designerdecimal

commit 19d4f883de87bdeb59dbdd9f2f3dec00dfd898af
Author: Ryan Wolf <rwolf@borderstylo.com>
Date:   Sun, 28 Feb 2010 19:05:17 -0800

first commit

Diffstat:
A
READEME.md
 | 
20
++++++++++++++++++++
A
designerdecimal.pl
 | 
77
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

2 files changed, 97 insertions(+), 0 deletions(-)
diff --git a/READEME.md b/READEME.md
@@ -0,0 +1,20 @@
+Designer Decimal
+====
+
+TODO: summary
+
+Usage
+-----
+
+TODO: sample input/output
+
+Attribution
+-----
+* John Kennedy, "Algorithm to Convert a Decimal to a Fraction" <http://homepage.smc.edu/kennedy_john/DEC2FRAC.PDF>
+  For describing a performant rational approximation algorithm.
+
+License
+-----
+designerdecimal is associated with CC0.
+
+To the extent possible under law, the person who associated CC0 with this work has waived all copyright and related or neighboring rights to this work. See http://creativecommons.org/publicdomain/zero/1.0/ for more details.
diff --git a/designerdecimal.pl b/designerdecimal.pl
@@ -0,0 +1,77 @@
+#!/usr/bin/perl
+# This file is part of designerdecimal and is associated with CC0.
+# To the extent possible under law, the person who associated CC0 with this work
+# has waived all copyright and related or neighboring rights to this work.
+# See http://creativecommons.org/publicdomain/zero/1.0/ for more details.
+
+# For describing a performant rational approximation algorithm, thanks to:
+# John Kennedy, "Algorithm to Convert a Decimal to a Fraction"
+# http://homepage.smc.edu/kennedy_john/DEC2FRAC.PDF
+
+use strict;
+use warnings;
+use Math::BigFloat;
+
+my $name = "Santa Claus";
+if (scalar(@ARGV) > 0) {
+  $name = join(' ', @ARGV);
+}
+
+my @chars = split(//, $name);
+my $ascii = '';
+foreach my $char (@chars) {
+  $ascii .= sprintf('%03s' , ord($char));
+}
+$ascii = '1.' . $ascii . '004';
+my $accuracy = length($ascii) - 1;
+
+my $x = Math::BigFloat->new($ascii);
+$x->accuracy($accuracy);
+my $z = $x->copy();
+my $dMinTwo = Math::BigFloat->bzero();
+my $d = Math::BigFloat->bone();
+my $n = Math::BigFloat->bzero();
+my $i = 0;
+
+print "For the name '$name':\n  Your designer decimal is $ascii\n";
+my $q = $n->copy();
+$q->bdiv($d);
+$q->accuracy($accuracy);
+
+until (BigFloatEqual($q, $x)) {
+  my $zInt = $z->copy();
+  $zInt->bfloor();
+  my $zSub = $z->copy();
+  $zSub->bsub($zInt);
+  $z = Math::BigFloat->bone();
+  $z->bdiv($zSub);
+
+  $zInt = $z->copy();
+  $zInt->bfloor();
+
+  my $dCpy = $d->copy();
+  $d->bmul($zInt->as_int());
+  $d->badd($dMinTwo);
+  $dMinTwo = $dCpy->copy();
+
+  $n = $d->copy();
+  $n->bmul($x);
+  $n->bfround(0, 'odd');
+
+  $q = $n->copy();
+  $q->round_mode('trunc');
+  $q->accuracy($accuracy);
+  $q->bdiv($d);
+
+  $i++;
+}
+
+print "  Your rational approximation is $n / $d\nTook $i iterations.\n";
+
+sub BigFloatEqual {
+  my ($x, $y) = @_;
+  if ($x->bcmp($y) == 0) {
+    return 1;
+  }
+  return 0;
+}