John Peacock >
Math-FixedPrecision >
Math::FixedPrecision

Module Version: 0.21
Math::FixedPrecision - Decimal Math without Floating Point Errors

use Math::FixedPrecision; $height = Math::FixedPrecision->new(12.362); # 3 decimal places $width = Math::FixedPrecision->new(9.65); # 2 decimal places $area = $height * $width; # area is now 119.29 not 119.2933 $length = Math::FixedPrecision->new("100.00"); # 2 decimal places $section = $length / 9; # section is now 11.11 not 11.1111111...

There are numerous instances where floating point math is unsuitable, yet the data does not consist solely of integers. This module employs new features in Math::BigFloat to automatically maintain precision during math operations. This is a convenience module, since all of the operations are handled by Math::BigFloat internally. You could do everything this module does by setting some attributes in Math::BigFloat. This module simplifies that task by assuming that if you specify a given number of decimal places in the call to new() then that should be the precision for that object going forward.

Please examine assumptions you are operating under before deciding between this module and Math::BigFloat. With this module the assumption is that your data is not very accurate and you do not want to overstate any resulting values; Math::BigFloat can unintentially inflate the apparent accuracy of a calculation.

The constructor accepts either a number or a string that looks like a number. But if you want to enforce a specific precision, you either need to pass an exact string or include the second term. In other words, all of the following variables have different precisions:

$var1 = Math::FixedPrecision->new(10); # 10 to infinite decimals $var2 = Math::FixedPrecision->new(10,2); # 10.00 to 2 decimals $var3 = Math::FixedPrecision->new("10.000"); # 10.000 to 3 decimals

All calculations will return a value rounded to the level of precision of the least precise datum. A number which looks like an integer (like $var1 above) has infinite precision (no decimal places). This is important to note since Perl will happily truncate all trailing zeros from a number like 10.000 and the code will get 10 no matter how many zeros you typed. If you need to assert a specific precision, you need to either explicitly state that like $var2 above, or quote the number like $var3. For example:

$var4 = $var3 * 2; # 20.000 to 3 decimals $var5 = Math::FixedPrecision->new("2.00"); # 2.00 to 2 decimals $var6 = $var3 * $var 5; # 20.00 to 2 decimals, not 3

John Peacock <jpeacock@rowman.com>

syntax highlighting: