In How Many Transformations, Daniel came to the realization that linear recurrence of the form

$\left\{\begin{array}{l}x_{n+1}=a_1x_n+b_1y_n,\\ y_{n+1}=a_2x_n+b_2y_n,\end{array}\right.$

could easily be solved by setting up the matrix interpretation, and then diagonalize the matrix (assuming that's possible) which would allow us quick exponentiation, and hence obtain the Nth term directly.

We've seen this in the context of the Fast Fibonacci Transform. Specifically, set $x_n = f_{n+1}$ and $y_n = f_n$, and you get the system of equations

$\begin{cases} x_{n+1} = 1x_n + 1y_n \\ y_{n+1} = 1x_n + 0 y_n \\ \end{cases}$

Hence, $\begin{pmatrix} x_n \\ y_n \\ \end{pmatrix} = \begin{pmatrix} 1 & 1 \\ 1 & 0 \\ \end{pmatrix}^n \begin{pmatrix} 1 \\ 1 \\ \end{pmatrix}$

Using this, show the following:

1. Performing the eigenvalue decomposition, prove Binet's formula.

2. Using only matrix properties, conclude that

$f_{2n+1} = f_{n+1} ^2 + f_n ^2.$

Hint: $A^{2n} = A^n \times A^n$.

3. Find a similar formula for $f_{2n}$.

4. Express $f_{3n}$ in terms of $f_{n}, f_{n+1}$.

No vote yet

1 vote

Easy Math Editor

This discussion board is a place to discuss our Daily Challenges and the math and science related to those challenges. Explanations are more than just a solution — they should explain the steps and thinking strategies that you used to obtain the solution. Comments should further the discussion of math and science.

When posting on Brilliant:

`*italics*`

or`_italics_`

italics`**bold**`

or`__bold__`

boldNote: you must add a full line of space before and after lists for them to show up correctlyparagraph 1

paragraph 2

`[example link](https://brilliant.org)`

`> This is a quote`

Remember to wrap math in`\(`

...`\)`

or`\[`

...`\]`

to ensure proper formatting.`2 \times 3`

`2^{34}`

`a_{i-1}`

`\frac{2}{3}`

`\sqrt{2}`

`\sum_{i=1}^3`

`\sin \theta`

`\boxed{123}`

## Comments

Sort by:

TopNewestThis kinda gave away the method one uses to solve my problem (if you don't use eigenvalue decomposition). However, I would still like to see how eigenvalue decomposition works, so it would be nice if a brilliant user could post a solution to the problems suggested in this note. Thanks.

Log in to reply

Added a solution. It's pretty basic matrix manipulation, and would have been the approach that I thought you used.

Log in to reply

Let $\rho =\frac{1+\sqrt{5}}2$ and ${\overline \rho} = \frac{1-\sqrt{5}}2$. Let $D = \begin{pmatrix} \rho&0 \\ 0&{\overline \rho} \end{pmatrix}$. Let $A = \begin{pmatrix} \rho&{\overline \rho} \\ 1&1 \end{pmatrix}$. Then $\begin{pmatrix} 1&1\\1&0 \end{pmatrix} = ADA^{-1}$.

Log in to reply

(1) So $\begin{pmatrix} f_{n+1} \\ f_n \end{pmatrix} = (ADA^{-1})^n \begin{pmatrix} 1\\0 \end{pmatrix} = AD^nA^{-1} \begin{pmatrix} 1\\0 \end{pmatrix}$. Some painful computations yield $\begin{pmatrix} f_{n+1}\\f_n \end{pmatrix} = \frac1{\sqrt{5}} \begin{pmatrix} \rho^{n+1}-{\overline \rho}^{n+1} \\ \rho^n - {\overline \rho}^n \end{pmatrix}$; Binet's formula falls out of this.

Log in to reply

(2) and (3): let $M = \begin{pmatrix} 1 & 1 \\ 1 & 0 \end{pmatrix} = ADA^{-1}$. Then $M^n \begin{pmatrix} 1 \\ 0 \end{pmatrix} = \begin{pmatrix} f_{n+1} \\ f_n \end{pmatrix}$, and $M^n \begin{pmatrix} 0 \\ 1 \end{pmatrix} = M^{n-1} \begin{pmatrix} 1 \\ 0 \end{pmatrix} = \begin{pmatrix} f_n \\ f_{n-1} \end{pmatrix}$. So $M^n = \begin{pmatrix} f_{n+1} & f_n \\ f_n & f_{n-1} \end{pmatrix}$.

Now then, $\begin{pmatrix} f_{2n+1} \\ f_{2n} \end{pmatrix} = M^{2n} \begin{pmatrix} 1 \\ 0 \end{pmatrix} = M^n \begin{pmatrix} f_{n+1} \\ f_n \end{pmatrix} = \begin{pmatrix} f_{n+1} & f_n \\ f_n & f_{n-1} \end{pmatrix} \begin{pmatrix} f_{n+1} \\ f_n \end{pmatrix}$.

So we get $\begin{pmatrix} f_{2n+1} \\ f_{2n} \end{pmatrix} = \begin{pmatrix} f_{n+1}^2 + f_n^2 \\ f_{n+1}f_n + f_n f_{n-1} \end{pmatrix}$, and the formulas we want can be read off from there.

Log in to reply

$f_{3n} = f_n(3f_{n+1}^2-3f_nf_{n+1}+2f_n^2)$. (Basically the same process as the previous, but I had to substitute $f_{n-1} = f_{n+1}-f_n$ in some places.)

(4) Similar computations giveLog in to reply

I wish I saw this earlier

Log in to reply

@Trevor B. @Trevor Arashiro @Agnishom Chattopadhyay @Michael Mendrin @Sharky Kesa

Could you guys help fill out the Fast Fibonacci Transform Wiki Page? Thanks!

Log in to reply

Okay! Will go through this

Log in to reply

For

Fast Fibonacci Transform, here's a python code for calculating fibonacci numbers:Log in to reply