Checking whether a number is int or float

[punjabian]

I am working on a program that could calculate to calculate pth percentile from an array of size n. I have to check whether the product p*n gives an int or a float...i want to do doing

percentile = p*n
if(percentile == int)

cant really figure out what to put instead of this int so that it would tell me whether percentile is an int or a float. thanx

[Moschops]

It's your code. When you made percentile, did you do it like this:

int percentile;

or like this?

float percentile;


Do you actually mean, is percentile a whole number (such as 12.0) or does it have a fractional amount (such as 12.23)?

[punjabian]

i declared percentile as float

float percentile;

but obviously when i do

percentile = p*n;

answer can be a whole number too like 4.0.
So, yes, i want to check whether its a whole number or a fractional number?

[Nisheeth]

Maybe:
int(percentile) == percentile?

[punjabian]

o yes it worked. thanx ..... but i found another way too..

int percentile1;
float difference;
percentile1 == percentile;
difference = percentile - percentile1;
if(difference==0)
cout << "whole number" << endl;
else
cout << "fractional number" << endl;

[Disch]

These techniques are iffy at best. Remember that floating points are approximations and 10 might actually be 10.0000000001, in which case your above methods will fail.

[punjabian]

you are right Disch, i just checked that. what can be the solution for that??

[Mathhead200]

Don't use float. Use something with better precision. (Even double would be way better.)

[R0mai]

double would be still inaccurate on it's lower bits.
You can test if two floating point numbers are approximately equal by testing if they are very close to each other :

bool is_float_equal(float x, float y) {    
    const float epsilon = 1.e-5; //very small number, margin of rounding error
    return std::abs(x-y) < epsilon;
}

(not tested)