Given an array of **n** elements. Make strictly increasing and strictly decreasing subsequences from the array such that each array element belongs to increasing subsequence or decreasing subsequence, but not both, or can be part of none of the subsequence. Minimize the number of elements which are not part of any of the subsequences and find the count of such elements.

Examples:

Input :arr[] = { 7, 8, 1, 2, 4, 6, 3, 5, 2, 1, 8, 7 }Output :2 Increasing sequence can be { 1, 2, 4, 5, 8 }. Decreasing sequence can be { 7, 6, 3, 2, 1 }. So, only 2 (8, 7) element is left which are not part of either of the subsequences.Input :arr[] = { 1, 4, 2, 3, 3, 2, 4, 1 }Output :0 Increasing sequence can be { 1, 2, 3, 4 }. Decreasing sequence can be { 4, 3, 2, 1 }. So, no element is left which is not part of either of the subsequences.

The idea is to make a decision on each index, starting from index 0, one by one. For each index there can be three possibilities, first, it can belong to increasing sequence, second, it can belong to decreasing sequence, third, it does not belong to any of these sequences.

So, for each index, check for the optimal answer (minimum element which is not part of any of the subsequences) by considering it once as a part of increasing subsequence or as a part of decreasing subsequence. If the optimal answer cannot be achieved by them then leave it as the element which is not part of any of the sequence.

To decrease the complexity (using Dynamic Programming), we can store the number of elements which are not part of any of the subsequences using 3D array dp[x][y][z], where x indicates the decision index, y indicates the last index of decreasing sequence, z indicates the last index of increasing sequence.

Below is C++ implementation of this approach:

## C++

`// C++ program to return minimum number of elements which ` `// are not part of increasing or decreasing subsequences. ` `#include<bits/stdc++.h> ` `#define MAX 102 ` `using` `namespace` `std; ` ` ` `// Return minimum number of elements which is not part of ` `// any of the sequence. ` `int` `countMin(` `int` `arr[], ` `int` `dp[MAX][MAX][MAX], ` `int` `n, ` `int` `dec, ` ` ` `int` `inc, ` `int` `i) ` `{ ` ` ` `// If already calculated, return value. ` ` ` `if` `(dp[dec][inc][i] != -1) ` ` ` `return` `dp[dec][inc][i]; ` ` ` ` ` `// If whole array is traversed. ` ` ` `if` `(i == n) ` ` ` `return` `0; ` ` ` ` ` `// calculating by considering element as part of ` ` ` `// decreasing sequence. ` ` ` `if` `(arr[i] < arr[dec]) ` ` ` `dp[dec][inc][i] = countMin(arr, dp, n, i, inc, i + 1); ` ` ` ` ` `// calculating by considering element as part of ` ` ` `// increasing sequence. ` ` ` `if` `(arr[i] > arr[inc]) ` ` ` `{ ` ` ` `// If cannot be calculated for decreasing sequence. ` ` ` `if` `(dp[dec][inc][i] == -1) ` ` ` `dp[dec][inc][i] = countMin(arr, dp, n, dec, i, i + 1); ` ` ` ` ` `// After considering once by decreasing sequence, now try ` ` ` `// for increasing sequence. ` ` ` `else` ` ` `dp[dec][inc][i] = min(countMin(arr, dp, n, dec, i, i + 1), ` ` ` `dp[dec][inc][i]); ` ` ` `} ` ` ` ` ` `// If element cannot be part of any of the sequence. ` ` ` `if` `(dp[dec][inc][i] == -1) ` ` ` `dp[dec][inc][i] = 1 + countMin(arr, dp, n, dec, inc, i + 1); ` ` ` ` ` `// After considering element as part of increasing and ` ` ` `// decreasing sequence trying as not part of any of the ` ` ` `// sequence. ` ` ` `else` ` ` `dp[dec][inc][i] = min(1 + countMin(arr, dp, n, dec, inc, i + 1), ` ` ` `dp[dec][inc][i]); ` ` ` ` ` `return` `dp[dec][inc][i]; ` `} ` ` ` `// Wrapper Function ` `int` `wrapper(` `int` `arr[], ` `int` `n) ` `{ ` ` ` `// Adding two number at the end of array, so that ` ` ` `// increasing and decreasing sequence can be made. ` ` ` `// MAX - 2 index is assigned INT_MAX for decreasing sequence ` ` ` `// because/ next number of sequence must be less than it. ` ` ` `// Similarly, for Increasing sequence INT_MIN is assigned to ` ` ` `// MAX - 1 index. ` ` ` `arr[MAX - 2] = INT_MAX; ` ` ` `arr[MAX - 1] = INT_MIN; ` ` ` ` ` `int` `dp[MAX][MAX][MAX]; ` ` ` `memset` `(dp, -1, ` `sizeof` `dp); ` ` ` ` ` `return` `countMin(arr, dp, n, MAX - 2, MAX - 1, 0); ` `} ` ` ` `// Driven Program ` `int` `main() ` `{ ` ` ` `int` `n = 12; ` ` ` `int` `arr[MAX] = { 7, 8, 1, 2, 4, 6, 3, 5, 2, 1, 8, 7 }; ` ` ` ` ` `cout << wrapper(arr, n) << endl; ` ` ` `return` `0; ` `} ` |

