Error while comparing Float variable

[cijothomas]

Dear friends,
Please take a look at the following code. Its storing .7 to b and comparing b with .7 . But getting just the opposite of expected result!!

Please explain!

(Three is expected o/p but we get two when run)

#include<stdio.h>
int main(int argc,char **argv)
{
 float a=0.5,b=0.7;
 if(b<0.7)
 {
	 
	if(a<0.5)
	 {
   printf("\nONe");
	}
     else
	 {
   printf("\nTwo");
	 }
 }
 else
 {
  printf("\nThree");
 }
 return 0;
}

[laserlight]

This is probably due to floating point inaccuracy, so you might check if (b < 0.7 + epsilon) instead, where epsilon is some small value.

By the way, please post your properly indented code in [code][/code] bbcode tags.

[cijothomas]

This is probably due to floating point inaccuracy, so you might check if (b < 0.7 + epsilon) instead, where epsilon is some small value.

By the way, please post your properly indented code in [code][/code] bbcode tags.

Sorry about not useing <code> tags.

One hint : If we use .7f in all places, this works fine! Anyone explain?

[laserlight]

If we use .7f in all places, this works fine! Anyone explain?

As far as I can tell, 0.7 cannot be precisely represented in binary, so floating point inaccuracy would still apply. My guess is that once you correctly switch to using float literals, the inaccuracy is the same in both cases, so your answer is consistent. Nonetheless, using an epsilon would still be good idea, methinks.

[juanpast]

laserlight explain it...

so you might check if (b < 0.7 + epsilon) instead

using f you set the value 0.7 as float, in the same range of b

[Codeplug]

http://www.parashift.com/c++-faq-lit...html#faq-29.17
http://c-faq.com/fp/fpequal.html
What Every Computer-Scientist Should Know About Floating Point Arithmetic

gg

[cijothomas]

Thanks for replying

Now i understand the things better