class Node: pass class Edge: pass map = {} vertices = [] edgeList = [] def insertVertex(vertex): vertices.append(vertex) def insertEdge(vertex1, vertex2, edgeVal): edge = Edge() edge.startVertex = vertex1 edge.endVertex = vertex2 edge.value = edgeVal edgeList.append(edge) #Create groups with only one vertex. def createSet(data): node = Node() node.data = data node.rank = 0 node.parent = node map[data] = node #Combines two groups into one. Does union by rank. def union(vertex1, vertex2): node1 = map[vertex1] node2 = map[vertex2] parent1 = findSetNode(node1) parent2 = findSetNode(node2) #If they are in the same group, do nothing. if(parent1.data == parent2.data): return #Else whose rank is higher becomes parent of other. if (parent1.rank <= parent2.rank): #Increment rank only if both sets have same rank. parent1.rank = parent1.rank + 1 if parent1.rank == parent2.rank else parent1.rank parent2.parent = parent1 else: parent1.parent = parent2 # Find the representative of this set def findSet(data): return findSetNode(map[data]).data # Finds the leader/representative recursivly and does PATH COMPRESSION as well. def findSetNode(node): parent = node.parent if (parent == node): return parent node.parent = findSetNode(node.parent) return node.parent # Find Minimum Spanning Tree using Kruskal's Algorithm. def createMST(): #Sort all edges in Ascending order. edgeList.sort(key=lambda x: x.value) #Create as many groups as the total vertices. for vertex in vertices: createSet(vertex) resultEdge = [] for edge in edgeList: #Get the sets of two vertices of the edge. root1 = findSet(edge.startVertex) root2 = findSet(edge.endVertex) #check if the vertices are on the same or different set. #If vertices are in the same set then ignore the edge. if (root1 == root2): continue else: # If vertices are in different sets then add the edge to result and union # these two sets into one. resultEdge.append(edge) union(edge.startVertex, edge.endVertex) return resultEdge #Adding vertices one by one insertVertex("a") insertVertex("b") insertVertex("c") insertVertex("d") insertVertex("e") #Adding edges with values. insertEdge("a", "b", 1) insertEdge("b", "e", 8) insertEdge("e", "d", 3) insertEdge("d", "a", 4) insertEdge("a", "c", 5) insertEdge("c", "e", 9) insertEdge("c", "d", 12) mstList = createMST() for edge in mstList: print("The edge : ",edge.startVertex," -- ",edge.endVertex," with value : ",edge.value)
So, we will be dealing with the below Graph,
1st Group - {a} --- rank 0 2nd Group - {b} --- rank 0 3rd Group - {c} --- rank 0 4th Group - {d} --- rank 0 5th Group - {e} --- rank 0
a --- b
1st Group - {a} --- rank 0 2nd Group - {b} --- rank 0
1st Group - {a,b}
In the above code, we have a class called Node. Which looks a little different.
class Node: pass
Although it has nothing inside it. But in reality there are three attributes in the Node class. i.e. data, rank and parent.
data of string type rank of int type parent of Node type
You can think of the Node class as a representation of a Vertex.
And how is that so ?
Just think, a vertex should have a rank and its name. And so we have,
data of string type rank of int type
But the only mystery is, what is parent of Node type?
A Node object inside a Node class! Looks Strange! Right?
It is like a connection, we are trying to establish between Vertices.
So, when the vertices are created for the first time, they should the leader of their group and off course they should be their parent.
Say for example when vertex a and b is created, a is in group 1 and b is in group 2.
Node parent for vertex b is Node b (Since vertex b is the leader of group 2).
And Node parent for vertex a is Node a (Since vertex a is the leader of group 1).
Next, when b is put into group 1, the leader of group 1 is a.
So, vertex b is a child of a.
And Node parent of b is Node a now.
The vertex a is named as Node a and vetex b is named as Node b. We have seen how the connection is established.
As we have seen the Node class is used to store the contents of the Vertices.
Similarly, the Edge class is used to store the contents of the Edges.
class Edge: pass
Similarly, the Edge class has three attributes(Well! Once again that is not defined in Edge class),
The above one is the example of an edge that has,
startVertex = a endVertex = b value = 1
So, firstly we construct the graph by adding the vertices.
