Continued Fractions Homework: Find Explicit Formula

  • Thread starter Thread starter aznkangaroo3
  • Start date Start date
  • Tags Tags
    Fractions
AI Thread Summary
The discussion centers on deriving an explicit formula for a continued fraction defined recursively by tn+1 = 1 + (1/tn). Participants suggest looking for patterns in the terms generated from the initial value t1 = 1, leading to the realization that the terms correspond to ratios of consecutive Fibonacci numbers. A proposed solution format is t_n = F_(n+1)/F_n, which leads to a recurrence relation for Fibonacci numbers. Additionally, a more general approach to solving recursions is discussed, emphasizing the importance of finding distinct roots in quadratic equations. The conversation concludes with the suggestion of using induction to verify the formula derived from Fibonacci numbers.
aznkangaroo3
Messages
1
Reaction score
0

Homework Statement



Give the formula in terms of tn+1 for the continued fraction:
http://www.math.sunysb.edu/posterproject/www/images/continued-fraction.gif
and so on...

Homework Equations


The Attempt at a Solution


I got the recursive formula:tn+1=1+(1/tn), but I need the explicit formula
 
Last edited by a moderator:
Physics news on Phys.org
I am not brilliant at math, so I don't know if there is a systematic way to solve these problems. You know that t_1 = 1, so when you calculate t_2 and so on, you would plug 1 in for t_1. Don't. Treat all cases of t_1 as if it were a symbol from another planet. Write out a few terms in terms of t_1 and see if you can find a pattern.
 
Ok. After much confusion, I'm finally getting this straight. If you start writing out your t_n's you'll see that you are getting ratios of consecutive fibonacci numbers. This suggests that you look for solutions of the form t_n=F_(n+1)/F_n. Put this in and find a recurrence relation for the F's. (It should look familiar). Now you want to look for solutions to this recurrence relation of the form C*a^n. Using your recurrence relation, what value(s) can 'a' have? Can you add two solutions of this form? Do you have enough constants to reproduce t1=1, t2=2?
 
Last edited:
ha, yeah, you got the right idea. You need to solve a recursion.

actually, let me tell you something about a more general recursion:
x_{n+1}=\frac{a+bx_n}{c+dx_n}

How would you solve it?
well our goal is add constants to both sides to make
x_{n+1}+\alpha=a'\frac{1+\alpha x_n}{c+dx_n}
and
x_{n+1}+\beta=b'\frac{1+\beta x_n}{c+dx_n}

so that when I divide the first equation by the second equation, I'll get:
\frac{x_{n+1}+\alpha}{x_{n+1}+\beta}=\frac{a'}{b'}\left (\frac{1+\alpha x_n}{1+\beta x_n}\right )

a' and b' can be expressed in a,b,c,d, and alpha and beta respectively. Basically, all you need to do is find two solutions to a quadratic equation, if there are two distinct roots, then you are in business. The original problem can be solved using the exact method.

and of course, you can just guess and check and prove that the formula works by induction, and it will turn out to be some Fibonacci numbers...
 
Last edited:
I picked up this problem from the Schaum's series book titled "College Mathematics" by Ayres/Schmidt. It is a solved problem in the book. But what surprised me was that the solution to this problem was given in one line without any explanation. I could, therefore, not understand how the given one-line solution was reached. The one-line solution in the book says: The equation is ##x \cos{\omega} +y \sin{\omega} - 5 = 0##, ##\omega## being the parameter. From my side, the only thing I could...
Back
Top