Consider the Fibonacci sequence { F i } i ≥ 1 \{ F_i \}_{i \ge 1} { F i } i ≥ 1 :
1 , 1 , 2 , 3 , 5 , 8 , 13 , 21 , . . . .
1, 1, 2, 3, 5, 8, 13, 21, ....
1 , 1 , 2 , 3 , 5 , 8 , 13 , 21 , .... which is defined recursively as follows:
{ F 1 = 1 F 2 = 1 F n + 2 = F n + 1 + F n n ≥ 1
\begin{cases}
F_1 = 1 \\
F_2 = 1 \\
F_{n + 2} = F_{n + 1} + F_{n} \quad n \ge 1
\end{cases}
⎩ ⎨ ⎧ F 1 = 1 F 2 = 1 F n + 2 = F n + 1 + F n n ≥ 1 We discuss some techniques for computing the (approximate) closed form formula for F n F_n F n for n ≥ 1 n \ge 1 n ≥ 1 .
Method 1
We guess that F n F_n F n is of closed form as F n = z n F_n = z^n F n = z n for some z ∈ R z \in \R z ∈ R . As F n + 2 = F n + 1 + F n F_{n + 2} = F_{n + 1} + F_n F n + 2 = F n + 1 + F n , z z z needs to satisfies
z 2 = z + 1
z^2 = z + 1
z 2 = z + 1 The quadratic equation has two roots, z 1 = 1 + 5 2 z_1 = \frac{1 + \sqrt 5}{2} z 1 = 2 1 + 5 and z 1 = 1 − 5 2 z_1 = \frac{1 - \sqrt 5}{2} z 1 = 2 1 − 5 respectively.
Let a , b ∈ R a, b \in \R a , b ∈ R . F n = a z 1 n + b z 2 n F_n = a z_1^n + b z_2^n F n = a z 1 n + b z 2 n is also a possible solution that satisfies F n + 2 = F n + 1 + F n F_{n + 2} = F_{n + 1} + F_n F n + 2 = F n + 1 + F n . We can compute the exact value of a a a and b b b by the constraints F 1 = F 2 = 1 F_1 = F_2 = 1 F 1 = F 2 = 1 :
a z 1 + b z 2 = 1 a z 1 2 + b z 2 2 = 1
a z_1 + b z_2 = 1 \\
a z_1^2 + b z_2^2 = 1
a z 1 + b z 2 = 1 a z 1 2 + b z 2 2 = 1 i.e.,
a ( 1 + 5 ) + b ( 1 − 5 ) = 2 a ( 6 + 2 5 ) + b ( 6 − 2 5 ) = 4
a (1 + \sqrt 5) + b (1 - \sqrt 5) = 2 \\
a (6 + 2\sqrt 5) + b (6 - 2 \sqrt 5) = 4
a ( 1 + 5 ) + b ( 1 − 5 ) = 2 a ( 6 + 2 5 ) + b ( 6 − 2 5 ) = 4 Therefore a = 1 / 5 a = 1 / \sqrt 5 a = 1/ 5 , b = − 1 / 5 b = -1 / \sqrt 5 b = − 1/ 5 , and
F n = ( 1 + 5 2 ) n − ( 1 − 5 2 ) n 5
F_n = \frac{ (\frac{1 + \sqrt 5} {2})^n - (\frac{1 - \sqrt 5} {2})^n }{\sqrt 5}
F n = 5 ( 2 1 + 5 ) n − ( 2 1 − 5 ) n ■ \blacksquare ■
Method 2
Consider the intervals derived from the sequences:
1 , 1 , 2 , 3 , 5 , 8 , 13 , 21 , . . . .
1, 1, 2, 3, 5, 8, 13, 21, ....
1 , 1 , 2 , 3 , 5 , 8 , 13 , 21 , .... → [ 1 1 , 2 1 ] , [ 3 2 , 5 3 ] , [ 8 5 , 13 8 ] , [ 21 13 , 34 21 ] , . . .
