Graph and its representations

Graph is a data structure that consists of following two components:


1. A finite set of vertices also called as nodes.

2. A finite set of ordered pair of the form (u, v) called as edge. The pair is ordered because (u, v) is not same
as (v, u) in case of directed graph(di-graph). The pair of form (u, v) indicates that there is an edge from vertex
u to vertex v. The edges may contain weight/value/cost.

Following two are the most commonly used representations of graph.

1. Adjacency Matrix
2. Adjacency List

Adjacency Matrix:
Adjacency Matrix is a 2D array of size V x V where V is the number of vertices in a graph. Let the 2D array be
adj[][], a slot adj[i][j] = 1 indicates that there is an edge from vertex i to vertex j. 
Pros:  Representation is easier to implement and follow. Removing an edge takes O(1) time. Queries like
whether there is an edge from vertex ‘u’ to vertex ‘v’ are efficient and can be done O(1).

Cons:  Consumes more space O(V^2). Even if the graph is sparse(contains less number of edges), it
consumes the same space. Adding a vertex is O(V^2) time.

Adjacency List:
An array of linked lists is used. Size of the array is equal to number of vertices. Let the array be array[]. An
entry array[i] represents the linked list of vertices adjacent to the ith vertex. This representation can also be
used to represent a weighted graph. The weights of edges can be stored in nodes of linked lists. 
Pros:  Saves space O(|V|+|E|) . In the worst case, there can be C(V, 2) number of edges in a graph thus
consuming O(V^2) space. Adding a vertex is easier.

Cons: Queries like whether there is an edge from vertex u to vertex v are not efficient and can be done O(V).

Breadth First Traversal or BFS for a Graph

Breadth First Traversal (or Search) for a graph is similar to Breadth First Traversal of a tree (See method 2
of this post). The only catch here is, unlike trees, graphs may contain cycles, so we may come to the same
node again. To avoid processing a node more than once, we use a boolean visited array. For simplicity, it is
assumed that all vertices are reachable from the starting vertex.

Note that the above code traverses only the vertices reachable from a given source vertex. All the vertices
may not be reachable from a given vertex (example Disconnected graph). To print all the vertices, we can
modify the BFS function to do traversal starting from all nodes one by one (Like the DFS modified version) .

Time Complexity: O(V+E) where V is number of vertices in the graph and E is number of edges in the graph.

while (queue.size() != 0)
{
// Dequeue a vertex from queue and print it
s = queue.poll();
System.out.print(s+" ");
 // Get all adjacent vertices of the dequeued vertex s
// If a adjacent has not been visited, then mark it
// visited and enqueue it
Iterator<Integer> i = adj[s].listIterator();
while (i.hasNext())
{
int n = i.next();
if (!visited[n])
{
visited[n] = true;
queue.add(n);
}

Depth First Traversal or DFS for a Graph

Depth First Traversal (or Search) for a graph is similar to Depth First Traversal of a tree. The only catch here
is, unlike trees, graphs may contain cycles, so we may come to the same node again. To avoid processing a
node more than once, we use a boolean visited array. 

Note that the above code traverses only the vertices reachable from a given source vertex. All the vertices
may not be reachable from a given vertex (example Disconnected graph). To do complete DFS traversal of
such graphs, we must call DFSUtil() for every vertex. Also, before calling DFSUtil(), we should check if it is
already printed by some other call of DFSUtil(). 

Time Complexity: O(V+E) where V is number of vertices in the graph and E is number of edges in the graph.

void DFSUtil(int v,boolean visited[])

 {
// Mark the current node as visited and print it
visited[v] = true;
System.out.print(v+" ");

// Recur for all the vertices adjacent to this vertex

Iterator<Integer> i = adj[v].listIterator();
while (i.hasNext())
{
int n = i.next();
if (!visited[n])
DFSUtil(n,visited);
}
}

// The function to do DFS traversal. It uses recursive DFSUtil()

void DFS()
{
// Mark all the vertices as not visited(set as
// false by default in java)
boolean visited[] = new boolean[V];

// Call the recursive helper function to print DFS traversal

// starting from all vertices one by one
for (int i=0; i<V; ++i)
if (visited[i] == false)
DFSUtil(i, visited);
}

Applications of Depth First Search

Depth-first search (DFS) is an algorithm (or technique) for traversing a graph.

Following are the problems that use DFS as a bulding block.

1)  For an unweighted graph, DFS traversal of the graph produces the minimum spanning tree and all pair
shortest path tree.

2) Detecting cycle in a graph 


A graph has cycle if and only if we see a back edge during DFS. So we can run DFS for the graph and check
for back edges. (See this for details)

3) Path Finding

We can specialize the DFS algorithm to find a path between two given vertices u and z.

i) Call DFS(G, u) with u as the start vertex.

ii) Use a stack S to keep track of the path between the start vertex and the current vertex.

iii) As soon as destination vertex z is encountered, return the path as the

contents of the stack

See this for details.

4) Topological Sorting

Topological Sorting is mainly used for scheduling jobs from the given dependencies among jobs. In computer
science, applications of this type arise in instruction scheduling, ordering of formula cell evaluation when
recomputing formula values in spreadsheets, logic synthesis, determining the order of compilation tasks to
perform in makefiles, data serialization, and resolving symbol dependencies in linkers [2].

5) To test if a graph is bipartite


We can augment either BFS or DFS when we first discover a new vertex, color it opposited its parents, and
for each other edge, check it doesn’t link two vertices of the same color. The first vertex in any connected
component can be red or black! See this for details.

6) Finding  Strongly Connected Components  of a graph  A directed graph is called strongly connected if
there is a path from each vertex in the graph to every other vertex. (See this for DFS based algo for finding
Strongly Connected Components)

7) Solving puzzles with only one solution, such as mazes. (DFS can be adapted to find all solutions to a
maze by only including nodes on the current path in the visited set.)

Applications of Breadth First Traversal

We have earlier discussed Breadth First Traversal Algorithm for Graphs. We have also discussed Applications
of Depth First Traversal. In this article, applications of Breadth First Search are discussed.

1) Shortest Path and Minimum Spanning Tree for unweighted graph In unweighted graph, the shortest
path is the path with least number of edges. With Breadth First, we always reach a vertex from given source
using minimum number of edges. Also, in case of unweighted graphs, any spanning tree is Minimum
Spanning Tree and we can use either Depth or Breadth first traversal for finding a spanning tree.

2) Peer to Peer Networks. In Peer to Peer Networks like BitTorrent, Breadth First Search is used to find all
neighbor nodes.

3) Crawlers in Search Engines: Crawlers build index using Breadth First. The idea is to start from source
page and follow all links from source and keep doing same. Depth First Traversal can also be used for
crawlers, but the advantage with Breadth First Traversal is, depth or levels of built tree can be limited.

4) Social Networking Websites:  In social networks, we can find people within a given distance ‘k’ from a
person using Breadth First Search till ‘k’ levels.

5) GPS Navigation systems: Breadth First Search is used to find all neighboring locations.

6) Broadcasting in Network:  In networks, a broadcasted packet follows Breadth First Search to reach all
nodes.

7) In Garbage Collection:  Breadth First Search is used in copying garbage collection using  Cheney’s
algorithm. Refer  this  and for details. Breadth First Search is preferred over Depth First Search because of
better locality of reference:
8)  Cycle detection in undirected graph:  In undirected graphs, either Breadth First Search or Depth First
Search can be used to detect cycle. In directed graph, only depth first search can be used.

9)  Ford–Fulkerson algorithm  In Ford-Fulkerson algorithm, we can either use Breadth First or Depth First
Traversal to find the maximum flow. Breadth First Traversal is preferred as it reduces worst case time
complexity to O(VE2).

10) To test if a graph is Bipartite We can either use Breadth First or Depth First Traversal.

11) Path Finding We can either use Breadth First or Depth First Traversal to find if there is a path between
two vertices.

12) Finding all nodes within one connected component: We can either use Breadth First or Depth First
Traversal to find all nodes reachable from a given node.

Many algorithms like Prim’s Minimum Spanning Tree and Dijkstra’s Single Source Shortest Path use structure
similar to Breadth First Search.

Longest Path in a Directed Acyclic Graph

Given a Weighted Directed Acyclic Graph (DAG) and a source vertex s in it, find the longest distances from s
to all other vertices in the given graph.
The longest path problem for a general graph is not as easy as the shortest path problem because the
longest path problem doesn’t have  optimal substructure property. In fact,  the Longest Path problem is NP-
Hard for a general graph. However, the longest path problem has a linear time solution for directed acyclic
graphs. The idea is similar to  linear time solution for shortest path in a directed acyclic graph., we
use Tological Sorting.

We initialize distances to all vertices as minus infinite and distance to source as 0, then we find a topological
sorting  of the graph. Topological Sorting of a graph represents a linear ordering of the graph (See below,
figure (b) is a linear representation of figure (a) ). Once we have topological order (or linear representation),
we one by one process all vertices in topological order. For every vertex being processed, we update
distances of its adjacent using distance of current vertex.

Following is complete algorithm for finding longest distances.

1) Initialize dist[] = {NINF, NINF, ….} and dist[s] = 0 where s is the source vertex. Here NINF means negative
infinite.
2) Create a toplogical order of all vertices.
3) Do following for every vertex u in topological order.
………..Do following for every adjacent vertex v of u
………………if (dist[v] < dist[u] + weight(u, v))
………………………dist[v] = dist[u] + weight(u, v)
Time Complexity: Time complexity of topological sorting is O(V+E). After finding topological order, the
algorithm process all vertices and for every vertex, it runs a loop for all adjacent vertices. Total adjacent
vertices in a graph is O(E). So the inner loop runs O(V+E) times. Therefore, overall time complexity of this
algorithm is O(V+E).

Find a Mother Vertex in a Graph

What is a Mother Vertex?

A mother vertex in a graph G = (V,E) is a vertex v such that all other vertices in G can be reached by a path
from v.

How to find mother vertex?

▪ Case 1:- Undirected Connected Graph : In this case, all the vertices are mother vertices as we can
reach to all the other nodes in the graph.
▪ Case 2:- Undirected/Directed Disconnected Graph : In this case, there is no mother vertices as we
cannot reach to all the other nodes in the graph.
▪ Case 3:- Directed Connected Graph : In this case, we have to find a vertex -v in the graph such that
we can reach to all the other nodes in the graph through a directed path.
A Naive approach : 

A trivial approach will be to perform a DFS/BFS on all the vertices and find whether we can reach all the
vertices from that vertex. This approach takes O(V(E+V)) time, which is very inefficient for large graphs.
Can we do better?

We can find a mother vertex in O(V+E) time. The idea is based on  Kosaraju’s Strongly Connected
Component Algorithm. In a graph of strongly connected components, mother verices are always vertices of
source component in component graph. The idea is based on below fact.

If there exist mother vertex (or vertices), then one of the mother vertices is the last finished vertex in DFS. (Or
a mother vertex has the maximum finish time in DFS traversal).

A vertex is said to be finished in DFS if a recursive call for its DFS is over, i.e., all descendants of the vertex
have been visited.

How does the above idea work?


Let the last finished vertex be v. Basically, we need to prove that there cannot be an edge from another vertex
u to v if u is not another mother vertex (Or there cannot exist a non-mother vertex u such that u-→v is an
edge). There can be two possibilities.

1. Recursive DFS call is made for u before v. If an edge u-→v exists, then v must have finished before u
because v is reachable through u and a vertex finishes after all its descendants.
2. Recursive DFS call is made for v before u. In this case also, if an edge u-→v exists, then either v must
finish before u (which contradicts our assumption that v is finished at the end) OR u should be reachable
from v (which means u is another mother vertex).
Algorithm :

1. Do DFS traversal of the given graph. While doing traversal keep track of last finished vertex ‘v’. This
step takes O(V+E) time.
2. If there exist mother vertex (or vetices), then v must be one (or one of them). Check if v is a mother
vertex by doing DFS/BFS from v. This step also takes O(V+E) time.
Time Complexity : O(V + E)

Transitive Closure of a Graph using DFS

Given a directed graph, find out if a vertex v is reachable from another vertex u for all vertex pairs (u, v) in the
given graph. Here reachable mean that there is a path from vertex u to v. The reach-ability matrix is called
transitive closure of a graph.

We have discussed a O(V3) solution for this here. The solution was based Floyd Warshall Algorithm. In this
post a O(V2) algorithm for the same is discussed.

Below are abstract steps of algorithm.

