Chapter 9 - Graph

A Graph G consists of a set V, whose members are called the vertices of G, together with a set E of pairs of distinct vertices from V. The pairs in E are called the edges of G.

If the pairs are unordered, G is called an undirected graph or a graph. Otherwise, G is called a directed graph or a digraph. Two vertices in an undirected graph are called adjacent if there is an edge from the first to the second.

Chapter 9 - Graph
A path is a sequence of distinct vertices, each adjacent to the next. A cycle is a path containing at least three vertices such that the last vertex on the path is adjacent to the first.

A graph is called connected if there is a path from any vertex to any other vertex. A free tree is defined as a connected undirected graph with no cycles.

Chapter 9 - Graph
In a directed graph a path or a cycle means always moving in the direction indicated by the arrows. A directed graph is called strongly connected if there is a directed path from any vertex to any other vertex.

If we suppress the direction of the edges and the resulting undirected graph is connected, we call the directed graph weakly connected

Examples of Graph

Examples of Graph

Digraph as an adjacency table

Directed graph

Adjacency set

Adjacency table

Digraph count <integer> // Number of vertices edge <array of <array of <boolean> > > // Adjacency table End Digraph

Weighted-graph as an adjacency table

Weighted-graph

vertex vector

adjacency table

WeightedGraph count <integer> edge<array of<array of<WeightType>>> End WeightedGraph

// Number of vertices // Adjacency table

7

Weighted-graph as an adjacency list

Digraph as an adjacency list

Directed graph

contiguous structure

linked structure

mixed structure

Digraph as an adjacency list (not using List ADT)

VertexNode first_edge <pointer to EdgeNode> next_vertex<pointer to VertexNode> End VertexNode
EdgeNode vertex_to <pointer to VertexNode> next_edge <pointer to EdgeNode> End EdgeNode

Directed graph
first_vertex

DiGraph first_vertex <pointer to VertexNode> End DiGraph

linked structure

10

Digraph as an adjacency list (using List ADT)

head

digraph

0 1 2 3

head
head head

1 2

2 3

GraphNode vertex <VertexType> // (key field) adjVertex<LinkedList of< VertexType >> indegree <int> are hidden outdegree <int> from the isMarked <boolean> image below End GraphNode

head

2
GraphNode vertex adjVertex

ADT List is linked list: DiGraph digraph <LinkedList<of<GraphNode>> End DiGraph

head
11

Digraph as an adjacency list (using List ADT)

GraphNode vertex <VertexType> // (key field) adjVertex<LinkedList of< VertexType >> indegree <int> outdegree <int> isMarked <boolean> End GraphNode

mixed list

ADT List is contiguous list:

DiGraph digraph <ContiguousList<of<GraphNode>> End DiGraph
12

Digraph as an adjacency list (using List ADT)

GraphNode vertex <VertexType> // (key field) adjVertex<ContiguousList of< VertexType >> indegree <int> outdegree <int> isMarked <boolean> End GraphNode

contiguous list

ADT List is contiguous list:

DiGraph digraph <ContiguousList<of<GraphNode>> End DiGraph
13

GraphNode
<void> GraphNode() // constructor of GraphNode 1. indegree = 0 2. outdegree = 0 3. adjVertex.clear() // By default, constructor of adjVertex made it empty. End GraphNode

GraphNode

vertex adjVertex
head

14

Operations for Digraph

Insert Vertex Delete Vertex Insert edge Delete edge Traverse

15

Digraph
Digraph private: digraph <List of <GraphNode> > // using of List ADT . <void> Remove_EdgesToVertex(val VertexTo <VertexType>) public: <ErrorCode> InsertVertex (val newVertex <VertexType>) <ErrorCode> DeleteVertex (val Vertex <VertexType>) <ErrorCode> InsertEdge (val VertexFrom <VertexType>, val VertexTo <VertexType>) <ErrorCode> DeleteEdge (val VertexFrom <VertexType>, val VertexTo <VertexType>)
// Other methods for Graph Traversal. End Digraph
16

Methods of List ADT

Methods of Digraph will use these methods of List ADT:
<ErrorCode> Insert (val DataIn <DataType>) // (success, overflow) <ErrorCode> Search (ref DataOut <DataType>) // (found, notFound) <ErrorCode> Remove (ref DataOut <DataType>) // (success , notFound) <ErrorCode> Retrieve (ref DataOut <DataType>) // (success , notFound) <ErrorCode> Retrieve (ref DataOut <DataType>, position <int>) // (success , range_error) <ErrorCode> Replace (val DataIn <DataType>, position <int>) // (success, range_error) <ErrorCode> Replace (val DataIn <DataType>, val DataOut <DataType>) // (success, notFound) <boolean> isFull() <boolean> isEmpty() <integer> Size()