\rightarrow [\frac{1}{1}, \frac{2}{1} ], \quad [ \frac{3}{2} , \frac{5}{3} ], \quad [ \frac{8}{5} , \frac{13}{8} ], \quad [ \frac{21}{13}, \frac{34}{21} ], ...
→ [ 1 1 , 1 2 ] , [ 2 3 , 3 5 ] , [ 5 8 , 8 13 ] , [ 13 21 , 21 34 ] , ... Define x n = F n + 1 F n x_n = \frac{F_{n + 1} }{F_n} x n = F n F n + 1 and denote I n I_n I n the n n n -th interval. Then
I n = [ x 2 n − 1 , x 2 n ]
I_n = [x_{2n - 1}, x_{2n } ]
I n = [ x 2 n − 1 , x 2 n ] By observation of the first few I n I_n I n 's, they seems to be nested closed intervals, with shrinking lengths. If the length of I n I_n I n converges to 0 0 0 , then the limit of x n x_n x n exists. Suppose that
lim n x n = ϕ
\lim_{n} x_n = \phi
n lim x n = ϕ Then
lim n F n = O ( ϕ n )
\lim_n F_n = O(\phi^n )
n lim F n = O ( ϕ n ) We will prove this in what follows.
The key transformation we performed is
x n = F n + 1 F n = F n + F n − 1 F n = 1 + 1 x n − 1 .
x_{n} = \frac{ F_{n + 1} }{F_{n} } = \frac{ F_{n } + F_{n - 1 } }{ F_{n} } = 1 + \frac{1}{ x_{ n - 1} }.
x n = F n F n + 1 = F n F n + F n − 1 = 1 + x n − 1 1 . Lemma 1
x 1 < x 3 < x 5 < x 7 < . . . x 2 > x 4 > x 6 > x 8 > . . .
x_1 < x_3 < x_5 < x_7 < ... \\
x_2 > x_4 > x_6 > x_8 > ... \\
x 1 < x 3 < x 5 < x 7 < ... x 2 > x 4 > x 6 > x 8 > ... Proof: For n ≥ 1 n \ge 1 n ≥ 1 ,
x 2 n + 2 − x 2 n = 1 x 2 n + 1 − 1 x 2 n − 1 = x 2 n − 1 − x 2 n + 1 x 2 n + 1 x 2 n − 1 .
x_{2n + 2} - x_{2n} = \frac{1}{ x_{2n + 1} } - \frac{1}{ x_{2n - 1} } = \frac{ x_{2n - 1} - x_{2n + 1} }{ x_{2n + 1} x_{2n - 1} }.
x 2 n + 2 − x 2 n = x 2 n + 1 1 − x 2 n − 1 1 = x 2 n + 1 x 2 n − 1 x 2 n − 1 − x 2 n + 1 . Similarly,
x 2 n + 1 − x 2 n − 1 = 1 x 2 n − 1 x 2 n − 2 = x 2 n − 2 − x 2 n x 2 n x 2 n − 2 .
x_{2n + 1} - x_{2n - 1} = \frac{1}{ x_{2n} } - \frac{1}{ x_{2n - 2} } = \frac{ x_{2n - 2} - x_{2n} }{ x_{2n} x_{2n - 2} }.
x 2 n + 1 − x 2 n − 1 = x 2 n 1 − x 2 n − 2 1 = x 2 n x 2 n − 2 x 2 n − 2 − x 2 n . As x 3 − x 1 = 3 / 2 − 1 > 0 x_3 - x_1 = 3 /2 - 1 > 0 x 3 − x 1 = 3/2 − 1 > 0 , and x 4 − x 2 = 5 / 3 − 2 < 0 x_4 - x_2 = 5 / 3 - 2 < 0 x 4 − x 2 = 5/3 − 2 < 0 , the lemma follows from induction.