1. Create a matrix tc[V][V] that would finally have transitive closure of given graph. Initialize all entries of
tc[][] as 0.
2. Call DFS for every node of graph to mark reachable vertices in tc[][]. In recursive calls to DFS, we don’t
call DFS for an adjacent vertex if it is already marked as reachable in tc[][].
void Graph::DFSUtil(int s, int v)
{
// Mark reachability from s to t as true.
tc[s][v] = true;

// Find all the vertices reachable through v

list<int>::iterator i;
for (i = adj[v].begin(); i != adj[v].end(); ++i)
if (tc[s][*i] == false)
DFSUtil(s, *i);
}

// The function to find transitive closure. It uses

// recursive DFSUtil()
void Graph::transitiveClosure()
{
// Call the recursive helper function to print DFS
// traversal starting from all vertices one by one
for (int i = 0; i < V; i++)
DFSUtil(i, i); // Every vertex is reachable from self.

for (int i=0; i<V; i++)

{
for (int j=0; j<V; j++)
cout << tc[i][j] << " ";
cout << endl;
}
}

Find k-cores of an undirected graph

Given a graph G and an integer K, K-cores of the graph are connected components that are left after all
vertices of degree less than k have been removed

The standard algorithm to find a k-core graph is to remove all the vertices that have degree less than- ‘K’ from
the input graph. We must be careful that removing a vertex reduces the degree of all the vertices adjacent to
it, hence the degree of adjacent vertices can also drop below-‘K’. And thus, we may have to remove those
vertices also. This process may/may not go until there are no vertices left in the graph.

To implement above algorithm, we do a modified DFS on the input graph and delete all the vertices having
degree less than ‘K’, then update degrees of all the adjacent vertices, and if their degree falls below ‘K’ we will
delete them too.

Time complexity of the above solution is O(V + E) where V is number of vertices and E is number of edges.

bool Graph::DFSUtil(int v, vector<bool> &visited,

vector<int> &vDegree, int k)
{
// Mark the current node as visited and print it
visited[v] = true;

// Recur for all the vertices adjacent to this vertex

list<int>::iterator i;
for (i = adj[v].begin(); i != adj[v].end(); ++i)
{
// degree of v is less than k, then degree of adjacent
// must be reduced
 if (vDegree[v] < k)
vDegree[*i]--;

// If adjacent is not processed, process it

if (!visited[*i])
{
// If degree of adjacent after processing becomes
// less than k, then reduce degree of v also.
if (DFSUtil(*i, visited, vDegree, k))
vDegree[v]--;
}
}

// Return true if degree of v is less than k

return (vDegree[v] < k);
}

Graph::Graph(int V)
{
this->V = V;
adj = new list<int>[V];
}

void Graph::addEdge(int u, int v)

{
adj[u].push_back(v);
adj[v].push_back(u);
}

// Prints k cores of an undirected graph

void Graph::printKCores(int k)
{
// INITIALIZATION
// Mark all the vertices as not visited and not
// processed.
vector<bool> visited(V, false);
vector<bool> processed(V, false);

int mindeg = INT_MAX;

int startvertex;

// Store degrees of all vertices

vector<int> vDegree(V);
for (int i=0; i<V; i++)
{
vDegree[i] = adj[i].size();

if (vDegree[i] < mindeg)

{
mindeg = vDegree[i];
startvertex=i;
}
}

DFSUtil(startvertex, visited, vDegree, k);

// DFS traversal to update degrees of all

// vertices.
for (int i=0; i<V; i++)
if (visited[i] == false)
DFSUtil(i, visited, vDegree, k);

// PRINTING K CORES
cout << "K-Cores : \n";
 for (int v=0; v<V; v++)
{
// Only considering those vertices which have degree
// >= K after BFS
if (vDegree[v] >= k)
{
cout << "\n[" << v << "]";

// Traverse adjacency list of v and print only

// those adjacent which have vDegree >= k after
// BFS.
list<int>::iterator itr;
for (itr = adj[v].begin(); itr != adj[v].end(); ++itr)
if (vDegree[*itr] >= k)
cout << " -> " << *itr;
}
}
}

Related Concepts : 


Degeneracy : Degeneracy of a graph is the largest value k such that the graph has a k-core. For example, the
above shown graph has a 3-Cores and doesn’t have 4 or higher cores. Therefore, above graph is 3-
degenerate.

Degeneracy of a graph is used to measure how sparse graph is.

Iterative Depth First Traversal of Graph

We have discussed recursive implementation of DFS in previous in previous post. In the post, iterative DFS is
discussed. The recursive implementation uses function call stack. In iterative implementation, an explicit stack
is used to hold visited vertices.

void DFSUtil(int s, Vector<Boolean> visited)

{
// Create a stack for DFS
Stack<Integer> stack = new Stack<>();

// Push the current source node

stack.push(s);

while(stack.empty() == false)
{
// Pop a vertex from stack and print it
s = stack.peek();
stack.pop();

// Stack may contain same vertex twice. So

// we need to print the popped item only
// if it is not visited.
if(visited.get(s) == false)
{
System.out.print(s + " ");
visited.set(s, true);
}

// Get all adjacent vertices of the popped vertex s

// If a adjacent has not been visited, then puah it
// to the stack.
Iterator<Integer> itr = adj[s].iterator();
 while (itr.hasNext())
{
int v = itr.next();
if(!visited.get(v))
stack.push(v);
}

}
}

// prints all vertices in DFS manner

void DFS()
{
Vector<Boolean> visited = new Vector<Boolean>(V);
// Mark all the vertices as not visited
for (int i = 0; i < V; i++)
visited.add(false);

for (int i = 0; i < V; i++)

if (!visited.get(i))
DFSUtil(i, visited);
}
Like recursive traversal, time complexity of iterative implementation is O(V + E).

Iterative Deepening Search(IDS) or Iterative Deepening

Depth First Search(IDDFS)
There are two common ways to traverse a graph,  BFS  and  DFS. Considering a Tree (or Graph) of huge
height and width, both BFS and DFS are not very efficient due to following reasons.

1. DFS first traverses nodes going through one adjacent of root, then next adjacent. The problem with this
approach is, if there is a node close to root, but not in first few subtrees explored by DFS, then DFS
reaches that node very late. Also, DFS may not find shortest path to a node (in terms of number of
edges).
1. BFS goes level by level, but requires more space. The space required by DFS is O(d) where d is depth
of tree, but space required by BFS is O(n) where n is number of nodes in tree (Why? Note that the last
level of tree can have around n/2 nodes and second last level n/4 nodes and in BFS we need to have
every level one by one in queue).
IDDFS combines depth-first search’s space-efficiency and breadth-first search’s fast search (for nodes closer
to root).

How does IDDFS work?


IDDFS calls DFS for different depths starting from an initial value. In every call, DFS is restricted from going
beyond given depth. So basically we do DFS in a BFS fashion.

Algorithm:

// Returns true if target is reachable from

// src within max_depth
bool IDDFS(src, target, max_depth)
for limit from 0 to max_depth
if DLS(src, target, limit) == true
return true
return false

bool DLS(src, target, limit)

if (src == target)
return true;

// If reached the maximum depth,

// stop recursing.
if (limit <= 0)
return false;

foreach adjacent i of src

if DLS(i, target, limit?1)
return true

return false
An important thing to note is, we visit top level nodes multiple times. The last (or max depth) level is visited
once, second last level is visited twice, and so on. It may seem expensive, but it turns out to be not so costly,
since in a tree most of the nodes are in the bottom level. So it does not matter much if the upper levels are
visited multiple times.
here can be two cases-

a) When the graph has no cycle: This case is simple. We can DFS multiple times with different height limits.

b) When the graph has cycles. This is interesting as there is no visited flag in IDDFS.

In an iterative deepening search, the nodes on the bottom level are expanded once, those on the next to
bottom level are expanded twice, and so on, up to the root of the search tree, which is expanded d+1 times.
So the total number of expansions in an iterative deepening search is-

(d)b + (d-1)b2 + .... + 3bd-2 + 2bd-1 + bd

That is,
Summation[(d + 1 - i) bi], from i = 0 to i = d
Which is same as O(bd)
After evaluating the above expression, we find that asymptotically IDDFS takes the same time as that of DFS
and BFS, but it is indeed slower than both of them as it has a higher constant factor in its time complexity
expression.

IDDFS is best suited for a complete infinite tree

(Branching factor for binary tree is 2).

Detect Cycle in a Directed Graph
Given a directed graph, check whether the graph contains a cycle or not. Your function should return true if
the given graph contains at least one cycle, else return false.

Depth First Traversal can be used to detect cycle in a Graph. DFS for a connected graph produces a tree.
There is a cycle in a graph only if there is a back edge present in the graph. A back edge is an edge that is
from a node to itself (selfloop) or one of its ancestor in the tree produced by DFS.
For a disconnected graph, we get the DFS forrest as output. To detect cycle, we can check for cycle in individual trees by
checking back edges.
To detect a back edge, we can keep track of vertices currently in recursion stack of function for DFS traversal. If we
reach a vertex that is already in the recursion stack, then there is a cycle in the tree. The edge that connects current vertex
to the vertex in the recursion stack is back edge. We have used recStack[] array(*nothing to do with a stack) to keep
track of vertices in the recursion stack.

Time Complexity of this method is same as time complexity of DFS traversal which is O(V+E).

bool Graph::isCyclicUtil(int v, bool visited[], bool *recStack)

{
if(visited[v] == false)
{
// Mark the current node as visited and part of recursion stack
visited[v] = true;
recStack[v] = true;

// Recur for all the vertices adjacent to this vertex

list<int>::iterator i;
for(i = adj[v].begin(); i != adj[v].end(); ++i)
{
if ( !visited[*i] && isCyclicUtil(*i, visited, recStack) )
return true;
else if (recStack[*i])
return true;
}

}
recStack[v] = false; // remove the vertex from recursion stack
return false;
}
// Returns true if the graph contains a cycle, else false.
// This function is a variation of DFS() in http://www.geeksforgeeks.org/archives/18212
bool Graph::isCyclic()
{
// Mark all the vertices as not visited and not part of recursion
// stack
bool *visited = new bool[V];
bool *recStack = new bool[V];
for(int i = 0; i < V; i++)
{
visited[i] = false;
recStack[i] = false;
}

// Call the recursive helper function to detect cycle in different

// DFS trees
for(int i = 0; i < V; i++)
if (isCyclicUtil(i, visited, recStack))
return true;

return false;
}

Disjoint Set (Or Union-Find) | Set 1 (Detect Cycle in an

Undirected Graph)
A disjoint-set data structure is a data structure that keeps track of a set of elements partitioned into a number
of disjoint (non-overlapping) subsets.  A  union-find algorithm  is an algorithm that performs two useful
operations on such a data structure:

Find: Determine which subset a particular element is in. This can be used for determining if two elements are
in the same subset.

Union: Join two subsets into a single subset.

In this post, we will discuss an application of Disjoint Set Data Structure. The application is to check whether a
given graph contains a cycle or not.

Union-Find Algorithm can be used to check whether an undirected graph contains cycle or not. Note that we
have discussed an  algorithm to detect cycle. This is another method based on  Union-Find. This method
assumes that graph doesn’t contain any self-loops.

We can keeps track of the subsets in a 1D array, lets call it parent[].

int find(int parent[], int i)

{
if (parent[i] == -1)
return i;
return find(parent, parent[i]);
}

// A utility function to do union of two subsets

void Union(int parent[], int x, int y)
{
int xset = find(parent, x);
int yset = find(parent, y);
parent[xset] = yset;
}

// The main function to check whether a given graph

// contains cycle or not
int isCycle( Graph graph)
{
// Allocate memory for creating V subsets
int parent[] = new int[graph.V];

// Initialize all subsets as single element sets

for (int i=0; i<graph.V; ++i)
parent[i]=-1;

// Iterate through all edges of graph, find subset of both

// vertices of every edge, if both subsets are same, then
// there is cycle in graph.
for (int i = 0; i < graph.E; ++i)
{
int x = graph.find(parent, graph.edge[i].src);
int y = graph.find(parent, graph.edge[i].dest);

if (x == y)
return 1;

graph.Union(parent, x, y);
}
 return 0;
}

Note that the implementation of union() and find() is naive and takes O(n) time in worst case. These methods
can be improved to O(Logn) using Union by Rank or Height. We will soon be discussing Union by Rankin a
separate post.

Detect cycle in an undirected graph

Given an undirected graph, how to check if there is a cycle in the graph?

We have discussed cycle detection for directed graph. We have also discussed a union-find algorithm for
cycle detection in undirected graphs. The time complexity of the union-find algorithm is O(ELogV). Like
directed graphs, we can use DFS to detect cycle in an undirected graph in O(V+E) time. We do a DFS
traversal of the given graph. For every visited vertex ‘v’, if there is an adjacent ‘u’ such that u is already visited
and u is not parent of v, then there is a cycle in graph. If we don’t find such an adjacent for any vertex, we say
that there is no cycle. The assumption of this approach is that there are no parallel edges between any two
vertices.

Boolean isCyclicUtil(int v, Boolean visited[], int parent)

{
// Mark the current node as visited
visited[v] = true;
Integer i;

// Recur for all the vertices adjacent to this vertex

Iterator<Integer> it = adj[v].iterator();
while (it.hasNext())
 {
i = it.next();

// If an adjacent is not visited, then recur for that

// adjacent
if (!visited[i])
{
if (isCyclicUtil(i, visited, v))
return true;
}

// If an adjacent is visited and not parent of current

// vertex, then there is a cycle.
else if (i != parent)
return true;
}
return false;
}

// Returns true if the graph contains a cycle, else false.

