Solve x in Periodic Continued Fraction Expansion

  • Thread starter Thread starter *FaerieLight*
  • Start date Start date
  • Tags Tags
    Fractions
Join the discussion
Registration is free. Ask a follow-up in this thread, or start your own.
4 replies · 2K views
*FaerieLight*
Messages
43
Reaction score
0

Homework Statement



Find the real positive number x with periodic continued fraction expansion [0;a,b,a,b,...] where a and b are positive integers.

Homework Equations





The Attempt at a Solution



I have determined that x definitely involves a root.
 
Physics news on Phys.org
It's been a while since I have done these kinds of problems but I think the notation "[0;a,b,a,b,...]" means a continued fraction of the form
[tex]x= \frac{a}{b+ \frac{a}{b+ \frac{a}{b+\cdot\cdot\cdot}}}[/tex]

If that is correct, then that is the same as
[tex]x= \frac{a}{b+ x}[/tex]
since the "continued" fraction in the denominator is still just "x".

Of course, that is the same as the quadratic equation [itex]x(b+ x)= x^2+ bx= a[/itex] or [itex]x^2+ bx- a= 0[/itex]. Use the quadratic formula to solve that.
 
If [tex]\alpha=[0;a,b,a,b,...][/tex], what do you get by computing

[tex]\frac{1}{\alpha} -a?[/tex]

Basically you want to continue performing operations on [tex]\alpha[/tex] until you get [tex]\alpha[/tex] back. The resulting equation will always be quadratic.

Edit: HallsofIvy, not quite, the continued fractions in the [tex][a_0;a_1,\ldots][/tex] notation are the so-called simple continued fractions, which always have unit numerators. So

[tex][0;a,b,a,b,...] = 0 + \frac{1}{a+\frac{1}{b+\frac{1}{a+\cdots}}}[/tex]

Your method still works for this of course, the result is just a bit more complicated.
 
Thanks HallsofIvy!

I was taught that [0;a,b,a,b,a,...] denotes
[tex]\frac{1}{a+\frac{1}{b+\frac{1}{a+...}}}[/tex]
The quadratic then becomes ax2+abx-b=0

Thank you very much! I actually looked at the continued fractions expansions for [tex]\sqrt{n}[/tex] for n=1,2,3,...,20 and tried to find patterns. I did find some interesting ones, but I just couldn't manage to use them to work out this problem. I suspected the solution was a simple one, and indeed thank you very much for showing me this beautifully simple solution.
 
Thanks to you too, fzero!