Double comparison for range is not working correctly

[jmcphers27]

Hello, I am trying to compare a double to be within various ranges, but the comparison is not working correctly. Rounding precision is not a concern, because if the value will match the boundaries so rarely, and it is rounding up to the higher range is acceptable. Below is the doe at issue:

if (kq == 2)
{
if (fundt[kq-1][ky-1][2-1] < 450,000,000.00) ktti[ky-1][0] = 1;
else if ((fundt[kq-1][ky-1][2-1] >= 450,000,000.00) && (fundt[kq-1][ky-1][2-1] < 525,000,000.00)) ktti[ky-1][0] = 2;
else if ((fundt[kq-1][ky-1][2-1] >= 525,000,000.00) && (fundt[kq-1][ky-1][2-1] < 650,000,000.00)) ktti[ky-1][0] = 3;
else if ((fundt[kq-1][ky-1][2-1] >= 650,000,000.00) && (fundt[kq-1][ky-1][2-1] < 750,000,000.00)) ktti[ky-1][0] = 4;
else if ((fundt[kq-1][ky-1][2-1] >= 750,000,000.00) && (fundt[kq-1][ky-1][2-1] < 850,000,000.00)) ktti[ky-1][0] = 5;
else ktti[ky-1][0] = 6;
printf("DEBUG9: The Quarter is %d, Trust Fund is %20.3f, the Tax Table is %d\n", kq, fundt[kq-1][ky-1][2-1], ktti[ky-1][0]);
}
else if (kq==4)
{
if (fundt[kq-1][ky-1][2-1] < 450,000,000.00) ktti[ky-1][1] = 1;
else if ((fundt[kq-1][ky-1][2-1] >= 450,000,000.00) && (fundt[kq-1][ky-1][2-1] < 525,000,000.00)) ktti[ky-1][1] = 2;
else if ((fundt[kq-1][ky-1][2-1] >= 525,000,000.00) && (fundt[kq-1][ky-1][2-1] < 650,000,000.00)) ktti[ky-1][1] = 3;
else if ((fundt[kq-1][ky-1][2-1] >= 650,000,000.00) && (fundt[kq-1][ky-1][2-1] < 750,000,000.00)) ktti[ky-1][1] = 4;
else if ((fundt[kq-1][ky-1][2-1] >= 750,000,000.00) && (fundt[kq-1][ky-1][2-1] < 850,000,000.00)) ktti[ky-1][1] = 5;
else ktti[ky-1][1] = 6;
printf("DEBUG10: The Quarter is %d, Trust Fund is %20.3f, the Tax Table is %d\n", kq, fundt[kq-1][ky-1][2-1], ktti[ky-1][1]);
}
else;

I know that a triple-dimension array looks complex, but I can guarantee that is a double. The literal constants I am using should automatically be doubles. The subscript of [2-1] looks odd, but this is code I was given to maintain, and that is how it is written elsewhere, so I kept it for consistency. The problem I am having is that when I run this code, all the data I input is less than the 450 million value, but the run falls through all the if and if-else conditions, and the else code is what is actually executed, i.e., I always get six as my result. I am running on a Sun Sparc, and the compiler used is SunStudio 12. I have tried using variables, with the values listed above assigned to each variable, but it does work either. When I use variables, if the input value is negative, the comparison for less than 450 million works, but any positive input values will through to the else and give me six. Thank you for any help you can give me with this.

[weaknessforcats]

Floating point comparisons fail when using the equality operators. Is 2.0000234 equal to 2? Well, no. But the compiler will convert 2 to a double and do the compare and based on the number of decimal places the numbers may well compare equal.

Read this: http://floating-point-gui.de/errors/comparison/