Skip to content
Related Articles

Related Articles

Improve Article
Construct tree from ancestor matrix
  • Difficulty Level : Hard
  • Last Updated : 18 Jan, 2021

Given an ancestor matrix mat[n][n] where Ancestor matrix is defined as below. 

mat[i][j] = 1 if i is ancestor of j
mat[i][j] = 0, otherwise

Construct a Binary Tree from a given ancestor matrix where all its values of nodes are from 0 to n-1.

  1. It may be assumed that the input provided the program is valid and tree can be constructed out of it.
  2. Many Binary trees can be constructed from one input. The program will construct any one of them.

Examples: 

Input: 0 1 1
       0 0 0 
       0 0 0 
Output: Root of one of the below trees.
    0                0
  /   \     OR     /   \
 1     2          2     1

Input: 0 0 0 0 0 0 
       1 0 0 0 1 0 
       0 0 0 1 0 0 
       0 0 0 0 0 0 
       0 0 0 0 0 0 
       1 1 1 1 1 0
Output: Root of one of the below trees.
      5              5               5
   /    \           / \            /   \
  1      2   OR    2   1    OR    1     2  OR ....
 /  \   /        /    /  \       / \    /
0   4  3        3    0    4     4   0  3
There are different possible outputs because ancestor
matrix doesn't store that which child is left and which
is right.

We strongly recommend you to minimize your browser and try this yourself first. 

Observations used in the solution: 



  1. The rows that correspond to leaves have all 0’s
  2. The row that corresponds to root has maximum number of 1’s.
  3. Count of 1’s in i’th row indicates number of descendants of node i.

The idea is to construct the tree in bottom up manner.

  1. Create an array of node pointers node[].
  2. Store row numbers that correspond to a given count. We have used multimap for this purpose.
  3. Process all entries of multimap from smallest count to largest (Note that entries in map and multimap can be traversed in sorted order). Do following for every entry.
    • Create a new node for current row number.
    • If this node is not a leaf node, consider all those descendants of it whose parent is not set, make current node as its parent.
  4. The last processed node (node with maximum sum) is root of tree.

Below is the implementation of the above approach:

C++




// Given an ancestor matrix for binary tree, construct
// the tree.
#include <bits/stdc++.h>
using namespace std;
   
# define N 6
   
/* A binary tree node */
struct Node
{
    int data;
    Node *left, *right;
};
   
/* Helper function to create a new node */
Node* newNode(int data)
{
    Node* node = new Node;
    node->data = data;
    node->left = node->right = NULL;
    return (node);
}
   
// Constructs tree from ancestor matrix
Node* ancestorTree(int mat[][N])
{
    // Binary array to determine whether
    // parent is set for node i or not
    int parent[N] = {0};
   
    // Root will store the root of the constructed tree
    Node* root = NULL;
   
    // Create a multimap, sum is used as key and row
    // numbers are used as values
    multimap<int, int> mm;
   
    for (int i = 0; i < N; i++)
    {
        int sum = 0; // Initialize sum of this row
        for (int j = 0; j < N; j++)
            sum += mat[i][j];
   
        // insert(sum, i) pairs into the multimap
        mm.insert(pair<int, int>(sum, i));
    }
   
    // node[i] will store node for i in constructed tree
    Node* node[N];
   
    // Traverse all entries of multimap.  Note that values
    // are accessed in increasing order of sum
    for (auto it = mm.begin(); it != mm.end(); ++it)
    {
      // create a new node for every value
      node[it->second] = newNode(it->second);
   
      // To store last processed node. This node will be
      // root after loop terminates
      root = node[it->second];
   
      // if non-leaf node
      if (it->first != 0)
      {
        // traverse row 'it->second' in the matrix
        for (int i = 0; i < N; i++)
        {
           // if parent is not set and ancestor exits
           if (!parent[i] && mat[it->second][i])
           {
             // check for unoccupied left/right node
             // and set parent of node i
             if (!node[it->second]->left)
               node[it->second]->left = node[i];
             else
               node[it->second]->right = node[i];
   
             parent[i] = 1;
           }
        }
      }
    }
    return root;
}
   
/* Given a binary tree, print its nodes in inorder */
void printInorder(Node* node)
{
    if (node == NULL)
        return;
    printInorder(node->left);
    printf("%d ", node->data);
    printInorder(node->right);
}
   
// Driver code
int main()
{
    int mat[N][N] = {{ 0, 0, 0, 0, 0, 0 },
        { 1, 0, 0, 0, 1, 0 },
        { 0, 0, 0, 1, 0, 0 },
        { 0, 0, 0, 0, 0, 0 },
        { 0, 0, 0, 0, 0, 0 },
        { 1, 1, 1, 1, 1, 0 }
    };
   