#Adding vertices one by one insertVertex("a") insertVertex("b") insertVertex("c") insertVertex("d") insertVertex("e")
Then adding the edges with its values,
#Adding edges with values. insertEdge("a", "b", 1) insertEdge("b", "e", 8) insertEdge("e", "d", 3) insertEdge("d", "a", 4) insertEdge("a", "c", 5) insertEdge("c", "e", 9) insertEdge("c", "d", 12)
Then we call the createMST() method to construct Minimum Spanning Tree using Kruskal's Algorithm.
mstList = createMST()
def createMST(): #Sort all edges in Ascending order. edgeList.sort(key=lambda x: x.value) #Create as many groups as the total vertices. for vertex in vertices: createSet(vertex) resultEdge = [] for edge in edgeList: #Get the sets of two vertices of the edge. root1 = findSet(edge.startVertex) root2 = findSet(edge.endVertex) #check if the vertices are on the same or different set. #If vertices are in the same set then ignore the edge. if (root1 == root2): continue else: # If vertices are in different sets then add the edge to result and union # these two sets into one. resultEdge.append(edge) union(edge.startVertex, edge.endVertex) return resultEdge
As we know, the first task is to sort the Edges based on its value,
edgeList.sort(key=lambda x: x.value)
The next task would be, to create individual groups for each Vertex using the createSet() method.
#Create as many groups as the total vertices. for vertex in vertices: createSet(vertex)
And as we know vertices
vertices = []
contains the List of Vertices. So, we take the vertices one by one using the for each loop and call the creteSet() method.
createSet(vertex)
node = Node() node.data = data node.rank = 0 node.parent = node map[data] = node
The parameter data contains the vertex(Say a for vertex a).
So, we create a Node object to hold the actual vertex(Say value a for vertex a)
node.data = data
the rank of the vertex.
node.rank = 0
rank is initialised to 0 because initially all the vertices would be the leader of their group and the leader is ranked 0.
The next line is extremely important,
node.parent = node
Now since, vertex a is the leader of their group. node.parent is vertex a itself.
Below is the example of vertex a.
Then we put the created node in the Dictionary named map,
map = {}
With the key as the value of vertex(a in this case) and value as the created node.
map[data] = node
All the Nodes/vertices are created in the same way and put in the map.
Let us come back to the createMST() method again,
And since we got all the groups created.
Now, it's time for us to implement the actual logic.
And we have declared resultEdge List that will store all the edges included in the Minimum Spanning Tree.
resultEdge = []
Now, we know that the List edgeList,
edgeList = []
Contains the List of all the edges in the Graph(Off course Sorted in ascending order done earlier).
And what will will do is take List edgeList one by one and try constructing the Minimum Spanning Tree.
for edge in edgeList: #Get the sets of two vertices of the edge. root1 = findSet(edge.startVertex) root2 = findSet(edge.endVertex) #check if the vertices are on the same or different set. #If vertices are in the same set then ignore the edge. if (root1 == root2): continue else: # If vertices are in different sets then add the edge to result and union # these two sets into one. resultEdge.append(edge) union(edge.startVertex, edge.endVertex)
So, the for each loop, takes the sorted edges one by one,
for edge in edgeList
And the first thing it does is, tries to find the leader of each vertex using the findSet() method,
root1 = findSet(edge.startVertex) root2 = findSet(edge.endVertex)
def findSet(data): return findSetNode(map[data]).data
Now, the cool thing to note is, the return statement,
return findSetNode(map[data]).data
Calls another findSetNode(...) method.
Sounds Interesting or Strange ?
Let's dive in,
map[data] from the return statement,
return findSetNode(map[data]).data
gives us a Node object.
How ?
Because, we have seen map holds Key as the Vertex name (Say a) and its Node as value.
And if we pass the value a to the Dictionary,
map["a"]
We get the Node object of a.
So, this time we will be calling the findSetNode(...) method, that accepts Node as parameter.
And luckily, we have the def findSetNode(node) method.
parent = node.parent if (parent == node): return parent node.parent = findSetNode(node.parent) return node.parent
So, in the first line, we have created a Node variable called parent and initialised the parent variable with the parent from the parameter node of Node type.
