Finding the golden ratio using C
From C++ to PHP, debugging to webhosting; help and discussion about writing your latest program to running your website. NOT for help when your PC won't work.
| Announcements | Posted on | |
|---|---|---|
-
- Reputation:
- Thread Starter
- Full Member
- Location: Northern Ireland
- Posts: 120
Finding the golden ratio using CHi guys.
I was writing a small program to calculate the golden ratio by comparing Fibonacci numbers. It appears to work perfectly when comparing, say, the 91st and 90th Fibonacci numbers, but suddenly unravels if you go much higher. For example:
fib(101) / fib(100) gives 0.3something, while
fib(201) / fib(200) gives a negative number!
Does anyone know what the problem is?
My code is shown below:
-
- Reputation:
- TSR Demigod
- Location: Scotland
- Posts: 5,838
Re: Finding the golden ratio using C
Mathematica:
If you want to do this sort of thing, see e.g. GMP.Code:In[1]:= N@Log[2,Fibonacci[200]] Out[1]= 137.687
-
- Reputation:
- TSR Idol
- Posts: 7,578
Re: Finding the golden ratio using CJust had a play with this... if it's any consolation you'll probably run out of precision in the long double well before you overflow the long long int.
on my box it appears to have converged at fib(48)/ fib(47) within the number of digits a long float can represent -
- Reputation:
- Overlord in Training
- Location: England
- Posts: 2,583
Re: Finding the golden ratio using CThat's not a buffer overflow. That's an integer overflow.(Original post by estel)
Has to be a buffer overflow.
unsigned long longs are only 2^64 -1, and I can definitely imagine the 200th fib term exceeding that -
- Reputation:
- TSR Idol
- Location: Bristol
- Posts: 9,352
Re: Finding the golden ratio using CYou're quite right. The words fail me(Original post by CJKay)
That's not a buffer overflow. That's an integer overflow.
