Re: Triangle Numbers Revisited

use strict;
use warnings;

my ($i, $j, $inum, $jnum, $left);

$| = 1; print "Enter the number to check : ";
my $inp = <STDIN>;

BING:
for ($i = tget($inp); $i; $i--) { # BUG ON THIS LINE - SEE BELOW FOR F
+IX
    $inum = tmake($i);
    $left = tget($inp - $inum - 1) + 1;
    for ($j = 1; $j < $left; $j++) {
        $jnum = tmake($j);
        last BING if (tis($inp - $inum - $jnum));
    }
}
print "       Your triangles are : $inum $jnum "
      . ($inp - $inum - $jnum) . "\n";

sub tget { return int(sqrt(1 + $_[0] * 2) - .5); }
sub tmake { return (.5 * $_[0] * ($_[0] + 1)); }
sub tis { my $n = shift; return ($n == tmake(tget($n))); }
[download]

This is blazingly fast with numbers even several orders of magnitude larger than 987654321, and requires almost no memory to run.

Comment on Re: Triangle Numbers Revisited Download Code

Replies are listed 'Best First'.
Re^2: Triangle Numbers Revisited by Limbic~Region (Chancellor) on Oct 14, 2004 at 11:29 UTC
TedPride, This is blazingly fast... I wanted to see just how fast, so I modified your code to fit into the benchmark FoxtrotUniform and I were playing with. It turns out your code is only about half as fast as my second version and it has a bug I didn't bother to track down. To see the bug, try entering 3133756 with your code. I say your code is only about half as fast because it got less than halfway through (line 2235 of 5000) when it blew up and it took approximately the same amount of time (3.619 seconds) to get there. In case you want to run some more tests yourself, take a look at gen_targets.pl and then use this modified code: Read more... (989 Bytes) Cheers - L~R	[reply] [d/l]
Re^3: Triangle Numbers Revisited by TedPride (Priest) on Oct 14, 2004 at 11:52 UTC
I found and fixed the bug. See my second version. As to it being blazingly fast, it is blazingly fast - just not as blazingly fast as it could be if I took the time to optimize the algorithm. Congrats if yours is faster.	[reply]