Round a Float to # of Significant Digits Following The Decimal Point
Unfortunately, this next couple of weeks I get to do some development in C, and have ran into the always lovely issue of rounding a floating point number within a given # of significant digits. First, the need to do this usually stems from the fact that atof(char*) or strtof(char*,char**) return values one wouldn’t want/expect.
Example:
char* sValue = "20.4323"; float fValue = atof(sValue); printf("%f", fValue); //Debugger reports 20.4323006, Prints 20.432301
Of course, this is only an issue if you are needing to store and perform math against this value, since [s]printf can easily clean up the value for display. Either way, I came across a nice helper function that would probably be useful for anyone who doesn’t have access to libraries with precision control (I.E. STL).
inline double fround(double n, unsigned d) { return floor(n * pow(10., d) + .5) / pow(10., d); }
Every day you learn something new, even if it is back to basics…
Popularity: 2% [?]
Similar Posts
Thursday, February 22nd, 2007 at 4:18 pm
Tags: C/C++
Categories: Software Development
RSS 2.0 feed
Trackback Link
Add a comment
May 7th, 2008 at 3:03 pm
Actually, this is not quite correct assuming you want significant figures instead of decimal places. Your function rounds to a spcified number of places after the decimal point. A definition of significant figures (or digits) can be found at http://en.wikipedia.org/wiki/Significant_figures. To retain only a specified number of significant figures, the function should be:
inline double fround(double n, unsigned d){
unsigned p;
p = d - (unsigned)log10(n);
return floor(n * pow(10., p) + .5) / pow(10., p);
}
Given fround(1703.12345689,6), the result will be 1703.12, whereas, your original function will return 1703.123157.
Here is a simple test to verify the results:
#include <stdio.h>#include <math.h>
inline double fround1(double n, unsigned d)
{
return floor(n * pow(10., d) + .5) / pow(10., d);
}
inline double fround(double n, unsigned d)
{
unsigned p;
p = d - (unsigned)log10(n);
return floor(n * pow(10., p) + .5) / pow(10., p);
}
int main() {
printf("old fround: %21.8fn", fround1(1703.12345689,6));
printf("new fround: %21.8fn", fround (1703.12345689,6));
}
Cheers.
May 9th, 2008 at 8:41 pm
Actually, your comment made me realize that I didn’t properly describe what it was that I had created. In my specific case, I didn’t care about the number of significant digits, I only cared about the number digits following the decimal point. So, I may even re-title the post since it is obviously incorrect. Thank you for pointing that out and not letting me lead others astray.