Boolean isCyclic()
{
// Mark all the vertices as not visited and not part of
// recursion stack
Boolean visited[] = new Boolean[V];
for (int i = 0; i < V; i++)
visited[i] = false;

// Call the recursive helper function to detect cycle in

// different DFS trees
for (int u = 0; u < V; u++)
if (!visited[u]) // Don't recur for u if already visited
 if (isCyclicUtil(u, visited, -1))
return true;

return false;
}

Time Complexity: The program does a simple DFS Traversal of graph and graph is represented using
adjacency list. So the time complexity is O(V+E)

*Now the important thing is why this method couldn’t be used for directed graph….here we have created
parent as -1 for every new component and try to search a cycle in this component…here we are sure that we
won’t get a edge to some other component and thus ( if visited[i] and i!=parent ) works….In case of directed it
is possible that we don't have a path from i to j but we have a directed edge from j to i…..so using this also
won’t be right.

Detect Cycle in a directed graph using colors

Given a directed graph, check whether the graph contains a cycle or not. Your function should return true if
the given graph contains at least one cycle, else return false. 

In the previous post, we have discussed a solution that stores visited vertices in a separate array which stores
vertices of current recursion call stack.

In this post a different solution is discussed. The solution is from CLRS book. The idea is to do DFS of given
graph and while doing traversal, assign one of the below three colors to every vertex.
WHITE : Vertex is not processed yet. Initially
all vertices are WHITE.

GRAY : Vertex is being processed (DFS for this

vertex has started, but not finished which means
that all descendants (ind DFS tree) of this vertex
are not processed yet (or this vertex is in function
call stack)

BLACK : Vertex and all its descendants are

processed.

While doing DFS, if we encounter an edge from current

vertex to a GRAY vertex, then this edge is back edge
and hence there is a cycle.
Time complexity of above solution is O(V + E) where V is number of vertices and E is number of edges in the
graph.

bool Graph::DFSUtil(int u, int color[])

{
// GRAY : This vertex is being processed (DFS
// for this vertex has started, but not
// ended (or this vertex is in function
// call stack)
color[u] = GRAY;

// Iterate through all adjacent vertices

list<int>::iterator i;
 for (i = adj[u].begin(); i != adj[u].end(); ++i)
{
int v = *i; // An adjacent of u

// If there is
if (color[v] == GRAY)
return true;

// If v is not processed and there is a back

// edge in subtree rooted with v
if (color[v] == WHITE && DFSUtil(v, color))
return true;
}

// Mark this vertex as processed

color[u] = BLACK;

return false;
}

// Returns true if there is a cycle in graph

bool Graph::isCyclic()
{
// Initialize color of all vertices as WHITE
int *color = new int[V];
for (int i = 0; i < V; i++)
color[i] = WHITE;

// Do a DFS traversal beginning with all

// vertices
for (int i = 0; i < V; i++)
if (color[i] == WHITE)
 if (DFSUtil(i, color) == true)
return true;

return false;
}

(* not to worry that the parent of the vertex we are visiting right now would be gray and we would always get a
cycle…..it’s a directed graph so parent won’t be adjacent to vertex until we do have a directed edge from the
vertex to the parent in which case there is gonna be a cycle).

Assign directions to edges so that the directed graph

remains acyclic
Given a graph with both directed and undirected edges. It is given that the directed edges don’t form cycle.
How to assign directions to undirected edges so that the graph (with all directed edges) remains acyclic even
after the assignment?

The idea is to use Topological Sorting. Following are two steps used in the algorithm.

1) Consider the subgraph with directed edges only and find topological sorting of the subgraph. In the above
example, topological sorting is {0, 5, 1, 2, 3, 4}. Below diagram shows topological sorting for the above
example graph.


2) Use above topological sorting to assign directions to undirected edges. For every undirected edge (u, v),
assign it direction from u to v if u comes before v in topological sorting, else assign it direction from v to u.

Below diagram shows assigned directions in the example graph.


Topological Sorting
Topological sorting for Directed Acyclic Graph (DAG) is a linear ordering of vertices such that for every
directed edge uv, vertex u comes before v in the ordering. Topological Sorting for a graph is not possible if the
graph is not a DAG.

For example, a topological sorting of the following graph is “5 4 2 3 1 0”. There can be more than one
topological sorting for a graph. For example, another topological sorting of the following graph is “4 5 2 3 1 0”.
The first vertex in topological sorting is always a vertex with in-degree as 0 (a vertex with no in-coming
edges).
Topological Sorting vs Depth First Traversal (DFS):
In DFS, we print a vertex and then recursively call DFS for its adjacent vertices. In topological sorting, we
need to print a vertex before its adjacent vertices. For example, in the given graph, the vertex ‘5’ should be
printed before vertex ‘0’, but unlike DFS, the vertex ‘4’ should also be printed before vertex ‘0’. So Topological
sorting is different from DFS. For example, a DFS of the shown graph is “5 2 3 1 0 4”, but it is not a
topological sorting

Algorithm to find Topological Sorting:

We recommend to first see implementation of DFS here. We can modify DFS to find Topological Sorting of a
graph. In DFS, we start from a vertex, we first print it and then recursively call DFS for its adjacent vertices. In
topological sorting, we use a temporary stack. We don’t print the vertex immediately, we first recursively call
topological sorting for all its adjacent vertices, then push it to a stack. Finally, print contents of stack. Note that
a vertex is pushed to stack only when all of its adjacent vertices (and their adjacent vertices and so on) are
already in stack.
void topologicalSortUtil(int v, boolean visited[],
Stack stack)
{
// Mark the current node as visited.
visited[v] = true;
Integer i;

// Recur for all the vertices adjacent to this

// vertex
Iterator<Integer> it = adj[v].iterator();
while (it.hasNext())
{
i = it.next();
if (!visited[i])
topologicalSortUtil(i, visited, stack);
 }

// Push current vertex to stack which stores result

stack.push(new Integer(v));
}

// The function to do Topological Sort. It uses

// recursive topologicalSortUtil()
void topologicalSort()
{
Stack stack = new Stack();

// Mark all the vertices as not visited

boolean visited[] = new boolean[V];
for (int i = 0; i < V; i++)
visited[i] = false;

// Call the recursive helper function to store

// Topological Sort starting from all vertices
// one by one
for (int i = 0; i < V; i++)
if (visited[i] == false)
topologicalSortUtil(i, visited, stack);

// Print contents of stack

while (stack.empty()==false)
System.out.print(stack.pop() + " ");
}

Time Complexity:  The above algorithm is simply DFS with an extra stack. So time complexity is same as
DFS which is O(V+E).
Applications:

Topological Sorting is mainly used for scheduling jobs from the given dependencies among jobs. In computer
science, applications of this type arise in instruction scheduling, ordering of formula cell evaluation when
recomputing formula values in spreadsheets, logic synthesis, determining the order of compilation tasks to
perform in makefiles, data serialization, and resolving symbol dependencies in linkers [2].

All Topological Sorts of a Directed Acyclic Graph

Topological sorting for  Directed  Acyclic  Graph (DAG) is a linear ordering of vertices such that for every
directed edge uv, vertex u comes before v in the ordering. Topological Sorting for a graph is not possible if the
graph is not a DAG.

Given a DAG, print all topological sorts of the graph.

Kahn’s algorithm for Topological Sorting

Topological sorting for  Directed  Acyclic  Graph (DAG) is a linear ordering of vertices such that for every
directed edge uv, vertex u comes before v in the ordering. Topological Sorting for a graph is not possible if the
graph is not a DAG.
The approach is based on the below fact :

A DAG G has at least one vertex with in-degree 0 and one vertex with out-degree 0.

Proof: There’s a simple proof to the above fact is that a DAG does not contain a cycle which means that all
paths will be of finite length. Now let S be the longest path from u(source) to v(destination). Since S is the
longest path there can be no incoming edge to u and no outgoing edge from v, if this situation had occurred
then S would not have been the longest path

=> indegree(u) = 0 and outdegree(v) = 0

Algorithm:

Steps involved in finding the topological ordering of a DAG:

Step-1:  Compute in-degree (number of incoming edges) for each of the vertex present in the DAG and
initialize the count of visited nodes as 0.

Step-2: Pick all the vertices with in-degree as 0 and add them into a queue (Enqueue operation)

Step-3: Remove a vertex from the queue (Dequeue operation) and then.

1. Increment count of visited nodes by 1.

2. Decrease in-degree by 1 for all its neighboring nodes.
3. If in-degree of a neighboring nodes is reduced to zero, then add it to the queue.
Step 5: Repeat Step 3 until the queue is empty.
Step 5: If count of visited nodes is not equal to the number of nodes in the graph then the topological sort is
not possible for the given graph.

Time Complexity: The outer for loop will be executed V number of times and the inner for loop will be
executed E number of times, Thus overall time complexity is O(V+E).

The overall time complexity of the algorithm is O(V+E)

Greedy Algorithms | Set 5 (Prim’s Minimum Spanning Tree

(MST))
We have discussed  Kruskal’s algorithm for Minimum Spanning Tree. Like Kruskal’s algorithm, Prim’s
algorithm is also a Greedy algorithm. It starts with an empty spanning tree. The idea is to maintain two sets of
vertices. The first set contains the vertices already included in the MST, the other set contains the vertices not
yet included. At every step, it considers all the edges that connect the two sets, and picks the minimum weight
edge from these edges. After picking the edge, it moves the other endpoint of the edge to the set containing
MST.

How does Prim’s Algorithm Work? The idea behind Prim’s algorithm is simple, a spanning tree means all
vertices must be connected. So the two disjoint subsets (discussed above) of vertices must be connected to
make a  Spanning  Tree. And they must be connected with the minimum weight edge to make it
a Minimum Spanning Tree.
Algorithm

1) Create a set mstSet that keeps track of vertices already included in MST.

2) Assign a key value to all vertices in the input graph. Initialize all key values as INFINITE. Assign key value
as 0 for the first vertex so that it is picked first.

3) While mstSet doesn’t include all vertices

….a) Pick a vertex u which is not there in mstSet and has minimum key value.

….b) Include u to mstSet.

….c)  Update key value of all adjacent vertices of  u. To update the key values, iterate through all adjacent
vertices. For every adjacent vertex v, if weight of edge u-v is less than the previous key value of v, update the
key value as weight of u-v

How to implement the above algorithm?

We use a boolean array mstSet[] to represent the set of vertices included in MST. If a value mstSet[v] is true,
then vertex v is included in MST, otherwise not. Array key[] is used to store key values of all vertices. Another
array parent[] to store indexes of parent nodes in MST. The parent array is the output array which is used to
show the constructed MST.
Time Complexity of the above program is O(V^2). If the input graph is represented using adjacency list, then
the time complexity of Prim’s algorithm can be reduced to O(E log V) with the help of binary heap.

Applications of Minimum Spanning Tree Problem

Minimum Spanning Tree (MST) problem: Given connected graph G with positive edge weights, find a min
weight set of edges that connects all of the vertices.

MST is fundamental problem with diverse applications.

Network design.
– telephone, electrical, hydraulic, TV cable, computer, road
The standard application is to a problem like phone network design. You have a business with several offices;
you want to lease phone lines to connect them up with each other; and the phone company charges different
amounts of money to connect different pairs of cities. You want a set of lines that connects all your offices with
a minimum total cost. It should be a spanning tree, since if a network isn’t a tree you can always remove
some edges and save money.
Approximation algorithms for NP-hard problems.

– traveling salesperson problem, Steiner tree

A less obvious application is that the minimum spanning tree can be used to approximately solve the traveling
salesman problem. A convenient formal way of defining this problem is to find the shortest path that visits
each point at least once.

Note that if you have a path visiting all points exactly once, it’s a special kind of tree. For instance in the
example above, twelve of sixteen spanning trees are actually paths. If you have a path visiting some vertices
more than once, you can always drop some edges to get a tree. So in general the MST weight is less than the
TSP weight, because it’s a minimization over a strictly larger set.
On the other hand, if you draw a path tracing around the minimum spanning tree, you trace each edge twice
and visit all points, so the TSP weight is less than twice the MST weight. Therefore this tour is within a factor
of two of optimal.

Indirect applications.

– max bottleneck paths

– LDPC codes for error correction

– image registration with Renyi entropy

– learning salient features for real-time face verification

– reducing data storage in sequencing amino acids in a protein

– model locality of particle interactions in turbulent fluid flows

– autoconfig protocol for Ethernet bridging to avoid cycles in a network

Cluster analysis

k clustering problem can be viewed as finding an MST and deleting the k-1 most

expensive edges.

Prim’s MST for Adjacency List Representation)

As discussed in the previous post, in Prim’s algorithm, two sets are maintained, one set contains list of
vertices already included in MST, other set contains vertices not yet included. With adjacency list
representation, all vertices of a graph can be traversed in O(V+E) time using BFS. The idea is to traverse all
vertices of graph using BFS and use a Min Heap to store the vertices not yet included in MST. Min Heap is
used as a priority queue to get the minimum weight edge from the cut. Min Heap is used as time complexity
of operations like extracting minimum element and decreasing key value is O(LogV) in Min Heap.

