# Find length of longest Fibonacci like subsequence

Given a strictly increasing array A of positive integers where, . The task is to find the length of the longest Fibonacci-like subsequence of A. If such subsequence does not exist, return 0.

Examples:

Input: A = [1, 3, 7, 11, 12, 14, 18]
Output:
Explanation:
The longest subsequence that is Fibonacci-like: [1, 11, 12]. Other possible subsequences are [3, 11, 14] or [7, 11, 18].

Input: A = [1, 2, 3, 4, 5, 6, 7, 8]
Output:
Explanation:
The longest subsequence that is Fibonacci-like: [1, 2, 3, 5, 8].

Naive Approach: A Fibonacci-like sequence is such that it has each two adjacent terms that determine the next expected term.

For example, with 1, 1, we expect that the sequence must continue 2, 3, 5, 8, 13, … and so on.

• Use Set or Map to determine quickly whether the next term of Fibonacci sequence is present in the array A or not. Because of the exponential growth of these terms, there will be not more than log(M) searches to get next element on each iteration.
• For each starting pair A[i], A[j], we maintain the next expected value y = A[i] + A[j] and the previously seen largest value x = A[j]. If y is in the array, then we can then update these values (x, y) -> (y, x+y) otherwise we stop immediately.

Below is the implementation of above approach:

## C++

 // CPP implementation of above approach #include  using namespace std;   // Function to return the max Length of // Fibonacci subsequence int LongestFibSubseq(int A[], int n) {     // Store all array elements in a hash     // table     unordered_set<int> S(A, A + n);       int maxLen = 0, x, y;       for (int i = 0; i < n; ++i) {         for (int j = i + 1; j < n; ++j) {               x = A[j];             y = A[i] + A[j];             int length = 2;               // check until next fib element is found             while (S.find(y) != S.end()) {                   // next element of fib subseq                 int z = x + y;                 x = y;                 y = z;                 maxLen = max(maxLen, ++length);             }         }     }       return maxLen >= 3 ? maxLen : 0; }   // Driver program int main() {     int A[] = { 1, 2, 3, 4, 5, 6, 7, 8 };     int n = sizeof(A) / sizeof(A);     cout << LongestFibSubseq(A, n);     return 0; }   // This code is written by Sanjit_Prasad

## Java

 // Java implementation of above approach  import java.util.*; public class GFG {   // Function to return the max Length of  // Fibonacci subsequence      static int LongestFibSubseq(int A[], int n) {         // Store all array elements in a hash          // table          TreeSet S = new TreeSet<>();         for (int t : A) {             // Add each element into the set              S.add(t);         }         int maxLen = 0, x, y;           for (int i = 0; i < n; ++i) {             for (int j = i + 1; j < n; ++j) {                   x = A[j];                 y = A[i] + A[j];                 int length = 3;                   // check until next fib element is found                  while (S.contains(y) && (y != S.last())) {                       // next element of fib subseq                      int z = x + y;                     x = y;                     y = z;                     maxLen = Math.max(maxLen, ++length);                 }             }         }         return maxLen >= 3 ? maxLen : 0;     }   // Driver program      public static void main(String[] args) {         int A[] = {1, 2, 3, 4, 5, 6, 7, 8};         int n = A.length;         System.out.print(LongestFibSubseq(A, n));     } } // This code is contributed by 29AjayKumar

## Python3

 # Python3 implementation of the  # above approach    # Function to return the max Length  # of Fibonacci subsequence  def LongestFibSubseq(A, n):        # Store all array elements in      # a hash table      S = set(A)      maxLen = 0       for i in range(0, n):          for j in range(i + 1, n):                x = A[j]              y = A[i] + A[j]              length = 2               # check until next fib              # element is found              while y in S:                    # next element of fib subseq                  z = x + y                  x = y                  y = z                 length += 1                 maxLen = max(maxLen, length)                    return maxLen if maxLen >= 3 else 0   # Driver Code if __name__ == "__main__":       A = [1, 2, 3, 4, 5, 6, 7, 8]      n = len(A)      print(LongestFibSubseq(A, n))        # This code is contributed  # by Rituraj Jain

## C#

 // C# implementation of above approach using System; using System.Collections.Generic;   class GFG  {        // Function to return the max Length of      // Fibonacci subsequence      static int LongestFibSubseq(int []A, int n)     {          // Store all array elements in a hash          // table          SortedSet<int> S = new SortedSet<int>();          foreach (int t in A)          {              // Add each element into the set              S.Add(t);          }          int maxLen = 0, x, y;            for (int i = 0; i < n; ++i)         {              for (int j = i + 1; j < n; ++j)              {                  x = A[j];                  y = A[i] + A[j];                  int length = 3;                    // check until next fib element is found                  while (S.Contains(y) && y != last(S))                 {                        // next element of fib subseq                      int z = x + y;                      x = y;                      y = z;                      maxLen = Math.Max(maxLen, ++length);                  }              }          }          return maxLen >= 3 ? maxLen : 0;      }            static int last(SortedSet<int> S)     {         int ans = 0;         foreach(int a in S)             ans = a;         return ans;     }           // Driver Code     public static void Main(String[] args)      {          int []A = {1, 2, 3, 4, 5, 6, 7, 8};          int n = A.Length;          Console.Write(LongestFibSubseq(A, n));      }  }    // This code is contributed by 29AjayKumar

Output

5

Time Complexity: O(N2 * log(M)), where N is the length of array and M is max(A).

Efficient Approach: To optimize the above approach the idea is to implement Dynamic Programming. Initialize a dp table, dp[a, b] that represents the length of Fibonacci sequence ends up with (a, b). Then update the table as dp[a, b] = (dp[b – a, a] + 1 ) or 2

Below is the implementation of the above approach:

## C++

 // CPP program for the above approach #include  using namespace std;   // Function to return the max Length of // Fibonacci subsequence int LongestFibSubseq(int A[], int n) {     // Initialize the unordered map     unordered_map<int, int> m;     int N = n, res = 0;       // Initialize dp table     int dp[N][N];       // Iterate till N     for (int j = 0; j < N; ++j) {         m[A[j]] = j;         for (int i = 0; i < j; ++i) {             // Check if the current integer             // forms a finonacci sequence             int k = m.find(A[j] - A[i]) == m.end()                         ? -1                         : m[A[j] - A[i]];               // Update the dp table             dp[i][j] = (A[j] - A[i] < A[i] && k >= 0)                            ? dp[k][i] + 1                            : 2;             res = max(res, dp[i][j]);         }     }       // Return the answer     return res > 2 ? res : 0; }   // Driver program int main() {     int A[] = { 1, 3, 7, 11, 12, 14, 18 };     int n = sizeof(A) / sizeof(A);     cout << LongestFibSubseq(A, n);     return 0; }

Output

3

Time Complexity: O(N2), where N is the length of the array.

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.

My Personal Notes arrow_drop_up Check out this Author's contributed articles.

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 Improve this article if you find anything incorrect by clicking on the "Improve Article" button below.