Floating point number is rounding off in C

Question

i started learning c. Today, while i am working on a program, i found an interesting thing and i made another small program (similar to my issue) to check it.

#include<stdio.h>
int main(void)
{
    float num1=867.0;
    float num2=.6921;
    printf("sum = %.4f \n",num1+num2);
    return 0;
}

if i run the above program, i am getting 867.6921 as the answer. but if i changed the .4%f to %f the answer is changing to sum = 867.692078. why's the change in the output? Also, is there any way that i can get the answer without using the .4%f?

http://stackoverflow.com/questions/21895756/why-are-floating-point-numbers-inaccurate — yano, Apr 15 '16 at 00:34
By the way, changing the `float` types to `double` produces the answer you expect. — Boo Radley, Apr 15 '16 at 00:39
@BooRadley it gives me `867.692100`. but the answer i need to get is `867.6921` — yash, Apr 15 '16 at 00:46

score 2 · Answer 1 · answered Apr 15 '16 at 00:35

2

but if i changed the .4%f to %f the answer is changing to sum = 867.692078

printf rounds the result to that many digits after the floating point you specified, the default is .6. While rounding, it uses the current floating point rounding mode, the default is round to the nearest.

answered Apr 15 '16 at 00:35

Maxim Egorushkin

131,725
17
180
271

that makes sense. thank you so much! i have another question if that's alright. in my program i am trying to convert a string to a floating point number. is there any way of doing the above without `.4%f` since the user doesn't know how many numbers are there after the decimal. – yash Apr 15 '16 at 00:39
...I don't think this is the point of confusion. This is a floating point representability issue. Naively we would expect `867.692100`, but that value is not representable as a `float`. – J... Apr 15 '16 at 00:39
@J... the output should be 867.6921 rather than 867.692100. is there any way of getting it? – yash Apr 15 '16 at 00:44

J... · Accepted Answer · 2016-04-15T00:57:17.670

0

The float type has about seven digits of precision (can be anywhere from 6-9 digits). One alternative is to use g instead of f. This fixes the number of precise digits in the number, not just those after the decimal.

  printf("sum = %.7g \n",num1+num2);

This produces output

sum = 867.6921

and will always give seven digits of precision for any input. Number formats would be like :

0.000000E+07
0000000
000000.0
00000.00
0000.000
000.0000

...etc

edited Apr 15 '16 at 00:57

answered Apr 15 '16 at 00:52

J...

30,968
6
66
143

@buggenerator I was guessing here from your comments, but was this actually your question? How to fix the number of precise digits in `printf`? – J... Apr 15 '16 at 00:59
Yeah. should've asked like that @J... – yash Apr 15 '16 at 01:00
http://www.exploringbinary.com/number-of-digits-required-for-round-trip-conversions/ – Maxim Egorushkin Apr 15 '16 at 13:11

Floating point number is rounding off in C

2 Answers2