Say, we have passed Node a and the parent of Node a is Node a(Because Node a is the leader).
parent = node.parent
Then we check if the Node parent is equal to the node from parameter Node node.
if (parent == node): return parent
In this case Node parent and Node node are equal.
So, the Node parent is returned.
But ! Yes there is a BUT.
Say, the node b was passed, when it was a child of a.
Now, let us relook the steps for Node b,
parent = node.parent
So, Node parent would be Node a.
And Node parent would be holding the Node a.
Then we check if the Node parent is equal to the node from parameter Node node.
if (parent == node): return parent
And if condition is not satisfied as
Node parent is holding Node a and node is holding Node b.
Fair enough !
Then we go to the next line, where findSet(Node node) method is called recursively,
node.parent = findSetNode(node.parent)
Until it finds the parent and finally node.parent is returned.
return node.parent
So, let us come back to createMST() method again.
Let us take the example of edge
root1 = findSet(edge.startVertex) root2 = findSet(edge.endVertex)
Where, edge.startVertex is a and edge.endVertex is b.
And we found the parent of a is a itself and parent of b is b.
So ,
root1 = "a"
and
root2 = "b"
And, if we see the code again,
for edge in edgeList: #Get the sets of two vertices of the edge. root1 = findSet(edge.startVertex) root2 = findSet(edge.endVertex) #check if the vertices are on the same or different set. #If vertices are in the same set then ignore the edge. if (root1 == root2): continue else: # If vertices are in different sets then add the edge to result and union # these two sets into one. resultEdge.append(edge) union(edge.startVertex, edge.endVertex)
There is this if statement that checks,
if (root1 == root2)
But in this case root1 is a and root2 is b.
So, we come to the else part.
else: # If vertices are in different sets then add the edge to result and union # these two sets into one. resultEdge.append(edge) union(edge.startVertex, edge.endVertex)
Now since, we are in the else part. This edge,
must be included in the Minimum Spanning Tree.
resultEdge.append(edge)
And thus, we have added it to the resultEdge list.
So, the next task would be to combine a and b in the same group.
union(edge.startVertex, edge.endVertex)
And this task of combining is done by the union(...) method.
node1 = map[vertex1] node2 = map[vertex2] parent1 = findSetNode(node1) parent2 = findSetNode(node2) #If they are in the same group, do nothing. if(parent1.data == parent2.data): return #Else whose rank is higher becomes parent of other. if (parent1.rank <= parent2.rank): #Increment rank only if both sets have same rank. parent1.rank = parent1.rank + 1 if parent1.rank == parent2.rank else parent1.rank parent2.parent = parent1 else: parent1.parent = parent2
Let us take the example of
to explain the union(...) method.
node1 = map[vertex1] node2 = map[vertex2]
And
node2 will have the contents of Node b and node1 will have the contents of Node a.
In the next two lines,
parent1 = findSetNode(node1) parent2 = findSetNode(node2)
We try to find the parents/leaders of node1 and node2 using findSet(node) method which we have already seen.
And in this case, the parent of a is a and the parent of b is b.
Next, we check, if both the parent of the vertices are equal or not,
if(parent1.data == parent2.data): return
If both the parents are equal, we do nothing.
But in this case, they are not equal as
parent1.data = "a"
and
parent2.data = "b"
So, we come to the next if statement.
if (parent1.rank <= parent2.rank): #Increment rank only if both sets have same rank. parent1.rank = parent1.rank + 1 if parent1.rank == parent2.rank else parent1.rank parent2.parent = parent1 else: parent1.parent = parent2
In the if statement, we check if parent1.rank is less than or equal to parent2.rank.
if (parent1.rank <= parent2.rank)
And in this case,
parent1.rank and parent2.rank has the same rank i.e. 0 (From the above diagram).
And in the next line,
parent1.rank = parent1.rank + 1 if parent1.rank == parent2.rank else parent1.rank
We increment the parent1.rank only if parent1.rank is equal to parent2.rank.
And in this case parent1.rank is equal to parent2.rank,
So, we increment parent1.rank by 1.
parent1.rank + 1
And parent1.rank becomes,
parent1.rank = 1
Then, we initialise parent2.parent with parent1.
parent2.parent = parent1
And we can see that the connection between a and b is established,
And continuing this way, we get the Minimum Spanning Tree,