Search element in a sorted matrix

Given a sorted matrix mat[n][m] and an element ‘x’. Find the position of x in the matrix if it is present, else print -1. Matrix is sorted in a way such that all elements in a row are sorted in increasing order and for row ‘i’, where 1 <= i <= n-1, the first element of row ‘i’ is greater than or equal to the last element of row ‘i-1’. The approach should have O(log n + log m) time complexity.

Examples: 

Input : mat[][] = { {1, 5, 9},
                    {14, 20, 21},
                    {30, 34, 43} }
        x = 14
Output : Found at (1, 0)

Input : mat[][] = { {1, 5, 9, 11},
                    {14, 20, 21, 26},
                    {30, 34, 43, 50} }
        x = 42
Output : -1

The Brute Force and Easy Way To do it:

 The Approach: the approach is very simple that we use to have linear search/mapping thing.

Found at (1,4)
This is how we can found using mapping: 
(1,4)

Time complexity: O(n+m), For traversing.
Auxiliary Space: O(n+m), For mapping.

Please note that this problem is different from Search in a row-wise and column-wise sorted matrix. Here matrix is more strictly sorted as the first element of a row is greater than the last element of the previous row.

A Simple Solution is to one by one compare x with every element of the matrix. If matches, then return the position. If we reach the end, return -1. The time complexity of this solution is O(n x m).

An efficient solution is to typecast a given 2D array to a 1D array, then apply binary search on the typecasted array but will require extra space to store this array.

Another efficient approach that doesn’t require typecasting is explained below. 

1) Perform binary search on the middle column 
   till only two elements are left or till the
   middle element of some row in the search is
   the required element 'x'. This search is done
   to skip the rows that are not required
2) The two left elements must be adjacent. Consider
   the rows of two elements and do following
   a) check whether the element 'x' equals to the 
      middle element of any one of the 2 rows
   b) otherwise according to the value of the 
      element 'x' check whether it is present in 
      the 1st half of 1st row, 2nd half of 1st row, 
      1st half of 2nd row or 2nd half of 2nd row. 

Note: This approach works for the matrix n x m 
      where 2 <= n. The algorithm can be modified
      for matrix 1 x m, we just need to check whether
      2nd row exists or not      

Example:  

Consider:    | 1  2  3  4| 
x = 3, mat = | 5  6  7  8|   Middle column:
             | 9 10 11 12|    = {2, 6, 10, 14}
             |13 14 15 16|   perform binary search on them
                             since, x < 6, discard the 
                             last 2 rows as 'a' will 
                             not lie in them(sorted matrix)
Now, only two rows are left
             | 1  2  3  4| 
x = 3, mat = | 5  6  7  8|   Check whether element is present
                             on the middle elements of these
                             rows = {2, 6}
                             x != 2 or 6
If not, consider the four sub-parts
1st half of 1st row = {1}, 2nd half of 1st row = {3, 4}
1st half of 2nd row = {5}, 2nd half of 2nd row = {7, 8}

According the value of 'x' it will be searched in the
2nd half of 1st row = {3, 4} and found at (i, j): (0, 2)                              

Implementation:

Found at (0,2)

Time complexity: O(log n + log m). O(Log n) time is required to find the two desired rows. Then O(Log m) time is required for binary search in one of the four parts with a size equal to m/2.
Auxiliary Space: O(1)

This method is contributed by Ayush Jauhari. 

Method 2: Using binary search in 2 dimensions

This method also has the same time complexity: O(log(m) + log(n)) and auxiliary space: O(1), but the algorithm is much easier and the code way cleaner to understand.

Approach: We can observe that any number (say k) that we want to find, must exist within a row, including the first and last elements of the row (if it exists at all). So we first find the row in which k must lie using binary search ( O(log N) ) and then use binary search again to search in that row( O(log M) ).

Algorithm:

1) first we’ll find the correct row, where k=2 might exist. To do this we will simultaneously apply binary search on the first and last column.   

    low=0, high=n-1

     i) if( k< first element of row(a[mid][0]) ) => k must exist in the row above

                                              => high=mid-1;

     ii) if( k> last element of row(a[mid][m-1])) => k must exist in the row below

                                               => low=mid+1;                                                   

     iii) if( k> first element of row(a[mid][0]) &&  k< last element of row(a[mid][m-1]))

                                               => k must exist in this row

                                               => apply binary search in this row like in a 1-D array

     iv) i) if( k== first element of row(a[mid][0]) ||  k== last element of row(a[mid][m-1])) => found                                       

Example:
let k=2; n=3,m=4;
    matrix a: [0, 1, 2, 3 ]
              [10,11,12,13]
              [20,21,22,23]

1) low=0, high=n-1(=2) => mid=1 //check 1st row     [0....3]
                                                 -->[10...13]<-- 
                                                    [20...23]                                                        
   k < a[mid][0] => high = mid-1;(=1)
2) low=0, high=1; =>mid=0; //check 0th row    -->[0...3]<--  
    k>a[mid][0] && k<a[mid][m-1]  => k must exist in this row
    
    now simply apply binary search in 1-D array: [0,1,2,3]                              

Below is the implementation of the above algorithm:

Found at (1,4)

Time Complexity: O(log n + log m). 
Auxiliary Space: O(1)

This method is contributed by Ayushwant Gaurav. If you like neveropen and would like to contribute, you can also write an article using write.geeksforgeeks.org or mail your article to review-team@geeksforgeeks.org. See your article appearing on the neveropen main page and help other Geeks.
Please write comments if you find anything incorrect, or if you want to share more information about the topic discussed above.