Rounding a floating point number

[jacob navia]

Hi

"How can I round a number to x decimal places" ?

This question keeps appearing. I would propose the following
solution

#include <float.h>
#include <math.h>

double roundto(double x, int digits)
{
int sgn=1;

if (x == 0)
return 0;
if (digits DBL_DIG)
digits = DBL_DIG;
else if (digits < 1)
digits = 1;

if(x < 0.0) {
sgn = -sgn;
x = -x;
}
double p = floorl(log10l(x ));
p--;
p = digits-p;
double pow10 = pow(10.0, p);
return sgn*floor(x*pow 10+0.5)/pow10;
}

long double roundtol(long double x, int digits)
{
int sgn=1;

if (x == 0)
return 0;
if (digits LDBL_DIG)
digits = LDBL_DIG;
else if (digits < 1)
digits = 1;

if(x < 0.0) {
sgn = -sgn;
x = -x;
}
long double p = floorl(log10l(x ));
p--;
p = digits-p;
long double pow10 = powl(10.0, p);
return sgn*floorl(x*po w10+0.5)/pow10;
}
#include <stdio.h>
int main(void)
{
double d = 1.7888889988996 678;
long double ld = 1.7888889988996 678998877L;

for (int i = 0; i<=DBL_DIG;i++ ) {
printf("i=%d: %.15g\n",i,roun dto(d,i));
}
printf("\n 1.7888889988996 678\n");
for (int i = 0; i<=LDBL_DIG;i++ ) {
printf("i=%d: %.18Lg\n",i,rou ndtol(ld,i));
}
printf("\n 1.7888889988996 678998877L\n");
return 0;
}

--------------------------------------------------------------------
I would propose it to add it to the FAQ.
--
jacob navia
jacob at jacob point remcomp point fr
logiciels/informatique
http://www.cs.virginia.edu/~lcc-win32

[Bartc]

"CBFalconer " <cb********@yah oo.comwrote in message
news:47******** ******@yahoo.co m...

user923005 wrote:

>CBFalconer <cbfalco...@yah oo.comwrote:

>>user923005 wrote:

... snip float rounding discussion ...

You will still have to multiply and divide by powers of 10 and
use floor() to achieve the same thing because the C99 round()
functions round to nearest integer. They do not round to
nearest k decimal places.

Are you claiming that such multiplication and division by 10 is
too complex for the average reader of c.l.c? :-)

No, but I am claiming that using the C99 functions to round to N
digits will take twice as much work as a function that does it
directly, since the multiplication and division by a power of
ten and the floor function are all that is necessary in either
case. And so while you can round to N digits using the C99
rounding functions, it really does not make a lot of sense to do
it that way.

Let me also point out that such rounding is rarely needed. Most of
the time maintaining the original alleged precision and rounding
only the output, via printf, is needed.

There was a long discussion about this stuff last November in c.l.c with
various examples of why such rounding can be useful. One example was
cleaning noisy data.

--
Bart