Following are the detailed steps.


1) Create a Min Heap of size V where V is the number of vertices in the given graph. Every node of min heap
contains vertex number and key value of the vertex.

2)  Initialize Min Heap with first vertex as root (the key value assigned to first vertex is 0). The key value
assigned to all other vertices is INF (infinite).

3) While Min Heap is not empty, do following

…..a) Extract the min value node from Min Heap. Let the extracted vertex be u.

…..b) For every adjacent vertex v of u, check if v is in Min Heap (not yet included in MST). If v is in Min Heap
and its key value is more than weight of u-v, then update the key value of v as weight of u-v.

Time Complexity: The time complexity of the above code/algorithm looks O(V^2) as there are two nested
while loops. If we take a closer look, we can observe that the statements in inner loop are executed O(V+E)
times (similar to BFS). The inner loop has decreaseKey() operation which takes O(LogV) time. So overall time
complexity is O(E+V)*O(LogV) which is O((E+V)*LogV) = O(ELogV) (For a connected graph, V = O(E))

void PrimMST(struct Graph* graph)

{
int V = graph->V;// Get the number of vertices in graph
int parent[V]; // Array to store constructed MST
int key[V]; // Key values used to pick minimum weight edge in cut

// minHeap represents set E

struct MinHeap* minHeap = createMinHeap(V);

// Initialize min heap with all vertices. Key value of

// all vertices (except 0th vertex) is initially infinite
for (int v = 1; v < V; ++v)
{
parent[v] = -1;
key[v] = INT_MAX;
minHeap->array[v] = newMinHeapNode(v, key[v]);
minHeap->pos[v] = v;
}

// Make key value of 0th vertex as 0 so that it

// is extracted first
key[0] = 0;
minHeap->array[0] = newMinHeapNode(0, key[0]);
minHeap->pos[0] = 0;

// Initially size of min heap is equal to V

minHeap->size = V;

// In the followin loop, min heap contains all nodes

// not yet added to MST.
while (!isEmpty(minHeap))
{
// Extract the vertex with minimum key value
struct MinHeapNode* minHeapNode = extractMin(minHeap);
int u = minHeapNode->v; // Store the extracted vertex number
 // Traverse through all adjacent vertices of u (the extracted
// vertex) and update their key values
struct AdjListNode* pCrawl = graph->array[u].head;
while (pCrawl != NULL)
{
int v = pCrawl->dest;

// If v is not yet included in MST and weight of u-v is

// less than key value of v, then update key value and
// parent of v
if (isInMinHeap(minHeap, v) && pCrawl->weight < key[v])
{
key[v] = pCrawl->weight;
parent[v] = u;
decreaseKey(minHeap, v, key[v]);
}
pCrawl = pCrawl->next;
}
}

// print edges of MST

printArr(parent, V);

Kruskal’s Minimum Spanning Tree Algorithm)

Below are the steps for finding MST using Kruskal’s algorithm

1. Sort all the edges in non-decreasing order of their weight.

2. Pick the smallest edge. Check if it forms a cycle with the spanning tree
formed so far. If cycle is not formed, include this edge. Else, discard it.

3. Repeat step#2 until there are (V-1) edges in the spanning tree.
The step#2 uses Union-Find algorithm to detect cycle

The algorithm is a Greedy Algorithm. The Greedy Choice is to pick the smallest weight edge that does not
cause a cycle in the MST constructed so far. Let us understand it with an example: Consider the below input
graph.

See code http://www.geeksforgeeks.org/greedy-algorithms-set-2-kruskals-minimum-spanning-tree-mst/

Time Complexity: O(ElogE) or O(ElogV). Sorting of edges takes O(ELogE) time. After sorting, we iterate
through all edges and apply find-union algorithm. The find and union operations can take atmost O(LogV)
time. So overall complexity is O(ELogE + ELogV) time. The value of E can be atmost O(V2), so O(LogV) are
O(LogE) same. Therefore, overall time complexity is O(ElogE) or O(ElogV)

Boruvka’s algorithm)
We have discussed following topics on Minimum Spanning Tree.

Like Prim’s and Kruskal’s, Boruvka’s algorithm is also a Greedy algorithm. Below is complete algorithm.

1) Input is a connected, weighted and directed graph.

2) Initialize all vertices as individual components (or sets).
3) Initialize MST as empty.
4) While there are more than one components, do following
for each component.
a) Find the closest weight edge that connects this
component to any other component.
b) Add this closest edge to MST if not already added.
5) Return MST.
Below is the idea behind above algorithm (The idea is same as Prim’s MST algorithm).

A spanning tree means all vertices must be connected. So the two disjoint subsets (discussed above) of
vertices must be connected to make a Spanning Tree. And they must be connected with the minimum weight
edge to make it a Minimum Spanning Tree.

Interesting Facts about Boruvka’s algorithm:


1) Time Complexity of Boruvka’s algorithm is O(E log V) which is same as Kruskal’s and Prim’s algorithms.

2) Boruvka’s algorithm is used as a step in a faster randomized algorithm that works in linear time O(E).

3) Boruvka’s algorithm is the oldest minimum spanning tree algorithm was discovered by Boruuvka in 1926,
long before computers even existed. The algorithm was published as a method of constructing an efficient
electricity network.

void boruvkaMST(struct Graph* graph)

{
// Get data of given graph
int V = graph->V, E = graph->E;
Edge *edge = graph->edge;

// Allocate memory for creating V subsets.

struct subset *subsets = new subset[V];

// An array to store index of the cheapest edge of

// subset. The stored index for indexing array 'edge[]'
int *cheapest = new int[V];

// Create V subsets with single elements

for (int v = 0; v < V; ++v)
{
subsets[v].parent = v;
subsets[v].rank = 0;
cheapest[v] = -1;
}

// Initially there are V different trees.

// Finally there will be one tree that will be MST
int numTrees = V;
int MSTweight = 0;

// Keep combining components (or sets) until all

// compnentes are not combined into single MST.
while (numTrees > 1)
{
// Traverse through all edges and update
// cheapest of every component
for (int i=0; i<E; i++)
{
// Find components (or sets) of two corners
// of current edge
 int set1 = find(subsets, edge[i].src);
int set2 = find(subsets, edge[i].dest);

// If two corners of current edge belong to

// same set, ignore current edge
if (set1 == set2)
continue;

// Else check if current edge is closer to previous

// cheapest edges of set1 and set2
else
{
if (cheapest[set1] == -1 ||
edge[cheapest[set1]].weight > edge[i].weight)
cheapest[set1] = i;

if (cheapest[set1] == -1 ||
edge[cheapest[set2]].weight > edge[i].weight)
cheapest[set2] = i;
}
}

// Consider the above picked cheapest edges and add them

// to MST
for (int i=0; i<V; i++)
{
// Check if cheapest for current set exists
if (cheapest[i] != -1)
{
int set1 = find(subsets, edge[cheapest[i]].src);
int set2 = find(subsets, edge[cheapest[i]].dest);
 if (set1 == set2)
continue;
MSTweight += edge[cheapest[i]].weight;
printf("Edge %d-%d included in MST\n",
edge[cheapest[i]].src, edge[cheapest[i]].dest,
edge[cheapest[i]].weight);

// Do a union of set1 and set2 and decrease number

// of trees
Union(subsets, set1, set2);
numTrees--;
}
}

Steiner Tree Problem

What is Steiner Tree?

Given a graph and a  subset  of vertices in the graph, a steiner tree spans though the given subset.  The
Steiner Tree may contain some vertices which are not in given subset but are used to connect the vertices of
subset.

The given set of vertices is called Terminal Vertices and other vertices that are used to construct Steiner tree
are called Steiner vertices.

The Steiner Tree Problem is to find the minimum cost Steiner Tree.
If given subset (or terminal) vertices is equal to set of all vertices in Steiner Tree problem, then the problem
becomes Minimum Spanning Tree problem. And if the given subset contains only two vertices, then it shortest
path problem between two vertices.

Finding out Minimum Spanning Tree is polynomial time solvable, but Minimum Steiner Tree problem is NP
Hard and related decision problem is NP-Complete.

Applications of Steiner Tree


Any situation where the task is minimize cost of connection among some important locations like VLSI
Design, Computer Networks, etc.

Shortest Path based Approximate Algorithm


Since Steiner Tree problem is NP-Hard, there are no polynomial time solutions that always give optimal cost.
Therefore, there are approximate algorithms to solve the same. Below is one simple approximate algorithm.

1) Start with a subtree T consisting of

one given terminal vertex
2) While T does not span all terminals
a) Select a terminal x not in T that is closest
to a vertex in T.
b) Add to T the shortest path that connects x with T
The above algorithm is (2-2/n) approximate, i.e., it guarantees that solution produced by this algorithm is not
more than this ration of optimized solution for a given graph with n vertices.(Similar to Dijkstra can be used).
Dijkstra’s shortest path algorithm)
Given a graph and a source vertex in graph, find shortest paths from source to all vertices in the given graph.

Dijkstra’s algorithm is very similar to  Prim’s algorithm for minimum spanning tree. Like Prim’s MST, we
generate a  SPT (shortest path tree)  with given source as root. We maintain two sets, one set contains
vertices included in shortest path tree, other set includes vertices not yet included in shortest path tree. At
every step of the algorithm, we find a vertex which is in the other set (set of not yet included) and has
minimum distance from source.

Below are the detailed steps used in Dijkstra’s algorithm to find the shortest path from a single source vertex
to all other vertices in the given graph.

Algorithm

1) Create a set sptSet (shortest path tree set) that keeps track of vertices included in shortest path tree, i.e.,
whose minimum distance from source is calculated and finalized. Initially, this set is empty.

2) Assign a distance value to all vertices in the input graph. Initialize all distance values as INFINITE. Assign
distance value as 0 for the source vertex so that it is picked first.

3) While sptSet doesn’t include all vertices

….a) Pick a vertex u which is not there in sptSetand has minimum distance value.

….b) Include u to sptSet.

….c)  Update distance value of all adjacent vertices of u. To update the distance values, iterate through all
adjacent vertices. For every adjacent vertex v, if sum of distance value of u (from source) and weight of edge
u-v, is less than the distance value of v, then update the distance value of v.

if (!sptSet[v] && graph[u][v]!=0 &&

dist[u] != Integer.MAX_VALUE &&
dist[u]+graph[u][v] < dist[v])
dist[v] = dist[u] + graph[u][v];

Notes:

1) The code calculates shortest distance, but doesn’t calculate the path information. We can create a parent
array, update the parent array when distance is updated (like  prim’s implementation) and use it show the
shortest path from source to different vertices.

2) The code is for undirected graph, same dijkstra function can be used for directed graphs also.

3) The code finds shortest distances from source to all vertices. If we are interested only in shortest distance
from source to a single target, we can break the for loop when the picked minimum distance vertex is equal to
target (Step 3.a of algorithm).

4) Time Complexity of the implementation is O(V^2). If the input graph is represented using adjacency list, it
can be reduced to O(E log V) with the help of binary heap. Please see

Dijkstra’s Algorithm for Adjacency List Representation for more details.
5) Dijkstra’s algorithm doesn’t work for graphs with negative weight edges. For graphs with negative weight
edges, Bellman–Ford algorithm can be used, we will soon be discussing it as a separate post.

Dijkstra’s Algorithm for Adjacency List Representation

Time Complexity: The time complexity of the above code/algorithm looks O(V^2) as there are two nested
while loops. If we take a closer look, we can observe that the statements in inner loop are executed O(V+E)
times (similar to BFS). The inner loop has decreaseKey() operation which takes O(LogV) time. So overall time
complexity is O(E+V)*O(LogV) which is O((E+V)*LogV) = O(ELogV)
Note that the above code uses Binary Heap for Priority Queue implementation. Time complexity can be
reduced to O(E + VLogV) using Fibonacci Heap. The reason is, Fibonacci Heap takes O(1) time for decrease-
key operation while Binary Heap takes O(Logn) time.

Bellman–Ford Algorithm
Given a graph and a source vertex src in graph, find shortest paths from src to all vertices in the given graph.
The graph may contain negative weight edges.
We have discussed Dijkstra’s algorithm for this problem. Dijksra’s algorithm is a Greedy algorithm and time
complexity is O(VLogV) (with the use of Fibonacci heap). Dijkstra doesn’t work for Graphs with negative
weight edges, Bellman-Ford works for such graphs. Bellman-Ford is also simpler than Dijkstra and suites well
for distributed systems. But time complexity of Bellman-Ford is O(VE), which is more than Dijkstra.
Input: Graph and a source vertex src

Output: Shortest distance to all vertices from src. If there is a negative weight cycle, then shortest distances
are not calculated, negative weight cycle is reported.
1)  This step initializes distances from source to all vertices as infinite and distance to source itself as 0.
Create an array dist[] of size |V| with all values as infinite except dist[src] where src is source vertex.