17

Insert New Vertex into Digraph

<ErrorCode> InsertVertex (val newVertex <VertexType>)
Inserts new vertex into digraph. Pre Post newVertex is a vertex needs to be inserted. if the vertex is not in digraph, it has been inserted and no edge is involved with this vertex. Return success, overflow, or duplicate_error

18

Insert New Vertex into Digraph

<ErrorCode> InsertVertex (val newVertex <VertexType>) 1. DataOut.vertex = newVertex 2. if (digraph.Search(DataOut) = success) 1. return duplicate_error 3. else 1. return digraph.Insert(DataOut) // success or overflow End InsertVertex
GraphNode vertex <VertexType> // (key field) adjVertex<List of< VertexType >> indegree <int> outdegree <int> isMarked <boolean> 19 End GraphNode

Delete Vertex from Digraph

<ErrorCode> DeleteVertex (val Vertex <VertexType>)
Deletes an existing vertex. Pre Post Vertex is the vertex needs to be removed . if Vertex 's indegree <>0, the edges ending at this vertex have been removed. Finally, this vertex has been removed. Return success, or notFound Uses Function Remove_EdgeToVertex.

20

Delete Vertex from Digraph

<ErrorCode> DeleteVertex (val Vertex <VertexType>)
1. 2. DataOut.vertex = Vertex if (digraph.Retrieve(DataOut) = success) 1. if (DataOut.indegree>0) 1. digraph.Remove_EdgeToVertex(Vertex) 2. digraph.Remove(DataOut) 3. return success 3. else 1. return notFound GraphNode End DeleteVertex vertex <VertexType> // (key field) adjVertex<List of< VertexType >> indegree <int> outdegree <int> isMarked <boolean> 21 End GraphNode

Auxiliary function Remove all Edges to a Vertex

<void> Remove_EdgesToVertex(val VertexTo <VertexType>)
Removes all edges from any vertex to VertexTo if exist. 1. position = 0 2. loop (digraph.Retrieve(DataFrom, position) = success) 1. if (DataFrom.outdegree>0) 1. if (DataFrom.adjVertex.Remove(VertexTo) = success) 1. DataFrom.outdegree = DataFrom.outdegree - 1 2. digraph.Replace(DataFrom, position) 2. position = position + 1 GraphNode End Remove_EdgesToVertex vertex <VertexType> // (key field) adjVertex<List of< VertexType >> indegree <int> outdegree <int> isMarked <boolean> 22 End GraphNode

Insert new Edge into Digraph

<ErrorCode> InsertEdge (val VertexFrom<VertexType>, val VertexTo <VertexType>)
Inserts new edge into digraph.

Post

if VertexFrom and VertexTo are in the digraph, and the edge from VertexFrom to VertexTo is not in the digraph, it has been inserted. Return success, overflow, notFound_VertexFrom , notFound_VertexTo or duplicate_error

23

1. 2. 3.

DataFrom.vertex = VertexFrom DataTo.vertex = VertexTo if ( digraph.Retrieve(DataFrom) = success ) 1. if ( digraph.Retrieve(DataTo) = success ) 1. newData = DataFrom 2. if ( newData.adjVertex.Search(VertexTo) = found ) 1. return duplicate_error 3. if ( newData.adjVertex.Insert(VertexTo) = success ) 1. newData.outdegree = newData.outdegree +1 2. digraph.Replace(newData, DataFrom) 3. return success GraphNode 4. else vertex <VertexType> // (key field) 1. return overflow adjVertex<List of< VertexType >> 2. else indegree <int> 1. return notFound_VertexTo outdegree <int> 4. else isMarked <boolean> 1. return notFound_VertexFrom 24 End GraphNode End InsertEdge

Delete Edge from Digraph

<ErrorCode> DeleteEdge (val VertexFrom <VertexType>, val VertexTo <VertexType>)
Deletes an existing edge in the digraph.

Post

if VertexFrom and VertexTo are in the digraph, and the edge from VertexFrom to VertexTo is in the digraph, it has been removed Return success, notFound_VertexFrom , notFound_VertexTo or notFound_Edge

25

1. 2. 3.

