tye,
I expanded the code I came up with for combinations of a fixed size:
#!/usr/bin/perl use strict; use warnings; my $iter = combo( 30..50 ); while ( my @combo = $iter->() ) { print "@combo\n"; } sub combo { my @list = @_; return sub { () } if ! @_; my (@position, @stop, $end_pos, $done); my ($by, $next) = (0, 1); return sub { return () if $done; if ( $next ) { $by++; return () if $by > @list; @position = (0 .. $by - 2, $by - 2); @stop = @list - $by .. $#list; $end_pos = $#position; $next = undef; } my $cur = $end_pos; { if ( ++$position[ $cur ] > $stop[ $cur ] ) { $position[ --$cur ]++; redo if $position[ $cur ] > $stop[ $cur ]; my $new_pos = $position[ $cur ]; @position[ $cur .. $end_pos ] = $new_pos .. $new_pos + + $by; } } if ( $position[0] == $stop[0] ) { $position[0] == @list ? $done = 1 : $next = 1; } return @list[ @position ]; } }
I haven't analyzed its O factor or spent a lot of time with good benchmarks, but it does appear to be considerably faster for combinations of 30..50.

Cheers - L~R


In reply to Re^2: Finding all Combinations by Limbic~Region
in thread Finding all Combinations by narse

Title:
Use:  <p> text here (a paragraph) </p>
and:  <code> code here </code>
to format your post; it's "PerlMonks-approved HTML":