2)  This step calculates shortest distances. Do following |V|-1 times where |V| is the number of vertices in
given graph.

…..a) Do following for each edge u-v

………………If dist[v] > dist[u] + weight of edge uv, then update dist[v]

………………….dist[v] = dist[u] + weight of edge uv

3) This step reports if there is a negative weight cycle in graph. Do following for each edge u-v

……If dist[v] > dist[u] + weight of edge uv, then “Graph contains negative weight cycle”

The idea of step 3 is, step 2 guarantees shortest distances if graph doesn’t contain negative weight cycle. If
we iterate through all edges one more time and get a shorter path for any vertex, then there is a negative
weight cycle

void BellmanFord(Graph graph,int src)

{
int V = graph.V, E = graph.E;
int dist[] = new int[V];

// Step 1: Initialize distances from src to all other

// vertices as INFINITE
for (int i=0; i<V; ++i)
dist[i] = Integer.MAX_VALUE;
dist[src] = 0;
 // Step 2: Relax all edges |V| - 1 times. A simple
// shortest path from src to any other vertex can
// have at-most |V| - 1 edges
for (int i=1; i<V; ++i)
{
for (int j=0; j<E; ++j)
{
int u = graph.edge[j].src;
int v = graph.edge[j].dest;
int weight = graph.edge[j].weight;
if (dist[u]!=Integer.MAX_VALUE &&
dist[u]+weight<dist[v])
dist[v]=dist[u]+weight;
}
}

// Step 3: check for negative-weight cycles. The above

// step guarantees shortest distances if graph doesn't
// contain negative weight cycle. If we get a shorter
// path, then there is a cycle.
for (int j=0; j<E; ++j)
{
int u = graph.edge[j].src;
int v = graph.edge[j].dest;
int weight = graph.edge[j].weight;
if (dist[u]!=Integer.MAX_VALUE &&
dist[u]+weight<dist[v])
System.out.println("Graph contains negative weight cycle");
}
printArr(dist, V);
}
Floyd Warshall Algorithm)
The  Floyd Warshall Algorithm  is for solving the All Pairs Shortest Path problem. The problem is to find
shortest distances between every pair of vertices in a given edge weighted directed Graph.

We initialize the solution matrix same as the input graph matrix as a first step. Then we update the solution
matrix by considering all vertices as an intermediate vertex. The idea is to one by one pick all vertices and
update all shortest paths which include the picked vertex as an intermediate vertex in the shortest path. When
we pick vertex number k as an intermediate vertex, we already have considered vertices {0, 1, 2, .. k-1} as
intermediate vertices. For every pair (i, j) of source and destination vertices respectively, there are two
possible cases.
1) k is not an intermediate vertex in shortest path from i to j. We keep the value of dist[i][j] as it is.
2) k is an intermediate vertex in shortest path from i to j. We update the value of dist[i][j] as dist[i][k] + dist[k][j].

Time Complexity: O(V^3)

The above program only prints the shortest distances. We can modify the solution to print the shortest paths
also by storing the predecessor information in a separate 2D matrix.

for (k = 0; k < V; k++)

{
// Pick all vertices as source one by one
for (i = 0; i < V; i++)
{
// Pick all vertices as destination for the
// above picked source
for (j = 0; j < V; j++)
 {
// If vertex k is on the shortest path from
// i to j, then update the value of dist[i][j]
if (dist[i][k] + dist[k][j] < dist[i][j])
dist[i][j] = dist[i][k] + dist[k][j];
}

Johnson’s algorithm for All-pairs shortest paths

The problem is to find shortest paths between every pair of vertices in a given weighted directed Graph and
weights may be negative. We have discussed Floyd Warshall Algorithm for this problem. Time complexity of
Floyd Warshall Algorithm is Θ(V3). Using Johnson’s algorithm, we can find all pair shortest paths in O(V2log V
+ VE) time. Johnson’s algorithm uses both Dijkstra and Bellman-Ford as subroutines.

If we apply  Dijkstra’s Single Source shortest path algorithm  for every vertex, considering every vertex as
source, we can find all pair shortest paths in O(V*VLogV) time. So using Dijkstra’s single source shortest path
seems to be a better option than Floyd Warshell, but the problem with Dijkstra’s algorithm is, it doesn’t work
for negative weight edge.

The idea of Johnson’s algorithm is to re-weight all edges and make them all positive, then apply Dijkstra’s
algorithm for every vertex.

How to transform a given graph to a graph with all non-negative weight edges?

One may think of a simple approach of finding the minimum weight edge and adding this weight to all edges.
Unfortunately, this doesn’t work as there may be different number of edges in different paths (See this for an
example). If there are multiple paths from a vertex u to v, then all paths must be increased by same amount,
so that the shortest path remains the shortest in the transformed graph.

The idea of Johnson’s algorithm is to assign a weight to every vertex. Let the weight assigned to vertex u be
h[u]. We reweight edges using vertex weights. For example, for an edge (u, v) of weight w(u, v), the new
weight becomes w(u, v) + h[u] – h[v]. The great thing about this reweighting is, all set of paths between any
two vertices are increased by same amount and all negative weights become non-negative. Consider any
path between two vertices s and t, weight of every path is increased by h[s] – h[t], all h[] values of vertices on
path from s to t cancel each other.

How do we calculate h[] values? Bellman-Ford algorithm is used for this purpose. Following is the complete
algorithm. A new vertex is added to the graph and connected to all existing vertices. The shortest distance
values from new vertex to all existing vertices are h[] values.

Algorithm:

1) Let the given graph be G. Add a new vertex s to the graph, add edges from new vertex to all vertices of G.
Let the modified graph be G’.

2) Run Bellman-Ford algorithm on G’ with s as source. Let the distances calculated by Bellman-Ford be h[0],
h[1], .. h[V-1]. If we find a negative weight cycle, then return. Note that the negative weight cycle cannot be
created by new vertex s as there is no edge to s. All edges are from s.
3) Reweight the edges of original graph. For each edge (u, v), assign the new weight as “original weight + h[u]
– h[v]”.

4) Remove the added vertex s and run Dijkstra’s algorithm for every vertex.

How does the transformation ensure nonnegative weight edges?


The following property is always true about h[] values as they are shortest distances.

h[v] <= h[u] + w(u, v)

The property simply means, shortest distance from s to v must be smaller than or equal to shortest distance
from s to u plus weight of edge (u, v). The new weights are w(u, v) + h[u] - h[v]. The value of the new weights
must be greater than or equal to zero because of the inequality "h[v] <= h[u] + w(u, v)"

Time Complexity: The main steps in algorithm are Bellman Ford Algorithm called once and Dijkstra called V
times. Time complexity of Bellman Ford is O(VE) and time complexity of Dijkstra is O(VLogV). So overall time
complexity is O(V2log V + VE).

The time complexity of Johnson's algorithm becomes same as Floyd Warshell when the graphs is complete
(For a complete graph E = O(V2). But for sparse graphs, the algorithm performs much better than  Floyd
Warshell.

Shortest Path in Directed Acyclic Graph

Given a Weighted Directed Acyclic Graph and a source vertex in the graph, find the shortest paths from given
source to all other vertices.

For a general weighted graph, we can calculate single source shortest distances in O(VE) time
using Bellman–Ford Algorithm. For a graph with no negative weights, we can do better and calculate single
source shortest distances in O(E + VLogV) time using Dijkstra’s algorithm. Can we do even better for Directed
Acyclic Graph (DAG)? We can calculate single source shortest distances in O(V+E) time for DAGs. The idea
is to use Topological Sorting.

We initialize distances to all vertices as infinite and distance to source as 0, then we find a topological sorting
of the graph. Topological Sorting of a graph represents a linear ordering of the graph (See below, figure (b) is
a linear representation of figure (a) ). Once we have topological order (or linear representation), we one by
one process all vertices in topological order. For every vertex being processed, we update distances of its
adjacent using distance of current vertex.

Following is complete algorithm for finding shortest distances.

1) Initialize dist[] = {INF, INF, ….} and dist[s] = 0 where s is the source vertex.
2) Create a toplogical order of all vertices.
3) Do following for every vertex u in topological order.
………..Do following for every adjacent vertex v of u
………………if (dist[v] > dist[u] + weight(u, v))
………………………dist[v] = dist[u] + weight(u, v)
Time Complexity:  Time complexity of topological sorting is O(V+E). After finding topological order, the
algorithm process all vertices and for every vertex, it runs a loop for all adjacent vertices. Total adjacent
vertices in a graph is O(E). So the inner loop runs O(V+E) times. Therefore, overall time complexity of this
algorithm is O(V+E).

Some interesting shortest path questions | Set 1

Question 1:  Given a directed weighted graph. You are also given the shortest path from a source
vertex ‘s’ to a destination vertex ‘t’.  If weight of every edge is increased by 10 units, does the shortest
path remain same in the modified graph?

The shortest path may change. The reason is, there may be different number of edges in different paths from
s to t. For example, let shortest path be of weight 15 and has 5 edges. Let there be another path with 2 edges
and total weight 25. The weight of the shortest path is increased by 5*10 and becomes 15 + 50. Weight of the
other path is increased by 2*10 and becomes 25 + 20. So the shortest path changes to the other path with
weight as 45.

Question 2:  This is similar to above question. Does the shortest path change when weights of all
edges are multiplied by 10?

If we multiply all edge weights by 10, the shortest path doesn’t change. The reason is simple, weights of all
paths from s to t get multiplied by same amount. The number of edges on a path doesn’t matter. It is like
changing unit of weights.
Question 3:  Given a directed graph where every edge has weight as either 1 or 2, find the shortest
path from a given source vertex ‘s’ to a given destination vertex ‘t’. Expected time complexity is
O(V+E).

If we apply Dijkstra’s shortest path algorithm, we can get a shortest path in O(E + VLogV) time. How to do it in
O(V+E) time? The idea is to use BFS . One important observation about BFS is, the path used in BFS always
has least number of edges between any two vertices. So if all edges are of same weight, we can use BFS to
find the shortest path. For this problem, we can modify the graph and split all edges of weight 2 into two
edges of weight 1 each. In the modified graph, we can use BFS to find the shortest path. How is this
approach O(V+E)? In worst case, all edges are of weight 2 and we need to do O(E) operations to split all
edges, so the time complexity becomes O(E) + O(V+E) which is O(V+E).

Dial’s Algorithm (Optimized Dijkstra for small range

weights)
Can we optimize Dijkstra’s shortest path algorithm to work better than O(E log V) if maximum weight
is small (or range of edge weights is small)?

For example, in the above diagram, maximum weight is 14. Many a times the range of weights on edges in is
in small range (i.e. all edge weight can be mapped to 0, 1, 2.. w where w is a small number). In that case,
Dijkstra’s algorithm can be modified by using different data structure, buckets, which is called dial
implementation of dijkstra’s algorithm. time complexity is  O(E + WV)  where W is maximum weight on any
edge of graph, so we can see that, if W is small then this implementation runs much faster than traditional
algorithm. Following are important observations.

▪ Maximum distance between any two node can be at max w(V – 1) (w is maximum edge weight and we
can have at max V-1 edges between two vertices).
▪ In Dijkstra algorithm, distances are finalized in non-decreasing, i.e., distance of the closer (to given
source) vertices is finalized before the distant vertices.

Algorithm

Below is complete algorithm:

1. Maintains some buckets, numbered 0, 1, 2,…,wV.

2. Bucket k contains all temporarily labeled nodes with distance equal to k.
3. Nodes in each bucket are represented by list of vertices.
4. Buckets 0, 1, 2,..wV are checked sequentially until the first non-empty bucket is found. Each node
contained in the first non-empty bucket has the minimum distance label by definition.
5. One by one, these nodes with minimum distance label are permanently labeled and deleted from the
bucket during the scanning process.
 6. Thus operations involving vertex include:
▪ Checking if a bucket is empty
▪ Adding a vertex to a bucket
▪ Deleting a vertex from a bucket.
7. The position of a temporarily labeled vertex in the buckets is updated accordingly when the distance
label of a vertex changes.
8. Process repeated until all vertices are permanently labeled (or distances of all vertices are finalized).
http://www.geeksforgeeks.org/dials-algorithm-optimized-dijkstra-for-small-range-weights/
class Graph
{
int V; // No. of vertices

// In a weighted graph, we need to store vertex

// and weight pair for every edge
list< pair<int, int> > *adj;

Printing Paths in Dijkstra’s Shortest Path Algorithm

Given a graph and a source vertex in graph, find shortest paths from source to all vertices in the given graph.

The idea is to create a separate array parent[]. Value of parent[v] for a vertex v stores parent vertex of v in
shortest path tree. Parent of root (or source vertex) is -1. Whenever we find shorter path through a vertex u,
we make u as parent of current vertex.
Once we have parent array constructed, we can print path using below recursive function.

void printPath(int parent[], int j)

{
// Base Case : If j is source
if (parent[j]==-1)
return;

printPath(parent, parent[j]);

printf("%d ", j);

}

Shortest Path in a weighted Graph where weight of an edge is 1

or 2
Given a directed graph where every edge has weight as either 1 or 2, find the shortest path from a given
source vertex ‘s’ to a given destination vertex ‘t’. Expected time complexity is O(V+E).