DataFrom.vertex = VertexFrom DataTo.vertex = VertexTo if ( digraph.Retrieve(DataFrom) = success ) 1. if ( digraph.Retrieve(DataTo) = success ) 1. newData = DataFrom 2. if ( newData.adjVertex.Remove(VertexTo) = success ) 1. newData.outdegree = newData.outdegree -1 2. digraph.Replace(newData, DataFrom) 3. return success 3. else 1. return notFound_Edge GraphNode 2. else vertex <VertexType> // (key field) 1. return notFound_VertexTo adjVertex<List of< VertexType >> 4. else indegree <int> 1. return notFound_VertexFrom outdegree <int> End DeleteEdge isMarked <boolean> 26 End GraphNode

Graph Traversal
Depth-first traversal: analogous to preorder traversal of an oredered tree.
Breadth-first traversal: analogous to level-by-level traversal of an ordered tree.

27

Depth-first traversal

28

Breadth-first traversal

29

Depth-first traversal
<void> DepthFirst (ref <void> Operation ( ref Data <DataType>))
Traverses the digraph in depth-first order. Post Uses The function Operation has been performed at each vertex of the digraph in depth-first order. Auxiliary function recursiveTraverse to produce the recursive depth-first order.

30

Depth-first traversal
<void> DepthFirst (ref <void> Operation ( ref Data <DataType>))
1. loop (more vertex v in Digraph) 1. unmark (v) 2. loop (more vertex v in Digraph) 1. if (v is unmarked) 1. recursiveTraverse (v, Operation) End DepthFirst

31

Depth-first traversal
<void> recursiveTraverse (ref v <VertexType>, ref <void> Operation ( ref Data <DataType>) )
Traverses the digraph in depth-first order. Pre Post v is a vertex of the digraph. The depth-first traversal, using function Operation, has been completed for v and for all vertices that can be reached from v. Uses function recursiveTraverse recursively.

32

Depth-first traversal
<void> recursiveTraverse(ref v <VertexType>, ref <void> Operation ( ref Data <DataType>) )
1. mark(v) 2. Operation(v) 3. loop (more vertex w adjacent to v) 1. if (vertex w is unmarked) 1. recursiveTraverse (w, Operation) End Traverse

33

Breadth-first traversal
<void> BreadthFirst (ref <void> Operation ( ref Data <DataType>) )
Traverses the digraph in breadth-first order. Post Uses The function Operation has been performed at each vertex of the digraph in breadth-first order. Queue ADT.

34

// BreadthFirst 1. queueObj <Queue> 2. loop (more vertex v in digraph) 1. unmark(v) 3. loop (more vertex v in Digraph) 1. if (vertex v is unmarked) 1. queueObj.EnQueue(v) 2. loop (NOT queueObj .isEmpty()) 1. queueObj.QueueFront(w) 2. queueObj.DeQueue() 3. if (vertex w is unmarked) 1. mark(w) 2. Operation(w) 3. loop (more vertex x adjacent to w) 1. queueObj.EnQueue(x) End BreadthFirst

35

Topological Order
A topological order for G, a directed graph with no cycles, is a sequential listing of all the vertices in G such that, for all vertices v, w G, if there is an edge from v to w, then v precedes w in the sequential listing.

36

Topological Order

37

Topological Order

38

Applications of Topological Order

Topological order is used for:
Courses available at a university, Vertices: course. Edges: (v,w), v is a prerequisite for w. A topological order is a listing of all the courses such that all perequisites for a course appear before it does.

A glossary of technical terms: no term is used in a definition before it is itself defined. The topics in the textbook.
39

Topological Order
<void> DepthTopoSort (ref TopologicalOrder <List>)
Traverses the digraph in depth-first order and made a list of topological order of digraph's vertices. Pre Acyclic digraph. Post The vertices of the digraph are arranged into the list TopologicalOrder with a depth-first traversal of those vertices that do not belong to a cycle. Uses List ADT and function recursiveDepthTopoSort to perform depth-first traversal. Idea: Starts by finding a vertex that has no successors and place it last in the list. Repeatedly add vertices to the beginning of the list. By recursion, places all the successors of a vertex into the topological order. Then, place the vertex itself in a position before any of its successors.

40

Topological Order

41

Topological Order
<void> DepthTopoSort (ref TopologicalOrder <List>)
1. loop (more vertex v in digraph) 1. unmark(v) 2. TopologicalOrder.clear() 3. loop (more vertex v in Digraph) 1. if (vertex v is unmarked) 1. recursiveDepthTopoSort(v, TopologicalOrder) End DepthTopoSort

42

Topological Order
<void> recursiveDepthTopoSort (val v <VertexType>, ref TopologicalOrder <List>)
Pre Post
Uses

Vertex v in digraph does not belong to the partially completed list TopologicalOrder. All the successors of v and finally v itself are added to TopologicalOrder with a depth-first order traversal. List ADT and the function recursiveDepthTopoSort.