    Node* root = ancestorTree(mat);
   
    cout << "Inorder traversal of tree is \n";
     
    // Function call
    printInorder(root);
   
    return 0;
}

Java




// Given an ancestor matrix for binary tree, construct
// the tree.
import java.util.HashMap;
import java.util.LinkedHashMap;
import java.util.Map;
import java.util.stream.Collectors;
 
/* A binary tree node */
class Node {
    int data;
    Node left, right;
 
    Node(int d)
    {
        data = d;
        left = right = null;
    }
}
 
public class TreeFromAncestorMatrix {
 
    static Node ancestorNode(int[][] mat)
    {
        int n = mat.length;
 
        // Binary array to determine whether parent is set
        // for node i or not
        int[] parent = new int[n];
        Node root = null;
 
        // Map to store row numbers as key and
        // their sum as their values
        Map<Integer, Integer> map = new HashMap<>();
 
        for (int i = 0; i < n; i++) {
            int sum = 0;
 
            for (int j = 0; j < n; j++)
                sum += mat[i][j];
 
            map.put(i, sum);
        }
 
        // node[i] will store node for i in
        //constructed tree
        Node node[] = new Node[n];
 
        // Sorting the map according to its
        //values
        Map<Integer, Integer> sorted
            = map.entrySet()
                  .stream()
                  .sorted(Map.Entry.comparingByValue())
                  .collect(Collectors.toMap(
                      e
                      -> e.getKey(),
                      e
                      -> e.getValue(),
                      (e1, e2) -> e2, LinkedHashMap::new));
 
        // Traverse all entries of sorted Map.
        // Note that values are accessed in increasing order
        // of sum
        for (Map.Entry<Integer, Integer> entry :
             sorted.entrySet()) {
            int key = entry.getKey();
            int value = entry.getValue();
 
            // create a new node for every
            // value
            node[key] = new Node(key);
 
            // if its is an internal node
            if (value != 0) {
 
                // Traverse row
                //corresponding to the node
                for (int i = 0; i < n; i++)
                {
 
                    // if parent is not set and
                    //ancestor exits
                    if (parent[i] == 0
                        && mat[key][i] != 0)
                    {
                        // check for unoccupied
                        // left/right node and
                        //set parent of node i
                        if (node[key].left == null)
                            node[key].left = node[i];
                        else
                            node[key].right = node[i];
                        parent[i] = 1;
                    }
 
                    // To store last processed
                    // node. This node will be root after
                    // loop terminates
                    root = node[key];
                }
            }
        }
 
        return root;
    }
  
    /* Given a binary tree, print its nodes in inorder */
    static void inOrder(Node root)
    {
        if (root == null)
            return;
        inOrder(root.left);
        System.out.print(root.data + " ");
        inOrder(root.right);
    }
 
    // Driver code
    public static void main(String[] args)
    {
        int mat[][] = {
            { 0, 0, 0, 0, 0, 0 }, { 1, 0, 0, 0, 1, 0 },
            { 0, 0, 0, 1, 0, 0 }, { 0, 0, 0, 0, 0, 0 },
            { 0, 0, 0, 0, 0, 0 }, { 1, 1, 1, 1, 1, 0 }
        };
 
        Node root = ancestorNode(mat);
 
         // Function call
        inOrder(root);
    }
}
 
// contribute by amarsomani

Python3




# key structure to store a binary tree node
class Node:
    def __init__(self, key, left = None, right = None):
        self.key = key
        self.left = left
        self.right = right
 
# Utility function to print binary tree nodes in-order fashion
def inorder(node):
    if node:
        inorder(node.left)
        print(node.key, end = ' ')
        inorder(node.right)
 
# Function to construct a binary tree
# from specified ancestor matrix
def constructBT(mat):
   
    # get number of rows in the matrix
    N = len(mat)
 
    # create an empty multi-dict
    dict = {}
 
    # Use sum as key and row numbers as values in the multi-dict
    for i in range(N):
 
        # find the sum of the current row
        total = sum(mat[i])
 
        # insert the sum and row number into the dict
        dict.setdefault(total, []).append(i)
 
    # node[i] will store node for i in constructed tree
    node = [Node(-1)] * N
    last = 0
 
    # the value of parent[i] is true if parent is set for i'th node
    parent = [False] * N
 
    # Traverse the dictionary in sorted order (default behavior)
    for key in dict.keys():
        for row in dict.get(key):
            last = row
             
