Actually I'm doing a puzzle..
where I need to find last 3 digits of the number before decimal.
Number is [ 3 + sqrt(5) ] ^ P
where P is taken as input.. This questions looks simple.
but the problem here comes when the number becomes bigger than "long long double"
for example .. take P = 28
it showing -808 as output.. which is impossible.
how should I deal with this kind of problem.
Get an arbitrary precision library like GMP: The GNU MP Bignum Library.
(3+sqrt(5))^28 is approximately 1.35565e20, which is within the range of a float, let alone a long double (there is no long long double that I know of). However, you may not have all the decimal digits accurately, which is what you want.
If all you need is the last three digits of the number before the decimal point, then that just means you should never store a number larger than 1000 (and if you go over 1000, then start subtracting 1000 until you aren't).
hmmm.. I compiled with P =28 by using long double..
this is answer. "-2147483648.000000" negative.. which is wrong.
I think GMP lib is one possible way.. but can't I get off with C lib's..
Last edited by suryak; 08-18-2011 at 01:13 PM. Reason: typo
What compiler / OS are you using?Originally Posted by suryak
Warning: Some or all of my posted code may be non-standard and as such should not be used and in no case looked at.
Ok... out of curiosity did you try that as ... pow(3+sqrt(5.0),n)
If you're using Windows and MinGW, then you have to do some work for printing long double -- I think I've posted this code here before as a quick-and-dirty way to see the numbers:
Now that that's out of the way, you actually don't want to do this in the way you are doing -- if you ever keep a number bigger than 1000, you're doing it wrong.Code:char *printf_ld(long double arg, char answer[15]) { /* For my own sake: scientific notation with six digits */ double mantissa, exponent, lg; lg = log10l(fabs(arg)); exponent = floor(lg); mantissa = pow(10, lg-exponent); mantissa = copysign(mantissa, arg); sprintf(answer, "%.6fe%+04g", mantissa, exponent); return answer;
> hmmm.. I compiled with P =28 by using long double..
> this is answer. "-2147483648.000000" negative.. which is wrong.
And where's your code?
Frankly, your terse posts with snippets of information are getting tiresome.
How do we know you didn't mess up the printf format, or anything else for that matter.
If you dance barefoot on the broken glass of undefined behaviour, you've got to expect the occasional cut.
If at first you don't succeed, try writing your phone number on the exam paper.
Print your long doubles using printf and "%Lf" as a format specifier. You should also be using sqrtl and powl (instead of just sqrt and pow), for extra precision. You'll notice that even with long double, your answers are not exact, and may be off by tens of thousands. That's because of floating point inaccuracy, which is exacerbated by the algorithm used in powl(). Read this article for details: What Every Computer Scientist Should Know About Floating-Point Arithmetic.
Not quite sure what you mean here. Could you not get it to work or are you not allowed to use 3rd party libraries? Maybe this is an assignment to illustrate the issues with floating point calculations...