*chevron_right*

*filter_none*

## Python3

`# Python3 program to return minimum number of elements which ` `# are not part of increasing or decreasing subsequences. ` `MAX` `=` `102` ` ` `# Return minimum number of elements which is not part of ` `# any of the sequence. ` ` ` `def` `countMin(arr,dp,n,dec,inc,i): ` ` ` ` ` `# If already calculated, return value. ` ` ` `if` `dp[dec][inc][i] !` `=` `-` `1` `: ` ` ` `return` `dp[dec][inc][i] ` ` ` ` ` `# If whole array is traversed. ` ` ` `if` `i` `=` `=` `n: ` ` ` `return` `0` ` ` ` ` `# calculating by considering element as part of ` ` ` `# decreasing sequence. ` ` ` `if` `arr[i]<arr[dec]: ` ` ` `dp[dec][inc][i] ` `=` `countMin(arr, dp, n, i, inc, i ` `+` `1` `) ` ` ` ` ` `# calculating by considering ` ` ` `# element as part of ` ` ` `# increasing sequence. ` ` ` `if` `arr[i] > arr[inc]: ` ` ` ` ` `# If cannot be calculated for ` ` ` `# decreasing sequence. ` ` ` `if` `dp[dec][inc][i] ` `=` `=` `-` `1` `: ` ` ` `dp[dec][inc][i] ` `=` `countMin(arr, dp, n, dec, i, i ` `+` `1` `) ` ` ` ` ` `# After considering once by ` ` ` `# decreasing sequence, now try ` ` ` `# for increasing sequence. ` ` ` `else` `: ` ` ` `dp[dec][inc][i] ` `=` `min` `(countMin(arr,dp,n,dec,i,i` `+` `1` `),dp[dec][inc][i]) ` ` ` ` ` `# If element cannot be part ` ` ` `# of any of the sequence. ` ` ` `if` `dp[dec][inc][i] ` `=` `=` `-` `1` `: ` ` ` `dp[dec][inc][i] ` `=` `1` `+` `countMin(arr, dp, n, dec, inc, i ` `+` `1` `) ` ` ` ` ` `# After considering element as part of increasing and ` ` ` `# decreasing sequence trying as not part of any of the ` ` ` `# sequence. ` ` ` `else` `: ` ` ` `dp[dec][inc][i]` `=` `min` `(` `1` `+` `countMin(arr,dp,n,dec,inc,i` `+` `1` `),dp[dec][inc][i]) ` ` ` ` ` `return` `dp[dec][inc][i] ` ` ` `# Wrapper Function ` `def` `wrapper(arr,n) : ` ` ` ` ` `# Adding two number at the end of array, so that ` ` ` `# increasing and decreasing sequence can be made. ` ` ` `# MAX - 2 index is assigned INT_MAX for decreasing sequence ` ` ` `# because/ next number of sequence must be less than it. ` ` ` `# Similarly, for Increasing sequence INT_MIN is assigned to ` ` ` `# MAX - 1 index. ` ` ` `arr[` `MAX` `-` `2` `]` `=` `1000000000` ` ` `arr[` `MAX` `-` `1` `]` `=` `-` `1000000000` ` ` `dp` `=` `[[[` `-` `1` `for` `i ` `in` `range` `(` `MAX` `)] ` `for` `i ` `in` `range` `(` `MAX` `)] ` `for` `i ` `in` `range` `(` `MAX` `)] ` ` ` `return` `countMin(arr,dp,n,` `MAX` `-` `2` `,` `MAX` `-` `1` `,` `0` `) ` ` ` `# Driver code ` `if` `__name__` `=` `=` `'__main__'` `: ` ` ` `n` `=` `12` ` ` `arr` `=` `[ ` `7` `, ` `8` `, ` `1` `, ` `2` `, ` `4` `, ` `6` `, ` `3` `, ` `5` `, ` `2` `, ` `1` `, ` `8` `, ` `7` `] ` ` ` `for` `i ` `in` `range` `(` `MAX` `): ` ` ` `arr.append(` `0` `) ` ` ` `print` `(wrapper(arr,n)) ` ` ` `# This code is contributed by sahilshelangia ` |

*chevron_right*

*filter_none*

Output:

2

**Time Complexity :** O(n^{3})

This article is contributed by ** Anuj Chauhan**. If you like GeeksforGeeks and would like to contribute, you can also write an article using contribute.geeksforgeeks.org or mail your article to contribute@geeksforgeeks.org. See your article appearing on the GeeksforGeeks main page and help other Geeks.

Please write comments if you find anything incorrect, or you want to share more information about the topic discussed above.

## Recommended Posts:

- Non-decreasing subsequence of size k with minimum sum
- Pyramid form (increasing then decreasing) consecutive array using reduce operations
- Number of Permutations such that no Three Terms forms Increasing Subsequence
- Minimum sum subsequence such that at least one of every four consecutive elements is picked
- Convert to Strictly increasing array with minimum changes
- Minimum number of increasing subsequences
- Longest Decreasing Subsequence
- Maximum Sum Decreasing Subsequence
- Find longest bitonic sequence such that increasing and decreasing parts are from two different arrays
- Longest Increasing Odd Even Subsequence
- Maximum Sum Increasing Subsequence | DP-14
- Longest Increasing Subsequence | DP-3
- Longest Common Increasing Subsequence (LCS + LIS)
- Printing Maximum Sum Increasing Subsequence
- Longest Increasing consecutive subsequence