A Simple Solution is to use Dijkstra’s shortest path algorithm, we can get a shortest path in O(E + VLogV)
time.

How to do it in O(V+E) time?  The idea is to use  BFS. One important observation about BFS is, the path
used in BFS always has least number of edges between any two vertices. So if all edges are of same weight,
we can use BFS to find the shortest path. For this problem, we can modify the graph and split all edges of
weight 2 into two edges of weight 1 each. In the modified graph, we can use BFS to find the shortest path.

How many new intermediate vertices are needed?  We need to add a new intermediate vertex for every
source vertex. The reason is simple, if we add a intermediate vertex x between u and v and if we add same
vertex between y and z, then new paths u to z and y to v are added to graph which might have note been
there in original graph. Therefore in a graph with V vertices, we need V extra vertices.

How is this approach O(V+E)? In worst case, all edges are of weight 2 and we need to do O(E) operations to
split all edges and 2V vertices, so the time complexity becomes O(E) + O(V+E) which is O(V+E).

Find if there is a path between two vertices in a directed

graph
Given a Directed Graph and two vertices in it, check whether there is a path from the first given vertex to
second.
We can either use Breadth First Search (BFS) or Depth First Search (DFS) to find path between two vertices.
Take the first vertex as source in BFS (or DFS), follow the standard BFS (or DFS). If we see the second
vertex in our traversal, then return true. Else return false.
private LinkedList<Integer> adj[]; //Adjacency List

//Constructor
Graph(int v)
{
V = v;
 adj = new LinkedList[v];
for (int i=0; i<v; ++i)
adj[i] = new LinkedList();
}
while (queue.size()!=0)
{
// Dequeue a vertex from queue and print it
s = queue.poll();

int n;
i = adj[s].listIterator();

// Get all adjacent vertices of the dequeued vertex s

// If a adjacent has not been visited, then mark it
// visited and enqueue it
while (i.hasNext())
{
n = i.next();

// If this adjacent node is the destination node,

// then return true
if (n==d)
return true;

// Else, continue to do BFS

if (!visited[n])
{
visited[n] = true;
queue.add(n);
}
Check if a graph is strongly connected | Set 1 (Kosaraju
using DFS)
Given a directed graph, find out whether the graph is strongly connected or not. A directed graph is strongly
connected if there is a path between any two pair of vertices.
It is easy for undirected graph, we can just do a BFS and DFS starting from any vertex. If BFS or DFS visits
all vertices, then the given undirected graph is connected. This approach won’t work for a directed graph.
A simple idea is to use a all pair shortest path algorithm like Floyd Warshall or find Transitive Closure of
graph. Time complexity of this method would be O(v3).
A better idea can be Strongly Connected Components (SCC) algorithm. We can find all SCCs in O(V+E)
time. If number of SCCs is one, then graph is strongly connected. The algorithm for SCC does extra work as it
finds all SCCs.
Following is Kosaraju’s DFS based simple algorithm that does two DFS traversals of graph:
1) Initialize all vertices as not visited.
2) Do a DFS traversal of graph starting from any arbitrary vertex v. If DFS traversal doesn’t visit all vertices,
then return false.
3) Reverse all arcs (or find transpose or reverse of graph)
4) Mark all vertices as not-visited in reversed graph.
5) Do a DFS traversal of reversed graph starting from same vertex v (Same as step 2). If DFS traversal
doesn’t visit all vertices, then return false. Otherwise return true.
Graph getTranspose()
{
Graph g = new Graph(V);
for (int v = 0; v < V; v++)
{
// Recur for all the vertices adjacent to this vertex
Iterator<Integer> i = adj[v].listIterator();
 while (i.hasNext())
g.adj[i.next()].add(v);
}
return g;
}
Time Complexity: Time complexity of above implementation is sane as Depth First Search which is O(V+E)
if the graph is represented using adjacency list representation.
The above approach requires two traversals of graph. We can find whether a graph is strongly connected or
not in one traversal using Tarjan’s Algorithm to find Strongly Connected Components.

Articulation Points (or Cut Vertices) in a Graph

A vertex in an undirected connected graph is an articulation point (or cut vertex) iff removing it (and edges
through it) disconnects the graph. Articulation points represent vulnerabilities in a connected network – single
points whose failure would split the network into 2 or more disconnected components. They are useful for
designing reliable networks.
For a disconnected undirected graph, an articulation point is a vertex removing which increases number of
connected components.
How to find all articulation points in a given graph?
A simple approach is to one by one remove all vertices and see if removal of a vertex causes disconnected
graph. Following are steps of simple approach for connected graph.
1) For every vertex v, do following
…..a) Remove v from graph
..…b) See if the graph remains connected (We can either use BFS or DFS)
…..c) Add v back to the graph
Time complexity of above method is O(V*(V+E)) for a graph represented using adjacency list. Can we do
better?
A O(V+E) algorithm to find all Articulation Points (APs)
The idea is to use DFS (Depth First Search). In DFS, we follow vertices in tree form called DFS tree. In DFS
tree, a vertex u is parent of another vertex v, if v is discovered by u (obviously v is an adjacent of u in graph).
In DFS tree, a vertex u is articulation point if one of the following two conditions is true.
1) u is root of DFS tree and it has at least two children.
2) u is not root of DFS tree and it has a child v such that no vertex in subtree rooted with v has a back edge to
one of the ancestors (in DFS tree) of u.
void APUtil(int u, boolean visited[], int disc[],
int low[], int parent[], boolean ap[])
{

// Count of children in DFS Tree

int children = 0;

// Mark the current node as visited

visited[u] = true;

// Initialize discovery time and low value

disc[u] = low[u] = ++time;

// Go through all vertices aadjacent to this

Iterator<Integer> i = adj[u].iterator();
while (i.hasNext())
{
int v = i.next(); // v is current adjacent of u

// If v is not visited yet, then make it a child of u

// in DFS tree and recur for it
if (!visited[v])
{
 children++;
parent[v] = u;
APUtil(v, visited, disc, low, parent, ap);

// Check if the subtree rooted with v has a connection to

// one of the ancestors of u
low[u] = Math.min(low[u], low[v]);

// u is an articulation point in following cases

// (1) u is root of DFS tree and has two or more chilren.

if (parent[u] == NIL && children > 1)
ap[u] = true;

// (2) If u is not root and low value of one of its child

// is more than discovery value of u.
if (parent[u] != NIL && low[v] >= disc[u])
ap[u] = true;
}

// Update low value of u for parent function calls.

else if (v != parent[u])
low[u] = Math.min(low[u], disc[v]);
}
}

// The function to do DFS traversal. It uses recursive function APUtil()

void AP()
{
// Mark all the vertices as not visited
boolean visited[] = new boolean[V];
int disc[] = new int[V];
 int low[] = new int[V];
int parent[] = new int[V];
boolean ap[] = new boolean[V]; // To store articulation points

// Initialize parent and visited, and ap(articulation point)

// arrays
for (int i = 0; i < V; i++)
{
parent[i] = NIL;
visited[i] = false;
ap[i] = false;
}

// Call the recursive helper function to find articulation

// points in DFS tree rooted with vertex 'i'
for (int i = 0; i < V; i++)
if (visited[i] == false)
APUtil(i, visited, disc, low, parent, ap);

// Now ap[] contains articulation points, print them

for (int i = 0; i < V; i++)
if (ap[i] == true)
System.out.print(i+" ");
}
Time Complexity: The above function is simple DFS with additional arrays. So time complexity
is same as DFS which is O(V+E) for adjacency list representation of graph.

Biconnected graph
An undirected graph is called Biconnected if there are two vertex-disjoint paths between any two vertices. In a
Biconnected Graph, there is a simple cycle through any two vertices.
By convention, two nodes connected by an edge form a biconnected graph, but this does not verify the above
properties. For a graph with more than two vertices, the above properties must be there for it to be
Biconnected.
A connected graph is Biconnected if it is connected and doesn’t have any Articulation Point. We mainly need
to check two things in a graph.
1) The graph is connected.
2) There is not articulation point in graph.
We start from any vertex and do DFS traversal. In DFS traversal, we check if there is any articulation point. If
we don’t find any articulation point, then the graph is Biconnected. Finally, we need to check whether all
vertices were reachable in DFS or not. If all vertices were not reachable, then the graph is not even
connected.
Time Complexity: The above function is a simple DFS with additional arrays. So time complexity is same as
DFS which is O(V+E) for adjacency list representation of graph.

Bridges in a graph
An edge in an undirected connected graph is a bridge iff removing it disconnects the graph.  For a
disconnected undirected graph, definition is similar, a bridge is an edge removing which increases number of
connected components.

Like Articulation Points,bridges represent vulnerabilities in a connected network and are useful for designing
reliable networks. For example, in a wired computer network, an articulation point indicates the critical
computers and a bridge indicates the critical wires or connections.

A O(V+E) algorithm to find all Bridges

The idea is similar to O(V+E) algorithm for Articulation Points. We do DFS traversal of the given graph. In
DFS tree an edge (u, v) (u is parent of v in DFS tree) is bridge if there does not exit any other alternative to
reach u or an ancestor of u from subtree rooted with v. As discussed in the previous post, the value low[v]
indicates earliest visited vertex reachable from subtree rooted with v. The condition for an edge (u, v) to be a
bridge is, “low[v] > disc[u]”.
while (i.hasNext())
{
int v = i.next(); // v is current adjacent of u

// If v is not visited yet, then make it a child

// of u in DFS tree and recur for it.
// If v is not visited yet, then recur for it
if (!visited[v])
{
parent[v] = u;
bridgeUtil(v, visited, disc, low, parent);

// Check if the subtree rooted with v has a

// connection to one of the ancestors of u
low[u] = Math.min(low[u], low[v]);

// If the lowest vertex reachable from subtree

// under v is below u in DFS tree, then u-v is
// a bridge
if (low[v] > disc[u])
System.out.println(u+" "+v);
}

// Update low value of u for parent function calls.

else if (v != parent[u])
low[u] = Math.min(low[u], disc[v]);
 }

Eulerian path and circuit for undirected graph

Eulerian Path is a path in graph that visits every edge exactly once. Eulerian Circuit is an Eulerian Path which
starts and ends on the same vertex.

The problem is same as following question. “Is it possible to draw a given graph without lifting pencil from the
paper and without tracing any of the edges more than once”.

A graph is called Eulerian if it has an Eulerian Cycle and called Semi-Eulerian if it has an Eulerian Path. The
problem seems similar to Hamiltonian Path which is NP complete problem for a general graph. Fortunately,
we can find whether a given graph has a Eulerian Path or not in polynomial time. In fact, we can find it in
O(V+E) time.

Following are some interesting properties of undirected graphs with an Eulerian path and cycle. We can use
these properties to find whether a graph is Eulerian or not.

Eulerian Cycle

An undirected graph has Eulerian cycle if following two conditions are true.

….a) All vertices with non-zero degree are connected. We don’t care about vertices with zero degree because
they don’t belong to Eulerian Cycle or Path (we only consider all edges).

….b) All vertices have even degree.
Eulerian Path

An undirected graph has Eulerian Path if following two conditions are true.

….a) Same as condition (a) for Eulerian Cycle

….b) If zero or two vertices have odd degree and all other vertices have even degree. Note that only one
vertex with odd degree is not possible in an undirected graph (sum of all degrees is always even in an
undirected graph)

Note that a graph with no edges is considered Eulerian because there are no edges to traverse

Time Complexity: O(V+E)

boolean isConnected()
{
// Mark all the vertices as not visited
boolean visited[] = new boolean[V];
int i;
for (i = 0; i < V; i++)
visited[i] = false;

// Find a vertex with non-zero degree

for (i = 0; i < V; i++)
if (adj[i].size() != 0)
break;

// If there are no edges in the graph, return true

if (i == V)
return true;
 // Start DFS traversal from a vertex with non-zero degree
DFSUtil(i, visited);

// Check if all non-zero degree vertices are visited

for (i = 0; i < V; i++)
if (visited[i] == false && adj[i].size() > 0)
return false;

return true;
}

/* The function returns one of the following values

0 --> If grpah is not Eulerian
1 --> If graph has an Euler path (Semi-Eulerian)
2 --> If graph has an Euler Circuit (Eulerian) */
int isEulerian()
{
// Check if all non-zero degree vertices are connected
if (isConnected() == false)
return 0;

// Count vertices with odd degree

int odd = 0;
for (int i = 0; i < V; i++)
if (adj[i].size()%2!=0)
odd++;

// If count is more than 2, then graph is not Eulerian

if (odd > 2)
return 0;
 // If odd count is 2, then semi-eulerian.
// If odd count is 0, then eulerian
// Note that odd count can never be 1 for undirected graph
return (odd==2)? 1 : 2;
}

Fleury’s Algorithm for printing Eulerian Path or Circuit

Eulerian Path is a path in graph that visits every edge exactly once. Eulerian Circuit is an Eulerian Path which
starts and ends on the same vertex.

