You are on page 1of 14

Fibonacci Search

Fibonacci search is used to search an element

of a sorted array with the help of Fibonacci
Fibonacci number is subtracted from the index
thereby reducing the size of the list.
Fibonacci search is a process of searching a
sorted array by utilizing divide and conquer
Fibonacci search has a complexity of O(log(x))
Fibonacci Search
The Fibonacci number sequence is given by
{0,1,1,2,3,5,8,13,21,.} and is generated by
the following recurrence relation
F0 = 0
F1 = 1
Fi = Fi-1 + Fi-2
The fibonacci sequence finds an application in
a search technique termed fibonacci search
Fibonacci Search
Binary search selects the median of the sublist
as its next element for comparison, the
fibonacci search determines the next element
of comparison as dictated by the fibonacci
number sequence
Fibonacci search works only on ordered lists
The number of elements in the list is one less
than a fibonacci number
n = Fk-1
Decision Tree for Fibonacci Search
The decision tree from Fibonacci search satisfies
the following characteristics
If we consider a grandparent, parent and its child
nodes and if the difference in index between the
grandparent and parent is Fk then
(i) If the parent is a left child node then the
difference in index between the parent and its
child nodes is Fk-1, whereas
(ii) If the parent is a right child node then the
difference in index between the parent and the
child nodes is Fk-2
Decision Tree for Fibonacci Search
Consider an ordered list L={K1, K2,Kn} where
K1 < K2 < Kn where n = Fk-1
The fibonacci search decision tree for n=20
where 20 = (F8-1) is shown in figure
The root of the decision tree which is the first
element in the list to be compared with key K
during the search is that key Ki whose index i is
the closest fibonacci sequence number to n
In the case of n=20, K13 is the root since the
closest fibonacci number to n=20 is 13.
Decision Tree for Fibonacci Search
K8 K18

K5 K11 K16 K20

K3 K7 K10 K12 K15 K17 K19

K2 K4 K6 K9 K14

Decision Tree for Fibonacci Search
Algorithm illustarates the procedure for
fibonacci search
N is the number of data elements is such that
(i) Fk+1 > (n+1) and
(ii) Fk = (n+1) for some m>=0 where Fk+1 and Fk
are two consecutive fibonacci numbers
Algorithm for Fibonacci Search
Procedure Fibonacci_search(L,n,K)
// L(1:n) is a linear ordered (non decreasing) list of data
elements. N is such that Fk+1>(n+1). Also Fk+m=(n+1). K
is the key to be searched in the list //
Obtain the largest fibonacci number Fk closest to n+1
p= Fk-1
m=(n+1) (p+q)
If (K>L(p)) then p=p+m
Algorithm for Fibonacci Search
While ((P<>0) and (not found)) do
: K=L[p] { print (key found);
: K <L[p] : if (r=0) then p=0
else {p=p-r; t=q; q=r; r=t-r;}
: K > L[p] : if (q=1) then p=0
else {p=p+r; q=q-r; r=r-q}
If (found=false) then print (key not found);
End fibonacci_search.
Algorithm for Fibonacci Search
Let us search for the key K=434
N=20, the number of elements is such that (i)
F9 > n+1 and (ii) F8+m = (n+1) where m=0 and
The algorithm for fibonacci search first obtains
the largest fibonacci number closest to n+1. ie
F8 in this case.
Algorithm for Fibonacci Search
It compares K=434 with the data element with index F7
Since K>L[13] the search list is reduced to
Now K compares itself with L[18]=401. Since K>L[18]
the search list is further reduced to L[19:20]={434,536}
Now K is compared with L[20]=536. Since K<L[20] is
true it results in the search list {434} which when
searched yields the search key.
The key is successfully found
Following a similar procedure, searching for 66 in the
list yields an unsuccessful search
Algorithm for Fibonacci Search
Search < t p q r Remarks
key K K=L[p]
434 13 8 5 n=20
m=0 since
K > L[13] = 221 13 8 5 Since K > L[p]
K > L[13] = 221 18 3 2
K > L[18} = 401 20 1 1
K < L[20] = 536 1 19 1 0
K = L[19] = 434 Key is Found
Algorithm for Fibonacci Search
Search < t p q r Remarks
key K K=L[p]
66 13 8 5 n=20
K > L[13] = 221 8 8 5 3
K > L[8] = 55 11 2 1
K < L[11] = 94 2 10 1 1
K < L[10] = 84 1 9 1 0
K < L[9] = 81 Since (r=0) p is set to 0
Key is not found
Fibonacci Search
An advantage of fibonacci search over binary
search is that while binary search involves
division which is computationally expensive,
during the selection of next element for key
comparison, fibonacci search involves only
addition and subtraction