            # create a new node
            node[row] = Node(row)
 
            # if leaf node, do nothing
            if key == 0:
                continue
 
            # traverse row
            for i in range(N):
               
                # do if parent is not set and ancestor exits
                if not parent[i] and mat[row][i] == 1:
                   
                    # check for the unoccupied node
                    if node[row].left is None:
                        node[row].left = node[i]
                    else:
                        node[row].right = node[i]
 
                    # set parent for i'th node
                    parent[i] = True
 
    # last processed node is the root
    return node[last]
 
# Construct a Binary Tree from Ancestor Matrix
if __name__ == '__main__':
 
    mat = [[0, 0, 0, 0, 0, 0],
        [1, 0, 0, 0, 1, 0],
        [0, 0, 0, 1, 0, 0],
        [0, 0, 0, 0, 0, 0],
        [0, 0, 0, 0, 0, 0],
        [1, 1, 1, 1, 0, 0]]
 
    root = constructBT(mat)
    inorder(root)
 
# This code is contributed by Priyadarshini Kumari

C#




// Given an ancestor matrix for binary tree, construct
// the tree.
using System;
using System.Collections.Generic;
using System.Linq;
 
/* A binary tree node */
public
  class Node
  {
    public
      int data;
    public
      Node left, right;
    public
      Node(int d)
    {
      data = d;
      left = right = null;
    }
  }
 
public class TreeFromAncestorMatrix
{
 
  static Node ancestorNode(int[,] mat)
  {
    int n = mat.GetLength(0);
 
    // Binary array to determine whether parent is set
    // for node i or not
    int[] parent = new int[n];
    Node root = null;
 
    // Map to store row numbers as key and
    // their sum as their values
    Dictionary<int, int> map = new Dictionary<int, int>();
    for (int i = 0; i < n; i++)
    {
      int sum = 0;
      for (int j = 0; j < n; j++)
        sum += mat[i, j];
      map.Add(i, sum);
    }
 
    // node[i] will store node for i in
    //constructed tree
    Node []node = new Node[n];
 
    var sorted = from entry in map orderby entry.Value ascending select entry;
 
    // Traverse all entries of sorted Map.
    // Note that values are accessed in increasing order
    // of sum
    foreach (KeyValuePair<int, int> entry in
             sorted)
    {
      int key = entry.Key;
      int value = entry.Value;
 
      // create a new node for every
      // value
      node[key] = new Node(key);
 
      // if its is an internal node
      if (value != 0)
      {
 
        // Traverse row
        //corresponding to the node
        for (int i = 0; i < n; i++)
        {
 
          // if parent is not set and
          //ancestor exits
          if (parent[i] == 0
              && mat[key,i] != 0)
          {
            // check for unoccupied
            // left/right node and
            //set parent of node i
            if (node[key].left == null)
              node[key].left = node[i];
            else
              node[key].right = node[i];
            parent[i] = 1;
          }
 
          // To store last processed
          // node. This node will be root after
          // loop terminates
          root = node[key];
        }
      }
    }
    return root;
  }
 
  /* Given a binary tree, print its nodes in inorder */
  static void inOrder(Node root)
  {
    if (root == null)
      return;
    inOrder(root.left);
    Console.Write(root.data + " ");
    inOrder(root.right);
  }
 
  // Driver code
  public static void Main(String[] args)
  {
    int [,]mat = {
      { 0, 0, 0, 0, 0, 0 }, { 1, 0, 0, 0, 1, 0 },
      { 0, 0, 0, 1, 0, 0 }, { 0, 0, 0, 0, 0, 0 },
      { 0, 0, 0, 0, 0, 0 }, { 1, 1, 1, 1, 1, 0 }
    };
    Node root = ancestorNode(mat);
 
    // Function call
    inOrder(root);
  }
}
 
// This code is contributed by Rajput-Ji
Output
0 1 4 5 3 2

Note that we can also use an array of vectors in place of multimap. We have used multimap for simplicity. Array of vectors would improve performance as inserting and accessing elements would take O(1) time.

This article is contributed by Aditya Goel. Please write comments if you find anything incorrect, or you want to share more information about the topic discussed above

Attention reader! Don’t stop learning now. Get hold of all the important DSA concepts with the DSA Self Paced Course at a student-friendly price and become industry ready.  To complete your preparation from learning a language to DS Algo and many more,  please refer Complete Interview Preparation Course.

In case you wish to attend live classes with industry experts, please refer Geeks Classes Live 




My Personal Notes arrow_drop_up
Recommended Articles
Page :