In this post, an algorithm to print Eulerian trail or circuit is discussed.

Following is Fleury’s Algorithm for printing Eulerian trail or cycle (Source Ref1).

1. Make sure the graph has either 0 or 2 odd vertices.

2. If there are 0 odd vertices, start anywhere. If there are 2 odd vertices, start at one of them.

3. Follow edges one at a time. If you have a choice between a bridge and a non-bridge, always choose the
non-bridge.

4. Stop when you run out of edges.

The idea is, “don’t burn bridges“ so that we can come back to a vertex and traverse remaining edges. 
We first find the starting point which must be an odd vertex (if there are odd vertices) and store it in variable
‘u’. If there are zero odd vertices, we start from vertex ‘0’. We call printEulerUtil() to print Euler tour starting
with u. We traverse all adjacent vertices of u, if there is only one adjacent vertex, we immediately consider it.
If there are more than one adjacent vertices, we consider an adjacent v only if edge u-v is not a bridge. How
to find if a given is edge is bridge? We count number of vertices reachable from u. We remove edge u-v and
again count number of reachable vertices from u. If number of reachable vertices are reduced, then edge u-v
is a bridge. To count reachable vertices, we can either use BFS or DFS, we have used DFS in the above
code. The function DFSCount(u) returns number of vertices reachable from u.

Once an edge is processed (included in Euler tour), we remove it from the graph. To remove the edge, we
replace the vertex entry with -1 in adjacency list. Note that simply deleting the node may not work as the code
is recursive and a parent call may be in middle of adjacency list.

http://www.geeksforgeeks.org/fleurys-algorithm-for-printing-eulerian-path/
Note that the above code modifies given graph, we can create a copy of graph if we don’t want the given
graph to be modified.

Time Complexity: Time complexity of the above implementation is O ((V+E)2). The function printEulerUtil() is
like DFS and it calls isValidNextEdge() which also does DFS two times. Time complexity of DFS for adjacency
list representation is O(V+E). Therefore overall time complexity is O((V+E)*(V+E)) which can be written as
O(E2) for a connected graph.
Strongly Connected Components
A directed graph is strongly connected if there is a path between all pairs of vertices. A strongly connected
component (SCC) of a directed graph is a maximal strongly connected subgraph. For example, there are 3
SCCs in the following graph.

We can find all strongly connected components in O(V+E) time using  Kosaraju’s algorithm. Following is
detailed Kosaraju’s algorithm.

1) Create an empty stack ‘S’ and do DFS traversal of a graph. In DFS traversal, after calling recursive DFS for
adjacent vertices of a vertex, push the vertex to stack. In the above graph, if we start DFS from vertex 0, we
get vertices in stack as 1, 2, 4, 3, 0.

2) Reverse directions of all arcs to obtain the transpose graph.

3) One by one pop a vertex from S while S is not empty. Let the popped vertex be ‘v’. Take v as source and
do DFS (call  DFSUtil(v)). The DFS starting from v prints strongly connected component of v. In the above
example, we process vertices in order 0, 3, 4, 2, 1 (One by one popped from stack).

How does this work?

The above algorithm is DFS based. It does DFS two times. DFS of a graph produces a single tree if all
vertices are reachable from the DFS starting point. Otherwise DFS produces a forest. So DFS of a graph with


only one SCC always produces a tree. The important point to note is DFS may produce a tree or a
forest when there are more than one SCCs depending upon the chosen starting point. For example, in
the above diagram, if we start DFS from vertices 0 or 1 or 2, we get a tree as output. And if we start
from 3 or 4, we get a forest. To find and print all SCCs, we would want to start DFS from vertex 4
(which is a sink vertex), then move to 3 which is sink in the remaining set (set excluding 4) and finally
any of the remaining vertices (0, 1, 2). So how do we find this sequence of picking vertices as starting
points of DFS? Unfortunately, there is no direct way for getting this sequence. However, if we do a
DFS of graph and store vertices according to their finish times, we make sure that the finish time of a
vertex that connects to other SCCs (other that its own SCC), will always be greater than finish time of
vertices in the other SCC (See this for proof). For example, in DFS of above example graph, finish time
of 0 is always greater than 3 and 4 (irrespective of the sequence of vertices considered for DFS). And
finish time of 3 is always greater than 4. DFS doesn’t guarantee about other vertices, for example
finish times of 1 and 2 may be smaller or greater than 3 and 4 depending upon the sequence of
vertices considered for DFS. So to use this property, we do DFS traversal of complete graph and push
every finished vertex to a stack. In stack, 3 always appears after 4, and 0 appear after both 3 and 4.

In the next step, we reverse the graph. Consider the graph of SCCs. In the reversed graph, the edges that
connect two components are reversed. So the SCC {0, 1, 2} becomes sink and the SCC {4} becomes source.
As discussed above, in stack, we always have 0 before 3 and 4. So if we do a DFS of the reversed graph
using sequence of vertices in stack, we process vertices from sink to source (in reversed graph). That is what
we wanted to achieve and that is all needed to print SCCs one by one

Time Complexity: The above algorithm calls DFS, fins reverse of the graph and again calls DFS. DFS takes
O(V+E) for a graph represented using adjacency list. Reversing a graph also takes O(V+E) time. For
reversing the graph, we simple traverse all adjacency lists.
The above algorithm is asymptotically best algorithm, but there are other algorithms like  Tarjan’s
algorithmand  path-based  which have same time complexity but find SCCs using single DFS. The Tarjan’s
algorithm is discussed in the following post.

Applications:
SCC algorithms can be used as a first step in many graph algorithms that work only on strongly connected
graph.
In social networks, a group of people are generally strongly connected (For example, students of a class or
any other common place). Many people in these groups generally like some common pages or play common
games. The SCC algorithms can be used to find such groups and suggest the commonly liked pages or
games to the people in the group who have not yet liked commonly liked a page or played a game.
http://www.geeksforgeeks.org/strongly-connected-components/

Transitive closure of a graph

Given a directed graph, find out if a vertex j is reachable from another vertex i for all vertex pairs (i, j) in the
given graph. Here reachable mean that there is a path from vertex i to j.  The reach-ability matrix is called
transitive closure of a graph.

The graph is given in the form of adjacency matrix say ‘graph[V][V]’ where graph[i][j] is 1 if there is an edge from
vertex i to vertex j or i is equal to j, otherwise graph[i][j] is 0.
Floyd Warshall Algorithm can be used, we can calculate the distance matrix dist[V][V] using Floyd Warshall, if dist[i][j]
is infinite, then j is not reachable from i, otherwise j is reachable and value of dist[i][j] will be less than V.

Instead of directly using Floyd Warshall, we can optimize it in terms of space and time, for this particular problem.
Following are the optimizations:

1) Instead of integer resultant matrix (dist[V][V] in floyd warshall), we can create a boolean reach-ability matrix
reach[V][V] (we save space). The value reach[i][j] will be 1 if j is reachable from i, otherwise 0.

2) Instead of using arithmetic operations, we can use logical operations. For arithmetic operation ‘+’, logical and ‘&&’ is
used, and for min, logical or ‘||’ is used. (We save time by a constant factor. Time complexity is same though)

for (k = 0; k < V; k++)

{
// Pick all vertices as source one by one
for (i = 0; i < V; i++)
{
// Pick all vertices as destination for the
// above picked source
for (j = 0; j < V; j++)
{
// If vertex k is on a path from i to j,
// then make sure that the value of reach[i][j] is 1
reach[i][j] = (reach[i][j]!=0) ||
((reach[i][k]!=0) && (reach[k][j]!=0))?1:0;
}
}
Time Complexity: O(V3) where V is number of vertices in the given graph.

Find the number of islands

Given a boolean 2D matrix, find the number of islands. A group of connected 1s forms an island. For
example, the below matrix contains 5 islands

Input : mat[][] = {{1, 1, 0, 0, 0},

{0, 1, 0, 0, 1},
{1, 0, 0, 1, 1},
{0, 0, 0, 0, 0},
{1, 0, 1, 0, 1}
Output : 5
This is an variation of the standard problem: “Counting number of connected components in a undirected
graph”.

A graph where all vertices are connected with each other, has exactly one connected component, consisting
of the whole graph. Such graph with only one connected component is called as Strongly Connected Graph.

The problem can be easily solved by applying DFS() on each component. In each DFS() call, a component or
a sub-graph is visited. We will call DFS on the next un-visited component. The number of calls to DFS() gives
the number of connected components. BFS can also be used.
A cell in 2D matrix can be connected to 8 neighbors. So, unlike standard DFS(), where we recursively call for all
adjacent vertices, here we can recursive call for 8 neighbors only. We keep track of the visited 1s so that they are not
visited again.

void DFS(int M[][], int row, int col, boolean visited[][])

{
// These arrays are used to get row and column numbers
// of 8 neighbors of a given cell
int rowNbr[] = new int[] {-1, -1, -1, 0, 0, 1, 1, 1};
int colNbr[] = new int[] {-1, 0, 1, -1, 1, -1, 0, 1};

// Mark this cell as visited

visited[row][col] = true;

// Recur for all connected neighbours

for (int k = 0; k < 8; ++k)
if (isSafe(M, row + rowNbr[k], col + colNbr[k], visited) )
DFS(M, row + rowNbr[k], col + colNbr[k], visited);
}

// The main function that returns count of islands in a given

// boolean 2D matrix
int countIslands(int M[][])
{
// Make a bool array to mark visited cells.
// Initially all cells are unvisited
boolean visited[][] = new boolean[ROW][COL];
 // Initialize count as 0 and travese through the all cells
// of given matrix
int count = 0;
for (int i = 0; i < ROW; ++i)
for (int j = 0; j < COL; ++j)
if (M[i][j]==1 && !visited[i][j]) // If a cell with
{ // value 1 is not
// visited yet, then new island found, Visit all
// cells in this island and increment island count
DFS(M, i, j, visited);
++count;

Time complexity: O(ROW x COL)

Euler Circuit in a Directed Graph

Eulerian Path is a path in graph that visits every edge exactly once. Eulerian Circuit is an Eulerian Path which
starts and ends on the same vertex.

A graph is said to be eulerian if it has eulerian cycle. We have discussed  eulerian circuit for an undirected
graph. In this post, same is discussed for a directed graph.

How to check if a directed graph is eulerian?


A directed graph has an eulerian cycle if following conditions are true (Source: Wiki)

1) All vertices with nonzero degree belong to a single strongly connected component.

2) In degree and out degree of every vertex is same.
We can detect singly connected component using Kosaraju’s DFS based simple algorithm.

To compare in degree and out degree, we need to store in degree an out degree of every vertex. Out degree
can be obtained by size of adjacency list. In degree can be stored by creating an array of size equal to
number of vertices.

Time complexity of the above implementation is O(V + E) as Kosaraju’s algorithm takes O(V + E) time. After
running  Kosaraju’s algorithm  we traverse all vertices and compare in degree with out degree which takes
O(V) time.

Check if a given graph is tree or not

Write a function that returns true if a given undirected graph is tree and false otherwise. 

An undirected graph is tree if it has following properties.

1) There is no cycle.
2) The graph is connected.

For an undirected graph we can either use BFS or DFS to detect above two properties.

How to detect cycle in an undirected graph?


We can either use BFS or DFS. For every visited vertex ‘v’, if there is an adjacent ‘u’ such that u is already
visited and u is not parent of v, then there is a cycle in graph. If we don’t find such an adjacent for any vertex,
we say that there is no cycle (See Detect cycle in an undirected graph for more details).
How to check for connectivity?

Since the graph is undirected, we can start BFS or DFS from any vertex and check if all vertices are
reachable or not. If all vertices are reachable, then graph is connected, otherwise not.

Boolean isCyclicUtil(int v, Boolean visited[], int parent)

{
// Mark the current node as visited
visited[v] = true;
Integer i;

// Recur for all the vertices adjacent to this vertex

Iterator<Integer> it = adj[v].iterator();
while (it.hasNext())
{
i = it.next();

// If an adjacent is not visited, then recur for

// that adjacent
if (!visited[i])
{
if (isCyclicUtil(i, visited, v))
return true;
}

// If an adjacent is visited and not parent of

// current vertex, then there is a cycle.
else if (i != parent)
return true;
}
return false;
 }

// Returns true if the graph is a tree, else false.

Boolean isTree()
{
// Mark all the vertices as not visited and not part
// of recursion stack
Boolean visited[] = new Boolean[V];
for (int i = 0; i < V; i++)
visited[i] = false;

// The call to isCyclicUtil serves multiple purposes

// It returns true if graph reachable from vertex 0
// is cyclcic. It also marks all vertices reachable
// from 0.
if (isCyclicUtil(0, visited, -1))
return false;

// If we find a vertex which is not reachable from 0

// (not marked by isCyclicUtil(), then we return false
for (int u = 0; u < V; u++)
if (!visited[u])
return false;

return true;
}

Karger’s algorithm for Minimum Cut

Given an undirected and unweighted graph, find the smallest cut (smallest number of edges that disconnects
the graph into two components).

