I'm Losing Pennies!

[RyanV]

Hello,

I'm trying to make a program that takes an amount of money from the user
and returns the optimal way to convert it into change.

So, if the user was to input $1.76 the optimal change would be 7 quarters
and 1 penny. For $3.48 the optimal change would be 13 quarters, 2 dime,
and 3 pennies. Etc, etc.

Here is my code for determining change:

if(toBeChanged >= .25)
{
numberOfQuarters = (int) (toBeChanged / .25);
toBeChanged = fmod(toBeChanged, .25);
}

if(toBeChanged >= .1)
{
numberOfDimes = (int) (toBeChanged / .1);
toBeChanged = fmod(toBeChanged, .1);
}

if(toBeChanged >= .05)
{
numberOfNickels = (int) (toBeChanged / .05);
toBeChanged = fmod(toBeChanged, .05);
}

if(toBeChanged >= .01)
{
numberOfPennies = (int) (toBeChanged / .01);
toBeChanged = fmod(toBeChanged, .01);
}

toBeChanged is a double, and the rest of the variables are ints.

The problem is that sometimes it produces the correct amount of change,
but sometimes it's a penny off. Here are some sample outputs to demonstrate
this.


Totally fine, everything adds up:

How much shall I make change for?:.56
Total number of quarters is: 2
Total number of dimes is: 0
Total number of nickels is: 1
Total number of pennies is: 1

Not fine, the change is 1 penny short:

How much shall I make change for?:.57
Total number of quarters is: 2
Total number of dimes is: 0
Total number of nickels is: 1
Total number of pennies is: 1
Total amount of change is:4

Does anyone know why this is happening?

Thanks.

[Ben Duncan]

Try adding in the 0's with your doubles.

Imagine that you are the computer. What does .25 mean to you?
It just means .25 and not 0.25. You must change it to that!

[ne555]

^ Nope. .25 is a valid number

I bet to machine precision. Use integers for money.

[Catfish]

Does anyone know why this is happening?

When you do floating point math, tiny errors will creep in. So maybe that's why.
How to counter this... do less floating point math, by extracting the decimals as integers?

http://en.wikipedia.org/wiki/Floating_point#Accuracy_problems

[histrungalot]

For this project you should convert all monetary values to integers. That way you avoid floating point errors. This would change the what you have coded but would fix your problem. The errors that creep in to the calculations of floating point numbers will always be there and will therefore always cause an issue.

[Mats]

You can mutiply the numbers by a multiple of 10 and do all the arithmatic with integers, then divide them down again to a correct floating point number.

[RyanV]

Huh, so floating point math is inherently imprecise?

Thanks for the help, guys, my program ended up working perfectly.

[sohguanh]

To represent floating point number in computers is not precise. Use Mats strategy. It is commonly used in programs. Although it also have floating point imprecision but since you multiply by 10 before doing the arithmetic, the final floating point result after conversion should be quite ok.

Since floating point is imprecise, try not to do if (x == 1.0f) such statement most of the times it returns false.