Thread: mysterious numbers in a float

Hi everyone,
I just began to learn c, and here's the first problem I encountered,
the text is about the 3 data types (int, float, char). Here particularly, it talks about precision in floats. Just to test it, I set up a simple code, Here:

Code:

adrian@Adriansys:~/ccode$ cat floating.c
#include <stdio.h>

main()
{

  float num1 = 91.273;

  printf("\n%.1f\n", num1);
  printf("%.2f\n", num1);
  printf("%.3f\n", num1);
  printf("%.6f\n", num1);
  printf("%f\n", num1);
  printf("%.14f\n", num1);

}
adrian@Adriansys:~/ccode$ ./floating

91.3
91.27
91.273
91.273003
91.273003
91.27300262451172
adrian@Adriansys:~/ccode$

And the question is, if num1 was made to be equal to 91.273, where are these numbers after .273 coming from? (The 0's are obvious, but the 26245........)

Thanks for all the help,

This is a classic issue - every programmer encounters it.

It's been covered so many times, you're likely to get links to some reference on the material everywhere you ask.

They key summary is that floats and doubles are approximations. Occasionally they return an exact value (like 3.0), and on other occasions they seem to refuse to accept an exact value (like 2.99). While there are means of dealing with the issue, there is no means of "correcting" a double or float (in a way that's consistent - it can be maddening).

Two caveats in dealing with this fact. First, if you require precision, you'll need to use other means (large integers, such as storing values in pennies instead of dollars with decimals), or you'll have to accept the minor error. This means, two, that when comparing values, you often must consider a small range instead of a simple equality.

Thanks for the help guys,
Now everything seems to be clear.