The input graph may have parallel edges.

For example consider the following example, the smallest cut has 2 edges.


A Simple Solution use Max-Flow based s-t cut algorithm to find minimum cut. Consider every pair of vertices
as source ‘s’ and sink ‘t’, and call minimum s-t cut algorithm to find the s-t cut. Return minimum of all s-t cuts.
Best possible time complexity of this algorithm is O(V5) for a graph. [How? there are total possible V2pairs
and s-t cut algorithm for one pair takes O(V*E) time and E = O(V2)].
Below is simple Karger’s Algorithm for this purpose. Below Karger’s algorithm can be implemented in O(E) =
O(V2) time.

1) Initialize contracted graph CG as copy of original graph

2) While there are more than 2 vertices.
a) Pick a random edge (u, v) in the contracted graph.
b) Merge (or contract) u and v into a single vertex (update
the contracted graph).
c) Remove self-loops
3) Return cut represented by two vertices.
Now graph has two vertices, so we stop. The number of edges in the resultant graph is the cut produced by
Karger’s algorithm.

Karger’s algorithm is a  Monte Carlo algorithm  and cut produced by it may not be minimum.  For
example, the following diagram shows that a different order of picking random edges produces a min-cut of
size 3.

Note that the above program is based on outcome of a random function and may produce different
output.

In this post, we have discussed simple Karger’s algorithm and have seen that the algorithm doesn’t always
produce min-cut. The above algorithm produces min-cut with probability greater or equal to that 1/(n2).
Find if there is a path of more than k length from a source
Given a graph, a source vertex in the graph and a number k, find if there is a simple path (without any cycle)
starting from given source and ending at any other vertex.

One important thing to note is, simply doing BFS or DFS and picking the longest edge at every step would not
work. The reason is, a shorter edge can produce longer path due to higher weight edges connected through
it.

The idea is to use Backtracking. We start from given source, explore all paths from current vertex. We keep
track of current distance from source. If distance becomes more than k, we return true. If a path doesn’t
produces more than k distance, we backtrack.

How do we make sure that the path is simple and we don’t loop in a cycle? The idea is to keep track of
current path vertices in an array. Whenever we add a vertex to path, we check if it already exists or not in
current path. If it exists, we ignore the edge.

Time Complexity: O(n!)

Word Ladder (Length of shortest chain to reach a target

word)
Given a dictionary, and two words ‘start’ and ‘target’ (both of same length). Find length of the smallest chain
from ‘start’ to ‘target’ if it exists, such that adjacent words in the chain only differ by one character and each
word in the chain is a valid word i.e., it exists in the dictionary. It may be assumed that the ‘target’ word exists
in dictionary and length of all dictionary words is same.

Example:

Input: Dictionary = {POON, PLEE, SAME, POIE, PLEA, PLIE, POIN}

start = TOON
target = PLEA
Output: 7
Explanation: TOON - POON - POIN - POIE - PLIE - PLEE - PLEA
The idea is to use BFS. We start from the given start word, traverse all words that adjacent (differ by one
character) to it and keep doing so until we find the target word or we have traversed all words.
Time Complexity of the above code is O(n²m) where n is the number of entries originally in the dictionary and
m is the size of the string
queue<QItem> Q;
QItem item = {start, 1}; // Chain length for start word is 1
Q.push(item);

Q.push(item);

// While queue is not empty

while (!Q.empty())
{
// Take the front word
QItem curr = Q.front();
 Q.pop();

// Go through all words of dictionary

for (set<string>::iterator it = D.begin(); it != D.end(); it++)
{
// Process a dictionary word if it is adjacent to current
// word (or vertex) of BFS
string temp = *it;
if (isadjacent(curr.word, temp))
{
// Add the dictionary word to Q
item.word = temp;
item.len = curr.len + 1;
Q.push(item);

// Remove from dictionary so that this word is not

// processed again. This is like marking visited
D.erase(temp);

// If we reached target
if (temp == target)
return item.len;
}
}

Print all paths from a given source to a destination

Given a directed graph, a source vertex ‘s’ and a destination vertex ‘d’, print all paths from given ‘s’ to ‘d’.
The idea is to do Depth First Traversal of given directed graph. Start the traversal from source. Keep storing
the visited vertices in an array say ‘path[]’. If we reach the destination vertex, print contents of path[]. The
important thing is to mark current vertices in path[] as visited also, so that the traversal doesn’t go in a cycle.
void Graph::printAllPaths(int s, int d)
{
// Mark all the vertices as not visited
bool *visited = new bool[V];

// Create an array to store paths

int *path = new int[V];
int path_index = 0; // Initialize path[] as empty

// Initialize all vertices as not visited

for (int i = 0; i < V; i++)
visited[i] = false;

// Call the recursive helper function to print all paths

printAllPathsUtil(s, d, visited, path, path_index);
}

// A recursive function to print all paths from 'u' to 'd'.

// visited[] keeps track of vertices in current path.
// path[] stores actual vertices and path_index is current
// index in path[]
void Graph::printAllPathsUtil(int u, int d, bool visited[],
int path[], int &path_index)
{
// Mark the current node and store it in path[]
visited[u] = true;
path[path_index] = u;
path_index++;
 // If current vertex is same as destination, then print
// current path[]
if (u == d)
{
for (int i = 0; i<path_index; i++)
cout << path[i] << " ";
cout << endl;
}
else // If current vertex is not destination
{
// Recur for all the vertices adjacent to current vertex
list<int>::iterator i;
for (i = adj[u].begin(); i != adj[u].end(); ++i)
if (!visited[*i])
printAllPathsUtil(*i, d, visited, path, path_index);
}

// Remove current vertex from path[] and mark it as unvisited

path_index--;
visited[u] = false;
}

Tarjan’s Algorithm to find Strongly Connected Components

A directed graph is strongly connected if there is a path between all pairs of vertices. A strongly connected
component (SCC) of a directed graph is a maximal strongly connected subgraph. 
We have discussed  Kosaraju’s algorithm for strongly connected components. The previously discussed
algorithm requires two DFS traversals of a Graph. In this post, Tarjan’s algorithm is discussed that requires
only one DFS traversal.

Tarjan Algorithm is based on following facts:


1. DFS search produces a DFS tree/forest

2. Strongly Connected Components form subtrees of the DFS tree.

3. If we can find head of such subtrees, we can print/store all the nodes in that subtree (including head) and
that will be one SCC.

4. There is no back edge from one SCC to another (There can be cross edges, but cross edges will not be
used while processing the graph).

To find head of a SCC, we calculate desc and low array (as done for articulation point, bridge, biconnected
component). As discussed in the previous posts, low[u] indicates earliest visited vertex (the vertex with
minimum discovery time) that can be reached from subtree rooted with u. A node u is head if disc[u] = low[u].

Disc: This is the time when a node is visited 1st time while DFS traversal. For nodes A, B, C, .., J in DFS tree,
Disc values are 1, 2, 3, .., 10.
Low:Low” value of a node tells the topmost reachable ancestor (with minimum possible Disc value)
via the subtree of that node.
Case1 (Tree Edge): If node v is not visited already, then after DFS of v is complete, then minimum of low[u]
and low[v] will be updated to low[u].
low[u] = min(low[u], low[v]);
Case 2 (Back Edge): When child v is already visited, then minimum of low[u] and Disc[v] will be updated to
low[u].
low[u] = min(low[u], disc[v]);
Same Low and Disc values help to solve other graph problems like articulation point, bridge and biconnected
component.

To track the subtree rooted at head, we can use a stack (keep pushing node while visiting). When a head
node found, pop all nodes from stack till you get head out of stack.

To make sure, we don’t consider cross edges, when we reach a node which is already visited, we should
process the visited node only if it is present in stack, else ignore the node.

Time Complexity: The above algorithm mainly calls DFS, DFS takes O(V+E) for a graph represented using
adjacency list.

http://www.geeksforgeeks.org/tarjan-algorithm-find-strongly-connected-components/

void Graph::SCCUtil(int u, int disc[], int low[], stack<int> *st,

bool stackMember[])
{
// A static variable is used for simplicity, we can avoid use
// of static variable by passing a pointer.
static int time = 0;

// Initialize discovery time and low value

disc[u] = low[u] = ++time;
st->push(u);
stackMember[u] = true;

// Go through all vertices adjacent to this

list<int>::iterator i;
for (i = adj[u].begin(); i != adj[u].end(); ++i)
{
int v = *i; // v is current adjacent of 'u'

// If v is not visited yet, then recur for it

if (disc[v] == -1)
{
SCCUtil(v, disc, low, st, stackMember);

// Check if the subtree rooted with 'v' has a

// connection to one of the ancestors of 'u'
// Case 1 (per above discussion on Disc and Low value)
low[u] = min(low[u], low[v]);
}

// Update low value of 'u' only of 'v' is still in stack

// (i.e. it's a back edge, not cross edge).
// Case 2 (per above discussion on Disc and Low value)
else if (stackMember[v] == true)
low[u] = min(low[u], disc[v]);
}

// head node found, pop the stack and print an SCC

int w = 0; // To store stack extracted vertices
if (low[u] == disc[u])
{
while (st->top() != u)
{
 w = (int) st->top();
cout << w << " ";
stackMember[w] = false;
st->pop();
}
w = (int) st->top();
cout << w << "n";
stackMember[w] = false;
st->pop();
}
}

// The function to do DFS traversal. It uses SCCUtil()

void Graph::SCC()
{
int *disc = new int[V];
int *low = new int[V];
bool *stackMember = new bool[V];
stack<int> *st = new stack<int>();

// Initialize disc and low, and stackMember arrays

for (int i = 0; i < V; i++)
{
disc[i] = NIL;
low[i] = NIL;
stackMember[i] = false;
}

// Call the recursive helper function to find strongly

// connected components in DFS tree with vertex 'i'
for (int i = 0; i < V; i++)
if (disc[i] == NIL)
 SCCUtil(i, disc, low, st, stackMember);
}

Biconnected Components
A biconnected component is a maximal biconnected subgraph.

Biconnected Graph is already discussed here. In this article, we will see how to find biconnected componentin
a graph using algorithm by John Hopcroft and Robert Tarjan.

Algorithm is based on Disc and Low Values discussed in Strongly Connected Components Article.

Idea is to store visited edges in a stack while DFS on a graph and keep looking for  Articulation
Points (highlighted in above figure). As soon as an Articulation Point u is found, all edges visited while DFS
from node u onwards will form one  biconnected component. When DFS completes for one  connected
component, all edges present in stack will form a biconnected component.

If there is no  Articulation Point  in graph, then graph is biconnected and so there will be one biconnected
component which is the graph itself.

void BCCUtil(int u, int disc[], int low[], LinkedList<Edge>st,

int parent[])
{

// Initialize discovery time and low value

disc[u] = low[u] = ++time;
int children = 0;
// Go through all vertices adjacent to this
Iterator<Integer> it = adj[u].iterator();
while (it.hasNext())
{
int v = it.next(); // v is current adjacent of 'u'

// If v is not visited yet, then recur for it

if (disc[v] == -1)
{
children++;
parent[v] = u;

// store the edge in stack

st.add(new Edge(u,v));
BCCUtil(v, disc, low, st, parent);

// Check if the subtree rooted with 'v' has a

// connection to one of the ancestors of 'u'
// Case 1 -- per Strongly Connected Components Article
if (low[u] > low[v])
low[u] = low[v];

// If u is an articulation point,
// pop all edges from stack till u -- v
if ( (disc[u] == 1 && children > 1) ||
(disc[u] > 1 && low[v] >= disc[u]) )
{
while (st.getLast().u != u || st.getLast().v != v)
{
System.out.print(st.getLast().u + "--" +
st.getLast().v + " ");
 st.removeLast();
}
System.out.println(st.getLast().u + "--" +
st.getLast().v + " ");
st.removeLast();

count++;
}
}

// Update low value of 'u' only of 'v' is still in stack

// (i.e. it's a back edge, not cross edge).
// Case 2 -- per Strongly Connected Components Article
else if (v != parent[u] && disc[v] < low[u])
{
if (low[u]>disc[v])
low[u]=disc[v];
st.add(new Edge(u,v));
}
}
}

// The function to do DFS traversal. It uses BCCUtil()

void BCC()
{
int disc[] = new int[V];
int low[] = new int[V];
int parent[] = new int[V];
LinkedList<Edge> st = new LinkedList<Edge>();

// Initialize disc and low, and parent arrays

for (int i = 0; i < V; i++)
 {
disc[i] = -1;
low[i] = -1;
parent[i] = -1;
}

for (int i = 0; i < V; i++)

{
if (disc[i] == -1)
BCCUtil(i, disc, low, st, parent);

int j = 0;

// If stack is not empty, pop all edges from stack

while (st.size() > 0)
{
j = 1;
System.out.print(st.getLast().u + "--" +
st.getLast().v + " ");
st.removeLast();
}
if (j == 1)
{
System.out.println();
count++;
}
}
}