Idea: Performs the recursion, based on the outline for the general function traverse. First, places all the successors of v into their positions in the topological order. Then, places v into the order.

43

Topological Order
<void> recursiveDepthTopoSort (val v <VertexType>, ref TopologicalOrder <List>)
1. mark(v) 2. loop (more vertex w adjacent to v) 1. if (vertex w is unmarked) 1. recursiveDepthTopoSort(w, TopologicalOrder) 3. TopologicalOrder.Insert(0, v) End recursiveDepthTopoSort

44

Topological Order
<void> BreadthTopoSort (ref TopologicalOrder <List>)
Traverses the digraph in depth-first order and made a list of topological order of digraph's vertices. Post The vertices of the digraph are arranged into the list TopologicalOrder with a breadth-first traversal of those vertices that do not belong to a cycle. Uses List and Queue ADT. Idea: Starts by finding the vertices that are not successors of any other vertex. Places these vertices into a queue of vertices to be visited. As each vertex is visited, it is removed from the queue and placed in the next available position in the topological order (starting at the beginning). Reduces the indegree of its successors by 1. The vertex having the zero value indegree is ready to processed and is places into the queue. 45

Topological Order

46

<void> BreadthTopoSort (ref TopologicalOrder <List>)

1. TopologicalOrder.clear() 2. queueObj <Queue> 3. loop (more vertex v in digraph) 1. if (indegree of v = 0) 1. queueObj.EnQueue(v) 4. loop (NOT queueObj.isEmpty()) 1. queueObj.QueueFront(v) 2. queueObj.DeQueue() 3. TopologicalOrder.Insert(TopologicalOrder.size(), v) 4. loop (more vertex w adjacent to v) 1. decrease the indegree of w by 1 2. if (indegree of w = 0) 1. queueObj.EnQueue(w) End BreadthTopoSort
47

Shortest Paths
Given a directed graph in which each edge has a nonnegative weight. Find a path of least total weight from a given vertex, called the source, to every other vertex in the graph.

A greedy algorithm of Shortest Paths: Dijkstra's algorithm (1959).

48

Dijkstra's algorithm
Let tree is the subgraph contains the shotest paths from the source vertex to all other vertices.
At first, add the source vertex to the tree.

Loop until all vertices are in the tree:

Consider the adjacent vertices of the vertices already in the tree. Examine all the paths from those adjacent vertices to the source vertex. Select the shortest path and insert the corresponding adjacent vertex into the tree.
49

 S: Set of vertices whose closest distances to the source are known. Add one vertex to S at each stage.
For each vertex v, maintain the distance from the source to v, along a path all of whose vertices are in S, except possibly the last one.

Dijkstra's algorithm in detail

To determine what vertex to add to S at each step, apply the greedy criterion of choosing the vertex v with the smallest distance.
Add v to S.

Update distance from the source for all w not in S, if the path through v and then directly to w is shorter than the previously recorded distance to w.

50

Dijkstra's algorithm

51

Dijkstra's algorithm

52

Dijkstra's algorithm

53

Dijkstra's algorithm

54

Dijkstra's algorithm

55

Dijkstra's algorithm

56

Dijkstra's algorithm
<void> ShortestPath (val source <VertexType>, ref listOfShortestPath <List of <DistanceNode>>)
Finds the shortest paths from source to all other vertices in digraph.

Post

Each node in listOfShortestPath gives the minimal path weight from vertex source to vertex destination in distance field.

DistanceNode destination <VertexType> distance <int> 57 End DistanceNode

// ShortestPath 1. listOfShortestPath.clear() 2. Add source to set S 3. loop (more vertex v in digraph) // Initiate all distances from source to v 1. distanceNode.destination = v 2. distanceNode.distance = weight of edge(source, v) // = infinity if
// edge(source,v) isn't in digraph.

3. listOfShortestPath.Insert(distanceNode) 4. loop (more vertex not in S) // Add one vertex v to S on each step. 1. minWeight = infinity // Choose vertex v with smallest distance. 2. loop (more vertex w not in S) 1. Find the distance x from source to w in listOfShortestPath 2. if (x < minWeight) DistanceNode destination <VertexType> 1. v = w distance <int> 2. minWeight = x 58 End DistanceNode 3. Add v to S.