□ \square □
Lemma 2. For n ≥ 1 n \ge 1 n ≥ 1 ,
x 2 n − x 2 n − 1 ≥ 0
x_{2n} - x_{2n - 1} \ge 0
x 2 n − x 2 n − 1 ≥ 0 Proof: Clearly this holds for n = 1 n = 1 n = 1 . Consider n ≥ 2 n \ge 2 n ≥ 2 ,
x 2 n − x 2 n − 1 = 1 x 2 n − 1 − 1 x 2 n − 2 = − x 2 n − 1 − x 2 n − 2 x 2 n − 1 ⋅ x 2 n − 2 = x 2 n − 2 − x 2 n − 3 x 2 n − 1 ⋅ x 2 n − 2 ⋅ x 2 n − 2 ⋅ x 2 n − 3
x_{2n} - x_{2n - 1} = \frac{1}{x_{2n - 1} } - \frac{1}{x_{2n - 2} } = - \frac{ x_{2n - 1} - x_{2n - 2} }{x_{2n - 1} \cdot x_{2n - 2}} = \frac{ x_{2n - 2} - x_{2n - 3} }{x_{2n - 1} \cdot x_{2n - 2} \cdot x_{2n - 2} \cdot x_{2n - 3} }
x 2 n − x 2 n − 1 = x 2 n − 1 1 − x 2 n − 2 1 = − x 2 n − 1 ⋅ x 2 n − 2 x 2 n − 1 − x 2 n − 2 = x 2 n − 1 ⋅ x 2 n − 2 ⋅ x 2 n − 2 ⋅ x 2 n − 3 x 2 n − 2 − x 2 n − 3 The claim follows from induction.
Corollary 2. For n ≥ 2 n \ge 2 n ≥ 2 ,
I n = x 2 n − x 2 n − 1 = x 2 n − 2 − x 2 n − 3 x 2 n − 1 ⋅ x 2 n − 2 ⋅ x 2 n − 2 ⋅ x 2 n − 3 ≤ x 2 n − 2 − x 2 n − 3 x 2 n − 1 ≤ x 2 n − 2 − x 2 n − 3 x 3 = 2 3 ( x 2 n − 2 − x 2 n − 3 ) = 2 3 I n − 1
\begin{aligned}
I_n &= x_{2n} - x_{2n - 1} \\
&= \frac{ x_{2n - 2} - x_{2n - 3} }{x_{2n - 1} \cdot x_{2n - 2} \cdot x_{2n - 2} \cdot x_{2n - 3} } \\
&\le \frac{ x_{2n - 2} - x_{2n - 3} }{ x_{2n - 1} } \\
&\le \frac{ x_{2n - 2} - x_{2n - 3} }{ x_3 } \\
&= \frac{2}{3} (x_{2n - 2} - x_{2n - 3}) \\
&= \frac{2}{3} I_{n - 1}
\end{aligned}
I n = x 2 n − x 2 n − 1 = x 2 n − 1 ⋅ x 2 n − 2 ⋅ x 2 n − 2 ⋅ x 2 n − 3 x 2 n − 2 − x 2 n − 3 ≤ x 2 n − 1 x 2 n − 2 − x 2 n − 3 ≤ x 3 x 2 n − 2 − x 2 n − 3 = 3 2 ( x 2 n − 2 − x 2 n − 3 ) = 3 2 I n − 1 Hence lim n ∣ I n ∣ = 0 \lim_n |I_n| = 0 lim n ∣ I n ∣ = 0 .
□ \square □
Combining lemma 1 and 2, we conclude that lim x n = ϕ \lim x_n = \phi lim x n = ϕ exists. As
lim x n = ϕ = lim ( 1 + 1 x n − 1 ) = 1 + 1 ϕ
\lim x_n = \phi = \lim (1 + \frac{1}{x_{n - 1} } ) = 1 + \frac{1}{\phi }
lim x n = ϕ = lim ( 1 + x n − 1 1 ) = 1 + ϕ 1 We conclude that ϕ = 1 + 5 2 \phi = \frac{ 1 + \sqrt{5} }{2 } ϕ = 2 1 + 5 .
← Back to all notes Back to top ↑