- IMPLEMENTED OPERATIONS
- CAVEATS & GOTCHAS
- SEE ALSO
- COPYRIGHT AND LICENSE
use Math::FixedPoint; my $num = Math::FixedPoint->new(1.23); $num += 3.1234; # $num = 4.35 # you can specifying the radix in the constructor my $num = Math::FixedPoint->new(1.23,3); $num += 3.1234; # $num = 4.353
This module implements fixed point arithmetic for Perl. There are applications, such as currency/money handling, where floating point numbers are not the best fit due to it's limited precision.
$ perl -e 'print int(37.73*100)' 3772
This problem is unacceptable in some applications. Some of those cases are better handled using fixed point math as precision is determined by the number of decimal places. To circumvent inherit problems with floating point numbers Math::BigFloat module is typically used, still problem exist, but precision is improved.
Now the problem with Math::BigFloat is that it is 3 or more orders of magnitude slower than Perl's floating point numbers, Math::FixedPoint on the other hand is 2 orders of magnitude slower than Perl's native numbers which is a huge gain over Math::BigFloat. That performance boost comes from the fact that most of the math is done internally using integer arithmetic.
Math::FixedPoint - fixed-point arithmetic for Perl
Creates a new object representing the
$number provided. If
$radix is not specified it will use the
$number's radix. If
$radix is provided number will be rounded to the specified decimal places
The following operations are implemented by Math::FixedPoint are +, +=, -, -=, *, *=, /, /=, =, <=>, cmp, "", int, abs
CAVEATS & GOTCHAS
This module still ALPHA, feedback and patches are welcome.
NUMBERS WITH DIFFERENT RADIX
It is not intuitive what it is going to happen when two numbers with different radix are used together
my $num1 = Math::FixedPoint->new(1.23,2); my $num2 = Math::FixedPoint->new(1.234,3); my $res = $num1 + $num2; # $res = 2.46 my $res = $num2 + $num1; # $res = 2.464
Due to the way that Perl handles overloaded methods, it will call the "add" method on the first object and will pass the second object as parameter. The "add" method will preserve the radix of the first object
INTEGRATING WITH OTHER NUMBER CLASSES
Due to similar reasons when combining different classes it is not obvious which will be the class of the result object
my $num1 = Math::FixedPoint->new(1.23); my $num2 = Math::BigFloat->new(1.24); my $res = $num1 + $num2; # ref $res = 'Math::FixedPoint' my $res = $num2 + $num1; # ref $res = 'Math::BigFloat'
It's critically important to have this in mind to prevent surprises
Although this module is implemented in pure Perl, it is still 5-10 times faster than Math::BigFloat (even more depending on Math::BigInt's backed).
Mariano Wahlmann <dichoso _at_ gmail.com>
COPYRIGHT AND LICENSE
This software is copyright (c) 2013 by Mariano Wahlmann.
This is free software; you can redistribute it and/or modify it under the same terms as the Perl 5 programming language system itself.