// ShortestPath (continue) 4. loop (more vertex w not in S) // Update distances from source // to all w not in S 1. Find the distance x from source to w in listOfShortestPath 2. if ( (minWeight + weight of edge from v to w) < x ) 1. Update distance from source to w in listOfShortestPath to (minWeight + weight of edge from v to w) End ShortestPath

DistanceNode destination <VertexType> distance <int> 59 End DistanceNode

Another example of Shortest Paths

Select the adjacent vertex having minimum path to the source vertex
60

Minimum spanning tree

DEFINITION:
Spanning tree: tree that contains all of the vertices in a connected graph. Minimum spanning tree: spanning tree such that the sum of the weights of its edges is minimal.

61

Spanning Trees

Two spanning trees in a network

62

A greedy Algorithm: Minimum Spanning Tree

Shortest path algorithm in a connected graph found an its spanning tree.
What is the algorithm finding the minimum spanning tree? A small change to shortest path algorithm can find the minimum spanning tree, that is Prim's algorithm since 1957.

63

Prim's algorithm
Let tree is the minimum spanning tree.
At first, add one vertex to the tree.

Loop until all vertices are in the tree:

Consider the adjacent vertices of the vertices already in the tree. Examine all the edges from each vertices already in the tree to those adjacent vertices. Select the smallest edge and insert the corresponding adjacent vertex into the tree.

64

Prim's algorithm in detail

Let S is the set of vertices already in the minimum spanning tree. At first, add one vertex to S.
For each vertex v not in S, maintain the distance from a vertex x to v, where x is a vertex in S and the edge(x,v) is the smallest in all edges from another vertices in S to v (this edge(x,v) is called the distance from S to v). As usual, all edges not being in graph have infinity value.

To determine what vertex to add to S at each step, apply the greedy criterion of choosing the vertex v with the smallest distance from S. Add v to S.
Update distances from S to all vertices v not in S if they are smaller than the previously recorded distances.

65

Prim's algorithm

Select the adjacent vertex having minimum edge to the vertices 66 already in the tree.

Prim's algorithm
<void> MinimumSpanningTree (val source <VertexType>, ref tree <Graph>)
Finds the minimum spanning tree of a connected component of the original graph that contains vertex source.

Post

tree is the minimum spanning tree of a connected component of the original graph that contains vertex source.

Uses local variables: Set S listOfDistanceNode continue <boolean>

DistanceNode vertexFrom <VertexType> vertexTo <VertexType> distance <WeightType> 67 End DistanceNode

1. 2. 3. 4. 5. 6.

tree.clear() tree.InsertVertex(source) Add source to set S listOfDistanceNode.clear() distanceNode.vertexFrom = source loop (more vertex v in graph)//Initiate all distances from source to v 1. distanceNode.vertexTo = v 2. distanceNode.distance = weight of edge(source, v) // = infinity if // edge(source,v) isn't in graph. 3. listOfDistanceNode.Insert(distanceNode)
DistanceNode vertexFrom <VertexType> vertexTo <VertexType> distance <WeightType> 68 End DistanceNode

7. continue = TRUE 8. loop (more vertex not in S) and (continue) //Add one vertex to S on // each step 1. minWeight = infinity //Choose vertex v with smallest distance toS 2. loop (more vertex w not in S) 1. Find the node in listOfDistanceNode with vertexTo is w 2. if (node.distance < minWeight) 1. v = w 2. minWeight = node.distance

DistanceNode vertexFrom <VertexType> vertexTo <VertexType> distance <WeightType> 69 End DistanceNode

3. if (minWeight < infinity) 1. Add v to S. 2. tree.InsertVertex(v) 3. tree.InsertEdge(v,w) 4. loop (more vertex w not in S) // Update distances from v to // all w not in S if they are smaller than the // previously recorded distances in listOfDistanceNode 1. Find the node in listOfDistanceNode with vertexTo is w 2. if ( node.distance > weight of edge(v,w) ) 1. node.vertexFrom = v 2. node.distance = weight of edge(v,w) ) 3. Replace this node with its old node in listOfDistance DistanceNode 4. else vertexFrom <VertexType> 1. continue = FALSE vertexTo <VertexType> End MinimumSpanningTree
distance <WeightType> End DistanceNode
70

Maximum flows

71

Maximum flows

72

Maximum flows

73

Maximum flows

74

Maximum flows

75

Matching

76

Matching

77

Matching

78

Matching

79

Matching

80

Graph coloring

81

Graph coloring

82

Graph coloring

83

Graph coloring

84

Graph coloring

85