John Holden

[email protected]
Inverting the Fibonacci Sequence
Mathematicians have been fascinated by the majestic simplicity of the Fibonacci
Sequence for centuries. It starts as a simple 1, 1, 2, 3, 5, 8, 13, ... computed recursively,
each term is equivalent to the sum of the previous two terms. This can be expressed
algebraically as Fn2  Fn1  Fn provided n  1 .

Fibonacci is so simple that children in their first algebra classes are drawn to ponder the
existence of a formula that more concisely defines the sequence. Graphing it indicates an
exponential correlation, and indeed nineteenth century mathematician J. P. M. Binet
discovered that the Golden Mean was related to the Fibonacci Sequence by proving that
n
 n  1 5
Fn  , provided that  is the Golden Mean and equal in value to .  is
  2
1 5
the conjugate, . This result is well known so we need not go further into Binet’s
2
formula here.

When looking at a graph of this sequence, I pondered the existence of an inverse function
that could compute the value of n, the index number which defines each term’s position
among the sequence, from the original Fibonacci term, Fn .

Finding an inverse for Binet’s formula is an algebraic nightmare, and it seems obvious
that there cannot be a perfect inverse function because each ordinate does not have a
unique abscissa-specifically F1  F2  1 . So the inverse function will have some
restrictions to its domain because Binet’s formula does not provide a one-to-one function.

By chance I stumbled across this theoretic inverse function. It reads

Try a few Fibonacci numbers yourself. Use Binet’s formula to find the nth term, then use
the new inverse function to find the index number, n , which should be the same as the
first n . We already understand that the inverse function will not work for n  1 because
F1  F2  1 (notice that this means the function does work for n  2 ).

How can we justify this formula for all integers n greater than 1 ?

Proof
The proof is a simple induction argument that hinges on the recursive properties of both
the Fibonacci Sequence itself and the Golden Mean.
What recursive property of the Golden Mean? Solve the equation x n2  x n1  x n for all
integer n . The exact values of the Golden Mean and its conjugate are given by this
formula, so  n 2   n1   n for all integers n .

If we deconstruct the rounding in our inverse function, n  log  Fn   2 , we will find

This justification is simple when using Mathematical Induction knowing the recursive
properties of phi and the Fibonacci Sequence. To do this, we tear the inequality into two
parts of  n2  Fn and Fn   n1 to justify separately (it is obvious that  n2   n1 for all
n and we do not need to discuss this).

1. When n  2 , we see that this inequality holds with  n2  Fn . We will quickly see

that from this point on  n2  Fn .

1   0  F2   1  1.618

2. Now let us prove  n2  Fn for all integer n  3 by the Principle of Mathematical

Induction.

Basis Step:

When n  3 , then

 1  1.618  2  F3 ,

and when n  4 , then

 2  2.618  3  F4 .

Induction Step:

Our induction for this theorem hinges on the recursive properties of the Golden

Mean and Fibonacci sequence which state that  n   n1   n2 and
 Fn  Fn1  Fn2 . With these properties and the demonstration of the inequality

for two consecutive values of n we can establish  n2  Fn for all integer n  3 .

Let n be a positive integer greater than 4 and consider  n 2 . The recursion

formula for  gives

 n  2   n 3   n  4

Appling the induction hypothesis,  k 2  Fk , is true for n  3, 4 . For k  5 it

follows that

 k  2  Fk  2
  k 1   Fk 1
 k  Fk

Which is what we were trying to establish. This recursion can be repeated for all

n  3 , thus this inequality is true on that interval. We have shown that for all

integer n  3 ,  n2  Fn by the Principle of Mathematical Induction.

3. Now we must establish Fn   n1 for all integer n  2 through mathematical

induction.

Basis Step:

When n  2 , then

F2  1  1.618   1

and when n  3 , then

F3  2  2.618   2 .

Induction Step:
 Let n be a positive integer greater than 3. Appling the Induction Hypothesis,

Fk   k 1 , and consider k  n  1 and k  n  2 . For k  4 , F4   41 because

Fk  2   k 3
 Fk 1   k  2
Fk   k 1

Again, we can repeat this recursion to establish Fk   k 1 for all integer k  5 .

Therefore, by the Principle of Mathematical Induction, it follows that, for all

n  3,

Fn   n1

holds.

4. These induction arguments prove the theorem.

Using this Formula

This formula is primarily an inverse function for Fibonacci numbers. Analogous to this

operation are inverse trigonometric functions. Just as sin(arcsin( x))  x , we could

express Binet’s Formula as Fib(n)  x and let the Inverse Formula be expressed in

function notation as Fib 1 ( x)  n , so Fib(Fib1 ( x))  x and Fib-1 (Fib(n))  n provided

n  2, 3, 4,... .

If some number x , not in the Fibonacci sequence, is entered into the inverse

function an integer n corresponding to the greatest Fibonacci number less than that x

will be given, which immediately brings us to another use for this theorem.

Characteristic Function for the Fibonacci sequence

be used to determine if a number is a Fibonacci number, or characteristic of the

Fibonacci sequence.

If Fib(Fib1 ( x))  x then x is a Fibonacci number.

Counting Function for the Fibonacci sequence

Another possible use for this function is to enter a number x into the inverse function

and determine whether x is characteristic of the Fibonacci sequence. If it is not, then the

output is equal to the number of Fibonacci numbers less than x . If x is characteristic of

the Fibonacci sequence then the number of Fibonacci numbers less than it would be the

output n would be one greater than the number of Fibonacci numbers less than x

because that n includes x . This is analogous to the  Counting Function in the Prime

Number Theorem.

Conclusion

We now have an inverse formula, a characteristic function, and a counting function for

the Fibonacci sequence. This is a new operation that can be used along with Binet’s

Formula any time one teaches or researches the Fibonacci sequence. This summer I used

the inverse formula when researching Fibonacci Primes. I wonder if Fibonacci himself

might have used this formula with his rabbits to see how many generations had gone by.

References

Knott, Ron. “A Formula for Fib(n).” 2003. http://www.mcs.surrey.ac.uk/Personal

/R.Knott/ Fibonacci/fibtable.html

Weisstein, Eric W. “Fibonacci Number.” Mathworld. 1999.

http://mathworld.wolfram.com/Fibonacci Number.html
Acknowledgments

Words cannot express my gratitude towards Rose Anna Wade, William Bauldry, Mark

Ginn, Ron Howard, Gregory Foley, and Brian Felkel for their comments and direction.