Back to Course


0% Complete
0/82 Steps
  1. Getting Started with Algorithm
    What is an Algorithm?
  2. Characteristics of Algorithm
    1 Topic
  3. Analysis Framework
  4. Performance Analysis
    3 Topics
  5. Mathematical Analysis
    2 Topics
  6. Sorting Algorithm
    Sorting Algorithm
    10 Topics
  7. Searching Algorithm
    6 Topics
  8. Fundamental of Data Structures
  9. Queues
  10. Graphs
  11. Trees
  12. Sets
  13. Dictionaries
  14. Divide and Conquer
    General Method
  15. Binary Search
  16. Recurrence Equation for Divide and Conquer
  17. Finding the Maximum and Minimum
  18. Merge Sort
  19. Quick Sort
  20. Stassen’s Matrix Multiplication
  21. Advantages and Disadvantages of Divide and Conquer
  22. Decrease and Conquer
    Insertion Sort
  23. Topological Sort
  24. Greedy Method
    General Method
  25. Coin Change Problem
  26. Knapsack Problem
  27. Job Sequencing with Deadlines
  28. Minimum Cost Spanning Trees
    2 Topics
  29. Single Source Shortest Paths
    1 Topic
  30. Optimal Tree Problem
    1 Topic
  31. Transform and Conquer Approach
    1 Topic
  32. Dynamic Programming
    General Method with Examples
  33. Multistage Graphs
  34. Transitive Closure
    1 Topic
  35. All Pairs Shortest Paths
    6 Topics
  36. Backtracking
    General Method
  37. N-Queens Problem
  38. Sum of Subsets problem
  39. Graph Coloring
  40. Hamiltonian Cycles
  41. Branch and Bound
    2 Topics
  42. 0/1 Knapsack problem
    2 Topics
  43. NP-Complete and NP-Hard Problems
    1 Topic
Lesson 35, Topic 2
In Progress

Optimal Binary Search Trees

Lesson Progress
0% Complete

A Binary Search Tree (BST) is a tree where the key values are stored in the internal nodes. The external nodes are null nodes. The keys are ordered lexicographically, i.e. for each internal node all the keys in the left sub-tree are less than the keys in the node, and all the keys in the right sub-tree are greater.

When we know the frequency of searching each one of the keys, it is quite easy to compute the expected cost of accessing each node in the tree. An optimal binary search tree is a BST, which has minimal expected cost of locating each node.

Basically, we know the key values of the each node, but let say we also know the frequencies of each node in the process of searching, means that how much time particular node is being searched. So, by knowing the frequencies of each node and their key values we can find the overall cost of searching.

In many applications the cost of searching is very important. So, it is required that the overall cost of searching should be as less as possible. And we know that search time of BST is more than the Balanced Binary Search Tree, as Balanced Binary Search tree has less number of levels than the BST. And there is one way which can further reduce the cost than the Balanced BST, which is Optimal Binary Search Tree . Let us understand that by following example.

KodNest Capture22

As there are 3 different keys, so we can have total 5 various BST by changing order of keys. And which van be found by, where “n” is the number of keys.

KodNest Capture23

So following are the various possible Binary Search Trees of the above data. And also the overall cost for searching for each BST.

KodNest Capture24

The cost is computed by multiplying the each node’s frequency with the level of tree( Here we are assuming that the tree starts from level 1 ) and then add them to compute the overall cost of BST.

As we can see in Figure 2 there are various 5 types of arrangements are possible. And as we have discussed earlier that, it is possible to further reduce the cost of Balanced BST which is specified in following figure 3.

KodNest Capture25

As it is shown in above figure that 2nd BST is balanced and the 4th BST is not balanced, though it’s cost is less than the cost of Balanced BST and its cost is the least among all, so it is our Optimal Binary Search Tree for the given data in Figure 1. So here our idea to generate the Optimal Binary Search Tree is that, the nodes whose frequencies are more should appear in the lower levels of Tree .i.e, In our example node with key 40 is having highest frequency and which appears at level 1 which is our Optimal BST.

Here note that: if the number of nodes are less then we can find optimal BST by checking all possible arrangements, but if the nodes are greater than 3 like 4,5,6….. then respectively 14,42,132….. , different BSTs are possible so by checking all arrangements to find Optimal Cost may lead to extra overhead. So we will see now another approach to solve the problem of Optimal BST using Dynamic Programming Approach.

Using Dynamic Approach:-

KodNest image

where c[ i , j ] in which c stands for cost, “ pi ” stands for frequency of node i. We will use matrix form to solve this problem.Problem Statement is as following and we have to find Optimal BST for that:

KodNest Capture27


Step: 1 :: According to above formula: 2 we derived the following matrix, that c[i,i] = Pi and for i>j → c[i,j] = 0;

KodNest Capture28

Step: 2 :: So now compute value of C[0,1] as following and put it in table.

KodNest Capture29

and the table after this step will look like…

KodNest Capture30

Step: 3:: Do same for all elements like c[1,2],c[2,3],c[0,2],c[1,3],c[0,3] and check your answer with the following table……..

KodNest Capture31

Step:4 :: So now we want to arrange the nodes in such a way that they will cost the minimum value which we derived previously. For that we have to choose root(parent node)in every group of nodes. See in step 2: for node 0 and node 1 the parent node will be decided by the value of K for which we took the minimum cost, in our example for value of k=0, we are able to find the minimum value SO parent node will be node 0 and similarly do for others and make table as following which will help you to arrange nodes very easily.

KodNest Capture32

Step: 5:: After creating table as shown above go from BOTTOM → UP. As shown in table Node 2 appears at bottom of the table so it is root of our BST. In our example the keys are sorted so we can say that node 3 will appear at the right sub-tree of node 2 , and node 0 and node 1 will appear at the left sub-tree of the node 2 as shown below.

KodNest Capture33

So now there is one question that “ Which node is parent node between node 0 and node 1??” For that again Go from Bottom →Up and see that node 0 is parent node , because when we were computing value of c[0,1] at that time we came to know that between 0,1 node 0 is parent node. And as in our example the keys are sorted so node 1 will appear in the right sub-tree of node 0(because key value of node 1 is higher than the key value of node 0). So our Optimal Binary Search Tree for our example will look like…….

KodNest Capture34

Above is our Optimal Binary Search Tree with cost 26.

Algorithm to Compute the minimum cost:

KodNest image 1

New Report