimes abbreviated topsort or toposort) ortopological ordering of a directed graph is a linear ordering of its verticessuch that for every directed edge (u, v) from vertex u to vertex y , u comesbefore v in the ordering.

From Industrial EngineeringThe canonical application of topological sorting (topological order) isin scheduling a sequence of jobs or tasks based on their dependencies.Topological sorting algorithms were first studied in the early 1960s inthe context of the PERT technique for scheduling in projectmanagement (Jarnagin 1960).

88 / 91

We have thatThe jobs are represented by vertices, and there is an edge from x to y ifjob x must be completed before job y can be started.

ExampleWhen washing clothes, the washing machine must finish before we put theclothes to dry.

ThenA topological sort gives an order in which to perform the jobs.

89 / 91

Algorithm

Pseudo CodeTOPOLOGICAL-SORT

1 Call DFS(G) to compute finishing times v .f for each vertex v .2 As each vertex is finished, insert it onto the front of a linked list3 Return the linked list of vertices

90 / 91

Algorithm

90 / 91

Algorithm

90 / 91

Algorithm

90 / 91

Example

Dressing

91 / 91

Preparation Data Structures 11 graphs

Education

Transcript of Preparation Data Structures 11 graphs