Why is the output different in both cases comparing floats

Question

PYTHON PROGRAM:

a = 0.2
if a == 0.2:
    print('*')

OUTPUT:

C PROGRAM:

#include <stdio.h>

int main(void) 
{
    float a = 0.2;
    if(a == 0.2)
    {
        puts("*");
    }
}

OUTPUT:

Why is the output different in both cases? Is there a difference between the working of == operator?

In the `a == 0.2` in C, 0.2 is a _double_ while a is a _float_. Change it to `if(a == 0.2f) { .. }`. If that works / changes the behavior, search for "Is floating point math broken?" — user2864740, Jun 15 '20 at 17:59
[Very related question](https://stackoverflow.com/questions/588004/is-floating-point-math-broken). Short summary: Don't ever do "exact comparison" using floating point values. Always use an *epsilon*. — Some programmer dude, Jun 15 '20 at 18:01
Also, it's often hard to do direct comparison between different languages with different semantics. Even for small limited areas where syntax might seem the same, the behavior might be vastly different. — Some programmer dude, Jun 15 '20 at 18:02
Alse [Why Are Floating Point Numbers Inaccurate?](https://stackoverflow.com/questions/21895756/why-are-floating-point-numbers-inaccurate) — Weather Vane, Jun 15 '20 at 18:03
@NSR No not even in Python can it be reliable. Lets say you have made two different functions that take different input and do some calculation on the input. Using a specific set of inputs should mathematically produce an exact equal result, but due to inaccuracies in representation and possibly compounded rounding errors, the two results are not *exactly* equal, just very close. — Some programmer dude, Jun 15 '20 at 18:12
@Some programmer dude Yeah, but while printing Python just approximates, right? — , Jun 15 '20 at 18:22
@Someprogrammerdude "Don't ever do "exact comparison" using floating point values. Always use an epsilon." is an overstatement. Comment space too small for a thorough explanation. — chux - Reinstate Monica, Jun 15 '20 at 19:30

score 3 · Accepted Answer · edited Jun 15 '20 at 18:14

It is because the types float and double have different width reserved for the mantissa. The type double can represent a floating number more precisely. In this case that matters as 0.2 can't be represented exactly and has a very slightly different representation when stored as a double vs. a float.

In the condition

if(a == 0.2)

the left operand has the type float while the right operand has the type double, as the default type of a number literal with a "." in C is a double.

So change the declaration the following way

double a = 0.2;

Or alternatively change the condition like

if(a == 0.2f)

Here is a demonstrative program

#include <stdio.h>

int main(void) 
{
    float a1 = 0.2;

    if ( a1 == 0.2f )
    {
        puts( "*" );
    }

    double a2 = 0.2;

    if ( a2 == 0.2 )
    {
        puts( "*" );
    }
}

Its output is

*
*

Why is the output different in both cases comparing floats

1 Answers1

Linked