Given an n x n matrix mat[n][n] of integers, find the maximum value of mat(c, d) – mat(a, b) over all choices of indexes such that both c > a and d > b.
Example:
Input:
mat[N][N] = {{ 1, 2, -1, -4, -20 },
{ -8, -3, 4, 2, 1 },
{ 3, 8, 6, 1, 3 },
{ -4, -1, 1, 7, -6 },
{ 0, -4, 10, -5, 1 }};
Output: 18
The maximum value is 18 as mat[4][2]
- mat[1][0] = 18 has maximum difference.
The program should do only ONE traversal of the matrix. i.e. expected time complexity is O(n2)
A simple solution would be to apply Brute-Force. For all values mat(a, b) in the matrix, we find mat(c, d) that has maximum value such that c > a and d > b and keeps on updating maximum value found so far. We finally return the maximum value.
C++ Program
// A Naive method to find maximum value of mat1[d]
// - ma[a][b] such that c > a and d > b
#include <bits/stdc++.h>
using namespace std;
#define N 5
// The function returns maximum value A(c,d) - A(a,b)
// over all choices of indexes such that both c > a
// and d > b.
int findMaxValue(int mat[][N])
{
// stores maximum value
int maxValue = INT_MIN;
// Consider all possible pairs mat[a][b] and
// mat1[d]
for (int a = 0; a < N - 1; a++)
for (int b = 0; b < N - 1; b++)
for (int c = a + 1; c < N; c++)
for (int d = b + 1; d < N; d++)
if (maxValue < (mat1[d] - mat[a][b]))
maxValue = mat1[d] - mat[a][b];
return maxValue;
}
// Driver program to test above function
int main()
{
int mat[N][N] = {
{ 1, 2, -1, -4, -20 },
{ -8, -3, 4, 2, 1 },
{ 3, 8, 6, 1, 3 },
{ -4, -1, 1, 7, -6 },
{ 0, -4, 10, -5, 1 }
};
cout << "Maximum Value is "
<< findMaxValue(mat);
return 0;
}
[ad type=”banner”]
Output:
Maximum Value is 18
The above program runs in O(n^4) time which is nowhere close to expected time complexity of O(n^2)
An efficient solution uses extra space. We pre-process the matrix such that index(i, j) stores max of elements in matrix from (i, j) to (N-1, N-1) and in the process keeps on updating maximum value found so far. We finally return the maximum value.
C++ Program
// An efficient method to find maximum value of mat1[d]
// - ma[a][b] such that c > a and d > b
#include <bits/stdc++.h>
using namespace std;
#define N 5
// The function returns maximum value A(c,d) - A(a,b)
// over all choices of indexes such that both c > a
// and d > b.
int findMaxValue(int mat[][N])
{
//stores maximum value
int maxValue = INT_MIN;
// maxArr[i][j] stores max of elements in matrix
// from (i, j) to (N-1, N-1)
int maxArr[N][N];
// last element of maxArr will be same's as of
// the input matrix
maxArr[N-1][N-1] = mat[N-1][N-1];
// preprocess last row
int maxv = mat[N-1][N-1]; // Initialize max
for (int j = N - 2; j >= 0; j--)
{
if (mat[N-1][j] > maxv)
maxv = mat[N - 1][j];
maxArr[N-1][j] = maxv;
}
// preprocess last column
maxv = mat[N - 1][N - 1]; // Initialize max
for (int i = N - 2; i >= 0; i--)
{
if (mat[i][N - 1] > maxv)
maxv = mat[i][N - 1];
maxArr[i][N - 1] = maxv;
}
// preprocess rest of the matrix from bottom
for (int i = N-2; i >= 0; i--)
{
for (int j = N-2; j >= 0; j--)
{
// Update maxValue
if (maxArr[i+1][j+1] - mat[i][j] >
maxValue)
maxValue = maxArr[i + 1][j + 1] - mat[i][j];
// set maxArr (i, j)
maxArr[i][j] = max(mat[i][j],
max(maxArr[i][j + 1],
maxArr[i + 1][j]) );
}
}
return maxValue;
}
// Driver program to test above function
int main()
{
int mat[N][N] = {
{ 1, 2, -1, -4, -20 },
{ -8, -3, 4, 2, 1 },
{ 3, 8, 6, 1, 3 },
{ -4, -1, 1, 7, -6 },
{ 0, -4, 10, -5, 1 }
};
cout << "Maximum Value is "
<< findMaxValue(mat);
return 0;
}
[ad type=”banner”]
Output:
Maximum Value is 18
If we are allowed to modify of the matrix, we can avoid using extra space and use input matrix instead.