Help rounding to two decimal places

[p1kn1c]

Hi, what is the best way to round a double to two decimal places?

ie, if i had the number 77.786 i would want it rounded to 77.79.


I dont want a way to just display this number as two decimal places but to keep it like that so any time i use it it will be 77.79.

Also i cant know anything about the number, so there is no way of knowing if it is 77.786...or 7777777.554.

Thanks for any help.

 

[cdraycott]

p1kn1c,

Here is a possible answer:

http://www.codeproject.com/cpp/significant_figures.asp

 

[ArkM]

Float point numbers (float and double) are not decimal numbers: they are binary approximations only. So you can't round these numbers exactly. Strictly speaking, 0.1, 0.01, 0.001 etc (double constants) are not equal to rational numbers 1/10, 1/100, 1/1000. No native decimal arithmetics in C and C++ languages. Have no illusions about this matter.

You may use this logic to round a double to two decimal places (for positive numbers):

double fr100(double x)
{
  double f, xi, xf;

  xf = modf(x,&xi);
  f = floor(xf*100+0.5)/100.0;
  return xi + f;
}

 

[OlafBogus]

Here is another way.
for 2dp set PRECISION to 0.01 for 4dp 0.0001 etc.

double myRemainder = fmod(myValue, PRECISION);
if (myRemainder > ( 0.5 * PRECISION ))
{
    myValue = (myValue - myRemainder + PRECISION);
}
else
{
    myValue = (myValue  - myRemainder);
}

O.B.

 

[Salem]

> I dont want a way to just display this number as two decimal places but to keep it like that so any time i use it it will be 77.79.
Then you're out of luck, floating point numbers are not the thing to use for the reason ArkM states.

http://en.wikipedia.org/wiki/Floating_point

In particular, seek out David Goldberg's paper at the end.

If you really want exactly 2dp all the time, then multiply everything by 100, then use %100 and /100 to extract the relevant quantities. Addition remains easy, but multiply becomes more interesting.

http://en.wikipedia.org/wiki/Fixed-point_arithmetic

--