Why is my code returning a number lower than what it should return?

Question

So I am doing an assignment for a programming course and I am having trouble with one line of my code.

the answer should be (if my math is correct) 40 but the program returns 39.

example: 18.33 * 40 *52 = 38126.4

I then use this code

double wage = 0;
wage = wage * 40 * 52;
wage = wage - floor(wage);
wage = wage * 100;
printf(" %d cents", (int)wage);

But the program returns 39. If I don't change it to int it is 40.0000000

What is happening?

...and what's exactly the content of `wage` just before your piece of code ? — Jabberwocky, Jan 16 '17 at 07:59

StoryTeller - Unslander Monica · Accepted Answer · 2017-01-16T08:55:07.877

4

Floating point numbers are incapable of representing every real number to infinite precision. And 40.0 is no exception. Instead, if your calculation is done in a correct and numerically stable way, you'll get the closest number a double can represent.

Functions like printf make the output pretty by considering how close the operand they received is to a given number.

So your result may be 39.99998, the closest number a double can represent to your correct result. And it is shown as 40.0000000.

However casting to an int performs truncation. So it becomes 39.

For your purposes (you seem to be dealing in money), an approach which will work best is to get rid of floating point numbers completely. You only need two integers. One for cents (which grows only up to 100), and one for dollars¹.

_{¹ I assume dollars and cents because of the multiplication by 100 in your code. Naturally it applies for any currency}

edited Jan 16 '17 at 08:55

answered Jan 16 '17 at 07:58

StoryTeller - Unslander Monica

165,132
21
377
458

Thanks. I solved my need for autograder by adding 0.01 – AncientOwl Jan 16 '17 at 08:04
But i'll keep your solution in mind also. Thanks again – AncientOwl Jan 16 '17 at 08:05
1

@AncientOwl - That's a brittle fix. It may skew *correct* calculations. I assure you some employers will not like paying an extra cent. – StoryTeller - Unslander Monica Jan 16 '17 at 08:06
Adding 0.01 sounds more like a kludge than a fix, the solution proposed by @StoryTeller is a much better idea. – Colin Jan 16 '17 at 08:07
True but for my current needs it works. This is just a introduction class so i'll learn more as I continue – AncientOwl Jan 16 '17 at 08:08
2

@AncientOwl-- I imagine that the assignment was to solve a programming problem, not to find a way to trick the autograder. – ad absurdum Jan 16 '17 at 08:24

Why is my code returning a number lower than what it should return?

1 Answers1